fix_metal wrote on 2026-05-23, 12:10:

I will later attach pics, but the HDD is a HP c5270a with a Maxtor PCB. The model is so little known that you don't even hit any meaningful Google results. I've just got [0] from the bay.

From a Google search, and [1] SMD marking handbook, this component is either a digital npn transistor, or a diode.

Not sure what else to test out here. I'll attach pics. Any ideas?

The drive is very likely just a HP branded Maxtor 71084A. The PCB on The Retro Web has all transistors marked "Q" and all diodes marked "CR", which should help you find out whether the component is/was a diode or a transistor.

Since it's AT, I'm assuming it and the drive have/use the old-style 4-pin molex power connectors.

Have you checked what the PSU is actually outputting ... the two center wires (normally black) should be ground.
One side wire (usually red) will have +5
One side wire (usually yellow) will have +12

There should be no other voltage(s) on this connector.

Assuming it checks out OK, compare the PSU connector wiring to that of your original working AT supply and make sure the red(+5) and yellow(+12) wires are on the same side of the connector on both.

Only ways I can think of for a PSU to "smoke" a hard drive is 1) overvolting one (or both) power rails or 2) if the previous owner removed the connector for some reason and re-inserted the wires on the wrong sides (which would seriously overvolt the +5 while undervolting the +12)

On your working PSU, make a note of exactly which side is +5 and which is +12, and ANY time you are testing an unknown/used PSU, always check these before connecting it to anything (note that a PSU may require some load to properly regulate and may show slightly undervolt rails with no load)

DaveDDS wrote on 2026-05-23, 15:59:

Since it's AT, I'm assuming it and the drive have/use the old-style 4-pin molex power connectors.

Yes, AT with blue/brown white/blue wires and 4 pins main button switch.

Have you checked what the PSU is actually outputting ... the two center wires (normally black) should be ground. One side wire ( […]
Show full quote

Have you checked what the PSU is actually outputting ... the two center wires (normally black) should be ground.
One side wire (usually red) will have +5
One side wire (usually yellow) will have +12

There should be no other voltage(s) on this connector.

Assuming it checks out OK, compare the PSU connector wiring to that of your original working AT supply and make sure the red(+5) and yellow(+12) wires are on the same side of the connector on both.

I've checked the voltages at the mainboard connector side. With no load I would expect overvoltage, instead, I get:

* -4V instead of -5V
* -10V instead of -12V
* +10V instead of +12V

The +5V and +12V are good, but didn't measure down the actual molex cables. Will do. Although, the mainboard and the other devices are still working fine (using the known to be working PSU). So, I assume voltages are ok at molex side as well.

Only ways I can think of for a PSU to "smoke" a hard drive is 1) overvolting one (or both) power rails or 2) if the previous owner removed the connector for some reason and re-inserted the wires on the wrong sides (which would seriously overvolt the +5 while undervolting the +12)

On your working PSU, make a note of exactly which side is +5 and which is +12, and ANY time you are testing an unknown/used PSU, always check these before connecting it to anything

Sigh. That's the same reasoning I've had. Though, I can't see how I've ended up this way.

(note that a PSU may require some load to properly regulate and may show slightly undervolt rails with no load)

Which is confirming what I've been observing, and wrote above about.

mkarcher wrote on 2026-05-23, 14:55:

fix_metal wrote on 2026-05-23, 12:10:

I will later attach pics, but the HDD is a HP c5270a with a Maxtor PCB. The model is so little known that you don't even hit any meaningful Google results. I've just got [0] from the bay.

From a Google search, and [1] SMD marking handbook, this component is either a digital npn transistor, or a diode.

Not sure what else to test out here. I'll attach pics. Any ideas?

The drive is very likely just a HP branded Maxtor 71084A. The PCB on The Retro Web has all transistors marked "Q" and all diodes marked "CR", which should help you find out whether the component is/was a diode or a transistor.

!!!!! It's that exact model!!! Let me google for a PCB <3 THANK YOU

I've also filled a ticket to The Retroweb to augment the existing Maxtor page.

Here are the pics of mine anyway.

