Hi all, since the question came now about 4000 times ;) I'll just pin this answer here: my oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which is a lie and remains just a dream. This thing works reliably for up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to for repairs. If the correct voltage is not important, but the waveform and the frequency, you can even use it for up to 75MHz. After all, even if it doesn't deliver 100MHz performance as promised, I'm still glad, that I have it and it's enough for what I'm doing.
@@Arachnoid_of_the_underverse Yes, I know. He basically confirms what I said. But I wouldn't call it bad, as he said, just because it doesn't support 100MHz as advertised. For many use cases it is still good enough for it's size and price. You just have to keep in mind, that it is reliable only up to about 35MHz... enough for retro repairs.
That is typical, you only get 1/3 of rating in scope or leeds for digital signals. If you were measuring sine waves only, it would do 100Mhz. Maybe some better probes or use a short spirng ground lead will give you better signals.
As a rule of thumb the sampling frequency should be 10x the max frequency. Best to my knowledge that scope samples at 250Mhz, so realiable bandwidth iz 25Mhz.
Very unusual failure! I've also never seen a non working clock affect the system that way. I suppose we just don't know exactly how that chipset is using the 32khz signal -- it must be using it for other timers. And that tantalum being bad pulling down the rail. How interesting! Great fix!
There's probably a time-of-day IC on that power-rail as well. When it loses power, it probably can't keep holding some /IRQ line high, which in turn causes an endless stream of interrupt requests.
After watching this video, I had the "inspiration" to fix my old IBM Aptiva P166 motherboard. The machine was posting VERY VERY slowly ( the memory test took more than 5 minutes, for example...). Using only my cheapo multimeter, poking through one of the clock circuits, I discovered a bad cap that was shorting (maybe grounding?) the 5v rail from the PS/2 connectors( a trace was also broken in this area). Trace fixed, cap removed, and now my board is working again! Thank you so much for sharing your knowledge with us! Greetings from Brazil!
I guess it's different when you have a problem to fix because you absolutely need the motherboard working for some other reason. When the process is more important than the final result, such as in a hobby, a teaching environment or a youtube video, it's certainly fun to troubleshoot things =)
I saw a comment you made on an Adrian's Video and decided to check your channel. It's marvelous! Good to see someone still trying to work on old PC hardware like you. Sticking here now too. Good work!
That was a really exciting repair with an even more surprising result. Even though I'm an electronics technician myself, I've learnt something. Liked and subscribed.
Trivia: These clock crystals are usually 32 768 Hz, which is 2^15 That can be halved all the way down to 1 Hz if one so desire. They're quite universal and can be found in all kinds of stuff that only need a few kilohertz clock frequencies.
The reason that the tantalum didn't explode in this case is likely because it's not tied to the PSU's 5V rail directly, but only through one of the clock power switching transistors (as you mentioned). As the dead tantalum heats up it becomes more conductive, which normally causes the even more power to be dissipated in the capacitor, leading to thermal runaway and eventually magic smoke release. But in this case the transistor has limited gain and is acting as a current source when overloaded, so the power dissipated in the capacitor actually decreases over time as the voltage drops, and the temperature reaches an equilibrium instead of increasing unbounded.
Some excellent diagnostic skills portrayed in this video! Those tants are of course always culprit #1, sabotaging old beasts like this. I've seen them produce some nice fireworks, short out, making the power good signal go low but I've never seen them slow down a machine to a crawl like this. Very interesting indeed! Keep up the good work!
It's interesting what some people can solve, but problem with that is that your life is going around you and solving such problems can take days or even weeks, so you will spend all your free time and a lot of money on that. I am also addicted to trying to find out where is the problem, but I think it's actualy similar as alcohol or smoking, you just have to do that even when you know it's pointless, it will cost money and all your time, I just have to somehow remove this addiction and live like a normal person again, because I spent literally years by solving some problems with old hardware. 😀
Very interesting! I wish this video had been out last week, I have just finished diagnosing and repairing the RTC power supply circuit on a 286 motherboard I own. One if the transistors had shorted in my case, discharging the battery overnight.
A good trick to measuring crystals with an oscilloscope is to add a 100k resistor in series with the 'scope tip. Just wrap one lead around the tip and probe with the resistor free end. You don't get the correct peak to peak voltage readings but you do get to see the waveform more correctly under a lighter load.
@@superchiaki Just look @ 19:46 he is using his MM in frequency mode! Mine doesn't. It shows 50.00 Hz (I'm located in Switzerland and here we have 50 Hz Power Net Frequency, not 60 Hz).
This was easily the most informative video I've seen on troubleshooting a slow voltage drop connected to the RTC. It always seems to be these filter caps by the power; the 5150/5160 has these fail a lot too these days. Excellent work!
Thank you very much. The tantalum caps are really a very common issue indeed, but in this case I was very surprised by the overall effect it had on the system. I was really keen to find out what's happening and I absolutely didn't expect what I found.
Kinda lucky there tantalums are not known to fail so gracefully. Very interesting repair, I had no idea the RTC was actually critical for the system beyond telling you the time
You did a great job in finding the issue! Like some others said, I would be annoyed and at the same time propelled to find the problem. In the 90s, I had sooo many old computer parts for cheap. Then some life changes happened, along with losing many computer parts. Now, when I want to buy some of those components again... The prices are out of hand. What a shame! I especially like 386SX and also DX boards/CPUs. They can run 32 bit software while still being somewhat simple and repairable.
I have a 386Dx40 board that will hang at startup when it's at full speed but works perfectly in "turbo" reduced speed. I think you may have lead me on a path to finally diagnosing it. Your videos are extremely educational. Thank you.
I have the exact same issue with a Quantel 286 motherboard. In my case it was battery acid damage near the switching circuit that killed some of the traces. Board works now :) Thanks for posting this video!
First time viewer - your video popped up in my feed. I really enjoyed the diagnostic approach you shared. Thanks for doing what you do! I'm slowly picking away at a faulty Apple 128k motherboard. I'll try some of the techniques you're demonstrating.
Wow, you did not promise too much, that was very weird and very interesting as well. Love the instructive bits with the graphics. Learning so much about how these things generally work, just on the side. One day I'll need to finally understand what Varta in which circuit I can replace how with which coin cell.
Haha... well, usually, if the battery is a rechargeable one -- NiCd or NiMH -- then you need a diode isolation circuit to keep the board from trying to charge a non-rechargeable replacement cell. This is often literally just a diode, like B+ -->|--- In+, where B+ is the battery + terminal, the diode is --->|---, and In+ is the + input of the clock circuit. If the RTC circuit uses a primary cell that is not rechargeable, like a Lithium 3V or 3.6V cell, then you can just drop in a CR2032 instead. The latter, with 3.6V, is a little higher than a coin cell, which is 3.0V nominal, so it could be borderline sufficient. But there's often quite a bit of tolerance. I'm building a 1/2-AA replacement module with a 3D printed cylinder, a vertical coin cell holder, and a PCB to boost 3.0 to 3.6V, to power some Macs that aren't quite happy with 3.0V. The downsides: 1) You need a very efficient boost regulator, since even a 10-100uA load from the converter circuit will drastically reduce the useful life of the cell. 2) A super high efficiency reg IC tends to be very very small (like 2x2mm.) 3) One such TI chip that I found is, like many things right now, in short supply at the moment.
