https://www.youtube.com/watch?v=ssL1DA_K0sI

classic...

Alright I actually think thats the adapter itself now, thanks. Besides it does find the SSD after I reseat the SATA when it trying to detect it, so... Also nope, I don't really use the ISA slot.

Btw, is it OK to use larger caps than the originals? I'm thinking about recapping an old AT power supply, which has a couple of 16V 1000µF caps, but I don't have that size at hand. However, I've got 16V 2200µF caps, so would they be fine?

Also, do you think the primary caps are going to be fine even after all these year, or should I replace them as well?

Mike_ wrote on 2026-09-13, 12:25:

Btw, is it OK to use larger caps than the originals? I'm thinking about recapping an old AT power supply, which has a couple of […]
Show full quote

Btw, is it OK to use larger caps than the originals? I'm thinking about recapping an old AT power supply, which has a couple of 16V 1000µF caps, but I don't have that size at hand. However, I've got 16V 2200µF caps, so would they be fine?

The attachment caps1.jpg is no longer available

Also, do you think the primary caps are going to be fine even after all these year, or should I replace them as well?

The attachment caps2.jpg is no longer available

There’s an application note on PSUs with a group stabilisation choke – from Fairchild Semiconductor, if I’m not mistaken.
It includes calculations for capacitance and ESR between the coupled power rails.
To be honest, I didn’t really understand any of it. 😀

The only thing I grasped was that AT/ATX power supplies are the collective brainchild of practitioners who didn’t bother with theory, but rigorously optimised the cost to the level of ‘it won’t burn out the first time it’s switched on’.
Most likely, 2,200 µF in the +12 V bus is a perfectly normal capacitance.
As for high-voltage capacitors – as far as I’m concerned, the stock ones are a bit on the small side, both in terms of capacitance and size.

shevalier wrote on 2026-09-14, 10:50:

There’s an application note on PSUs with a group stabilisation choke – from Fairchild Semiconductor, if I’m not mistaken. It inc […]
Show full quote

There’s an application note on PSUs with a group stabilisation choke – from Fairchild Semiconductor, if I’m not mistaken.
It includes calculations for capacitance and ESR between the coupled power rails.
To be honest, I didn’t really understand any of it. 😀

The only thing I grasped was that AT/ATX power supplies are the collective brainchild of practitioners who didn’t bother with theory, but rigorously optimised the cost to the level of ‘it won’t burn out the first time it’s switched on’.
Most likely, 2,200 µF in the +12 V bus is a perfectly normal capacitance.
As for high-voltage capacitors – as far as I’m concerned, the stock ones are a bit on the small side, both in terms of capacitance and size.

Yeah, power electronics theory is pretty difficult. 😁

I meant that would it be okay if I replace both 1000µF caps with 2200µF ones, but I guess just using one to replace both of them should work as well. Maybe I could put a smaller cap, like 470µF, in the other footprint so that ESR and ESL go a bit lower but capacitance won't be more than double the original.

Now that you mention it, 2x 330µF 200V is a bit low for capacitance on high-voltage side. That's effectively 165µF and if I calculated correctly, it would result in ripple voltage of more than 50V at full load. But if I decide to replace them, I need to order parts so I might just as well order 1000µF 16V caps for the 12V line. Clearly I didn't order enough of them last time, as I ended up using all of them for recapping other stuff, such as a motherboard for a friend. 😁

Also, there's only one 2200µF 10V cap for what has to be the 5V line. That also sounds a bit low-end even for a 20A line.

Mike_ wrote on 2026-09-13, 12:25:

... Btw, is it OK to use larger caps than the originals? I'm thinking about recapping an old AT power supply, which has a couple of 16V 1000µF caps, but I don't have that size at hand. However, I've got 16V 2200µF caps, so would they be fine? ...

You should use larger caps. Look at the diameter of the parking spot:

^ They knowingly put undersized caps there, in strict cumpliment of the prime directive explained by Shevalier:

shevalier wrote on 2026-09-14, 10:50:

... AT/ATX power supplies are the collective brainchild of practitioners who didn’t bother with theory, but rigorously optimised the cost to the level of ‘it won’t burn out the first time it’s switched on ...

