I have the same board - just a bit different. The VRM section of mine looks a bit different and there is an IC in the bottom left corner. It would be a shame if the south bridge has an issue. I haven't tested mine, but I will let you know once it is clean and we can take some measurements! Great video! Thanks Tony!
Hello Tony. You can check both component testers with a large new capacitor, something like 1000 uF or bigger. A good one should have an ESR well below 0.1 ohms. Now add in series one ohm resistor and measure it again. Now both testers should measure 1 ohm ESR if they are ok. Anyway the component tester have a calibration procedure (at least the cheap one from ebay). You can try to recalibrate it if needed. How to access the calibration procedure depends on the installed firmware, so you will need to do a bit of googling.
Hi - thanks, I wasn't aware of that procedure thanks! I did run the "calibration" before shooting the video but that didn't change much unfortunately. I'll try what you suggest later thanks!
@@tony359 The cheap tester is based on an open-source German design and is well documented and the large pdf manual is available. The calibration requires a stable small value, known good, known value ceramic cap as part of the calibration. Something like a common 100nF bypass cap is good. To start the calibration short all 3 pins on the tester together then follow the on-screen prompts. Then do the same thing in the Finirsi. If it doesn't have a calibration proceedure toss it in the bin.
@@g4z-kb7ct I found the manual yesterday. I'll see if I can trigger the calibration. The FNIRSI has a calibration which was done before shooting the video! I'll make another quick video to show some extra info the bin is where it belongs.
Yes, thanks, I'm planning of making some shields with some drink cans - I think North West Repair made some. I DO fear the BGA, I've been consistently burning everything for a decade! :D
@@tony359 I bought some USB stick PCB's on Ali and used the memory chips from old Samsung SSD's to produce working memory sticks. That was cheap and I learned how to unsolder, reball and reflow without blowing off other components... :-) Most of the time I use a second hot air blower from below to preheat the board to 150 degrees (Celsius) before I start unsoldering or resoldering a chip. You can also preheat the area with the normal hot air gun. I used a thermocouple in the beginning to make sure I do not overshoot too much and also to check the temperature during the soldering process. And I use a thin flexible metal spudger to gently go under and lift the chips. You find tons of tutorials on TH-cam but it needs practice.
Oh, I have all those tools. I have the big pre-heater, I have a good hot air station etc. To be honest, I've only worked on large board with massive ground planes (PS3, PS4) and those are difficult. Smaller boards should be simpler.
regarding southbridge pdf: "Power-Good Input. This signal comes from the power supply to indicate that power is available and stable. M1543C will use this signal to generate reset sequence for the system.", so there is "sequence", what about if it is syncronized with some clock, OSC14 for example. And if we lost input clock, entire chip is stuck.
@@ILRBW Hi - thanks. I checked (after the video) the small clock IC nearby and it works. Whether it reaches the IC under the package, I cannot say :) But thanks!
@Insomniator1 Sorry, it's taken a while. I do NOT have clocks on 20 and 30. I do on a different board as a comparison. That is the ISA clock coming from the southbridge though? Wouldn't it be affected by a dead SB? I cannot be sure but I think I saw clock on B30 for a split second on my tests. I could not replicate. There is a small crystal nearby the SB and that is outputting the proper clock. What do you think? Thanks!
I think there might be a bad cap and possibly a bad diode allowing feedback preventing the system to reach 3.3v. I don't see the south bridge being the issue.
I thought about that. I looked for any other place where the ISA reset line could be connected but I couldn't find any. One thing I forgot to check are the PCI reset and CPU reset. Those are coming from different pads under the SB. If those are 3.5V, maybe then is as you say.
Looking forward to seeing you replace that south bridge! Good luck! It's great to see videos on modern boards. I still feel quite a novice when it comes to anything surface mount, let alone BGA!
Well, a socket 7 a modern board? Surely more modern than an Apple 2 :) BGA is something I have been struggling to master for a while... but I am persistent (or stubborn if you want!)
If you have a new electrolytic with low ESR (like less than 1 ohm) you can just put a known small resistor in series with the cap and test that combination. That should tell you which tester is correct. Try several different series resistor values to get more confidence.
I have tried, I'll publish the result on the 2nd channel. Though those testers don't work with high frequency, they use some tricks so the resistor might actually confuse the process. But I'l show that process anyways. Thanks!
Also regarding tool longevity. Case in point: I have a Hitachi V665 analog oscilloscope, 2-channel 60MHz. It was made in 1975. I have had it for 20 years, and whoever owned it before me gave it pretty heavy use. It still works perfectly fine after almost 50 years. Even the CRT on it is still bright, sharp and nice.Well made tools, especially if you treat them right, can last you a lifetime.
@@tony359 haha I keep the Hitachi on the floor next to the bench, standing up on its rear feet (yes it has feet on the back so it's designed to be used standing up like this). It's not only large, but also heavy. About 15kg. :)
Oh I do remember keeping the Fluke on its back feet as well and my probes were barely long to reach the boards. And I'd systematically trip in all those wires and everything would fly across the room.
I have tested the same old one component tester you have and got 0,34 ESR on a 100uf capacitors and then tested with a DER EE LCR Meter DE-5000 and got 0,42 so i think the old one is more correct than the new you got.
You could try to isolate, brake pin between south bridge and inverter. Then try to manually push it high and low, a measure how motherboard will respond. About component testers You could try to power them with good battery, do measurement to component, then put almost used up battery and test the same component. If one tester will have about same values and other show something different, then You will know which one is better.
I'm always happy to see new videos from you. I'm a big fan of learning and teaching and despite all my experience I still manage to learn something from you!
Definitely I want to see you replacing the southbridge! But despite the end result is not the one we wanted, it was nice seeing you exploring the different possibilities for issues. I learned something and it is what I like. Have you tried to push the southbridge against the board due to a bad solder connection? I could explain the changes in voltages between days and prevent the board from posting.
I have - it didn't make it in the video. I warmed it up, cooled it down, pushed in many different ways: I never had any change whatsoever. The voltage change was due to ground - I normally use the isolation transformer just for convenience, it has a big ON OFF button. With that, I get 2.2/2.7. With mains, I get 1.7. It kinds of suggests that the line is floating maybe?
Good luck with the changing of the bga. I used to visit very high tek companies when I worked for BSI and UL in the South West of the UK. 😅 The placement of them is a very skilled job, mostly done by very expensive equipment. Look on TH-cam for the placement and checking they are in corret posioning of them. I would like to see if it can be done, with bench equipment. Thanks for your well decribed and intresting videos.
Take a look in the pci slots. Got two of these from ebay with similar behavior . Both ended up having smashed pins, grounding out inside the pci connector.
@@tony359 Bummer, I never had a short on the reset line. The data pins were the problem. My best guess was the shorted data lines kept the chipset in a weird in between state. Not letting it come out of reset. Best of luck!
I believe they are relevant to each other. Shorted VRM mosfet (drain and source shorted first) killed CPU (which is previous owners) with overvoltage first. Killed CPU is now shorted internally with some data pins and killed south bridge with it. While this can drain extreme current, VRM mosfet shorted with its base drain and source and killed PWM chip. I assume this happened, but i might be wrong. Do you have "original" CPU with this board? If so please check it for shorts.
Hi - I tried with hot air but not to the point of melting the solder. If I go that way I'd rather swap it or re-ball it as apparently re-flowing seldom works. Hot air didn't make any difference though.
Your LCR tester must have a certain test frequency that should be stated somewhere in the details of its specifications. The higher this test frequency, the better will the test results concerning the ESR will be. Just my humble opinion. Thanks for your tinkering job !
I looked into the original project and it's not using high frequencies - the ATMEGA can't deal with that. It's using some tricks to achieve a "good" result. I'll make another video to show that! Thanks for your comment!
I would cut the trace going from the south bridge to the 04 to at the edge of the bridge IC to make sure nobody else is conecting to the trace causing this verid voltage. If you measure the same voltage next to the bridge ic than no other way to figure out what is happening than desolder and replace the brige. After the removal of the BGA it is worth to measure the reset trace again, you never know who else is try to talk into this reset trace. Another idea: try to reflow once before replacement maybe it will help.
Today I checked the CPU reset line and it's also stuck at 1.7V - I know it comes from a different pad, I haven't checked if they happen to be electrically connected. Oh yes, if and when I remove the SB, I'll test without it first. Reflowing might help but it's never a good thing for the PCB and adjacent components to get to 200C multiple times so I'd be tempted to just swap it. But it's an idea thanks!
For those cheap component testers you usually do calibration where you connect pins 1,2,3 together and power up. Your old tester maybe too old to have that feature or need to have software update. For my tester it improved measurement accuracy. If I remember right it was showing too hi esr and Rds values before calibration. Edit: Also discharge all caps before putting them to tester. I have killed one tester that way.
Thanks - The Fnirsi was calibrated just before shooting the video unfortunately. Mine doesn't have that feature indeed - I'll see if I can update the FW!
I have this board, It works but its niggly, Windows 98 wont install nicely if USB (other BIOS setting ??) is enabled, I can't remember exactly but using ATX power, the soft-power/the power button doesn't work right if I recall correctly, if the instant power off is enabled in the BIOS it doesn't work, so its holding the button for 5 seconds, while not the end of world its gets annoying fast when troubleshoot a retro PC . Don't mess with JP2 near battery, as jumped to pin 2-3 will cause one of diodes near by to short, luckily the diode just de-soldered it self from the motherboard didn't do lasting damage.
interesting - you're saying that maybe all I have to do is to hold the ON button for 5 seconds or something like that like that weird Acer MoBo I had before? I've done most of my testing without the BIOS to be honest as I'd expect the Reset line to work without it - the board won't execute code in RESET I suppose... I think I tried that jumper in position 2-3...
@@tony359 R178 is indeed a 10 OHM (SMD 100) resistor on my board. sorry I messaged without completely watching. No, not at your level of troubleshooting so was just giving general advice on that Board... that ALI chipset also has issues It doesn't fully cache memory above 128MB or something like that as if recall correctly unless its chipset revisions G and H. I can test other bits on my own board (its in strip down state no battery and I stole its jumpers) if that would help, I do own a scope but digital testing is not my forte.
I see FNRSI products advertised all over TH-cam. If they cannot be bothered to create a brand name more complex than slamming their face into a keyboard.. Perhaps the warranty will not be worth the bytes it's written on?