I've had a similar issue on a different piece of hardware : a trident 8 bit vga video card The card was working but had a terrible video output. Some of that was due to the low quality ramdac, but when I swapped it and messed with the card I noticed that the image was getting darker and darker as I used it and there were some bad interferences in the brightness of the signal. I checked the tantalum caps and one reported a capacity way over its initial capacity (like 2-3times the initial value). When I replaced it, the card suddenly displayed a beautiful picture ! When they don't fail short tantalum caps definitely have some weird side effects !
high capacitance indicates leakage because capacitance testers measure the amount of time the cap takes to charge in order to calculate the capacitance...if its leaky it will take more time to charge....never seen a leaky tantalum though...i thought only electrolytics and old paper and wax or paper in oil caps could develop lekage current, uasually tantalums just short
It was almost 25-30 years ago and I remember nothing significant since that time, but somehow I still remember this weird BIOS seconds refresh in many of the old BIOS of that time.
Capacitors are a nightmare in IT for decades computers would have random issues I personally never noticed a clo0ck issue like that except battery's dying but I certainly learnt something new today. Remember the era of popping capacitors from particular Pentium 2 till after LGA 775 when they learn to use solid capacitors. Those jelly bean part capacitors I do know a site where you can buy them you can buy nearly anything EE related I will try and find the site amongst my millions of bookmarks and site lists. I would also support your channel I love your work and at the moment work is rare as rocking horse manure in Australia due to the Rona demolishing our economy but I am trying to do research on old computers and also start a charity building older computers for poor countries that don't have access to computers. Your knowledge helps us enthusiasts and IT pro's honestly I think you discovered a problem which has plagued IT since the late 80's and parts like motherboards were just ditched and a upgrade put in its place when a simple off the shelf jellybean EE part could of fixed it with the right trouble shooting techniques. One thing I do repair when I get my hands on them is old AT and ATX power supplies and power supplies are becoming short in supply they are a dangerous component to work on as you know and I do it on a EE hobby level despite my IT qualifications. I recommend preserving and or reconditioning any AT power supplies you have just make sure there is no residual power in them and know what you are doing as you would but for others reading this BE CAREFUL
Well... You've just answered my question as to how a real time clock can malfunction. And how to test for such a malfunction. Now to ask my boss for an Oscilloscope... Also +1 subscribed to your channel. Take care, stay safe, and have fun!
the cylindrical oscillator is just a crystal that oscillates at a specific frequency, it can't oscillate by itself so it needs a driver for it, one that continuously "bangs" that crystal to keep it ringing, while the 4 pin crystal oscillators have that driver built in. when you see some voltage slowly fade away, it could also be temperature related, like a faulty transistor that stops conducting when it gets hot.
It is interesting that the board even worked. I had a problem with mainboard where the crystal was gone so it didn't even boot. So it looks like that mainboard needs the RTC impulses only at POST.
Absolutely awesome root cause analysis! I enjoyed and learned from that video a lot! Thank you! btw, from where did you get this interesting LCR meter? Is this a self made kit? cheers, peter
Hi Peter! Thank you very much, where have you been so long? I'm missing some cool CPU videos ;) The oscilloscope seems to be very interesting for so many people. Every second question is about the oscilloscope. It is not a DIY device, since all what you can get there is couple of kHz, but no way to get into MHz. My oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which is a lie and remains just a dream. This thing works reliably for up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to for repairs. If the correct voltage is not important, but the waveform and the frequency, you can even use it for up to 75MHz. After all, even if it doesn't deliver 100MHz performance as promised, I'm still glad, that I have it and it's enough for what I'm doing.
@@necro_ware Hey. Yeah, I had some issues and were totally time overloaded. But now the wintertime is coming and many videos on my channel. ☺️. I was meaning the device with which you checked the tantalum cap. This pcb with display…. cheers, peter
@@CPUGalaxy Ah, that tester. I was asked so many times now about the oscilloscope, that I assumed you mean that one as well :D Well, the tester is from China, but it was dead on arrival, the power supply didn't work and there were other issues with it. I had to modify, fix it and add a 3D printed stand with a battery holder underneath. So, yes, it's kind of self made, if you want so....
Personally, I would have never thought to check the RTC circuit. I probably wouldn't have even noticed the stopped clock in BIOS. But as for your fix, bodging a line direct to a known good 3V power source would have been my first choice for diagnosing the issue too.
this brings back pc memories... the fast 386 was the way to go, then the 486, so... i really tuned in to see how the little Oscope works on troubleshooting PC pcb... thanks a lot...😉
Wieder einmal ein super tolles Video. Wieder viel gelernt. Die ersten Tests hätte ich auch gemacht, also Spannung und Takt. Danach keine Ahnung. Vielen Dank!
Mein primäres Ziel ist die Zuschauer mit Ideen zu versorgen, die helfen könnten eigene Hardware zu reparieren. So lange ich dem eigenen Anspruch gerecht werde ist alles gut :D
Once again a great video! Very useful information too. If I ever have this problem now... I will be repaired. BTW this was a leaky cap (Partially sorted). I hate these tantalum caps. I always go bad. Just replace them with electrolytic. These are extremely cheap too. Go online and get the full set of values. :)
Thank you. I have all kind of electrolytic caps too for my DIY projects, I don't know, why I just didn't use one of them :D I just had a can full of desoldered parts in front of me on the table and just grabbed something, what IMHO would fit. LOL
I remember I had a similar issue in a laptop, I didnt know what it was until I noticed the clock in the bios was running like a stopwatch... the laptop had liquid damage and the rtc circuit got affected by it, luckily a good soak with alcohol fixed it
i was about to comment that blue capacitor was a ceramic 22 pikofarads in thermal insolation jacket probably capable of handling up to 1 kv actually , but you was smart on your own power :] anyway very interesting problem
Yeah, me too! :-) For the audience: Ceramic caps are typically rated in picofarads. So, "22K" is 22p, with a K rating, not 22-thousand. If it were a 22nF cap, it would be rated at 22000, or 223 (22 + three zeroes). Now as to what the K stands for, it seems that's a little less clear. It could be tolerance, temperature, or voltage rating. Good luck! :-P
This was awesome! As someone who is also interested in retro hardware and as someone who is currently an electrical engineering student, this is super informative and helpful. I absolutely was baffled by this issue at the outset, but found your explanation to be so compelling, simple, and interesting. Old tantalum caps sure can act weird when they go bad! Thank you for sharing this! I have a VLB video card that's suffering from a weird issue where its image fades and darkens to nothing, and I am now somewhat convinced it's a bad capacitor. I'll for sure be using these techniques and skills as I go! By the way, I absolutely loved that tester you used to check the capacitor. Do you perhaps mind sharing more info about it? It seems particularly useful! P.S. Thank you for showing some basics of probing motherboards! I haven't seen any other video that really shows the easiest ways to maneuver an oscilloscope around and, while I have one, I haven't really used it for board repairs.
Thanks a lot! Glad you liked it. The tester was a half working part, which I ordered from China years ago. It had some problems in the power supply circuits, which I fixed back then, otherwise it works quite well, but I think you can buy something better, which would work out of the box :D
@@necro_ware My apologies for taking so long to respond, but thank you for your reply! I see, that makes sense. It’s a very neat device! I saw a few out there and may have to pick one up as I tinker more and more. I also just wanted to say thank you again! As someone who’s new to all this, your channel makes everything so easy to understand and you present everything in such an interesting way!
Nice detective work on a very strange fault. Personally, I always replace tantalum capacitors on very old boards because they become unreliable just due to their age it seems. (I'm not sure that actual use causes them to wear so much as just being old does.)