See "commented.jpg" to better understand where I've noticed the issue.

mkarcher wrote on 2026-05-23, 14:55:

... has all transistors marked "Q" and all diodes marked "CR", which should help you find out whether the component is/was a diode or a transistor.

This is quite interesting. Assuming ENGs were savvy enough back then, I would expect CR components acting as sacrifical lambs in case of spikes (overvoltage). Which, in turns, confirms the CR701 and CR702 I've swapped are diodes, and working ones, and the problem shoud be somewhere else entirely.

I'll check on that. Thank you.

It smoked right next to C710? That yellow component is a tantalum capacitor. These components are known to be prone to failure, especially when they are old and were unpowered for decades. In that case, they go short circuit, overheat and sometimes really burn down. It is quite likely that C710 failed, emitted smoke, and a fuse on the +12V line blew due to the short circuit. C710 no longer needs to be short now, the short may have burned away (in the very literal sense) inbetween. I suggest you desolder C710 and inspect that component for burn marks.

If it turns out that C710 indeed failed, it might just be bad luck that it failed in this specific situation, and possibly the "good" power supply is not the actual cause of the failure.

Couple of updates:
1. The "bad" PSU does not have any swapped cables at molex sides, 12V are at yellow where they are supposed to be
2. The CR components in finding all test good
3. The tantalum caps typically fail due to over voltage. Although, they literally burn or explode in that scenario. C710 I just tested good by measuring when unsoldered, and all the others seem good looking anyway
4. I've found the 5V line going through a dead 6.8uH inductance, but replacing it didn't solve. I would have been genuinely surprised if it did as it's 12V for spinning motor
5. There is a ~40Ohm reading between the molex 5V pin and most of the components that supposedly sit on that line. I still have to find where's that coming from
6. Last but (maybe) not least: the 5V line goes straight to a pin of the big Adaptec ASIC. Not sure why. If it got overvoltage to 12V, I think that's cooked and tombed.

@mkarcher: since 2024 I've been regularly playing with this PC and this HDD.

I'll second in finding a healthy Maxtor/HP identical drive and swap PCB, unless anyone else have better ideas

fix_metal wrote on 2026-05-24, 00:00:

3. The tantalum caps typically fail due to over voltage. Although, they literally burn or explode in that scenario. C710 I just tested good by measuring when unsoldered, and all the others seem good looking anyway

OK, so this removes failure of C710 from the list of possible issues.

fix_metal wrote on 2026-05-24, 00:00:

4. I've found the 5V line going through a dead 6.8uH inductance, but replacing it didn't solve. I would have been genuinely surprised if it did as it's 12V for spinning motor

This is a hint that there was a massive overcurrent on the +5V line that burned the old inductor. The massive overcurrent on the +5V line might have been caused by a significant overvoltage breaking down components, which are now gone for good. Nevertheless, fixing +5V stuff might actually help even on a drive that does not spin up. All IDE drivers of that age support power management in some way: The spindle motor can be controlled by the firmware, so if the firmware is not running, it is quite possible that the motor driver never gets enabled.

fix_metal wrote on 2026-05-24, 00:00:

6. Last but (maybe) not least: the 5V line goes straight to a pin of the big Adaptec ASIC. Not sure why. If it got overvoltage to 12V, I think that's cooked and tombed.

I expect that the thing that is cooked is on the other side of the 6.8uH inductor than the Molex connector. If the Adaptec ASIC is "behind" the inductor, it's not unlikely that it is the first component to break down if the +5V line is high (not even necessarily +12V, +8V might be sufficient for that), and be the component I mentioned as "now gone for good" in reply to item number four.

fix_metal wrote on 2026-05-24, 00:00:

I'll second in finding a healthy Maxtor/HP identical drive and swap PCB, unless anyone else have better ideas

Assuming the Adaptec ASIC might be cooked, a PCB swap seems like the best way to handle it. Possibly finding the source of the 40 ohm resistance in the +5V supply can give further hints on what actually happened.