Component terminals rejecting solder like this resistor it happens, when there's a thick oxide layer. It's an indication that the component is expired and hasn't been stored well. Also i hate components being arranged so close to a hole like that, because that's why that whole section is damaged and had to be reworked by previous owner. Is the hole needed? I don't think it's an AT/ATX mounting hole, i don't know why they put it in, did they need it for some sort of factory jig? It's terrible either way. Ahahah when you have one voltmeter, you always know what the voltage is. When you have two voltmeters, you never know what the voltage is! Doubly true for these ESR/LCR devices! I would rather trust the cheap one. It's a faithful reproduction of a German open source project, and the values it delivers, even if they are not what one might call exact, because really who knows, these are not straightforward values to measure, they just make sense for repair purposes, since it's quite sensitive to decayed parts. What TC3 shows doesn't make much sense! I too have the classic cheap LCR T4. Been thinking of hand building one of my own based on the upstream project.
The resistor is definitely damaged (though it works). I need go get some stock of SMD components for those cases... The cheapo tester has served me well over the years, I cannot be 100% sure but it felt accurate enough. That 22uF cap is completely shot - I know that for sure: the TC3 doesn't reflect that!
Hi Tony, I'm enjoy this video and o so the wiling to fix it after may be a total waste of time. For me is one of your best videos a ever see in your channel. talking about the meter , I bought the DER EE DE-5000 to see and you can test your caps at 120 Hz there are a lot of videos in TH-cam talking about this tester and how to use it way better then me explain it to you. The best of all is not that expensive. Great video!!!
When it happens to the board what CPU was in. I could imagine that the V-regulator grills this CPU and the 5V goes through to the ALI Chip and kills everything else. But who knows.
This is a nice board and I'd like to see you getting it to work eventually, but I'm also good with seeing another video of you fixing something else, so whatever you end up deciding is good with me. 🙂 For the ESR testers, they are probably not trustworthy when it comes to ESR, but a big part of the difference could simply be that they measure the ESR using different frequencies. You could look at their output signals using the oscilloscope and check if the Fnirsi uses a higher frequency. If a capacitor is leaking (electrically, that is), then the ESR and capacitance is more or less irrelevant and you should be able to measure the leakage current using a DC supply and multimeter. Then simply apply the rated DC voltage to the capacitor and check that the current goes to zero or very near to that (
Yes I understand that if the frequency is different then the ESR will read different. Good idea about checking it with the scope - won't the scope affect the reading at all?
@@tony359 The input capacitance of the scope should be in range of 5-15pF so that would be negligible in relation to a uF capacitor. But the idea was to mainly compare the frequency emitted from the devices. I assume they both use a fixed frequency for their capacitor measurements. Problem is whether the scope's input impedance of 10 MOhm will confuse the meter and make it think it is measuring a resistor, but of you set the probe to 10x the input impedance should be 10x higher, or even better use a 100x probe if you have that for minimal interference.
@@tony359 Diff probes are active devices so they buffer the signal and do not necessarily increase the input impedance. A typical input impedance for a diff probe is 5~10 MOhm so not higher than the regular probe on 10x, which is 10 MOhm (I miswrote before since the scope is only 10 MOhm with the probe on 10x). The scope itself is 1 MOhm and basically the probe adds a 9 MOhm resistor in series with the signal when you set it to 10x. But 10 MOhm is also reasonably high and I doubt those cheap component testers are sensitive enough to be disturbed by that. I think it is not much different from the resistance between the pins in the ZIF socket if you're in a moist environment. 🙂
With the component tester. If you get the version Tony has, make sure it has the metallic crystal on the other side! The other ones are junk and fail. They are a great little tool though. Helped me fix a graphics card a while ago.
I am aware there are 10000000 clones of that tester. I think I was lucky with mine! :) I had two actually - I blew up one when testing a large (still charged) capacitor...
loved the video, Tony! so sad for the MB, i have the exact one, aesthetically at least, but with the 430TX chipset, the 5TX52 model. it also has a problem, the FDD is not recognized at all, it's not the 40 code error, i think 3, which means, read online, FDD controller dead. you should repair it, the acorp logo when booting is magic, for me :)
I think I stumbled into your review while I was testing myself - you didn't compare with a proper ESR either if I remember right? Not complaining, just to make sure I haven't missed a review where the ESR was being compared with a proper tool.
@@tony359 I fished out the unit and will do some comparison in video on it, I’m just about to live stream and will do it in that and record some video at the same time to publish later.
@@tony359 Yes, I tested it this morning as part of my live stream, I did it in a format that I can put into a stand alone video also, I compared it to 2 other good capacitor testers, the DER DE5000 and the Shannon Tweezer ST42, I could have tested on a few other units also, but the results were clear just from these units, so I have some conclusive results there, I will try and do the video this week.
Tip for surface mount work around plastic: Captain tape. It won't make the plastic invincible but it increases your margin for error enough to be worth it than not. Using a preheater under the board to get the board warm will also help. Also do yourself a favor and don't use the $60 junkazon special. Get a Quick or a Atten.
I do have a (genuine) quick :) (Junkazon special LOL!) I do have a pre-heater! I will make a shield using some tin (from a drink can). Plus kapton as you say. Thanks for taking the time to send me some tips! :)
@@tony359 I don't trust those $60 stations. I don't think they have the "horsepower" of the more expensive ones and after seeing "EEVblog #1171 - Yihau WEP SMD Rework Station Meltdown!" I have concerns about fire safety.
I used to have a Aoyue and I know what you mean. They're junk and you can only burn things with them as the air temp is uneven. That said, dosdude1 does wonders with it :)
@@scotshabalam2432 I have a Katsu 8582D which has been faithfully serving me for years. I got another one recently and found a dodgy DiodeGoneWild special non Class Y1 safety capacitor inside. Now I'm 2nd guessing its safety!
Well, I would nee to pull it to ground, not just cutting it. And the CPU and PCI reset lines are elsewhere. One thing I did not check is what happens to those other reset lines though.
What's the resistance of reset to GND, ie, something may be pulling it down. There's probably is a bunch of things that have to be good before the chipset releases the reset, eg, crystal has to be oscillating. I have had a MB with a stuck crystal, and when I accidently did a diode test across it, ie, gave it voltage, it started working and fixed the MB.
That's a good idea, I thought about that just an hour ago! And yes, someone also suggested the oscillator. I might re-visit that board before I got for the replacement for sure. Thanks!
Great job Tony! I have several boards with stuck reset lines and I have no idea what is wrong with them maybe we can come to a conclusion of a common or most frequent problem with this symptom and hopefully it is not always a problem with south bridges...
@@tony359 I forgot to mention the Reset line voltage, which you said to be "all over the place." The voltage fluctuations, as seen on the meter 2.3-2.7V, are telltale signs of bad caps.
@@PavelKrejci-ik5uv It reads 1.7V when ground is connected. There are very little electrolytic caps on the board and I am confident there are none on data (reset) lines. It's difficult to probe the BGA and confirm that 5V is stable but I can check elsewhere. I appreciate caps are the evil but I doubt a cap is causing this on this board TBH. Thanks anyways!
FNIRSI TC3 displays absolute GARBAGE instead of ESR! Capacitances, however, are measured correctly. EXAMPLE. 220 μF 63 V Samwha SD: 0.12 Ohm. The same capacitor with a 8.98 Ohm 0.5% resistor in series: 1.5 Ohm (!!!). On an old Atmega based component tester: without the resistor 0.20 Ohm, with the same resistor 8.5 Ohm. Way closer to truth (the total resistance should be around 9.1 Ohm).
@@tony359 Tony, would be great if you'll do a video about the reliability of the TC-3 Multi Tester and/or use it more often. I got one for most less offer, because I have a serious lack of money and I have to repair/refurbish a lot of computer and audio stuff.
Amazingly educative video, as always. I do not want you to go through the pain of changing the ALI chip, so please do not count my view 🙂. Perhaps you could "force" the reset down with a 555 simple circuit after 1 second of power, isolating it from other circuit? BTW, if you need a good ESR and caps meter, I do recommend the Atlas ESR70 gold. A bit expensive but it is for life.
The ISA Reset line is one of the resets. Then there are the PCI line and the CPU line (which I haven't checked BTW!) so I don't think it's going to work. I could try doing that on the CPU reset line to see if the CPU starts executing code! Yes, I might get one of those proper meters - why aren't Fluke getting in touch for a review? :D
@@tony359 Tony, you'll get there. I understand the complexity, and you need a signal choreography there (as an old teacher used to say) 🙂 Ohhh Fluke.... I heard **somewhere** that their equipment "seems" to work 😀 Lucky you, getting Flukes without investment. My face shows envy right now 😀 Enjoy mate !!!. We'll be here to cheer you.
@@tony359 Oh, apologies. Long day... HEY @fluke !!!! ARE YOU LISTENING ??? Tony needs some of you merchandise !!!!! Don't be shy, the guy has 23K followers and all of them are potential customers !!!!
The difference in ESR and capacitance could be do to the frequency it tests the capacitor at. What does your Fluke say for the capacitance? Does either of your testers say the frequency for LCR testing? If you are going for a LCR meter go with the DER EE, DE-5000. Excellent LCR meter for the money. You may want to pay the extra now and get the meter with the tweezer option, expensive to add later
@@tony359 Use oscilloscope and measure the frequency both testers are using. Maybe thats why ESR differs so much. Btw 22u capacitor with ESR 10ohm is bad, but in transistor radio would still work OK for years if its used in AF coupling :-)
I'm looking into that, they don't use a traditional way, there is no "frequency". They use some tricks because the ATMega cannot deal with those frequencies. The original documentation of the "transistor tester" says the ESR should be comparable to the 10KHz one.
Motherboards with this chipset are the most expensive super socket 7 platforms to buy (at least in my country). If it will be really necessary I would like to see reballing of this south bridge.
Because those boards were unstable and weird even when were new. So not many boards survived. I like more Via MVP3, it has own issues too, but mostly works.
I've done some more tinkering and I cannot find much. I wish it wasn't BGA so I could poke the legs and make sure it gets power, clock etc. I found the CPU reset line is stuck in the same weird way (hint, maybe the CPU was fried and fried the reset line?) and the small 32KHz clock is working. But yes, it could be something else for sure.
Does this board have jumper to determine power supply type? My ss7 board behaves exactly the same way when that jumper is set incorrectly or is missing.
It does make sense that the south bridge is damaged. What I think happened the the Vcore mosfet shorted, took out the Vcore driver chip via the ugate signal, and when it killed that chip, it then sent 12V up the powergood out of that chip to the southbridge, damaging that too. That's my guess. Re the Component tester, have you done the calibration routine on it? You basically need 3 pins shorted together (I use a 3-pin header soldered together), and a capacitor greater than 100nf. You put shorted pins into 1-2-3 on the tester, press the button, and it'll come up selftest. it will scroll through for a bit, then ask you to remove the short, then after a bit it'll ask for the capacitor. once you fit the cap, it continues till it's done. It's worth doing, to see if it improves the accuracy. It's seems the TC3 might have this calibration mode too, so it might be worth trying, and see if it improves the readings.
