ฝัง
- เผยแพร่เมื่อ 10 ก.พ. 2025
- In this video I would like to talk about one of the most common reasons for broken electronics. Voltage regulators are used almost everywhere, let's talk about how they work and repair an old good 486 mainboard.
Mainboard model 486-PIO-3:
www.win3x.org/u...
Music by Model Povedeniya
modelp.bandcam...
Patreon:
/ necroware
Hi guys! There seem to be a lot of confusion around the type of the voltage regulator shown in this video. Many of you are talking about a switching voltage regulators controlled by PWM, as they often can be used in more modern devices. However, in this video I describe a 2-stage linear voltage regulator, which were widely used on the mainboards back in the days. I fully understand your confusion, since they are very similar to the switching voltage regulators, but not quite the same. If you are interested in technical background about the slightly older technology, which I'm talking about in this video, search for Darlington Transistor on Wikipedia or other sources you like, the idea is more or less the same. And if you are interested in a switching voltage regulator, I have another video in the pipeline about that as well. So stay tuned and I wish you an exciting tinkering!
I think everyone is just freaking out because when most English speaking electronics hobbyists say DC/DC converter, they specifically mean a buck or boost switching converter. Never mind that an LDO/linear regulator is indeed used to convert one DC voltage to another DC voltage, and could be considered a DC/DC converter. So, stop nit picking on terminology, enjoy the well produced video which goes to the time to explain the repair, and please, please give it a rest.
It was well put together, got the point across, explained the repair without making any errors that would cause someone to make a major mistake - in other words an explanation for those trying to learn or understand. And I think he did a good job!
@@bryanpratt3933 I think that’s because there’s a bit of the video around 6:00 that goes to some trouble to explain that this kind of regulator is more efficient than a linear regulator, and so doesn’t produce as much heat.
But the thing is, this looks like it IS a linear regulator. It has an external controller, in contrast to the LM78xx series where it’s all integrated into one IC, but the pass transistor is _exactly_ as efficient as one of those linear ICs - Ohm’s Law says it must be. Perhaps the dedicated PNP can be a little more robust, but the heat it generates will be a product of the current consumed and the voltage delta between supply and set output voltage - just like an LM78xx.
For it to be more efficient, it would have to be a switching topology. But for one, there’s no inductor, so it isn’t. The Darlington reference above just means the transistor is driven by another transistor to increase its current gain. But it’s still a linear reg. :-)
I don’t mean this to take anything away from the video, but we’re all here to learn something and enjoy the shared time on the bench, so I see no harm in clarifying where things appear to have been misunderstood. (And hey, I didn’t look up the controller IC, so if I’m wrong about something, fair game to put me in my place too!! I welcome the critique.)
@@nickwallette6201 Exactly. It is basically an adjustable linear regulator that supports a much lower minimum dropout voltage than i.e. the 3V of an LM317. I've replaced this kind of linear regulator on an early socket 7 board by a cheap chinese symmetrical step down module to make it support 2V K6 CPUs that would otherwise dissipate way too much power on that linear regulator. But still a very nice video on an interesting topic nobody touched so far!
@@nickwallette6201 Yes, it definitely is a linear regulator, it just has an external pass transistor. At 7:04 it's plain to see it's a LM2951 - look up the data sheet "The LP2950 and LP2951 devices are bipolar,low-dropout voltage regulators that can accommodate a wide input supply-voltage range of up to 30 V." It's just a linear regulator with an external pass transistor.
I believe you can have a switching power supply without an inductor but you do need a capacitor. The idea would be to bring the pass transistor to saturation for only as long as it takes to bring the capacitor to desired voltage. Once reached, switch it off until it goes below a threshold.
But more likely it is as others said, a linear regulator with external pass transistor where pass transistor is in its linear region, creating a resistance to drop the voltage at the given current.
Necroware's video is the most exciting moment of the week.
Definately 😆
That must have been a very boring week then :D
@@necro_ware Not at all. Your videos are such full of great contents that make our day.
So happy I found this channel.
thanks for the video.. great job
I love repair videos :D I found this channel only last week and I've been binge watching the old repairs video !
EDIT: Although here, I must point out your explanation of the dc/dc switching converter is incorrect. What you described _is_ a linear regulator, just one with an external pass-transistor. And AFAICT it's exactly what's here, a linear regulator with external pass transistor (and this has no efficiency advantage, it's just easier to dissipate higher power ...)