It seems you were able to clean the area that was near the smoke quite well, which means that the PCB is not burnt, but some smoke was just deposited on top of the PCB. We don't know the origin of the smoke yet. It obviously was not C710, and it also was not CR701. Looking at the PCB in detail, I don't think the Adaptec chip is related to the no-spin issue: That ASIC is most likely just handling the IDE interface. The core processor that runs the firmware seems to be U401, and the Symbios chip U501 is likely "actual hard drive stuff". The M56742FP is the motor driver chip, which very likely has an "enable" pin. Too bad I can't find the datasheet for that chip, so you could verify whether it gets enabled at all. In theroy, you could try to reverse engineer the pinout of the M56742FP, but I don't consider it's worth the effort.

I found a cheap working PCB of a Maxtor 71336AP. The layout is identical, they both have 301066101. My idea is to swap the firmware chip and give it a go. If I'm lucky that didn't fry. If I'm not ....I'll go with plan B and buy a working 71080 I found, which is a little more expensive, and sell the 71336AP.
It's real sad that a backup PSU became my nemesis.

fix_metal wrote on 2026-05-27, 19:26:

I found a cheap working PCB of a Maxtor 71336AP. The layout is identical, they both have 301066101. My idea is to swap the firmware chip and give it a go.

I'd try to just swap the whole Maxtor PCB on that drive. If you are lucky, there is nothing on the HP PCB that is required for the drive to work. Possibly there is a defect list in an EEPROM, but if you do a full read-write test with the Maxtor PCB and the HP drive, the actual HP defects would get added to the Maxtor PCB.

I actually swapped some 13GB hard drive PCB 20 years ago, and was able to get all data off the drive. So you can't even be sure that using a PCB from a different drive causes any kind of trouble.

mkarcher wrote on 2026-05-27, 19:42:

fix_metal wrote on 2026-05-27, 19:26:

I found a cheap working PCB of a Maxtor 71336AP. The layout is identical, they both have 301066101. My idea is to swap the firmware chip and give it a go.

I'd try to just swap the whole Maxtor PCB on that drive. If you are lucky, there is nothing on the HP PCB that is required for the drive to work. Possibly there is a defect list in an EEPROM, but if you do a full read-write test with the Maxtor PCB and the HP drive, the actual HP defects would get added to the Maxtor PCB.

I actually swapped some 13GB hard drive PCB 20 years ago, and was able to get all data off the drive. So you can't even be sure that using a PCB from a different drive causes any kind of trouble.

Well, the two disks model have different sizes/CHS, so I assume that information is hard codes in the firmware. But the PCB is here, so stay tuned

It worked. I swapped the PCBs, and the firmwares. Wow!
/Me happi boi

Now, I'm unsure whether to service the original PCB or toss it.

fix_metal wrote on 2026-05-28, 14:25:

Well, the two disks model have different sizes/CHS, so I assume that information is hard codes in the firmware. But the PCB is here, so stay tuned

Oops, I missed the fact that you ordered a PCB for a different drive from the same series. Indeed, trying only with swapped firmware makes more sense in that case. Nevertheless, I also heard about hard drives loading the size from a special part of the disk, resulting a non-working firmware if the mechanical part is broken or the PCB operated without the mechanical drive connected.

fix_metal wrote on 2026-05-28, 17:36:

Now, I'm unsure whether to service the original PCB or toss it.

I guess that just depends on how you want to spend your time. If you like to troubleshoot the original PCB, go for it. If you feel more like doing something else, toss it. I don't think it makes sense to sell it as a damaged PCB.

fix_metal wrote on 2026-05-28, 17:36:

It worked. I swapped the PCBs, and the firmwares.

I'm happy to hear that you were able to fix the drive that way. Have fun!

fix_metal wrote on 2026-05-28, 17:36:

It worked. I swapped the PCBs, and the firmwares. ...

Congrats! - good to hear!

Now, I'm unsure whether to service the original PCB or toss it.

It could serve as an "educational experience"... but as it wouldn't get you anything useful (like another whole working drive) - I personally would not bother, and just take the "win".

But... having a "big basement" - My tendency with things like this is to "hang on to it" - you probably will never need it... it will just sit in a box "forever"

BUT - you never know if your might have another failure and could use it either to scavenge parts from -or- to be able to more aggressively track connections. You can always decide to toss it at any time.