Hi - the powergood line of the VRM controller is not connected to anything, you might be able to see that when I removed it. Maybe the 5V got to the CPU, the CPU was fried and sent 5V to the Southbridge? But the Southbridge is behind the Northbridge... (that would be bad news!) Thank you for your idea!
@@tony359 You're right! my bad. I thought it would get a powergood signal from the VRM chip, but obviously not. I think we'd really need to look at the circuit diagram to see where the southbridge powergood is then. It's worth checking all other things connected to the PG and Reset lines before jumping into a BGA replacement.
Without schematics it's complicated. To be honest, this is a socket 7 board. I think the PSU powergood signal is the only one going to the chipset. Also, computerguy from TRW told me his "power good line" going to the SB is 3.5V. Mine is 3.5V. So, unless there is a BGA problem, powergood is not the problem. I think.
It could be, the white paper is several hundreds pages, I searched for "reset" and what I showed on video is the only mention. It could possibly be that something else is holding RESET - though I'd still expect RESET to be 3.5V, not 2.2V!
I think you forgot to test if the new voltage regulator IC is actually outputting the power good signal itself. in complex boards like this one single signal not good will drag down all the rest so maybe there is one single fault left somewhere making the reset line stuck. I am not sure if the southbridge itself output PG to something. I think it just receives?
It didn't make it in the video but if you watch the pads when I de-soldered the IC, you should be able to see that the PG signal is not connected - so unfortunately I doubt it's our problem. Yes, there might be something else which is preventing the SB from starting. Who knows. One thing I forgot to check are the CPU and PCI reset lines: are they doing the same weird thing than the ISA reset lines? Thanks for your help!
These component testers are very useful (I've had 2 that I've been using for the past 10 years or so) but you must be aware of their limitations. For instance take a JFET like J105 which has a very high Idss on the order of a few hundred mA. The tester will not identify it correctly. It reads it either as a resistor, or as a double diode. This is probably the most extreme case; there are other situations that fool the tester, for instance components that need to be tested at voltages higher than it can provide, which is around 8-9V (the 9V battery is an indication of that).
Of course. I don't expect a £5 tool to perform like a £200 one. I'm not expecting perfection. Just not 100x difference from a similar one :) Thank you for watching!
I haven't finished watching yet but the reason why your resistor wasn't taking solder was because it had corrosion on it; if you look closely, you can see the pitting. I would have just taken another from a donor board. Or, you can try and hold it with tweezers and clean it with a brass brush.
Hi Tony, please make a follow up video. Also, I would like your and from the guys the opinion for a similar issue that I have. I have a motherboard "Fujitsu Siemens FSC D1837-A21 W26361-W84-X-02" and I tried the below: I replaced the RAM with a same and functional one (from a same second motherboard that I have). I replaced the CPU with a same and functional one (from a same second motherboard that I have). I check the CPU socket for bend pins and all are ok. I have replace the BIOS chip with a same and functional one (from a same second motherboard that I have). I have replaced all the caps in the yellow circles with brand new capacitors. I place a buzzer on the motherboard and no beep even when I remove the RAM. On the coils of the CPU I have always 1.55V but at some point after testing I didn’t voltage at all in CPU coils and after I desoldered the resistor from the ICOMMON pin of the ISL6565 chip the it start to have again 1.55V on the CPU but still no image. Maybe this motherboard have issues with cold solder joints? The current from the BIOS battery is 0,54uA (in the faulty one). I tried some PCI tester cards and gave me the below codes: The one of them show nothing "--". The second show "0000". The third show "FF". Unfortunately, I can’t find any schematic of this motherboard. Do you know what else to check? Please I would very appreciate any advice. I have three months now that I am trying to repair this motherbord (I am not doing it for the money but because every problem give you new knowledge). Also, I have made many measurements with multimeter and oscilloscope and analysis of the phase chip on the eevblog forum (Thread name: "Mainboard with No Picture")... if you please take a look. Please I would very appreciate any advice. Also, what is the "reset sequence for the system"...? Where can I read more for this? I apologise if my test is big. Thank you very much for the sharing of your knowledge.
Well, I am told that the Power Good line from the inverter reads 3.5V on a working board. The reset line reads the same weird voltage without the inverter. Anything's possible but it feels unlikely, what do you think?
If you're not comfortable with doing BGA, I wouldn't want to push you. Or recommend you do it as anything other than a practice test, knowing this board is expendable. I know it wouldn't actually fix what's going on, but would it be advisable for testing purposes to bypass the southbridge's weird mid-voltage output and instead feed 3.5v directly into the place that expects it, in order to clear reset? That way you could see if it works otherwise .
I can try that but the SB has three reset lines: ISA, PCI, CPU. To be fair I forgot to check the other two lines but I’d need to force all three lines and not the ISA only. It’s an idea though.
@@tony359 ok 👌 sometimes i remove and reball the old one and try. It only cost the balls and time :) but in your case its a bad place mutch plastic around.
If you don't replace the southbridge, that would be not very satisfactory for you and for us, right? ^_~ And if the chip doesn't cost too much, it's free training for you, working on BGA chips! And an educational and possibly entertaining video for us :D So, if you feel like it, please do it :D
of course! Well, between the controller, the mosfet and the inverter, I've already spent more money this board is worth - that said, it's not for money, right? :)
I seem to be in the minority here, but i would NOT replace the southbridge, although I might let someone else make the attempt. Since you admit it's not in your wheelhouse, I would 100% understand if you just declared this a "parts board". That's what I'd do. It's not sufficiently rare to be worth saving at all costs, and I don't think I would have pursued the voltage problem because a parts killer is worse than no board at all. Once demonstrated to be dangerous to other components, I would have declared it "for parts only" and maybe gone so far as to remove the power connectors from the board to make damn sure nobody tries to use it. I guess I'm saying you crossed the line where I would have accepted failure a long time ago, and I honestly think you should accept failure at this point if you don't have a high degree of confidence in replacing the southbridge.
I see your point but "accepting failure" is not what I usually do! :) The board is broken anyways, you might want to see the attempt - if I ever decided to go for it - as BGA practice :) Thanks for your input!
You can think of the ESR as an AC resistance in series with the capacitor ( AC resistance because it manifests in AC ), and knowing that helps quite a bit, because ESR testing is done be passing an AC signal trough the capacitor and measuring the current and the voltage drop, an equivalent series resistance can be calculated. The problem is that the testing signal frequency must be high enough to make the capacitive reactance unsignificant, but not so high as to get an value influenced by the ESL. If testing on a lower frequency then the software must calculate firs the capacitive reactance from the frequency and the measured capacitance, and subtract that capacitive reactance from the final ESR value. Also, for the value to be a precise one, the rms current drawn by the capacitor under testing, must be accurately measured, and that way after measuring the AC voltage drop on the capacitor, an ESR value can be calculated. Granted, there may be many other ways to do it, but this is the most intuitive i know of ( easier to understand ). On a regular ESR tester ( meaning one on which you can select the ESR function before the actual testing ), you simply use a known resistor of a small value, and let say 1% or maybe something better if possible, and you test that resistor with the ESR function, if the read value matches the resistor value, then the ESR meter is well calibrated ( it is best to test this with SMD resistors for their low inductance ). On the component tester i do not know if you can do that, but you can replicate the conditions described above. Use an audio amplifier ( the power doesn't really matters, just use what you have ), give it a sine wave at it's input, and put a capacitor at it's output as it's load, making sure to use a series resistor with that capacitor, both because the amplifier will not enjoy the capacitive load, and to limit the load current, and measure the RMS current ( sine wave because it is easier to calculate the rms ). You need to use the scope for it's high bandwidth. Use whatever frequency your amplifier is happy with ( let say 10khz ), you first calculate the capacitive reactance using the testing frequency and the tested capacitive value with the Fluke ( i think it's way more accurate then the component tester ). Then measure the voltage drop on the series resistor ( well measured value resistor, with the Fluke of course ), to calculate the rms current, then lastly measure the voltage drop on the capacitor under test pins. What voltage drop you find you divide it by the rms current, and then you subtract from it the capacitive reactance calculated at the beginning, what is left is the ESR, and it should be quite reasonable near the real value. That way you have something to compare it with the component testers, and if one of them is way closer to the value measured in the improvised way, then that component tester is more accurate.
Thanks for the extensive explanation! Someone also suggested to measure a new cap, then add a resistor in series and measure again and that resistor's value should add to the ESR value?
@@tony359 you are welcomed, as always. I did thought about that way too, and it should add up, but it may be tricky to find the right value resistor. You do not know how the firmware is configured and how the tester reads the components, thus it may see it as a resistor instead of a capacitor, or it may not see it as anything... A small enough resistor could be ok, meaning it could not confuse the tester, but too small a resistor would make the comparison pointless ( 16Ohm read ESR in series with 1R means nothing for any good comparison ). A bigger value resistor would make the comparison much more precise because a known value resistor would dominate the reading ( for example 16Ohm read ESR in series with 100R, would be much more relevant because the 100R resistor dominates the total value ), but it could also confuse the firmware, so... Ah well, you can always test it, experiment with different resistors, see what you get. You can also test new and good quality capacitors which have their ESR already specified in their datasheets, you thus know about where the component tester should read, so you will find out which of them is better. You could also find that they both read close values of ESR on new and good brand caps, that means they actually are calibrated on some level. That would mean that they read different values on bad caps because they could have different firmware configurations for reading higher values... Just sayin', electronics really is fun, ain't it?...😄
I have now tested, adding a 34 Ohm resistor. The old tester reads ESR of 34 and the same (correct) capacitance. The FNIRSI reads something completely different (I forgot!) and the capacitance dropped 20% (from 1000 to 800) Do you think this test has some meaning?
@@tony359 It could suggest that the old tester is better at ESR reading ( at least on high values ). But to be sure there needs to be much more testing, with different value capacitors and resistors, with new and old capacitors, the more you test, the more relevant the comparison. You lay every single result on a spreadsheet, then you analyze the results and you will see which of the testers is better.
Thermal camera: can you swap to another color palette that doesn't have white as the hot-spots, so you can read the text when needed? Component testers: The ESR is likely being measured at different frequencies. Even a real LCR meter can keep you guessing a bit because the ESR on caps changes based on what frequency you chose to measure them at. 60hz/120hz/1khz/10khz/100khz. I have one of those Anatek "Blue ESR" meters that runs on a freaking zilog processor of all things, the dude who made the power circuit was brilliant, same 9v battery it's whole life so far. I often prefer my Siborg "LCR-Reader" tweezers, though they were twice the price, but capacity+ESR is a winning combination that you need to know and the form factor is nice. Their models like "LCR-Reader-MPA" have even more features, but are just far too rich for my blood. There are also ones like "Elite2" and "Pro1" by LCR Research.