What ISA VGA card would you recommend for a 286/386? All the ones I tried give me a vertical stripe pattern (sometimes its dim, sometimes more noticeable). Is it perhaps an incompatibility with LCD monitors? I noticed yours doesn't display these kind of vertical stripes. Do you know what I'm talking about?
Yes, this is an incompatibility with the TFT displays. You should take Cirrus Logic CL-GD542x, where x is 0,1,2.. those cards have good DAC, which has no problems with TFTs.
Как всегда захватывающе, настоящий триллер ) Думаю, никто и никогда не расчитывал, что железо будет жить так долго, поэтому "необычных" глюков со временем становится всё больше. И это всегда увлекательно! )
Tant beads were the bane of my life when repairing CNC controllers I had motherboard that wouldn't boot and wiped bios when the battery was flat, that battery cct can cause strange things
Great repair and very interesting fault! 👍 This tantalum cap didn't explode, because of low current in that rail, but after power up, it gets shorted internally.. Now You get potentially explosive capacitor, just solder wires, put it in front of slow motion camera and turn on 5V supply! 😂
Yeah my first guess would not have been a shorted tantalum, but it looks like that's exactly what this was. The supply was switching from battery to +5V, and is heavily current-limited, but the shorted tantalum was pulling it down. I definitely would've suspected one of those transistors first, assuming the tant was holding the charge while switching from the battery to effectively no power source. Really interesting. Nice sleuthing, Necroware!
On one hand I like the video quality and editing. Very good job. On the other hand I am shocked that you are fixing a mainboard and do not know about basic things like capacitor markings or procedures on fault finding. Some rules like never ever change two things at the same time, or if you remove a part always examine (measure) that part before putting a new one in. Did you figure out eventually what the problem was with the capacitor? Short circuit? High ESR? Low capacitance?
I suppose, this channel is not professional enough for your level. But don't be sad, if you try hard, you will find something, what fits your expectations in a better way. I wish you good luck, my friend.
@@necro_ware If you share a video publicly always be prepared to receive critics as well. Make sure you give at least the same level of respect as you were given by ones who are supporting you by commenting on your channel.
@@kralg I was not trying to be disrespectful. You were shocked by my incompetence and I just don't want to shock anybody with my hobby made videos. Also your comment was anything else, but a proper critics. I think you just really should search for professional level videos and should not waste your time on hobbyist. They are not worth the time.
Ok, I'm actually not an unfriendly person, so let me explain my behavior. I really appreciate constructive critics, which helps me to improve my skills, but telling someone, that you are shocked that the person is incompetent and does things, which the person should not do is simply insulting. Please read your comment once again and tell me, where is constructive critics hidden in there? Did you tell, where s.o. can find the datasheets to the parts? Did you think about, that what you see in the video is just a part of what happens behind the scene? Just imagine, that you didn't see everything, what happened and that I may be cut out some parts to avoid, that the video lasts 3 hours? If you like to give a good advise, learn how to give a good advise first. There is no need to waste your time and write, that you were shocked. Neither does that help you, nor me. That's why I absolutely seriously mean, that if the level of such a hobby channel is too low for you, search something, what meets your expectations. And if you like such hobby channels, then read your posts before you send them and ask yourself, how that can help the guy behind the camera to improve. And if you simply want to insult anybody, well that's another story....
i was totally not expecting the system to rely on the RTC for timing interrupts for userspace applications, i guess i'm much more used to modern systems having internal interrupts and not needing to rely on such a thing.
If the quartz oscillator starts up and stops in several seconds, replace the 10 M Ohm resistor in parallel to the oscillator. A very common problem in 40 series based generators.
I lost decades of collected parts in a house fire and haven't done any real soldering since college 20 years ago... what would you suggest as means to develop skill in diagnosing and repairing old mainboards? I do have some left but not so many I can risk destroying them without some knowledge.
Maaan, I had one of these motherboards. This is actually the same and the only motherboard I have ever killed by accident. Basically I was doing some work on it and by accident left the VGA card partially inserted and (almost asleep on my feet) I decided to push it in while the system was still on instead of powering it down first. This proved to be a terrible idea, for the first (and only) time in my life I killed a motherboard. Never again that board booted or posted in any sense. In any case that resulted in me upgrading my system to a 486SLC 40Mhz from my SX16 386, which was not the best upgrade choice I ever made, should had gone with a 386DX 40Mhz, compatibility issues with the SLC proved to be a pain in the arse to deal with.
I'm pretty sure, that your Hantek is light years better, than, what I have there. It is very basic, with only one channel and very limited capabilities. It is just a lot cheaper and even slightly smaller, than yours.
Just because it bugs me, the x10 mode when you enable it, is used to divide the signal by 10, to measure higher voltage (like 230V) with an input usually caped to 50V.
Yes, this is right, but the question is how it works. There is a resistor inside which is then turned on and when measuring very weak signals it prevents the disturbance of that signal. If you'll try to measure such a crystal with 1x your oscilloscope will pull the signal up or down and the crystal will be not strong enough to stand it.
@@necro_ware usually the input impedance of your oscilloscope is already pretty high (1MΩ/20pF) and should not be impacting your signal that much (as long as you're not playing with GHz frequencies or higher). Using the x10 mode will simply add a 9MΩ impedance in the loop to do a simple voltage divider. If 1MΩ has an effect on your circuit, 10MΩ will have about 10% of that effect. If you're in a circuit that is impacted by that kind of impedance, you might want to switch to an x100 probe which will then add a 99MΩ impedance in the loop, and then you will reduce the curent drawn by the oscilloscope to 1% of what it could have been. Note : I'm saying impedance instead of just resistance because you're not just adding a resistance, but a resistance & a capacitor.
@@necro_ware besides switching to x10 mode will also add the compensation capacitor that will introduce distortion on what you can see and might also impact the circuit. My electronics teacher always told us to use x10/x100 modes with a lot of caution.
Important note: That Crystal is not a crystal oscillator, just a crystal. It behaves electrically like a capacitor and can be seen like a pendulum in a clock. A pendulum that doesn't receive a pulse of energy will just stay in place. You need a circuit that keeps pushing the pendulum. There is likely an IC on the board that keeps "pushing" that crystal creating a quartz locked loop.
2 ปีที่แล้ว
the TC4069 buffer/inverter IC from the video is not only used to convert the sine->square wave but also to keep oscillating that crystal - it's a pretty common practice, there is a great TI article about it, just search for "Use of the CMOS Unbuffered Inverter in Oscillator Circuits"
It was already, you can find plenty of detailed videos about it. This oscilloscope is limited to s.t. around 35MHz despite, that it advertises 100MHz. However, for retro repairs like this one, it's more than enough. Even on the 486, the FSB is usually at 33MHz, or can be set to. And if you don't care about the exact voltages, but only frequency and waveform, this thing can handle up to 75MHz. But as I said, at that frequency you shouldn't believe the shown voltage values anymore ;)
Great stuff Necro! I also have some 386 motherboards that do not work really reliably... Will check this part of the Mobo for sure when I have some time to spare... Thank you! Keep on going! May I ask what is that oscilloscope you are using? Perfect for a hobbist, since it is so small.
The oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which remains just a dream. This thing works reliably up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to. If the correct voltage is not important, but the waveform and the frequency, you can use it for up to 75MHz. Even if it doesn't deliver 100MHz performance, I'm still glad, I have it and it's enough for what I'm doing. It is very small and handy.
thourghly enjoyable analysis and explainations! might i ask what model your oscilliscope is? i'm considering to buy my first osciliscope for my own hobby projects, and it's small formafactor intruiged me.