What makes switching regulator more efficient is that they don't "modulate" the base voltage. They either turn it entirely ON or entirely OFF to avoid the region where the transistor dissipates power. The "raw" output would not be wavy, it's be switching from 0 to 5V which is obviously unacceptable and that's why they need an inductor and capacitor on the output. The lack of an inductor is a big clue this is not a switching regulator.
Hehe :D Exactly! I don't describe a switching regulator in this video, because I describe a 2-stage linear regulator with external power transistor, which was used on that mainboard. If you are keen to see a switching regulator on my channel, then tune in in the one of my next videos, where I plan to talk about that ;) Thank you very much for watching!
@@necro_ware Ack, looking forward to it :D
@@necro_ware You did imply that it was more efficient than a LDO though, which it is not.
Well I said, that it is more efficient, than a simple LDO like LM7805 or similar. In fact this 2-stage linear voltage regulator can control a lot higher current and gets less hot. More power and less heat means that less energy will be dissipated (wasted). So, it is more efficient -> profit ;)
@@necro_ware No, it is not more efficient. It simply has an external pass transistor (whereas the pass transistor is internal in a monolithic linear voltage regulator). The external pass transistor may have a higher power dissipation rating, or even a lower thermal resistance junction-case, allowing it to operate at higher power dissipation levels, or use a smaller heat sink for the same power dissipation. However, It is still a linear regulator, and still is just as inefficient as a monolithic linear regulator. Power burned up in the regulator is the same in either case: Voltage dropped across the regulator * current through the regulator.
Don't confuse dropout voltage with efficiency. Monolithic regulators typically require 2V or more input voltage than output voltage to properly operate in regulation. That is the trade-off for an all-in-one, simple to use part. A regulator design with an external pass transistor can have significantly lower dropout voltage (Vdo), as then it is a function of the pass transistor its self - a pass transistor with a low Vce(sat) can result in a voltage regulator with a very low dropout voltage (0.1V), assuming the Hfe of the transistor is high enough, and the regulator IC can drive enough base current. But, even a low-dropout (LDO) regulator still causes a reduction in voltage by burning it up as heat
This channel is so underrated. Really interesting videos, keep it up! 😍
Super Video mit vielen tollen Grafiken und Erklärungen! Dankeschön.
Danke für das Lob!
I still have around a Soyo 5TCS mobo, which specifically failed in memtest v5.01 only when running a P54C @ 200MHz (66x3) and the PC case was closed. After using another CPU and a bigger heatsink/fan, the problem persisted, although the heatsink would get barely warm in any case. After a more careful examination, I found that the screw that kept the voltage regulator in good contact with the correspoding heatsink was loose... Putting some thermal paste and tightening the screw completely solved the problem! Thus, inadequate VRM cooling may produce intermittent failures when stress-testing even non-CPU components. A careful inspection may reveal that such simple things can cause system instabilities - just have a second look before resorting to soldering and desoldering components. Keep up the good work!
this is the best stuff on youtube!
I'm learning so much from these videos! Thanks!
Good job! There are so many "repair" videos where it still doesn't work at the end, so this is a refreshing change.
Great video, thanks for sharing! I have the same AMD DX-2 80, but the 5V version. It gets VERY hot and definitely requires a heatsink. My Pine PT-429 motherboard doesn't support 3V CPUs, unfortunately.
Watching your video's makes me want to go to my dads house and dig in his garages for all the 90's computers and boards and cards i used to drag home as a teen.
Thanks very much for posting this repair video. I had very similar symptoms on my 486 board where in the end the power transistor TIP31C was defective. My board is now back up and running.
I'm glad I could help! I read about your success already on dosreloaded.de ;)
Great repair video! I had a motherboard with the same issue - a bit more difficult to diagnose as the transistor would output the correct voltage with no load but dropped to a lower voltage with the CPU connected!
Holy two year reply, but can you recall what the issue was? I think i'm in the same boat as you, although I've replaced both the transistor and regulator IC.
Great explanations and repair video, thank you! This will help me check my old boards.
Glad I could help! And where you are here, I'd also like to thank you for all the sound cards samples. As a sound cards fan, I think, that I heard every one video on your channel ;)
@@necro_ware Thank you! And it is a mystery to me why I didn't find your channel earlier. So many great videos, it's Gold!