I could, good idea, though it should really be black as the text is miniscule on my 3K display. It's a DPI issue with the app. Yes, I understand that ESR depends on the frequency of the measurement. However, those components testers don't use high frequency to test ESR, I'll publish a follow up on the second channel to give some extra insight. I'm looking around for a proper tester, I might get one at some point. Thanks for your feedback!
@@tony359 I think I might actually partially understand what you're referring to with the parts tester now,(I needed to go look at some of the sourcecode and still only have a base understanding though!) they don't deal with measuring proper phase angles or actually push a proper waveform into the parts under test, rather they form a sort of RC network with a few different preset charge/discharge resistors and wiggle things around to take some readings. But I'd certainly watch a video with a proper explanation on how that all works.
I am not qualified to give a proper explanation! :) But I wanted to show that they don't work in the traditional way as you also discovered. The ESR reading is apparently approximating the 10KHz reading with a proper LCR meter.
@@tony359 I would just as well accept an unqualified explanation! Also fyi I really had to dig to find 259, YT has never recommended it, and it's not plainly linked anywhere, only via community tab.
Thanks for letting me know. It's the first line of the video description - but I know it can be lost in all the details. The fact that "Tony259" as a handle was taken didn't help I guess. So it's "Tony259" but the handle is "Tony359_2". Though if you search for "tony259" I think I show up. What would you recommend to make it more visible?
@@tony359 ment it Mark guy often is more masochistic than You, Tony, and Mark keeps his special grinning smile even things are running more worse than expected. For sure you as well know the Brit "Leraning Electronics Repair" living on the Canary Islands. I think doing such kind of "hobby", you need some kind of black humor of Monty Python
@@tony359remember the song Don't give up, from Peter Gabriel with Kate Bush. All my life I have had to face so many (unpleasant) problems, but even it hurds so much, especially the unnecessary ones done by evil, ignorant, lazy people, I take them as a challenge for myself. 😉
@@tony359 Power and reset sequencing in the southbridge runs from that. Another strange thing is voltage level on the ISA reset, try to check levels of other signals. May be a problem in the chipset power. There must be circuit that power chipset from +5V Vsb for ATX (and 3V3 generation) and some combination for AT power.
The document says that there should be a jumper to select AT or ATX - in AT mode, as soon as the chipset gets power, it powers up. In ATX mode, it gets in standby and waits for the power switch. But there is no such jumper on the board so...
@@tony359 The 3.3V power supply on this motherboard is generated by a passive type stabilizer placed on a massive radiator in the upper right corner. It is powered for AGP/PCI, memory, CPU I/O and most likely other 3.3V circuits. This power scheme is adopted, perhaps, on all such Baby-AT boards and on many ATX. I had exactly the same version 5ALi61 and there were no problems with it at all, despite the fact that there was an Intel 740 video accelerator, and he is very critical to the quality of the AGP implementation, and then GeForce4 MX440. Quartz and strapping on the south bridge are located between the ISA slot and the processor socket. A comment has already been advised to check the frequencies generated from it by the south bridge at the contacts OSC(b30) and CLOCK(b20) on ISA slot. But I read a comment here that probably the processor itself was damaged during the breakdown of the processor power converter, and in turn it damaged the south bridge, it looks reasonable. The north bridge may also be damaged - who knows. Damage to the power converter is, of course, strange, I would expect damage due to incorrectly installed memory, due to inappropriate or faulty ISA/PCI/AGP expansion cards, or damage to the south bridge upon arrival of static electricity discharge via USB or PS/2. The power converter controller chip has a protection circuit against incorrect load and short circuit.
Thanks - Ah yes, I forgot about the 3.3V. I'll check that too even though I believe the ALi datasheet says the core of the SB is powered by 5V. I saw the other comment - haven't responded yet because I want to check those pins first :) I checked the clock directly on the PCB and it works but checking on the ISA slot is better. I thought the 3.3V regulator would end up much closer to the AGP slot - weird! Thanks for the hints!
In my opinion you can't trust those component testers to measure ESR of a capacitor. You might find this recent video from "Defpom's Electronics Repair" interesting: th-cam.com/video/1BhfNQuzeLs/w-d-xo.html
Indeed, Scott made that video after watching mine and I made another follow up video after watching his :) The point though is that the original tester CAN be used to measure ESR, just won't be immensely accurate but perfectly fine for a "good" or "bad" answer. The FNIRSI is just useless.
In my 30 years experience in electronics to leave a board like that you would’ve been sacked ,what a mess track not repaired properly lamination damaged on the board 😂👍
Geez, "the algorithm" makes people do weird stuff - like waste time fixing purely generic and worthless socket 7 motherboards from 25 years ago - why?? Fix something that's actually worthwhile. 1996 called, they want their broken motherboard back. They also said if you really want a working socket 7 board, there's millions of them out there - just get another one. Unless of course this is just a ruse to learn BGA rework, then an old board is a perfect worthless platform to learn on that you can test in the end to see if you were successful.
I love a free internet and a free society: I am allowed to spend time fixing generic, unworthy socket 7 boards and learn something in the process and you are allowed to invest your time to tell me why I shouldn't do that and what I should invest my time doing instead! Thanks for watching!
I have the same board - just a bit different. The VRM section of mine looks a bit different and there is an IC in the bottom left corner. It would be a shame if the south bridge has an issue. I haven't tested mine, but I will let you know once it is clean and we can take some measurements! Great video! Thanks Tony!
Thanks Alex for your help - it would be great to compare our boards!
I also have a similar model, which was working until yesterday. Let me know if I can help. I may be doing the poor man reflow on it next week
pan fried "Socket 7" ?
Baked, not fried 😉
@@clintcolombin don't forget a sprinkle of olive oil a the end! ;)
Let’s replace that south bridge 🌉
👍
I’d love to see you replace the BGA chip. The board is dead without it anyway and you’re not making anything worse.
It's a good point - even if I burn it, it won't change much!
Hello Tony. You can check both component testers with a large new capacitor, something like 1000 uF or bigger. A good one should have an ESR well below 0.1 ohms. Now add in series one ohm resistor and measure it again. Now both testers should measure 1 ohm ESR if they are ok. Anyway the component tester have a calibration procedure (at least the cheap one from ebay). You can try to recalibrate it if needed. How to access the calibration procedure depends on the installed firmware, so you will need to do a bit of googling.
Hi - thanks, I wasn't aware of that procedure thanks! I did run the "calibration" before shooting the video but that didn't change much unfortunately. I'll try what you suggest later thanks!
@@tony359 The cheap tester is based on an open-source German design and is well documented and the large pdf manual is available. The calibration requires a stable small value, known good, known value ceramic cap as part of the calibration. Something like a common 100nF bypass cap is good. To start the calibration short all 3 pins on the tester together then follow the on-screen prompts. Then do the same thing in the Finirsi. If it doesn't have a calibration proceedure toss it in the bin.
@@g4z-kb7ct I found the manual yesterday. I'll see if I can trigger the calibration. The FNIRSI has a calibration which was done before shooting the video! I'll make another quick video to show some extra info the bin is where it belongs.
👍🏼do the replacement! BGA is fun! 😜
Super fun when you see everything melting away! :D
@@tony359 Kapton tape, aluminium tape or foil, anything to shield the slots will help tremendously. Don't fear the BGA!
Yes, thanks, I'm planning of making some shields with some drink cans - I think North West Repair made some. I DO fear the BGA, I've been consistently burning everything for a decade! :D
@@tony359 I bought some USB stick PCB's on Ali and used the memory chips from old Samsung SSD's to produce working memory sticks. That was cheap and I learned how to unsolder, reball and reflow without blowing off other components... :-) Most of the time I use a second hot air blower from below to preheat the board to 150 degrees (Celsius) before I start unsoldering or resoldering a chip. You can also preheat the area with the normal hot air gun. I used a thermocouple in the beginning to make sure I do not overshoot too much and also to check the temperature during the soldering process. And I use a thin flexible metal spudger to gently go under and lift the chips. You find tons of tutorials on TH-cam but it needs practice.
Oh, I have all those tools. I have the big pre-heater, I have a good hot air station etc. To be honest, I've only worked on large board with massive ground planes (PS3, PS4) and those are difficult. Smaller boards should be simpler.
regarding southbridge pdf: "Power-Good Input. This signal comes from the power supply to indicate that power is available and stable. M1543C will use this signal to generate reset sequence for the system.", so there is "sequence", what about if it is syncronized with some clock, OSC14 for example. And if we lost input clock, entire chip is stuck.
uhm interesting! I shall check that, you never know!
@@tony359 yes, the reason may be the lack of clocking of the south bridge. Quartz or strapping may be defective.
@@ILRBW Hi - thanks. I checked (after the video) the small clock IC nearby and it works. Whether it reaches the IC under the package, I cannot say :) But thanks!
@@tony359 check OSC(b30) and CLOCK(b20) on ISA slot, if both are presented, this is not clock issue
@Insomniator1 Sorry, it's taken a while. I do NOT have clocks on 20 and 30. I do on a different board as a comparison. That is the ISA clock coming from the southbridge though? Wouldn't it be affected by a dead SB? I cannot be sure but I think I saw clock on B30 for a split second on my tests. I could not replicate. There is a small crystal nearby the SB and that is outputting the proper clock. What do you think? Thanks!
18:12 Not if the tool is branded Apple. Such tools have expiration dates encoded. :)
ahahahah! Imagine an Apple Oscilloscope :)
I think there might be a bad cap and possibly a bad diode allowing feedback preventing the system to reach 3.3v. I don't see the south bridge being the issue.
I thought about that. I looked for any other place where the ISA reset line could be connected but I couldn't find any. One thing I forgot to check are the PCI reset and CPU reset. Those are coming from different pads under the SB. If those are 3.5V, maybe then is as you say.
Great socket 7 video. My old component tester needs to be 'self' calibrated regularly.
I have calibrated the FNIRSI before the video. I've never calibrated the old one - I didn't even know it was an option! Thank you!
Great video, as always. The production value is top notch
Thanks, appreciated!
Looking forward to seeing you replace that south bridge! Good luck!
It's great to see videos on modern boards. I still feel quite a novice when it comes to anything surface mount, let alone BGA!
Well, a socket 7 a modern board? Surely more modern than an Apple 2 :)
BGA is something I have been struggling to master for a while... but I am persistent (or stubborn if you want!)
If you have a new electrolytic with low ESR (like less than 1 ohm) you can just put a known small resistor in series with the cap and test that combination. That should tell you which tester is correct. Try several different series resistor values to get more confidence.