Why always the tantalum caps? My friend fixed my SS7 mobo and it also had one culprit on the board. It did not blow but nothing happened when i pressed the power button.
The good thing about tantalum caps is, that they don't leak and so don't degrade and don't damage the PCB over the time. They usually also have better ESR values and give better overall stability. On the bad side is, they either explode sometimes and/or leave us with a short circuit. It is a very rare case to see such a behavior as I showed in this video, where the tantalum is neither shorted nor exploded, but starts to be conductive as it gets warmer.
The oscilloscope is a FNIRSI-5012H, it advertises 100MHz, but remains just a dream. This thing works reliably up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is 33MHz. If the correct voltage is not important, but the waveform and the frequency, you can use it for up to 75MHz. Even if it doesn't deliver 100MHz performance, I'm still glad, I have it and it's enough for what I'm doing. It is very small and handy.
You mean why it is there? It is how such oscillators were fixed in place. Look at any mainboard from the early 90's, you will find it on all mainboards, where the oscillator is socketed and not soldered.
I have a 386 msinboard that dont reboot on reset and ctrl-alt-del. To restart I need to switch off, wait 10 seconds and switch on. If dont wait systems hangs on memory count, the less wait the less memory will be checked and system hangs. I think that a problem is in capacistors near memory modules.
Amazing job! Could you explain how the Switching circuit works ?(those two transistors) I mean the circuit which switch power from battery or PSU for BIOS. Thanks many times.
Thank you. I hope, I don't tell something wrong, but as far as know the two transistors are just a more effective diode replacement, which blocks the current from the battery into direction of non RTC circuits. All the other parts are simple voltage dividers to control the transistors. There is also at least one diode to prevent the external battery from charging, plus some bypass capacitors for stabilization, like the dead one in this video. You can find the schematics here, this is not exactly the same on all the mainboards, but mostly very similar: www.minuszerodegrees.net/images2/5170_battery_circuitry.jpg
This question seems to be very interesting for so many people. You are probably the 10th person who asks :D The oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which remains just a dream. This thing works reliably up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to. If the correct voltage is not important, but the waveform and the frequency, you can use it for up to 75MHz. Even if it doesn't deliver 100MHz performance, I'm still glad, I have it and it's enough for what I'm doing. It is very small and handy.
I have a couple of motherboards that are absolute conundrums to me. I wonder if you could isolate and fix them? One is a Packard bell model, I think a PB400, and the other is this interesting MarkVision FlexBoard, which I think was damaged too much by the battery and I think the BIOS is destroyed. Still it would be interesting to see if you could effect repairs on them.
I was hoping, that my videos would inspire and help other people to repair their hardware ;) I already showed some repair videos about battery leakage and how to proceed.
@@necro_ware I have already attempted to repair these two boards, and I was unsuccessful. I have successfully repaired about 2 dozen other systems in the past 2 years as a way to keep from going crazy during this human malware event, but these two have eluded me. I know that both have had encounters with leaking batteries, and all visible corrosion was cleaned. The Packard bell boots, and then crashes after boot and the Markvision doesn't seem to respond no matter what. These are just strange issues that go beyond my simple skills of cleaning battery corrosion and removing and replacing capacitors, resistors, diodes, and the odd transistor. From the outlay they both look like they should be simple fixes, but they have proven to be beyond my skill.
@@helldog3105 Well, unfortunately, not all of them can be repaired. At least not now. In such cases I just put the hardware, which I can't repair into a box and come back later to it. Sometimes, when I repair another piece, an idea strikes me like a lightning and suddenly I know, where to look. May be it's just not the right time... yet ;)
I believe Necroware's assessment at the end is correct. In code, it's quite common to poll the time every few hundred milliseconds. If you poll once per second, you will almost surely have a window that takes slightly over one RTC second. This means you eventually land at the very tail end of second N, and the next polled value will be at the very start of second N+2, skipping over N+1 entirely. So, you poll multiple times per second to ensure you don't miss one. But unless you're aligned perfectly with the RTC clock, you will end up with a lopsided, jittery count. The solution is to poll VERY frequently -- like every 100ms or less -- and just not update the screen if the time hasn't changed yet. But, when you're talking about low-stakes stuff like a Time-of-Day clock, it's really not important enough to worry about.
I may be simplified it a bit. You had actually three sources, a rechargable battery, an optional external battery and the PSU. With a simple diode on the external battery you would get a higher voltage drop, than with the transistor cascade. This would just dissipate the energy and would be not as efficient. The batteries had to survive years. On the other hand, this circuitry is also responsible for the control of reverse flow to the rechargeable battery.
Hi all, since the question came now about 4000 times ;) I'll just pin this answer here: my oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which is a lie and remains just a dream. This thing works reliably for up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to for repairs. If the correct voltage is not important, but the waveform and the frequency, you can even use it for up to 75MHz. After all, even if it doesn't deliver 100MHz performance as promised, I'm still glad, that I have it and it's enough for what I'm doing.
EEV ran a review of the oscilloscope th-cam.com/video/SIH48bIUU00/w-d-xo.html&ab_channel=EEVblog
@@Arachnoid_of_the_underverse Yes, I know. He basically confirms what I said. But I wouldn't call it bad, as he said, just because it doesn't support 100MHz as advertised. For many use cases it is still good enough for it's size and price. You just have to keep in mind, that it is reliable only up to about 35MHz... enough for retro repairs.
That is typical, you only get 1/3 of rating in scope or leeds for digital signals. If you were measuring sine waves only, it would do 100Mhz. Maybe some better probes or use a short spirng ground lead will give you better signals.
As a rule of thumb the sampling frequency should be 10x the max frequency. Best to my knowledge that scope samples at 250Mhz, so realiable bandwidth iz 25Mhz.
@@ergindemir7366 10x is already too good. From the sampling theory you need at least 3x of the base frequency to get OK'isch results.
Very unusual failure! I've also never seen a non working clock affect the system that way. I suppose we just don't know exactly how that chipset is using the 32khz signal -- it must be using it for other timers. And that tantalum being bad pulling down the rail. How interesting! Great fix!
There's probably a time-of-day IC on that power-rail as well. When it loses power, it probably can't keep holding some /IRQ line high, which in turn causes an endless stream of interrupt requests.
After watching this video, I had the "inspiration" to fix my old IBM Aptiva P166 motherboard. The machine was posting VERY VERY slowly ( the memory test took more than 5 minutes, for example...). Using only my cheapo multimeter, poking through one of the clock circuits, I discovered a bad cap that was shorting (maybe grounding?) the 5v rail from the PS/2 connectors( a trace was also broken in this area). Trace fixed, cap removed, and now my board is working again! Thank you so much for sharing your knowledge with us! Greetings from Brazil!
When I have motherboard issues I'm annoyed, Necroware: very exciting 😀
*rofl*
I'd feel the same way (intrigued and challenged to fix the issue) ;-)
I guess it's different when you have a problem to fix because you absolutely need the motherboard working for some other reason. When the process is more important than the final result, such as in a hobby, a teaching environment or a youtube video, it's certainly fun to troubleshoot things =)
I saw a comment you made on an Adrian's Video and decided to check your channel. It's marvelous! Good to see someone still trying to work on old PC hardware like you.
Sticking here now too. Good work!
Welcome! Glad you like it.
That was a really exciting repair with an even more surprising result. Even though I'm an electronics technician myself, I've learnt something.
Liked and subscribed.