Then they rinsed & repeated:

Mike_ wrote on 2026-09-14, 15:01:

... 2x 330µF 200V is a bit low for capacitance on high-voltage side ...

Mike_ wrote on 2026-09-14, 15:01:

... there's only one 2200µF 10V cap for what has to be the 5V line. That also sounds a bit low-end even for a 20A line ...

TELVM wrote on 2026-09-15, 08:26:

You should use larger caps. Look at the diameter of the parking spot: […]
Show full quote

You should use larger caps. Look at the diameter of the parking spot:

^ They knowingly put undersized caps there, in strict cumpliment of the prime directive explained by Shevalier:

Actually two 1000µF caps for 12V line is probably fine as it's just 8A max. But 5V line and high-volt caps do seem to be underrated...

This +5VSB line is rated at a measly 3A max. And yet Seasonic felt the need to overkill & bombproof it with a humongous 4700uF KZE:

If there is a filter with inductance at the output of the power rail, then the second capacitor is not included in the calculations. It is recommended to set its capacitance to either 50% of the first capacitor’s value for a flyback topology, or to the same value as the first capacitor for a forward-converter topology (but not exceeding the first capacitor’s capacitance).
This is a clear engineering recommendation that is consistent with my experience.

What follows also matches what I’ve observed, especially in “latest-generation” ATX power supplies with group regulation.
The first capacitors should be proportional to the square of the voltage ratio, i.e., (12/5)² ~6 times.
That is, a 3,300 μF capacitor is used on the +5 V rail, and a 470 μF capacitor on the +12 V rail.
The ESR should also be proportional—20 mOhm on the +5 V rail and 90 mOhm on the +12 V rail.
This prevents current overflow through the group choke and ripple redistribution.
I’ve noticed that toward the end of the era of such power supplies, regardless of power rating, a 3300 μF capacitor was used on the +5V rail and a 2200 μF capacitor on the +12V rail.
Apparently, there’s a technique for redistributing ripple by weakening the magnetic coupling between channels—either through intentionally incorrect winding of the group choke or by installing a small inductor before the first capacitor. I haven’t seen anything like that either.
That’s why we need someone who really understands this stuff. 😀

shevalier wrote on 2026-09-16, 04:33:

If there is a filter with inductance at the output of the power rail, then the second capacitor is not included in the calculations. It is recommended to set its capacitance to either 50% of the first capacitor’s value for a flyback topology, or to the same value as the first capacitor for a forward-converter topology (but not exceeding the first capacitor’s capacitance).
This is a clear engineering recommendation that is consistent with my experience.

Good point, you need to check whether the caps are actually in parallel or if there's a choke between them.

shevalier wrote on 2026-09-16, 04:33:

What follows also matches what I’ve observed, especially in “latest-generation” ATX power supplies with group regulation. The fi […]
Show full quote

What follows also matches what I’ve observed, especially in “latest-generation” ATX power supplies with group regulation.
The first capacitors should be proportional to the square of the voltage ratio, i.e., (12/5)² ~6 times.
That is, a 3,300 μF capacitor is used on the +5 V rail, and a 470 μF capacitor on the +12 V rail.
The ESR should also be proportional—20 mOhm on the +5 V rail and 90 mOhm on the +12 V rail.
This prevents current overflow through the group choke and ripple redistribution.
I’ve noticed that toward the end of the era of such power supplies, regardless of power rating, a 3300 μF capacitor was used on the +5V rail and a 2200 μF capacitor on the +12V rail.
Apparently, there’s a technique for redistributing ripple by weakening the magnetic coupling between channels—either through intentionally incorrect winding of the group choke or by installing a small inductor before the first capacitor. I haven’t seen anything like that either.
That’s why we need someone who really understands this stuff. 😀

It looks like the 470µF caps for the PSU are for -5V and -12V rails. That sounds like really small for +12V rail.