Hey thanks very useful info great work to !!!
Полезно было :) Как раз сейчас занимаюсь платой 486.
Тогда могу только пожелать удачи и успеха! :)
@@necro_ware Thank you :)
Nice video, thank you :)
If that 486 DX2-80 is cold, try an overclock of the multiplier to 2.5x or bus to 50MHz? Golden sample maybe?
The multiplier on a 486 is not really changeable. AMD 486 CPUs had a "hack", where one reserved leg was put on ground to force 2x multiplier and without that hack (which many mainboards don't support) the CPUs are running with 3x multiplier. I tried this CPU at 100MHz (3x33MHz) and it ran stable. Could be possible to get it to 120MHz with FSB of 40MHz I guess.... may be another day.
@@necro_ware I've been at this a long time and never heard nor read about that AMD 486 hack before.
@@dalecomer5951 Well it's not really a hack, that's why I wrote it quoted, it's just if you simply put an AMD CPU instead of the Intel you will get a 3x multiplier instead of 2x. I just looked into the datasheet and it was a CLKMUL pin, which you have to put on ground to get the proper multiplier on AMD. But on Intel CPUs this pin could (or had to?) be left floating, what many earlier 486 mainboards did. Later boards usually added a jumper to set that pin on ground. Some boards you have to mod to get an AMD CPU work there properly. It's btw regardless if the CPU is 3.3V or 5V.
@@necro_ware Yes, not a hack but an unadvertized "feature."
@@dalecomer5951 Yes, the usual problem - "is it a feature or bug?" :D
Excellent video. Very illustrative !!!
I have exactly the same problem on my 486 motherboard. But unfortunately I don't have time to replace the voltage regulator :D thank you for another great video.
Great video as always.
Very good. More repairs please!
thank you for sharing your work.I studied one point from your work.I am from sri lanka
Buys two when getting replacement parts. Smart.
Actually, LP2951 is just a LDO voltage regulator chip, not a switching DC-DC converter. Check the datasheet. Also, DC-DC converters usually require an inductor for operation. Regardless, very nice video and congratulations to the successful repair!
Thank you very much!
Love your videos thankyou from Australia
Love these videos
Your vieos are helping lot.
TIL lp2951 and PNP transistor act as a darlington transistor setup! neato!
Exactly! +1
I think it will be Sziklai pair connection, a PNP works as a external current booster of the regulator IC itself, rather than normal darlington.
@@ifgops I looked at the data sheet, and it appears a PNP transistor is inside the IC package, with the external PNP they appeared to form a darlington pair. unless I read it wrong and it was an NPN... still neato!
Love your repair videos !
Please, make a video repairing an AT power supply.
Excellent videos. Best regards
Thank you. If some interesting case about an AT PSU will pop up, I will definitely make a video about it.
Great video … been doing something similar except that my (socket 7) motherboard outputs the correct voltages when no cpu is inserted, but acts weird when a cpu is inserted. It came with a WinChip 2000 CPU (3.52v) but I can only get it to boot when the jumpers are set to 2.8v
Thank you! If I'm allowed to guess, in that case I'd double check the capacitor first and then the controller. I also assume, that on the S7 there could be already a switching regulator rather than a linear. If there is an inductor nearby, then it would be a switching one. In such case the PWM controller is a very common culprit, which leads to exactly that behavior. Less likely to be faulty, but still possible are the resistors, which act as a voltage divider to control the feedback line. If the resistors went out of spec, that can end up in a voltage slide which would confuse the PWM controller.
Good repair! :)
I had a similar issue with the vlb board I have (V4P895P3 / SMT V: 1.0). Spent ages trying to troubleshoot as only 5v cpus seemed to work. Came down to the regulator being used. Previous owner had obviously tried to solder one on and I didn't notice how awful of a job they did. Swapped it out and it works perfectly now. Also did the same thing, shoving some 24awg solid core into the vcc and ground pins to test voltage.
Ohhh this is something I don't see every day!
The description of DC/DC converter was a little oversimplified, because it was more adequate for explaining how a linear voltage regulator works ;) In DC/DC converter the transistor is never (or at least it is not supposed to be) partially open - it is either fully open, or fully closed. It may work as PWM or PDM.