Trivia:
These clock crystals are usually 32 768 Hz, which is 2^15
That can be halved all the way down to 1 Hz if one so desire. They're quite universal and can be found in all kinds of stuff that only need a few kilohertz clock frequencies.
yes.
Yep, their original intended purpose was, alas.. standalone RTCs! Those on walls and wrists!
The reason that the tantalum didn't explode in this case is likely because it's not tied to the PSU's 5V rail directly, but only through one of the clock power switching transistors (as you mentioned). As the dead tantalum heats up it becomes more conductive, which normally causes the even more power to be dissipated in the capacitor, leading to thermal runaway and eventually magic smoke release. But in this case the transistor has limited gain and is acting as a current source when overloaded, so the power dissipated in the capacitor actually decreases over time as the voltage drops, and the temperature reaches an equilibrium instead of increasing unbounded.
Awesome stuff. You’re the best at figuring out those old boards.
Some excellent diagnostic skills portrayed in this video!
Those tants are of course always culprit #1, sabotaging old beasts like this. I've seen them produce some nice fireworks, short out, making the power good signal go low but I've never seen them slow down a machine to a crawl like this. Very interesting indeed! Keep up the good work!
I recap most of them against high capacity SMDs in 1206 or what ever fits to the soldering pads. Never had an issue with that.
It's interesting what some people can solve, but problem with that is that your life is going around you and solving such problems can take days or even weeks, so you will spend all your free time and a lot of money on that. I am also addicted to trying to find out where is the problem, but I think it's actualy similar as alcohol or smoking, you just have to do that even when you know it's pointless, it will cost money and all your time, I just have to somehow remove this addiction and live like a normal person again, because I spent literally years by solving some problems with old hardware. 😀
Very interesting! I wish this video had been out last week, I have just finished diagnosing and repairing the RTC power supply circuit on a 286 motherboard I own. One if the transistors had shorted in my case, discharging the battery overnight.
A good trick to measuring crystals with an oscilloscope is to add a 100k resistor in series with the 'scope tip. Just wrap one lead around the tip and probe with the resistor free end. You don't get the correct peak to peak voltage readings but you do get to see the waveform more correctly under a lighter load.
works this too with an multimeter in freq. counter-mode?
@@superchiaki Just look @ 19:46 he is using his MM in frequency mode!
Mine doesn't. It shows 50.00 Hz (I'm located in Switzerland and here we have 50 Hz Power Net Frequency, not 60 Hz).
@@superchiaki The signal straight off the crystal would probably be too weak and unstable for a meter.
What a great didactic video. Thanks for sharing all this knowledge in such an educational way.
Necroware repair? I'm all into it!
Me too!
I have learned so much from you. Thank you sir and God bless.
This was easily the most informative video I've seen on troubleshooting a slow voltage drop connected to the RTC. It always seems to be these filter caps by the power; the 5150/5160 has these fail a lot too these days. Excellent work!
Thank you very much. The tantalum caps are really a very common issue indeed, but in this case I was very surprised by the overall effect it had on the system. I was really keen to find out what's happening and I absolutely didn't expect what I found.
Kinda lucky there tantalums are not known to fail so gracefully. Very interesting repair, I had no idea the RTC was actually critical for the system beyond telling you the time
You did a great job in finding the issue! Like some others said, I would be annoyed and at the same time propelled to find the problem. In the 90s, I had sooo many old computer parts for cheap. Then some life changes happened, along with losing many computer parts. Now, when I want to buy some of those components again... The prices are out of hand. What a shame!
I especially like 386SX and also DX boards/CPUs. They can run 32 bit software while still being somewhat simple and repairable.
Great work. Tantalum capacitors are notorious for failing after a number of years.
Thanks for sharing.
I have a 386Dx40 board that will hang at startup when it's at full speed but works perfectly in "turbo" reduced speed. I think you may have lead me on a path to finally diagnosing it. Your videos are extremely educational. Thank you.
Wow, who would guess? Very informative! Thanks a lot!
I have the exact same issue with a Quantel 286 motherboard. In my case it was battery acid damage near the switching circuit that killed some of the traces. Board works now :) Thanks for posting this video!
First time viewer - your video popped up in my feed. I really enjoyed the diagnostic approach you shared. Thanks for doing what you do! I'm slowly picking away at a faulty Apple 128k motherboard. I'll try some of the techniques you're demonstrating.
Thank you! Wish you a lot of success and fun ;)
Wow, you did not promise too much, that was very weird and very interesting as well. Love the instructive bits with the graphics. Learning so much about how these things generally work, just on the side.
One day I'll need to finally understand what Varta in which circuit I can replace how with which coin cell.
Haha... well, usually, if the battery is a rechargeable one -- NiCd or NiMH -- then you need a diode isolation circuit to keep the board from trying to charge a non-rechargeable replacement cell. This is often literally just a diode, like B+ -->|--- In+, where B+ is the battery + terminal, the diode is --->|---, and In+ is the + input of the clock circuit.
If the RTC circuit uses a primary cell that is not rechargeable, like a Lithium 3V or 3.6V cell, then you can just drop in a CR2032 instead. The latter, with 3.6V, is a little higher than a coin cell, which is 3.0V nominal, so it could be borderline sufficient. But there's often quite a bit of tolerance.
I'm building a 1/2-AA replacement module with a 3D printed cylinder, a vertical coin cell holder, and a PCB to boost 3.0 to 3.6V, to power some Macs that aren't quite happy with 3.0V. The downsides: 1) You need a very efficient boost regulator, since even a 10-100uA load from the converter circuit will drastically reduce the useful life of the cell. 2) A super high efficiency reg IC tends to be very very small (like 2x2mm.) 3) One such TI chip that I found is, like many things right now, in short supply at the moment.
I've had a similar issue on a different piece of hardware : a trident 8 bit vga video card
The card was working but had a terrible video output. Some of that was due to the low quality ramdac, but when I swapped it and messed with the card I noticed that the image was getting darker and darker as I used it and there were some bad interferences in the brightness of the signal.
I checked the tantalum caps and one reported a capacity way over its initial capacity (like 2-3times the initial value). When I replaced it, the card suddenly displayed a beautiful picture !
When they don't fail short tantalum caps definitely have some weird side effects !
high capacitance indicates leakage because capacitance testers measure the amount of time the cap takes to charge in order to calculate the capacitance...if its leaky it will take more time to charge....never seen a leaky tantalum though...i thought only electrolytics and old paper and wax or paper in oil caps could develop lekage current, uasually tantalums just short
maybe thats the issue i got with an old macintosh nubus-videocard.
It was almost 25-30 years ago and I remember nothing significant since that time, but somehow I still remember this weird BIOS seconds refresh in many of the old BIOS of that time.
Nice repair!
This video made me remember that slow clock update in bios on old boards :)
That was a good video.
Thank you for spelling everything out for us pea brains.