I checked another AT power supply, and it has three 2200µF caps, presumably two 10V caps are for 5V rail and and one 16V for 12V rail.

However, its high voltage caps aren't doing too well...

An example of a relatively late (~2014) group-regulated PSU (Be Quiet Purepower L8 300W):

^ Made by FSP, platform EPN. Three 3300uF (ESR= 28) on two +12V rails. Two 2200uF on +3.3V & another two 2200uF on +5V. One 1000uF on -12V. Two 1500uF on +5VSB. (All Teapos and Taicons - not jap caps, but reasonably decent for a budget PSU).

Though topology is two-transistor forward and passive Schottky diodes, this PSU is quite efficient at low loads (under 100W total) typical of single CPU retrocomps up to PIII era.

And one of the white pots to the left allows for some fiddling with the +12V/+5V voltage ratio in "+5V heavy / +12V light" scenarios. 😀

.

Mike_ wrote on 2026-09-16, 07:48:

... I checked another AT power supply, and it has three 2200µF caps, presumably two 10V caps are for 5V rail and and one 16V for 12V rail...

In old-style PSUs with just one cap on the +12V output filtering, sometimes you find they placed a fat 12.5mm 3300uF one, like this Crapxon:

Listen, I don't know.
My brain just short-circuited at that point, because the conclusions drawn from those calculations completely contradicted my experience tinkering with the PSU. 🙁

Shevalier, if you don't mind me asking ... what do you think about in-cable capacitors?

.

Buildzoid - Rambling about in-cable capacitors on ATX PSUs

TELVM wrote on 2026-07-08, 23:59:

Interesting video about fixing a classic cheap (TL494 controller) PSU: […]
Show full quote

Interesting video about fixing a classic cheap (TL494 controller) PSU:

Let's fix this DEAD ATX power supply!

Healthy 35A rated +5V line:

However most of the caps are FUBAR, to the point that the controller can't even get enough voltage to start up.

^ ".... not unexpected with such prestigious brands been used ... " 😀 😀 😀 😀 😀

.

After recapping he concludes: "It seems to be working fine, 12 volts is fine, 5 volts is fine ..."

But from 23:53 onwards notice how, as he increases the load amp after amp on the +5V line, +5V voltage progressively drops, while +12V climbs up simultaneously, until at 7A (35W) load on +5V they are at 4.64V (out of ATX spec) and 12.49V respectively:

^ That's the group-regulated topology being defeated by a "+5V heavy / +12V light" crossloading.

Even on a supposedly "~period correct, +5V heavy", though of not so great quality, PSU.

Little late reply for this post, but the less than stellar result for that particular PSU doesn't surprise me at all. Q-tec was absolutely bottom of the barrel chinese junk you could have and unfortunately people often mixed it with Q-technology which made silent and high quality PSUs.

I had one of their PSUs for a while, a "550W" unit which probably barely could deliver 400W on a good day. I didn't know jack about these PSUs those days, but even then the insides looked very cheap and suspect to failure compared to some other units I've seen. For a 550W unit it was also very lightweight, heatsinks were just made of couple of millimeter thin aluminium plate with some drilled holes in them.

At some point I sold my older Socket A stuff to one of my co-workers and he wante to buy this PSU from me, I had already replaced it with a better one. He had it couple of months when the damn thing let out the magic smoke on him. It ended well as nothing catched fire and it didn't take other hardware with it, but I was sort of expecting something like this to happen and said the same thing to him when he wanted to buy it. That PSU was only 1-2 year old by the time it decided to die.

TELVM wrote on 2026-09-16, 09:36:

Shevalier, if you don't mind me asking ... what do you think about in-cable capacitors? […]
Show full quote

Shevalier, if you don't mind me asking ... what do you think about in-cable capacitors?

.

Buildzoid - Rambling about in-cable capacitors on ATX PSUs

More often than not, this is pointless*, alas.

It is up to the user to worry about power supply. This applies to both USB and PCI(x/-e) devices.
Modern motherboards do not contain any capacitors for peripherals at all.