I have tried, I'll publish the result on the 2nd channel. Though those testers don't work with high frequency, they use some tricks so the resistor might actually confuse the process. But I'l show that process anyways. Thanks!
Also regarding tool longevity. Case in point: I have a Hitachi V665 analog oscilloscope, 2-channel 60MHz. It was made in 1975. I have had it for 20 years, and whoever owned it before me gave it pretty heavy use. It still works perfectly fine after almost 50 years. Even the CRT on it is still bright, sharp and nice.Well made tools, especially if you treat them right, can last you a lifetime.
They do indeed! I loved my Fluke scope - it just needed a whole bench to be used! :D
@@tony359 haha I keep the Hitachi on the floor next to the bench, standing up on its rear feet (yes it has feet on the back so it's designed to be used standing up like this). It's not only large, but also heavy. About 15kg. :)
Oh I do remember keeping the Fluke on its back feet as well and my probes were barely long to reach the boards. And I'd systematically trip in all those wires and everything would fly across the room.
I have tested the same old one component tester you have and got 0,34 ESR on a 100uf capacitors and then tested with a DER EE LCR Meter DE-5000 and got 0,42 so i think the old one is more correct than the new you got.
I've never got the feeling that that old component tester was telling me BS. I've been using it for many years... Thanks for double checking that!
You could try to isolate, brake pin between south bridge and inverter. Then try to manually push it high and low, a measure how motherboard will respond.
About component testers You could try to power them with good battery, do measurement to component, then put almost used up battery and test the same component. If one tester will have about same values and other show something different, then You will know which one is better.
Well the Fnirsi is rechargeable :)
Which line? The Power good line? That goes high and low as expected.
@@tony359 output line from south bridge.
I see your point. If the line is still at 2.2V, then something else is holding it in that state. Thanks!
I'm always happy to see new videos from you. I'm a big fan of learning and teaching and despite all my experience I still manage to learn something from you!
That is very kind of you, thank you!
Definitely I want to see you replacing the southbridge! But despite the end result is not the one we wanted, it was nice seeing you exploring the different possibilities for issues. I learned something and it is what I like.
Have you tried to push the southbridge against the board due to a bad solder connection? I could explain the changes in voltages between days and prevent the board from posting.
I have - it didn't make it in the video. I warmed it up, cooled it down, pushed in many different ways: I never had any change whatsoever.
The voltage change was due to ground - I normally use the isolation transformer just for convenience, it has a big ON OFF button. With that, I get 2.2/2.7. With mains, I get 1.7. It kinds of suggests that the line is floating maybe?
@@tony359 this means that the southbridge is damaged? I feel so...
Is it possible that 10 ohm isn't correct?
ahh... good point. I'll ask other owners! Thanks!
"white thing" is 10 ohm, it is correct, checked on my 5ali61 rev F. board, smd marking is "100" = 10 * 10^0
Thanks!
Good luck with the changing of the bga.
I used to visit very high tek companies when I worked for BSI and UL in the South West of the UK.
😅
The placement of them is a very skilled job, mostly done by very expensive equipment.
Look on TH-cam for the placement and checking they are in corret posioning of them.
I would like to see if it can be done, with bench equipment.
Thanks for your well decribed and intresting videos.
I wish the placement was my main issue! :D
Thank you, watch this space, it might happen at some point! :)
Take a look in the pci slots. Got two of these from ebay with similar behavior . Both ended up having smashed pins, grounding out inside the pci connector.
Thanks for the idea - I did check those pins and unfortunately nothing is shorted.
@@tony359 Bummer, I never had a short on the reset line. The data pins were the problem. My best guess was the shorted data lines kept the chipset in a weird in between state. Not letting it come out of reset. Best of luck!
I believe they are relevant to each other. Shorted VRM mosfet (drain and source shorted first) killed CPU (which is previous owners) with overvoltage first. Killed CPU is now shorted internally with some data pins and killed south bridge with it. While this can drain extreme current, VRM mosfet shorted with its base drain and source and killed PWM chip. I assume this happened, but i might be wrong. Do you have "original" CPU with this board? If so please check it for shorts.
I don't unfortunately. I just bought the CPU. Thanks for the hints!
Hi Tony, try re-flowing it mate, if the voltage changes there may be a solder break. Bob. UK
Hi - I tried with hot air but not to the point of melting the solder. If I go that way I'd rather swap it or re-ball it as apparently re-flowing seldom works. Hot air didn't make any difference though.
Your LCR tester must have a certain test frequency that should be stated somewhere in the details of its specifications. The higher this test frequency, the better will the test results concerning the ESR will be. Just my humble opinion. Thanks for your tinkering job !
I looked into the original project and it's not using high frequencies - the ATMEGA can't deal with that. It's using some tricks to achieve a "good" result. I'll make another video to show that!
Thanks for your comment!
I would cut the trace going from the south bridge to the 04 to at the edge of the bridge IC to make sure nobody else is conecting to the trace causing this verid voltage. If you measure the same voltage next to the bridge ic than no other way to figure out what is happening than desolder and replace the brige. After the removal of the BGA it is worth to measure the reset trace again, you never know who else is try to talk into this reset trace. Another idea: try to reflow once before replacement maybe it will help.
Today I checked the CPU reset line and it's also stuck at 1.7V - I know it comes from a different pad, I haven't checked if they happen to be electrically connected.
Oh yes, if and when I remove the SB, I'll test without it first.
Reflowing might help but it's never a good thing for the PCB and adjacent components to get to 200C multiple times so I'd be tempted to just swap it. But it's an idea thanks!
For those cheap component testers you usually do calibration where you connect pins 1,2,3 together and power up. Your old tester maybe too old to have that feature or need to have software update. For my tester it improved measurement accuracy. If I remember right it was showing too hi esr and Rds values before calibration.
Edit: Also discharge all caps before putting them to tester. I have killed one tester that way.
Thanks - The Fnirsi was calibrated just before shooting the video unfortunately.
Mine doesn't have that feature indeed - I'll see if I can update the FW!
I have this board, It works but its niggly, Windows 98 wont install nicely if USB (other BIOS setting ??) is enabled, I can't remember exactly but using ATX power, the soft-power/the power button doesn't work right if I recall correctly, if the instant power off is enabled in the BIOS it doesn't work, so its holding the button for 5 seconds, while not the end of world its gets annoying fast when troubleshoot a retro PC . Don't mess with JP2 near battery, as jumped to pin 2-3 will cause one of diodes near by to short, luckily the diode just de-soldered it self from the motherboard didn't do lasting damage.
interesting - you're saying that maybe all I have to do is to hold the ON button for 5 seconds or something like that like that weird Acer MoBo I had before? I've done most of my testing without the BIOS to be honest as I'd expect the Reset line to work without it - the board won't execute code in RESET I suppose...
I think I tried that jumper in position 2-3...
@@tony359 R178 is indeed a 10 OHM (SMD 100) resistor on my board. sorry I messaged without completely watching. No, not at your level of troubleshooting so was just giving general advice on that Board... that ALI chipset also has issues It doesn't fully cache memory above 128MB or something like that as if recall correctly unless its chipset revisions G and H. I can test other bits on my own board (its in strip down state no battery and I stole its jumpers) if that would help, I do own a scope but digital testing is not my forte.
Thanks for confirming that that resistor is actually correct!
I see FNRSI products advertised all over TH-cam.
If they cannot be bothered to create a brand name more complex than slamming their face into a keyboard..
Perhaps the warranty will not be worth the bytes it's written on?
ahahah - Well, the USB tester I purchased was good. I thought the component tester would be better tbh...
Component terminals rejecting solder like this resistor it happens, when there's a thick oxide layer. It's an indication that the component is expired and hasn't been stored well.
Also i hate components being arranged so close to a hole like that, because that's why that whole section is damaged and had to be reworked by previous owner. Is the hole needed? I don't think it's an AT/ATX mounting hole, i don't know why they put it in, did they need it for some sort of factory jig? It's terrible either way.
Ahahah when you have one voltmeter, you always know what the voltage is. When you have two voltmeters, you never know what the voltage is! Doubly true for these ESR/LCR devices!
I would rather trust the cheap one. It's a faithful reproduction of a German open source project, and the values it delivers, even if they are not what one might call exact, because really who knows, these are not straightforward values to measure, they just make sense for repair purposes, since it's quite sensitive to decayed parts. What TC3 shows doesn't make much sense! I too have the classic cheap LCR T4. Been thinking of hand building one of my own based on the upstream project.
The resistor is definitely damaged (though it works). I need go get some stock of SMD components for those cases...
The cheapo tester has served me well over the years, I cannot be 100% sure but it felt accurate enough. That 22uF cap is completely shot - I know that for sure: the TC3 doesn't reflect that!
Btw. this is NOT a mounting hole ☝️ and bssically you can find it on all corners. Possibly for production or emp- and deploying purposes.
Hi Tony, I'm enjoy this video and o so the wiling to fix it after may be a total waste of time. For me is one of your best videos a ever see in your channel. talking about the meter , I bought the DER EE DE-5000 to see and you can test your caps at 120 Hz there are a lot of videos in TH-cam talking about this tester and how to use it way better then me explain it to you. The best of all is not that expensive. Great video!!!
Hello - thanks for your kind words! I am thinking of the DE-5000 indeed. As someone said (!) a good tool lasts for a long time :)
When it happens to the board what CPU was in. I could imagine that the V-regulator grills this CPU and the 5V goes through to the ALI Chip and kills everything else. But who knows.
That's a possibility for sure...
Years ago I had an Acorp on Slot 1 and it worked caused a lot of problems :)
try hitting southbridge with some hot air to reflow before replacing it?
I have unfortunately. Hot, cold, apply pressure... I *never* had a glimpse of life out of it. Thanks for the idea!
This is a nice board and I'd like to see you getting it to work eventually, but I'm also good with seeing another video of you fixing something else, so whatever you end up deciding is good with me. 🙂
For the ESR testers, they are probably not trustworthy when it comes to ESR, but a big part of the difference could simply be that they measure the ESR using different frequencies. You could look at their output signals using the oscilloscope and check if the Fnirsi uses a higher frequency.
If a capacitor is leaking (electrically, that is), then the ESR and capacitance is more or less irrelevant and you should be able to measure the leakage current using a DC supply and multimeter. Then simply apply the rated DC voltage to the capacitor and check that the current goes to zero or very near to that (
Yes I understand that if the frequency is different then the ESR will read different. Good idea about checking it with the scope - won't the scope affect the reading at all?
@@tony359 The input capacitance of the scope should be in range of 5-15pF so that would be negligible in relation to a uF capacitor. But the idea was to mainly compare the frequency emitted from the devices. I assume they both use a fixed frequency for their capacitor measurements.