I'm not sure, sarcasm? :D
Great repair video. Very unexpected cause. Well done in trouble shooting this
Capacitors are a nightmare in IT for decades computers would have random issues I personally never noticed a clo0ck issue like that except battery's dying but I certainly learnt something new today. Remember the era of popping capacitors from particular Pentium 2 till after LGA 775 when they learn to use solid capacitors. Those jelly bean part capacitors I do know a site where you can buy them you can buy nearly anything EE related I will try and find the site amongst my millions of bookmarks and site lists. I would also support your channel I love your work and at the moment work is rare as rocking horse manure in Australia due to the Rona demolishing our economy but I am trying to do research on old computers and also start a charity building older computers for poor countries that don't have access to computers. Your knowledge helps us enthusiasts and IT pro's honestly I think you discovered a problem which has plagued IT since the late 80's and parts like motherboards were just ditched and a upgrade put in its place when a simple off the shelf jellybean EE part could of fixed it with the right trouble shooting techniques. One thing I do repair when I get my hands on them is old AT and ATX power supplies and power supplies are becoming short in supply they are a dangerous component to work on as you know and I do it on a EE hobby level despite my IT qualifications. I recommend preserving and or reconditioning any AT power supplies you have just make sure there is no residual power in them and know what you are doing as you would but for others reading this BE CAREFUL
Well... You've just answered my question as to how a real time clock can malfunction. And how to test for such a malfunction. Now to ask my boss for an Oscilloscope...
Also +1 subscribed to your channel.
Take care, stay safe, and have fun!
the cylindrical oscillator is just a crystal that oscillates at a specific frequency, it can't oscillate by itself so it needs a driver for it, one that continuously "bangs" that crystal to keep it ringing, while the 4 pin crystal oscillators have that driver built in.
when you see some voltage slowly fade away, it could also be temperature related, like a faulty transistor that stops conducting when it gets hot.
It is interesting that the board even worked. I had a problem with mainboard where the crystal was gone so it didn't even boot. So it looks like that mainboard needs the RTC impulses only at POST.
Awesome fault to stumble on, certainly a first for me too. It makes perfect sense but only once you know where the problem is coming from.
Interesting repair, i'd like to see more
Hey man, you deserve more subscribers, good video
I really hadn't considered how it would need to switch from mains to battery.
Every day's a school day!
That was really interesting, thanks for taking us along with you
Very interesting. Learning a lot from your videos.
Absolutely awesome root cause analysis! I enjoyed and learned from that video a lot! Thank you! btw, from where did you get this interesting LCR meter? Is this a self made kit? cheers, peter
Hi Peter! Thank you very much, where have you been so long? I'm missing some cool CPU videos ;) The oscilloscope seems to be very interesting for so many people. Every second question is about the oscilloscope. It is not a DIY device, since all what you can get there is couple of kHz, but no way to get into MHz. My oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which is a lie and remains just a dream. This thing works reliably for up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to for repairs. If the correct voltage is not important, but the waveform and the frequency, you can even use it for up to 75MHz. After all, even if it doesn't deliver 100MHz performance as promised, I'm still glad, that I have it and it's enough for what I'm doing.
@@necro_ware Hey. Yeah, I had some issues and were totally time overloaded. But now the wintertime is coming and many videos on my channel. ☺️. I was meaning the device with which you checked the tantalum cap. This pcb with display…. cheers, peter
@@CPUGalaxy Ah, that tester. I was asked so many times now about the oscilloscope, that I assumed you mean that one as well :D Well, the tester is from China, but it was dead on arrival, the power supply didn't work and there were other issues with it. I had to modify, fix it and add a 3D printed stand with a battery holder underneath. So, yes, it's kind of self made, if you want so....
Super interesting and a pleasure to watch! Thank you for sharing !
Hi Sebastian! Thank you very much and an extra thank you for the support! ;)
2 digits should be read as pF and K is the tolerance code. It is +/- 10%
I think that it should be a 22pF with +/- 10% tolerance.
Personally, I would have never thought to check the RTC circuit. I probably wouldn't have even noticed the stopped clock in BIOS.
But as for your fix, bodging a line direct to a known good 3V power source would have been my first choice for diagnosing the issue too.
Yeah, it's always a mix of an educated guess and a lot of luck ;)
Just found your channel and have already learned. That capacitor ringing tip I never thought of.
Great video, as usual !! love the repairing PC videos since pretty much no one does old PC repair stuff. !!!
Thank you! I wouldn't say no one, but there are definitely more home computers, which get repaired.
Neat, I don't think I've seen the RTC stop like that before.
this brings back pc memories... the fast 386 was the way to go, then the 486, so... i really tuned in to see how the little Oscope works on troubleshooting PC pcb... thanks a lot...😉
Wieder einmal ein super tolles Video. Wieder viel gelernt. Die ersten Tests hätte ich auch gemacht, also Spannung und Takt. Danach keine Ahnung. Vielen Dank!
Mein primäres Ziel ist die Zuschauer mit Ideen zu versorgen, die helfen könnten eigene Hardware zu reparieren. So lange ich dem eigenen Anspruch gerecht werde ist alles gut :D
Once again a great video!
Very useful information too.
If I ever have this problem now... I will be repaired.
BTW this was a leaky cap (Partially sorted).
I hate these tantalum caps. I always go bad.
Just replace them with electrolytic.
These are extremely cheap too.
Go online and get the full set of values. :)
Thank you. I have all kind of electrolytic caps too for my DIY projects, I don't know, why I just didn't use one of them :D I just had a can full of desoldered parts in front of me on the table and just grabbed something, what IMHO would fit. LOL
We need more of this.
Great find, learned a lot from this video
I remember I had a similar issue in a laptop, I didnt know what it was until I noticed the clock in the bios was running like a stopwatch... the laptop had liquid damage and the rtc circuit got affected by it, luckily a good soak with alcohol fixed it
great video!
Good bit of troubleshooting! 👍
i was about to comment that blue capacitor was a ceramic 22 pikofarads in thermal insolation jacket probably capable of handling up to 1 kv actually , but you was smart on your own power :]
anyway very interesting problem
Yeah, me too! :-) For the audience: Ceramic caps are typically rated in picofarads. So, "22K" is 22p, with a K rating, not 22-thousand. If it were a 22nF cap, it would be rated at 22000, or 223 (22 + three zeroes).
Now as to what the K stands for, it seems that's a little less clear. It could be tolerance, temperature, or voltage rating. Good luck! :-P
This was awesome! As someone who is also interested in retro hardware and as someone who is currently an electrical engineering student, this is super informative and helpful. I absolutely was baffled by this issue at the outset, but found your explanation to be so compelling, simple, and interesting. Old tantalum caps sure can act weird when they go bad! Thank you for sharing this! I have a VLB video card that's suffering from a weird issue where its image fades and darkens to nothing, and I am now somewhat convinced it's a bad capacitor. I'll for sure be using these techniques and skills as I go!
By the way, I absolutely loved that tester you used to check the capacitor. Do you perhaps mind sharing more info about it? It seems particularly useful!
P.S. Thank you for showing some basics of probing motherboards! I haven't seen any other video that really shows the easiest ways to maneuver an oscilloscope around and, while I have one, I haven't really used it for board repairs.
Thanks a lot! Glad you liked it. The tester was a half working part, which I ordered from China years ago. It had some problems in the power supply circuits, which I fixed back then, otherwise it works quite well, but I think you can buy something better, which would work out of the box :D
@@necro_ware My apologies for taking so long to respond, but thank you for your reply! I see, that makes sense. It’s a very neat device! I saw a few out there and may have to pick one up as I tinker more and more. I also just wanted to say thank you again! As someone who’s new to all this, your channel makes everything so easy to understand and you present everything in such an interesting way!
Nice detective work on a very strange fault. Personally, I always replace tantalum capacitors on very old boards because they become unreliable just due to their age it seems. (I'm not sure that actual use causes them to wear so much as just being old does.)
Thank you for the video!, I LOVE IT!
What ISA VGA card would you recommend for a 286/386? All the ones I tried give me a vertical stripe pattern (sometimes its dim, sometimes more noticeable). Is it perhaps an incompatibility with LCD monitors?