You are talking about switching voltage regulator and I was talking about 2-stage linear voltage regulator ;) If you are interested to see a switching voltage regulator, I'm planning to make a video about that one as well.... thank you for watching..
@@necro_ware DC-DC converter is a switching voltage regulator, lineal regulator doesn't use 'DC-DC converter' name.
@@volodumurkalunyak4651 Yes, my bad. They are just called sometimes DC-DC converters as well, if you go shopping for them (9:45). That's why I used the wrong naming. If you look at the functionality it is actually a DC-DC converter, since it converts DC to DC, but usually nobody means a linear converter, but a switching one by that name indeed....
Interesting tutorial TY
I believe it is also possible to mod the jumper section so it has a trim pot instead. Could be useful for overclocking and slight voltage increases. By the way, how hot does the transistor get? I would think that the heat is what probably killed the previous one...
After couple of doom runs it was slightly warm. May be around 30°C, but I didn't measure it exactly.
Very educational
Nice work , 486 board where my sort of time , where did you get the small transistor tester from ?
China, it was half not working as I got it. You could use it through USB power, but not with batteries. There were some power supply parts missing in the circuitry, so I had to solder some parts and 3D print the stand with the battery holder.
I want to get an AMD DX2 80 or DX4 133 and see if you can clock it at 66MHz FSB and not clock doubled or get the DX4 133 to work at 66MHz and clock doubled. If I remember, I did something like that with mine waaay back but just did my DX4 100 as a "DX2 100" just clock doubled but with a 50MHz FSB. It worked fine for the years that I used it and with the jumpers to set the PCI bus or VLB can't remember which, I could get really fast data transfer like when loading doom. I had a nice heat sink with fan to keep it from over heating. I think that it was faster than a P75 or P90 when I ran some tests. Just the Pentium FPU was much faster though.
I had similar problem back then, I had ~200Mhz CPU that stopped running on one board, different (slower) CPU did work, back then I had no clue how to fix it! :( :D
That was too easy! I have a Micronics M4Li for which the 3v regulator was an extra cost add-on. It wasn't a cheap board when new but the manufacturer was cheap. In photos the add-on looks like just the power transistor with a socket which fits onto a standard 3-pin header. My M4Pi (earlier rev.) had a super i/o chip which could not handle the faster (later) ATA drives.
It doesn't need to be hard, but if my planing becomes true, you will probably like one of my upcoming videos ;)
I have the predecessor, the PIO2. It's ultra stable, but also PCI transfer rates are quite low. I couldn't get a Cx 5x85 working stable in it, but then again it could overclock an AM 5x286 133 to 160 MHz using 40 MHz bus speed.
So concluding: If this board likes your CPU, it can do some tricks. If not, you better try another board (this one failed 2 5x86 which worked just fine with two other SIS based boards)
Hi, I have an old board, and i followed your video on making an ATX-AT adapter and plugging it into an ATX power supply with a -5V rail and sufficient wattage on the 5V line. When I short the green wire to ground, the board turns on, the power LED i plugged in lights up, and the Pentium Pro CPU warms up. And the board produces a long beep, pause, another same long beep, pause, same long beep, pause, and it repeats this forever. When I remove the only RAM module that I have to fit this board, the beeping goes away and theres no beep code at all. But when i plug in that same memory module again it produces the long beep pause and repeat. I tested my PCI video card in a Pentium 4 build and it works great. Does this indicate Memory Issues? Or is my board fried?
Yes, repeating long beeps is a signal for broken memory. Most Pentium boards need the memory modules to be used in pairs. For example, if you need 16MB RAM, you will need 2x8MB EDO RAM.
Klasse erklärt.
Vielen Dank!
Pozdrawiamy Marka!
???
Thoes amd 486s were great for overclocking since they ran so much cooler than intels...
The lower voltage issue could be due to the capacitor being a little warn out. That said the CPU might be running cooler as a result of the lower voltage. Even in modern systems undervolting the CPU can lower temperatures. While that can cause stability issues in newer systems it's far less of an issue on old processors like that one.
Thank you very much for the sharing your idea. In fact, I tested the capacitor off camera and it still looked quite good. I decided to leave it in place so far, since I don't like to change the parts, if they are not broken.