Problem is whether the scope's input impedance of 10 MOhm will confuse the meter and make it think it is measuring a resistor, but of you set the probe to 10x the input impedance should be 10x higher, or even better use a 100x probe if you have that for minimal interference.
Thanks, I'll try to measure that. I do have a 500x probe, but it's my differential probe. Would that help?
@@tony359 Diff probes are active devices so they buffer the signal and do not necessarily increase the input impedance. A typical input impedance for a diff probe is 5~10 MOhm so not higher than the regular probe on 10x, which is 10 MOhm (I miswrote before since the scope is only 10 MOhm with the probe on 10x). The scope itself is 1 MOhm and basically the probe adds a 9 MOhm resistor in series with the signal when you set it to 10x.
But 10 MOhm is also reasonably high and I doubt those cheap component testers are sensitive enough to be disturbed by that. I think it is not much different from the resistance between the pins in the ZIF socket if you're in a moist environment. 🙂
Gotcha thanks. I'll try!
Hope u can get it to work, it's a great motherboard.
Fingers crossed!
With the component tester. If you get the version Tony has, make sure it has the metallic crystal on the other side! The other ones are junk and fail. They are a great little tool though. Helped me fix a graphics card a while ago.
I am aware there are 10000000 clones of that tester. I think I was lucky with mine! :) I had two actually - I blew up one when testing a large (still charged) capacitor...
This board becomes very special when using Tillamook core CPU
I would love to see it working after all the work you have put into it already
Thanks!
I would really like to see a replacement for PCH. You can only get better through practice.
that's also true and an old board like that is simpler than a PS4 :)
I would definitely like to see you replace the Southbridge (10k or not ;)).
thanks!
go on DO IOT but get 2 of them incase you stuff the first one up - the older boards have softer solder so they dont need as much heat
Two! I've already spent more money on this board than it's worth :)
Yes, older board are leaded solder so definitely much much easier! Thanks!
ESR doesn't really tel you too much unless you know what frequency it is measured at. Maybe the two testers are using different testing frequencies?
Probably, I can check, thanks.
loved the video, Tony! so sad for the MB, i have the exact one, aesthetically at least, but with the 430TX chipset, the 5TX52 model. it also has a problem, the FDD is not recognized at all, it's not the 40 code error, i think 3, which means, read online, FDD controller dead. you should repair it, the acorp logo when booting is magic, for me :)
Oh, does it have an Acorp logo? Interesting! I hope I can get to that point!
@2:30 I reviewed the TC3 about 6 months ago in case anyone wants to see a review on it.
I think I stumbled into your review while I was testing myself - you didn't compare with a proper ESR either if I remember right? Not complaining, just to make sure I haven't missed a review where the ESR was being compared with a proper tool.
@@tony359 I fished out the unit and will do some comparison in video on it, I’m just about to live stream and will do it in that and record some video at the same time to publish later.
@@TheDefpom Cool - if you end up doing it, I'd like to share the outcome on a short video I'd like to make.
@@tony359 Yes, I tested it this morning as part of my live stream, I did it in a format that I can put into a stand alone video also, I compared it to 2 other good capacitor testers, the DER DE5000 and the Shannon Tweezer ST42, I could have tested on a few other units also, but the results were clear just from these units, so I have some conclusive results there, I will try and do the video this week.
@@tony359 I have just emailed you a link to the video I will be making public this week.
Tip for surface mount work around plastic: Captain tape. It won't make the plastic invincible but it increases your margin for error enough to be worth it than not.
Using a preheater under the board to get the board warm will also help.
Also do yourself a favor and don't use the $60 junkazon special. Get a Quick or a Atten.
I do have a (genuine) quick :) (Junkazon special LOL!)
I do have a pre-heater!
I will make a shield using some tin (from a drink can). Plus kapton as you say.
Thanks for taking the time to send me some tips! :)
@@tony359 I don't trust those $60 stations. I don't think they have the "horsepower" of the more expensive ones and after seeing "EEVblog #1171 - Yihau WEP SMD Rework Station Meltdown!" I have concerns about fire safety.
I used to have a Aoyue and I know what you mean. They're junk and you can only burn things with them as the air temp is uneven. That said, dosdude1 does wonders with it :)
@@scotshabalam2432 I have a Katsu 8582D which has been faithfully serving me for years. I got another one recently and found a dodgy DiodeGoneWild special non Class Y1 safety capacitor inside. Now I'm 2nd guessing its safety!
I try not to buy mains-operated equipment from Aliexpress/Ebay. They can have serious safety flaws...
Did you reflow the power connector on the MB?
No, why?
@@tony359 You might have a flaky solder connection to the pins on the power header? Just a thought. I'm sure you have checked this already.
Well, anything is possible. I haven't checked that I'm getting 5V on the board to be fair.
I'm waiting for replace the southbridge. It's not the end of the story, It's a begining :)
A beginning of a new disaster! :D
@@tony359 Yes, but we love it :)
I would try to cut reset trace from the ALi chip and see what it does, before replacing it. Also, that 10Ω resistor looks a little bit sus to me.
Well, I would nee to pull it to ground, not just cutting it. And the CPU and PCI reset lines are elsewhere. One thing I did not check is what happens to those other reset lines though.
What's the resistance of reset to GND, ie, something may be pulling it down.
There's probably is a bunch of things that have to be good before the chipset releases the reset, eg, crystal has to be oscillating. I have had a MB with a stuck crystal, and when I accidently did a diode test across it, ie, gave it voltage, it started working and fixed the MB.
That's a good idea, I thought about that just an hour ago! And yes, someone also suggested the oscillator. I might re-visit that board before I got for the replacement for sure. Thanks!
I'd love to have you try to replace the southbridge. I know I would be interested in seeing that video!
The motherboard is scared now! :) Thanks!
You musst start the automatic Calibration of the LCR. Short all 3 Probes and turn it on!
And: empty Battery/ Undervoltage triggers bad Readings!
I did before shooting the video.
I can try again with a full battery, thanks.
@@tony359 Send the FNIRSI back, it´s the best you can do. Try another one...
Really nice video! I like to watch videos like this. I hope this gets much more than 10k views in short time..
I'd be happy with 6K :)
Great job Tony!
I have several boards with stuck reset lines and I have no idea what is wrong with them
maybe we can come to a conclusion of a common or most frequent problem with this symptom and hopefully it is not always a problem with south bridges...
That would be great - though I start thinking this board was zapped somehow...
Randomness, fluctuations, cycling or instability are generally caused by bad aged capacitors. Logic and chips are consistent wheter faulty or good.
This board is consistently dead! :)
@@tony359 I forgot to mention the Reset line voltage, which you said to be "all over the place." The voltage fluctuations, as seen on the meter 2.3-2.7V, are telltale signs of bad caps.
@@PavelKrejci-ik5uv It reads 1.7V when ground is connected. There are very little electrolytic caps on the board and I am confident there are none on data (reset) lines. It's difficult to probe the BGA and confirm that 5V is stable but I can check elsewhere. I appreciate caps are the evil but I doubt a cap is causing this on this board TBH. Thanks anyways!
FNIRSI TC3 displays absolute GARBAGE instead of ESR! Capacitances, however, are measured correctly.
EXAMPLE. 220 μF 63 V Samwha SD: 0.12 Ohm. The same capacitor with a 8.98 Ohm 0.5% resistor in series: 1.5 Ohm (!!!).
On an old Atmega based component tester: without the resistor 0.20 Ohm, with the same resistor 8.5 Ohm. Way closer to truth (the total resistance should be around 9.1 Ohm).
amazing thanks, I am going to do this test later on!
@@tony359 Tony, would be great if you'll do a video about the reliability of the TC-3 Multi Tester and/or use it more often. I got one for most less offer, because I have a serious lack of money and I have to repair/refurbish a lot of computer and audio stuff.
I might make a short video where I run some tests on my second channel.
@42;00 replace the HEX inverter, there was damage next to it, so there could be corrosion under it, or the chip is damaged EDIT.. damn that’s a pity.
yep! :)
But thanks!
Amazingly educative video, as always. I do not want you to go through the pain of changing the ALI chip, so please do not count my view 🙂. Perhaps you could "force" the reset down with a 555 simple circuit after 1 second of power, isolating it from other circuit? BTW, if you need a good ESR and caps meter, I do recommend the Atlas ESR70 gold. A bit expensive but it is for life.
The ISA Reset line is one of the resets. Then there are the PCI line and the CPU line (which I haven't checked BTW!) so I don't think it's going to work. I could try doing that on the CPU reset line to see if the CPU starts executing code!
Yes, I might get one of those proper meters - why aren't Fluke getting in touch for a review? :D
@@tony359 Tony, you'll get there. I understand the complexity, and you need a signal choreography there (as an old teacher used to say) 🙂 Ohhh Fluke.... I heard **somewhere** that their equipment "seems" to work 😀 Lucky you, getting Flukes without investment. My face shows envy right now 😀 Enjoy mate !!!. We'll be here to cheer you.
I said "why AREN'T Fluke getting in touch" :) I wish!
@@tony359 Oh, apologies. Long day... HEY @fluke !!!! ARE YOU LISTENING ??? Tony needs some of you merchandise !!!!! Don't be shy, the guy has 23K followers and all of them are potential customers !!!!
@@tony359 @fluke Crappy TH-cam. No tagging allowed !!!!
The difference in ESR and capacitance could be do to the frequency it tests the capacitor at. What does your Fluke say for the capacitance? Does either of your testers say the frequency for LCR testing? If you are going for a LCR meter go with the DER EE, DE-5000. Excellent LCR meter for the money. You may want to pay the extra now and get the meter with the tweezer option, expensive to add later
Thanks - yes, I am eyeing the DE-5000. I think it's time I have a decent capacitor tester :) Thanks for the hint!
@@tony359 Use oscilloscope and measure the frequency both testers are using. Maybe thats why ESR differs so much.
Btw 22u capacitor with ESR 10ohm is bad, but in transistor radio would still work OK for years if its used in AF coupling :-)
I'm looking into that, they don't use a traditional way, there is no "frequency". They use some tricks because the ATMega cannot deal with those frequencies. The original documentation of the "transistor tester" says the ESR should be comparable to the 10KHz one.
¡Gracias!
Muchas gracias Paulo! :)
Motherboards with this chipset are the most expensive super socket 7 platforms to buy (at least in my country). If it will be really necessary I would like to see reballing of this south bridge.
Thanks!
Because those boards were unstable and weird even when were new. So not many boards survived. I like more Via MVP3, it has own issues too, but mostly works.