I noticed yours doesn't display these kind of vertical stripes. Do you know what I'm talking about?
Yes, this is an incompatibility with the TFT displays. You should take Cirrus Logic CL-GD542x, where x is 0,1,2.. those cards have good DAC, which has no problems with TFTs.
Как всегда захватывающе, настоящий триллер ) Думаю, никто и никогда не расчитывал, что железо будет жить так долго, поэтому "необычных" глюков со временем становится всё больше. И это всегда увлекательно! )
Да, я был в шоке, честно говоря, котда начало доходить, в чём причина тормозов этих. Вообще не ожидал такого поворота.
Yes refresh issue on the clock, my 386SX 16 had exactly the same issue on the exact same type of BIOS hehe
Tant beads were the bane of my life when repairing CNC controllers
I had motherboard that wouldn't boot and wiped bios when the battery was flat, that battery cct can cause strange things
Great repair and very interesting fault! 👍
This tantalum cap didn't explode, because of low current in that rail, but after power up, it gets shorted internally..
Now You get potentially explosive capacitor, just solder wires, put it in front of slow motion camera and turn on 5V supply! 😂
I already saw a lot of tantalum caps exploding, scares a shit out of me every time :D This one is already in the bin.
Yeah my first guess would not have been a shorted tantalum, but it looks like that's exactly what this was. The supply was switching from battery to +5V, and is heavily current-limited, but the shorted tantalum was pulling it down. I definitely would've suspected one of those transistors first, assuming the tant was holding the charge while switching from the battery to effectively no power source. Really interesting. Nice sleuthing, Necroware!
Very nice video! I liked it very much!
On one hand I like the video quality and editing. Very good job. On the other hand I am shocked that you are fixing a mainboard and do not know about basic things like capacitor markings or procedures on fault finding. Some rules like never ever change two things at the same time, or if you remove a part always examine (measure) that part before putting a new one in. Did you figure out eventually what the problem was with the capacitor? Short circuit? High ESR? Low capacitance?
I suppose, this channel is not professional enough for your level. But don't be sad, if you try hard, you will find something, what fits your expectations in a better way. I wish you good luck, my friend.
@@necro_ware If you share a video publicly always be prepared to receive critics as well. Make sure you give at least the same level of respect as you were given by ones who are supporting you by commenting on your channel.
@@kralg I was not trying to be disrespectful. You were shocked by my incompetence and I just don't want to shock anybody with my hobby made videos. Also your comment was anything else, but a proper critics. I think you just really should search for professional level videos and should not waste your time on hobbyist. They are not worth the time.
@@necro_ware If you think that this is the right way of seeing things and this is what help you and your viewers the most, I will not be against it.
Ok, I'm actually not an unfriendly person, so let me explain my behavior. I really appreciate constructive critics, which helps me to improve my skills, but telling someone, that you are shocked that the person is incompetent and does things, which the person should not do is simply insulting. Please read your comment once again and tell me, where is constructive critics hidden in there? Did you tell, where s.o. can find the datasheets to the parts? Did you think about, that what you see in the video is just a part of what happens behind the scene? Just imagine, that you didn't see everything, what happened and that I may be cut out some parts to avoid, that the video lasts 3 hours?
If you like to give a good advise, learn how to give a good advise first. There is no need to waste your time and write, that you were shocked. Neither does that help you, nor me. That's why I absolutely seriously mean, that if the level of such a hobby channel is too low for you, search something, what meets your expectations. And if you like such hobby channels, then read your posts before you send them and ask yourself, how that can help the guy behind the camera to improve. And if you simply want to insult anybody, well that's another story....
1st timer here.. Enjoyed your video immensely. SUB + 🔔 was a EZ decision. Thank you. Cheers from So.Ca.USA, 3rd House On the Right.
Just stumbled upon this channel. Great video and instant subscribe from me!
Really interesting.....thank you!
i was totally not expecting the system to rely on the RTC for timing interrupts for userspace applications, i guess i'm much more used to modern systems having internal interrupts and not needing to rely on such a thing.
And if you think, that back then there was no such thing like userspace/kernelspace, but only realmode DOS, you will be totally blown away :D
If the quartz oscillator starts up and stops in several seconds, replace the 10 M Ohm resistor in parallel to the oscillator. A very common problem in 40 series based generators.
That was interesting. Even if I directly thought when the clock stops that it is something with the powerdelivery.
Well, probably you are by far more experienced than me, that's why I always need my time for investigation.
I lost decades of collected parts in a house fire and haven't done any real soldering since college 20 years ago... what would you suggest as means to develop skill in diagnosing and repairing old mainboards? I do have some left but not so many I can risk destroying them without some knowledge.
Maaan, I had one of these motherboards. This is actually the same and the only motherboard I have ever killed by accident. Basically I was doing some work on it and by accident left the VGA card partially inserted and (almost asleep on my feet) I decided to push it in while the system was still on instead of powering it down first. This proved to be a terrible idea, for the first (and only) time in my life I killed a motherboard. Never again that board booted or posted in any sense. In any case that resulted in me upgrading my system to a 486SLC 40Mhz from my SX16 386, which was not the best upgrade choice I ever made, should had gone with a 386DX 40Mhz, compatibility issues with the SLC proved to be a pain in the arse to deal with.
Your oscilloscope seems to be so much better than my Hantek 2C42!
I'm pretty sure, that your Hantek is light years better, than, what I have there. It is very basic, with only one channel and very limited capabilities. It is just a lot cheaper and even slightly smaller, than yours.
Cool, i have the same Board but not Powered after years. Thanks for your work! I think, i must play Doom again :-)
I needed that info 30 years ago !!
I'm so sorry for not being there for you ;)
@@necro_ware Me crying ..... :)
GREAT GUESS WORK
Hi! What is the model of oscilloscope you use?
Just because it bugs me, the x10 mode when you enable it, is used to divide the signal by 10, to measure higher voltage (like 230V) with an input usually caped to 50V.
Yes, this is right, but the question is how it works. There is a resistor inside which is then turned on and when measuring very weak signals it prevents the disturbance of that signal. If you'll try to measure such a crystal with 1x your oscilloscope will pull the signal up or down and the crystal will be not strong enough to stand it.
@@necro_ware usually the input impedance of your oscilloscope is already pretty high (1MΩ/20pF) and should not be impacting your signal that much (as long as you're not playing with GHz frequencies or higher). Using the x10 mode will simply add a 9MΩ impedance in the loop to do a simple voltage divider. If 1MΩ has an effect on your circuit, 10MΩ will have about 10% of that effect.
If you're in a circuit that is impacted by that kind of impedance, you might want to switch to an x100 probe which will then add a 99MΩ impedance in the loop, and then you will reduce the curent drawn by the oscilloscope to 1% of what it could have been.
Note : I'm saying impedance instead of just resistance because you're not just adding a resistance, but a resistance & a capacitor.
@@necro_ware besides switching to x10 mode will also add the compensation capacitor that will introduce distortion on what you can see and might also impact the circuit. My electronics teacher always told us to use x10/x100 modes with a lot of caution.
Important note: That Crystal is not a crystal oscillator, just a crystal. It behaves electrically like a capacitor and can be seen like a pendulum in a clock. A pendulum that doesn't receive a pulse of energy will just stay in place. You need a circuit that keeps pushing the pendulum. There is likely an IC on the board that keeps "pushing" that crystal creating a quartz locked loop.
the TC4069 buffer/inverter IC from the video is not only used to convert the sine->square wave but also to keep oscillating that crystal - it's a pretty common practice, there is a great TI article about it, just search for "Use of the CMOS Unbuffered Inverter in Oscillator Circuits"
yeah, the difficulties of finding replacement 32K real time clock quartz oscillators :-D Cool video, and a nice troubleshooting mystery, though
As I said, the quarz is not a good example of a part, which is hard to find, but many others can be a pain in the back.