Speaking of the temperature, the CPU requires 3.3V and the board delivered around 3.26V, so by 0.04V difference I can't speak about real undervolting. I guess, it is normal for 3.3V 486 CPUs, that they are so much cooler, than their 5V predecessors. I guess, after a while in a closed case, the CPU would probably get much warmer, but still very surprising, that it stayed cool for such a long time.
Conversely, it could mean that CPU is just a silicon lottery winner and runs cool. That means it would probably run at 3x33 just fine, maybe with a bump to 3.45V, and _still_ not need a fan (although a heatsink is probably a good idea).
Have you ever stumbled upon POST errors "2015" and "2016"? Im getting it constantly on acer 386SX motherboard. BIOS seem to be AMI.
i remember repair a socket 3, the transistor was broken and i remplace it with c2314 and work very well, can i use 7805 too ?
Different Boards need different transistors, there are things like PNP vs NPN, pinout and electrical specs to be considered, so you can't just take any one. And 7805 is not a transistor, it is LDO to convert voltage above 7V into 5V. You need a transistor and you need 3.3V on the output, so 7805 will not work.
Hi. Curious as to what temperature you soldering iron is at for this board work. If it's not temperature controlled, then wattage. Thanks.
I use 250-300°C for leaded solder and 300-350°C for unleaded.
wait so you're saying if you need 3.3v you can just solder a wire onto the transistor heatsink!?
Not always, you have to see, which transistor is used. Some of them put the output voltage on the heat spreaded, some are putting ground on it. So, it depends.
what is the name of the transistor tester that you use if you don't mind my asking?
There is no name for it, I bought it years ago on Aliexpress. I got it non functional.... long story. But I think you can find plenty of similar devices on ebay or ali.
That ain't a DC/DC it's an LDO - there is no inductor on the board.
Maybe it's colder because of the slight undervolt and better regulator?
Well, you can't consider that 0.05V from 3.3V as an "undervolt". It is still within a proximity of 0.1V by specification. Also in my opinion, a better regulator should only play a role for the regulator itself, but not for the CPU.
with this board would an Intel DX2-66 actually work being driven at 4 volts?
Not every one, but some would work.
I wonder if the dx266 would run stable at 4v setting?
Not all, but many do run stable at 4V. I use it in one of my boards to keep it cold.
@@necro_ware That is pretty interesting. I have yet to try it myself. I bet it makes quite a difference.
Wrong pwm description. Main idea of pwm - no semi open or semi closed transistor. If transistior open it no consumes power because on zero resistance, if transistor closed it consumes no power because there is no current at all. Transistor opens / closes in loop, regulation works only due a time when transistor opened and when transistor closed. Change time of open/close - changes output voltage. After transtor connected a capacitor and coil, they smooth a spikes to direct current of needed level.
What you describe is a switching voltage regulator, what I described and was used on this mainboard is a 2-stage linear voltage regulator. A video about a switching regulators is in the pipeline as well...
Might have been deducted from the missing coil, taht it's not quite a modern VRM stage.
LP2951 is LDO, not DC-DC!
My bad, I just called it DC-DC because of 9:45. You never can make it right and people would ask, why DC-DC converter stands on the bag, despite that I'm talking about LDO or Darlington stage. I guess it is what it is.
@@necro_ware LDO is also a kind of DC converter :-)
Absolutely true :D
Or just tap the 3.3v rail from the ATX PSU adaptor you're using.
I think it stayed so cool only because Doom didn't use the FPU.
Well, the DX2-66 @5V was quite hot in the same board running Doom. The difference of almost 2V makes the CPU a lot more efficient, I just didn't expect that much difference. It certainly will heat up in a closed case on a warm day, but still, it is a huge difference.
@@necro_ware I know, sometimes running only the POST to check if the BIOS recognizes some socket 3/5/7 CPU without a heatsink is enough to make those burn your fingers.
I also have a K6-2 400 (2.2V Core 3.3V I/O) that I found on the street with twisted pins, someone had stepped on it, after straightening them very carefully I discovered that it was good and I have used it more than once on 3.3V mainboards to do those kind of compatibility tests before trying to mess with unofficial jumper combinations or regulator mods.
Once I used it to install Windows 98 and various games and it survived without any (apparent) issue but I used a half decent socket 370 heatsink with a copper core and a much faster fan.
Russian music in the last seconds?
Yes