The MVP3 of my 591P says "hi" - I still have shivers :)
By all means, give it a try, after all we will never know if you try the chip swap. Also, get a really ESR tester.
Yes I am considering a proper ESR tester :) Thanks!
I really doubt that bga chip is faulty. Chances are low... must be something else.
Still very interesting investigation!
I've done some more tinkering and I cannot find much. I wish it wasn't BGA so I could poke the legs and make sure it gets power, clock etc. I found the CPU reset line is stuck in the same weird way (hint, maybe the CPU was fried and fried the reset line?) and the small 32KHz clock is working. But yes, it could be something else for sure.
Does this board have jumper to determine power supply type? My ss7 board behaves exactly the same way when that jumper is set incorrectly or is missing.
No unfortunately. Which is weird but there is no selection for AT/ATX. I tried ATX and it powers up when the power on button is shorted.
I have never seen anyone attempt a BGA replacement. I would be interested to know how hard it is and what all it involves.
Thanks! The video is slowly approaching 10K so... :)
I will try to reflow this southbridge. Even if I don't like to see you suffer with these chipset buggers in BGA connection 😆
It's an option but if I go that way I'd rather replace it/reball it...
Come on 10k views! I wanna see you try to replace the southbridge.
Fingers crossed! (or not?) :)
It does make sense that the south bridge is damaged. What I think happened the the Vcore mosfet shorted, took out the Vcore driver chip via the ugate signal, and when it killed that chip, it then sent 12V up the powergood out of that chip to the southbridge, damaging that too. That's my guess.
Re the Component tester, have you done the calibration routine on it? You basically need 3 pins shorted together (I use a 3-pin header soldered together), and a capacitor greater than 100nf. You put shorted pins into 1-2-3 on the tester, press the button, and it'll come up selftest. it will scroll through for a bit, then ask you to remove the short, then after a bit it'll ask for the capacitor. once you fit the cap, it continues till it's done. It's worth doing, to see if it improves the accuracy. It's seems the TC3 might have this calibration mode too, so it might be worth trying, and see if it improves the readings.
Hi - the powergood line of the VRM controller is not connected to anything, you might be able to see that when I removed it. Maybe the 5V got to the CPU, the CPU was fried and sent 5V to the Southbridge? But the Southbridge is behind the Northbridge... (that would be bad news!) Thank you for your idea!
@@tony359 You're right! my bad. I thought it would get a powergood signal from the VRM chip, but obviously not. I think we'd really need to look at the circuit diagram to see where the southbridge powergood is then. It's worth checking all other things connected to the PG and Reset lines before jumping into a BGA replacement.
Without schematics it's complicated. To be honest, this is a socket 7 board. I think the PSU powergood signal is the only one going to the chipset. Also, computerguy from TRW told me his "power good line" going to the SB is 3.5V. Mine is 3.5V. So, unless there is a BGA problem, powergood is not the problem. I think.
Some power circuits have a power good signal that is used to hold reset low while it's stabilising. Could it be something like that acting up?
It could be, the white paper is several hundreds pages, I searched for "reset" and what I showed on video is the only mention. It could possibly be that something else is holding RESET - though I'd still expect RESET to be 3.5V, not 2.2V!
I think you forgot to test if the new voltage regulator IC is actually outputting the power good signal itself. in complex boards like this one single signal not good will drag down all the rest so maybe there is one single fault left somewhere making the reset line stuck. I am not sure if the southbridge itself output PG to something. I think it just receives?
It didn't make it in the video but if you watch the pads when I de-soldered the IC, you should be able to see that the PG signal is not connected - so unfortunately I doubt it's our problem.
Yes, there might be something else which is preventing the SB from starting. Who knows. One thing I forgot to check are the CPU and PCI reset lines: are they doing the same weird thing than the ISA reset lines? Thanks for your help!
i have similar form factor board , only slot 1 and also broken :P
They're cute :)
@@tony359 "bx cell 100"
could the component testers not have been calibrated to account for the extension leads you used?
It could be but it seems that I am not the only one having the same experience.
could that chipset have some sort of "bios eprom checking" to prevent the motherboard doing anything without it?
Maybe - I tried with the BIOS IC too though.
These component testers are very useful (I've had 2 that I've been using for the past 10 years or so) but you must be aware of their limitations. For instance take a JFET like J105 which has a very high Idss on the order of a few hundred mA. The tester will not identify it correctly. It reads it either as a resistor, or as a double diode. This is probably the most extreme case; there are other situations that fool the tester, for instance components that need to be tested at voltages higher than it can provide, which is around 8-9V (the 9V battery is an indication of that).
Of course. I don't expect a £5 tool to perform like a £200 one. I'm not expecting perfection. Just not 100x difference from a similar one :) Thank you for watching!
Also simple germanium diode is often not detected even germanium transistors are measured quite OK.
I haven't finished watching yet but the reason why your resistor wasn't taking solder was because it had corrosion on it; if you look closely, you can see the pitting. I would have just taken another from a donor board. Or, you can try and hold it with tweezers and clean it with a brass brush.
I need to buy one of those "books" of SMD components, I have some capacitors but not resistors. It definitely needs replacing! Thanks!
Surely your reference would be known good components for the component tester?
I forgot to test a new capacitor in both and see what they say.
Hi Tony, please make a follow up video. Also, I would like your and from the guys the opinion for a similar issue that I have. I have a motherboard "Fujitsu Siemens FSC D1837-A21 W26361-W84-X-02" and I tried the below:
I replaced the RAM with a same and functional one (from a same second motherboard that I have).
I replaced the CPU with a same and functional one (from a same second motherboard that I have).
I check the CPU socket for bend pins and all are ok.
I have replace the BIOS chip with a same and functional one (from a same second motherboard that I have).
I have replaced all the caps in the yellow circles with brand new capacitors.
I place a buzzer on the motherboard and no beep even when I remove the RAM.
On the coils of the CPU I have always 1.55V but at some point after testing I didn’t voltage at all in CPU coils and after I desoldered the resistor from the ICOMMON pin of the ISL6565 chip the it start to have again 1.55V on the CPU but still no image. Maybe this motherboard have issues with cold solder joints?
The current from the BIOS battery is 0,54uA (in the faulty one).
I tried some PCI tester cards and gave me the below codes:
The one of them show nothing "--".
The second show "0000".
The third show "FF".
Unfortunately, I can’t find any schematic of this motherboard. Do you know what else to check? Please I would very appreciate any advice. I have three months now that I am trying to repair this motherbord (I am not doing it for the money but because every problem give you new knowledge). Also, I have made many measurements with multimeter and oscilloscope and analysis of the phase chip on the eevblog forum (Thread name: "Mainboard with No Picture")... if you please take a look. Please I would very appreciate any advice. Also, what is the "reset sequence for the system"...? Where can I read more for this? I apologise if my test is big. Thank you very much for the sharing of your knowledge.
that board is a socket 775, it's way more modern than this one! :)
@@tony359 except of modern it is a big headache and a ghost too because I can't find any schematic.
Couldn't that be some bad cap or resistive path pulling the inverter's limited high output voltage down?
Well, I am told that the Power Good line from the inverter reads 3.5V on a working board. The reset line reads the same weird voltage without the inverter. Anything's possible but it feels unlikely, what do you think?
YES BGA SOUTH BRIDGE REPLACING.
thanks!
Nice! Enjoyed the video.
Glad to hear it thanks!
If you're not comfortable with doing BGA, I wouldn't want to push you. Or recommend you do it as anything other than a practice test, knowing this board is expendable.
I know it wouldn't actually fix what's going on, but would it be advisable for testing purposes to bypass the southbridge's weird mid-voltage output and instead feed 3.5v directly into the place that expects it, in order to clear reset? That way you could see if it works otherwise .
I can try that but the SB has three reset lines: ISA, PCI, CPU. To be fair I forgot to check the other two lines but I’d need to force all three lines and not the ISA only. It’s an idea though.
Replace southbridge, please! 🙂
Thanks!
Would like to see bga replaced
Thanks!
Tony, what P.O.S.T. tester card are you using?
Hi - it's a very common one you can find on Ebay, it doesn't have a name to be honest. I hope you can find one!
Come on Tony replace that southbtidge you can do it!
I am pretty sure there are a lot of sadists under the Tony channel watchers 😏
ahaha LOL - next video: "I burnt another board!" :D
Why not try reflow it? I wote for bga work love your channel
Because when I reflow bad things happen! If I go that way I'd rather re-ball or replace! :)
@@tony359 ok 👌 sometimes i remove and reball the old one and try. It only cost the balls and time :) but in your case its a bad place mutch plastic around.
It's an option for sure but I've damaged too many stuff when doing BGA to want to do that twice :)
Yes, replace the Southbridge
Thanks!
@@tony359 Your always welcome! :)
If you don't replace the southbridge, that would be not very satisfactory for you and for us, right? ^_~ And if the chip doesn't cost too much, it's free training for you, working on BGA chips! And an educational and possibly entertaining video for us :D So, if you feel like it, please do it :D
of course! Well, between the controller, the mosfet and the inverter, I've already spent more money this board is worth - that said, it's not for money, right? :)
I seem to be in the minority here, but i would NOT replace the southbridge, although I might let someone else make the attempt. Since you admit it's not in your wheelhouse, I would 100% understand if you just declared this a "parts board". That's what I'd do. It's not sufficiently rare to be worth saving at all costs, and I don't think I would have pursued the voltage problem because a parts killer is worse than no board at all. Once demonstrated to be dangerous to other components, I would have declared it "for parts only" and maybe gone so far as to remove the power connectors from the board to make damn sure nobody tries to use it.
I guess I'm saying you crossed the line where I would have accepted failure a long time ago, and I honestly think you should accept failure at this point if you don't have a high degree of confidence in replacing the southbridge.
I see your point but "accepting failure" is not what I usually do! :)
The board is broken anyways, you might want to see the attempt - if I ever decided to go for it - as BGA practice :)
Thanks for your input!
hello i have this issue also not with this brand of tester but in my case have calibration option on menu and this solve the problem
I calibrated this unit before shooting the video 🙂
What about trying a pullup resistor to 3.3V in case the output of a chip is weak?
on the Reset line? I did look for a pullup somewhere on the board but couldn't find any. Are you suggesting to pull-up the whole reset line?
inverter chip is after SB, so probably weak is SB, but it has 3 reset outputs - ISA, PCI and CPU
Indeed. I could try to pull up the CPU reset line and see if it starts?
bummer! 😓
You can think of the ESR as an AC resistance in series with the capacitor ( AC resistance because it manifests in AC ), and knowing that helps quite a bit, because ESR testing is done be passing an AC signal trough the capacitor and measuring the current and the voltage drop, an equivalent series resistance can be calculated.