@@necro_ware Yeah, I get that. It just made me giggle a little bit.
You should take that scope apart, and take a look at the input circuitry. It might have some bandwidth limitation.
It was already, you can find plenty of detailed videos about it. This oscilloscope is limited to s.t. around 35MHz despite, that it advertises 100MHz. However, for retro repairs like this one, it's more than enough. Even on the 486, the FSB is usually at 33MHz, or can be set to. And if you don't care about the exact voltages, but only frequency and waveform, this thing can handle up to 75MHz. But as I said, at that frequency you shouldn't believe the shown voltage values anymore ;)
Great stuff Necro! I also have some 386 motherboards that do not work really reliably... Will check this part of the Mobo for sure when I have some time to spare... Thank you! Keep on going! May I ask what is that oscilloscope you are using? Perfect for a hobbist, since it is so small.
The oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which remains just a dream. This thing works reliably up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to. If the correct voltage is not important, but the waveform and the frequency, you can use it for up to 75MHz. Even if it doesn't deliver 100MHz performance, I'm still glad, I have it and it's enough for what I'm doing. It is very small and handy.
thourghly enjoyable analysis and explainations!
might i ask what model your oscilliscope is? i'm considering to buy my first osciliscope for my own hobby projects, and it's small formafactor intruiged me.
Thank you! Pleas take a look at the pinned comment under that video, there you will find all the information you need.
do you have a link to your tools - the scope looks interesting.
Happy to click affiliate links 😀
whats the name of your component tester you use for testing the cap bad?
Why always the tantalum caps? My friend fixed my SS7 mobo and it also had one culprit on the board. It did not blow but nothing happened when i pressed the power button.
The good thing about tantalum caps is, that they don't leak and so don't degrade and don't damage the PCB over the time. They usually also have better ESR values and give better overall stability. On the bad side is, they either explode sometimes and/or leave us with a short circuit. It is a very rare case to see such a behavior as I showed in this video, where the tantalum is neither shorted nor exploded, but starts to be conductive as it gets warmer.
2:35 "I try to use simple equipment as much as possible". Puts the board on life support. We love you Germans!
Excellent video! By the way, what is the model of your oscilloscope? It's very small and neat.
Interested to know too.
The oscilloscope is a FNIRSI-5012H, it advertises 100MHz, but remains just a dream. This thing works reliably up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is 33MHz. If the correct voltage is not important, but the waveform and the frequency, you can use it for up to 75MHz. Even if it doesn't deliver 100MHz performance, I'm still glad, I have it and it's enough for what I'm doing. It is very small and handy.
@@necro_ware Thanks!
@@necro_ware thanks!
00:38 What's with the zip tie on the oscillator?
You mean why it is there? It is how such oscillators were fixed in place. Look at any mainboard from the early 90's, you will find it on all mainboards, where the oscillator is socketed and not soldered.
Thanks, it was a very interesting video!
I have a 386 msinboard that dont reboot on reset and ctrl-alt-del. To restart I need to switch off, wait 10 seconds and switch on. If dont wait systems hangs on memory count, the less wait the less memory will be checked and system hangs. I think that a problem is in capacistors near memory modules.
Sounds like caps, yes.
@necro_ware removed 3 bad caps, now reboots!
Amazing job! Could you explain how the Switching circuit works ?(those two transistors) I mean the circuit which switch power from battery or PSU for BIOS. Thanks many times.
Thank you. I hope, I don't tell something wrong, but as far as know the two transistors are just a more effective diode replacement, which blocks the current from the battery into direction of non RTC circuits. All the other parts are simple voltage dividers to control the transistors. There is also at least one diode to prevent the external battery from charging, plus some bypass capacitors for stabilization, like the dead one in this video. You can find the schematics here, this is not exactly the same on all the mainboards, but mostly very similar:
www.minuszerodegrees.net/images2/5170_battery_circuitry.jpg
@@necro_ware Yeah, that's it. Thank you many many times.
Good repair!
How much is that scope? I'd like to pick up a cheap mobile one like that for troubleshooting.
What scope are you using?
This question seems to be very interesting for so many people. You are probably the 10th person who asks :D The oscilloscope is a FNIRSI-5012H, it advertises 100MHz, which remains just a dream. This thing works reliably up to 35MHz, but it's more than enough to repair anything up to 486, since the FSB is usually up to 33MHz, or can be limited to. If the correct voltage is not important, but the waveform and the frequency, you can use it for up to 75MHz. Even if it doesn't deliver 100MHz performance, I'm still glad, I have it and it's enough for what I'm doing. It is very small and handy.
@@necro_ware thanks!
Damn, a 386. The good old days.
I have a couple of motherboards that are absolute conundrums to me. I wonder if you could isolate and fix them? One is a Packard bell model, I think a PB400, and the other is this interesting MarkVision FlexBoard, which I think was damaged too much by the battery and I think the BIOS is destroyed. Still it would be interesting to see if you could effect repairs on them.
I was hoping, that my videos would inspire and help other people to repair their hardware ;) I already showed some repair videos about battery leakage and how to proceed.
@@necro_ware I have already attempted to repair these two boards, and I was unsuccessful. I have successfully repaired about 2 dozen other systems in the past 2 years as a way to keep from going crazy during this human malware event, but these two have eluded me. I know that both have had encounters with leaking batteries, and all visible corrosion was cleaned. The Packard bell boots, and then crashes after boot and the Markvision doesn't seem to respond no matter what. These are just strange issues that go beyond my simple skills of cleaning battery corrosion and removing and replacing capacitors, resistors, diodes, and the odd transistor. From the outlay they both look like they should be simple fixes, but they have proven to be beyond my skill.
@@helldog3105 Well, unfortunately, not all of them can be repaired. At least not now. In such cases I just put the hardware, which I can't repair into a box and come back later to it. Sometimes, when I repair another piece, an idea strikes me like a lightning and suddenly I know, where to look. May be it's just not the right time... yet ;)
That behavior on the seconds in the bios is normal. I have some boards with the same behavior.
Yes, I think so too. Just never paid attention before :D
I believe Necroware's assessment at the end is correct.
In code, it's quite common to poll the time every few hundred milliseconds. If you poll once per second, you will almost surely have a window that takes slightly over one RTC second. This means you eventually land at the very tail end of second N, and the next polled value will be at the very start of second N+2, skipping over N+1 entirely. So, you poll multiple times per second to ensure you don't miss one.
But unless you're aligned perfectly with the RTC clock, you will end up with a lopsided, jittery count. The solution is to poll VERY frequently -- like every 100ms or less -- and just not update the screen if the time hasn't changed yet. But, when you're talking about low-stakes stuff like a Time-of-Day clock, it's really not important enough to worry about.
Im curious why the complex switching between battery and psu, why not just diode OR them into RTC_VCC?
I may be simplified it a bit. You had actually three sources, a rechargable battery, an optional external battery and the PSU. With a simple diode on the external battery you would get a higher voltage drop, than with the transistor cascade. This would just dissipate the energy and would be not as efficient. The batteries had to survive years. On the other hand, this circuitry is also responsible for the control of reverse flow to the rechargeable battery.