In the past, capacitors were placed next to each slot.
But nobody wants to pay ‘for other people’s problems’.
That’s why the problems faced by graphics card manufacturers are one thing, and the production costs of motherboards are quite another. 😀

* Except for some Radeon X8xx graphics cards.
ATI had a moment of madness, which is why there is only a 10 µF MLCC ceramic capacitor at the input of their VRM.

But that was a one-off incident; it seems no one else has been up to that sort of nonsense since.

We’ve come to this…

There are exactly two 140mm fans on the market, 25mm thick and with high static pressure, that can be used in a PSU.
These are the SILENT WINGS PRO 4 140mm PWM from Be-Quiet and another ‘industrial’ Noctua.
But they both have a starting voltage of 5 volts.
And they cost 1.5 to 2 times more than their equivalents.

So here I am, having to deal with this nonsense.
The CWT PSH (without any numbers) was sold under the Corsair, Thermaltake and Chiftec CFT brands.
Seems to work

PS Perhaps Google will index this page and stop showing me its ‘smart search’ suggestions, which have been driving me mad for two days running. 😀

shevalier wrote on 2026-09-16, 12:35:

We’ve come to this… […]
Show full quote

We’ve come to this…

There are exactly two 140mm fans on the market, 25mm thick and with high static pressure, that can be used in a PSU.
These are the SILENT WINGS PRO 4 140mm PWM from Be-Quiet and another ‘industrial’ Noctua.
But they both have a starting voltage of 5 volts.
And they cost 1.5 to 2 times more than their equivalents.

So here I am, having to deal with this nonsense.
The CWT PSH (without any numbers) was sold under the Corsair, Thermaltake and Chiftec CFT brands.
Seems to work

PS Perhaps Google will index this page and stop showing me its ‘smart search’ suggestions, which have been driving me mad for two days running. 😀

I mean no disrespect: What is the point?

ldeveraux wrote on 2026-09-16, 15:29:

shevalier wrote on 2026-09-16, 12:35:

We’ve come to this… […]
Show full quote

We’ve come to this…

There are exactly two 140mm fans on the market, 25mm thick and with high static pressure, that can be used in a PSU.
These are the SILENT WINGS PRO 4 140mm PWM from Be-Quiet and another ‘industrial’ Noctua.
But they both have a starting voltage of 5 volts.
And they cost 1.5 to 2 times more than their equivalents.

So here I am, having to deal with this nonsense.
The CWT PSH (without any numbers) was sold under the Corsair, Thermaltake and Chiftec CFT brands.
Seems to work

PS Perhaps Google will index this page and stop showing me its ‘smart search’ suggestions, which have been driving me mad for two days running. 😀

I mean no disrespect: What is the point?

The fact is that it’s already difficult to buy a standard power supply fan, and soon it will become impossible altogether. 🙁
Fans are now optimised for airflow (which is not the same as static pressure) and have almost entirely switched to PWM control.
However, PWM fans are reluctant to start at low voltages – they require a voltage of at least 5V. The Yate Loon D14BH-12, for which the fan control system was developed, starts at 3.5V.
The control system outputs constant 4 V for low loads, at which point the Be-Quiet simply stands still; it then starts to jerk and only spins up once the PSU is already noticeably warm.

So we end up having to reverse-engineer the circuit just to, for goodness’ sake, change a simple fan in PSU.

But Shevalier!

You are supposed to buy a new, 80$+ PSU, not a new 20$ fan!
(Dont you see the sticker that says 'No User Serviceable Parts' on the outside!? SO WHAT if you have an EE degree! CORPORATE PROFITS AND CORPORATE LIABILITY ARE AT STAKE!)

(Ducks)

wierd_w wrote on 2026-09-16, 18:25:

But Shevalier!
if you have an EE degree!

I’ve got it

P.S. That’s precisely the problem.
If I didn’t have the necessary skills, I’d buy a new power supply. But it’s such a shame to throw away a perfectly good one that just needs:
- and so on — a list of everything I’ve done to this power supply and everything I’ve bought.
Every time, I promised myself it would be the last time.