The problem is that the testing signal frequency must be high enough to make the capacitive reactance unsignificant, but not so high as to get an value influenced by the ESL.
If testing on a lower frequency then the software must calculate firs the capacitive reactance from the frequency and the measured capacitance, and subtract that capacitive reactance from the final ESR value.
Also, for the value to be a precise one, the rms current drawn by the capacitor under testing, must be accurately measured, and that way after measuring the AC voltage drop on the capacitor, an ESR value can be calculated.
Granted, there may be many other ways to do it, but this is the most intuitive i know of ( easier to understand ).
On a regular ESR tester ( meaning one on which you can select the ESR function before the actual testing ), you simply use a known resistor of a small value, and let say 1% or maybe something better if possible, and you test that resistor with the ESR function, if the read value matches the resistor value, then the ESR meter is well calibrated ( it is best to test this with SMD resistors for their low inductance ).
On the component tester i do not know if you can do that, but you can replicate the conditions described above.
Use an audio amplifier ( the power doesn't really matters, just use what you have ), give it a sine wave at it's input, and put a capacitor at it's output as it's load, making sure to use a series resistor with that capacitor, both because the amplifier will not enjoy the capacitive load, and to limit the load current, and measure the RMS current ( sine wave because it is easier to calculate the rms ).
You need to use the scope for it's high bandwidth.
Use whatever frequency your amplifier is happy with ( let say 10khz ), you first calculate the capacitive reactance using the testing frequency and the tested capacitive value with the Fluke ( i think it's way more accurate then the component tester ).
Then measure the voltage drop on the series resistor ( well measured value resistor, with the Fluke of course ), to calculate the rms current, then lastly measure the voltage drop on the capacitor under test pins.
What voltage drop you find you divide it by the rms current, and then you subtract from it the capacitive reactance calculated at the beginning, what is left is the ESR, and it should be quite reasonable near the real value.
That way you have something to compare it with the component testers, and if one of them is way closer to the value measured in the improvised way, then that component tester is more accurate.
Thanks for the extensive explanation! Someone also suggested to measure a new cap, then add a resistor in series and measure again and that resistor's value should add to the ESR value?
@@tony359 you are welcomed, as always.
I did thought about that way too, and it should add up, but it may be tricky to find the right value resistor.
You do not know how the firmware is configured and how the tester reads the components, thus it may see it as a resistor instead of a capacitor, or it may not see it as anything...
A small enough resistor could be ok, meaning it could not confuse the tester, but too small a resistor would make the comparison pointless ( 16Ohm read ESR in series with 1R means nothing for any good comparison ).
A bigger value resistor would make the comparison much more precise because a known value resistor would dominate the reading ( for example 16Ohm read ESR in series with 100R, would be much more relevant because the 100R resistor dominates the total value ), but it could also confuse the firmware, so...
Ah well, you can always test it, experiment with different resistors, see what you get.
You can also test new and good quality capacitors which have their ESR already specified in their datasheets, you thus know about where the component tester should read, so you will find out which of them is better.
You could also find that they both read close values of ESR on new and good brand caps, that means they actually are calibrated on some level. That would mean that they read different values on bad caps because they could have different firmware configurations for reading higher values...
Just sayin', electronics really is fun, ain't it?...😄
I have now tested, adding a 34 Ohm resistor. The old tester reads ESR of 34 and the same (correct) capacitance. The FNIRSI reads something completely different (I forgot!) and the capacitance dropped 20% (from 1000 to 800)
Do you think this test has some meaning?
@@tony359 It could suggest that the old tester is better at ESR reading ( at least on high values ).
But to be sure there needs to be much more testing, with different value capacitors and resistors, with new and old capacitors, the more you test, the more relevant the comparison.
You lay every single result on a spreadsheet, then you analyze the results and you will see which of the testers is better.
Thanks. I'll do some tests - but also understand that the final test is with a proper LCR meter :)
Thermal camera: can you swap to another color palette that doesn't have white as the hot-spots, so you can read the text when needed?
Component testers: The ESR is likely being measured at different frequencies. Even a real LCR meter can keep you guessing a bit because the ESR on caps changes based on what frequency you chose to measure them at. 60hz/120hz/1khz/10khz/100khz.
I have one of those Anatek "Blue ESR" meters that runs on a freaking zilog processor of all things, the dude who made the power circuit was brilliant, same 9v battery it's whole life so far. I often prefer my Siborg "LCR-Reader" tweezers, though they were twice the price, but capacity+ESR is a winning combination that you need to know and the form factor is nice. Their models like "LCR-Reader-MPA" have even more features, but are just far too rich for my blood. There are also ones like "Elite2" and "Pro1" by LCR Research.
I could, good idea, though it should really be black as the text is miniscule on my 3K display. It's a DPI issue with the app.
Yes, I understand that ESR depends on the frequency of the measurement. However, those components testers don't use high frequency to test ESR, I'll publish a follow up on the second channel to give some extra insight. I'm looking around for a proper tester, I might get one at some point. Thanks for your feedback!
@@tony359 I think I might actually partially understand what you're referring to with the parts tester now,(I needed to go look at some of the sourcecode and still only have a base understanding though!) they don't deal with measuring proper phase angles or actually push a proper waveform into the parts under test, rather they form a sort of RC network with a few different preset charge/discharge resistors and wiggle things around to take some readings. But I'd certainly watch a video with a proper explanation on how that all works.
I am not qualified to give a proper explanation! :) But I wanted to show that they don't work in the traditional way as you also discovered. The ESR reading is apparently approximating the 10KHz reading with a proper LCR meter.
@@tony359 I would just as well accept an unqualified explanation!
Also fyi I really had to dig to find 259, YT has never recommended it, and it's not plainly linked anywhere, only via community tab.
Thanks for letting me know. It's the first line of the video description - but I know it can be lost in all the details. The fact that "Tony259" as a handle was taken didn't help I guess. So it's "Tony259" but the handle is "Tony359_2". Though if you search for "tony259" I think I show up.
What would you recommend to make it more visible?
These component testers need to be calibrated with a known capacitor before you can reliably measure capacitance and ESR. Just saying...
Well, the "calibration procedure" is to short the leads and press a button.
What kind of calibration can you do with a known capacitor?
@@tony359 no idea. Maybe to set a baseline. On mine I need to measure a certain value, but the tester guides me through the steps.
So when you create your 3rd channel will it be tony159?
Tony3359 :)
I hope you love puzzles 😊
uhm... sometimes :)
@@tony359 ment it Mark guy often is more masochistic than You, Tony, and Mark keeps his special grinning smile even things are running more worse than expected. For sure you as well know the Brit "Leraning Electronics Repair" living on the Canary Islands. I think doing such kind of "hobby", you need some kind of black humor of Monty Python
I watch each of Mark's video with my jaw on the floor, then he proceeds to machine new screws or knobs...
@@tony359remember the song Don't give up, from Peter Gabriel with Kate Bush.
All my life I have had to face so many (unpleasant) problems, but even it hurds so much, especially the unnecessary ones done by evil, ignorant, lazy people, I take them as a challenge for myself. 😉
@@tony359 exactly! 😁
Verify clocking of the chip (14 Mhz). And I thing some of this boards require good 3V bios battery to dessert reset.
I tried with and without battery - but the clock is a good idea, thanks!
@@tony359 Power and reset sequencing in the southbridge runs from that. Another strange thing is voltage level on the ISA reset, try to check levels of other signals. May be a problem in the chipset power. There must be circuit that power chipset from +5V Vsb for ATX (and 3V3 generation) and some combination for AT power.
The document says that there should be a jumper to select AT or ATX - in AT mode, as soon as the chipset gets power, it powers up. In ATX mode, it gets in standby and waits for the power switch. But there is no such jumper on the board so...
@@tony359 The 3.3V power supply on this motherboard is generated by a passive type stabilizer placed on a massive radiator in the upper right corner. It is powered for AGP/PCI, memory, CPU I/O and most likely other 3.3V circuits. This power scheme is adopted, perhaps, on all such Baby-AT boards and on many ATX.
I had exactly the same version 5ALi61 and there were no problems with it at all, despite the fact that there was an Intel 740 video accelerator, and he is very critical to the quality of the AGP implementation, and then GeForce4 MX440.
Quartz and strapping on the south bridge are located between the ISA slot and the processor socket. A comment has already been advised to check the frequencies generated from it by the south bridge at the contacts OSC(b30) and CLOCK(b20) on ISA slot.
But I read a comment here that probably the processor itself was damaged during the breakdown of the processor power converter, and in turn it damaged the south bridge, it looks reasonable. The north bridge may also be damaged - who knows. Damage to the power converter is, of course, strange, I would expect damage due to incorrectly installed memory, due to inappropriate or faulty ISA/PCI/AGP expansion cards, or damage to the south bridge upon arrival of static electricity discharge via USB or PS/2. The power converter controller chip has a protection circuit against incorrect load and short circuit.
Thanks - Ah yes, I forgot about the 3.3V. I'll check that too even though I believe the ALi datasheet says the core of the SB is powered by 5V. I saw the other comment - haven't responded yet because I want to check those pins first :) I checked the clock directly on the PCB and it works but checking on the ISA slot is better. I thought the 3.3V regulator would end up much closer to the AGP slot - weird! Thanks for the hints!
In my opinion you can't trust those component testers to measure ESR of a capacitor. You might find this recent video from "Defpom's Electronics Repair" interesting: th-cam.com/video/1BhfNQuzeLs/w-d-xo.html
Indeed, Scott made that video after watching mine and I made another follow up video after watching his :)
The point though is that the original tester CAN be used to measure ESR, just won't be immensely accurate but perfectly fine for a "good" or "bad" answer. The FNIRSI is just useless.
@@tony359 I agree 100 percent...
Replace south bridge
the video is slowly approaching 10K - which is a surprise for me - so maybe... :)
In my 30 years experience in electronics to leave a board like that you would’ve been sacked ,what a mess track not repaired properly lamination damaged on the board 😂👍
well I suppose that was not really a professional repair - unless you are referring to me! 😂
Geez, "the algorithm" makes people do weird stuff - like waste time fixing purely generic and worthless socket 7 motherboards from 25 years ago - why?? Fix something that's actually worthwhile. 1996 called, they want their broken motherboard back. They also said if you really want a working socket 7 board, there's millions of them out there - just get another one. Unless of course this is just a ruse to learn BGA rework, then an old board is a perfect worthless platform to learn on that you can test in the end to see if you were successful.
I love a free internet and a free society: I am allowed to spend time fixing generic, unworthy socket 7 boards and learn something in the process and you are allowed to invest your time to tell me why I shouldn't do that and what I should invest my time doing instead!
Thanks for watching!