shevalier wrote on 2026-10-01, 06:39:

It seems that noise from the reference has started to affect the dead-time control. Not because the capacitance is greater, but […]
Show full quote

It seems that noise from the reference has started to affect the dead-time control.
Not because the capacitance is greater, but because that’s how it was designed from the outset; it’s just that the old capacitor has simply dried out.
I’d clean up the power of the TL494 itself and see what happens, because, manufacturers cut corners on every aspect of the TL494’s powering itself.

Ideally, a Cvref bypass capacitor of ~0.1 µF should be fitted (this is missing in 99% of power supplies).
Either a film capacitor or an X5/7 rated for 25+ volts will do.
And the Rvcc+Cvcc circuit for the TL494 itself should be located in the immediate vicinity of its pins.
In 80% of cases, this is missing; in 10%, only Rvcc is present
R – 4.7–22 Ω (SMD, 0805/1206)
C – 4.7–1 µF X5/7 rated for 50+ volts.
Radd in series with C_dt is often present. It should be approximately up to 5% of R_dt. It will slightly reduce the high-frequency noise passing through C_dt. SMD 0805/0603.

Cvcc and Cvref – either SMD, if there is a ground trace nearby, or for through-hole mounting. ‘A roof over the house’, capacitor simply on top of the chip. This was a common method for decoupling memory chips during the ZX Spectrum era 😀.

It turned out that there actually is a Rvcc+Cvcc circuit for the controller, it's that 4.7µF cap next to the 0.68µF cap in the earlier pictures. And there's also a 22Ω resistor as Rvcc. So I just added a 680nF cap between Vcc and ground to filter out high-frequency stuff that the electrolytic cap can't handle, and a 100nF cap between Vref and gnd. It looks like adding a resistor between Vref and the cap would have been a bit messy, so I skipped it for now. Then I powered it on again, and the whine was still there, but the odd thing happening on 12V rail was reduced.

I also decided to probe 12V line and Vcc at the same time, and it appears that there's something happening at the same time.

So far everything was good, but then things got weird. I heard some crackling and saw smoke rising from the PSU, after which I removed the power cord from wall. It looks like a big film cap next to the high-voltage capacitors was the source of the smoke. Also, it looks like one of the primary switching BJTs is now shorted between base and emitter.

I wonder what the heck just happened? The recap or the modifications done shouldn't have caused the PSU to blow up...

EDIT: Actually Vcc is ramping like that at 100Hz frequency, not 1kHz. So that comes from rectified mains frequency. But the question still remains that why was that visible from output? It doesn't explain why it blew up, either.

Mike_ wrote on 2026-10-01, 16:02:

So far everything was good, but then things got weird. I heard some crackling and saw smoke rising from the PSU, after which I r […]
Show full quote

So far everything was good, but then things got weird. I heard some crackling and saw smoke rising from the PSU, after which I removed the power cord from wall. It looks like a big film cap next to the high-voltage capacitors was the source of the smoke. Also, it looks like one of the primary switching BJTs is now shorted between base and emitter.

The attachment blown_cap.jpg is no longer available

I wonder what the heck just happened? The recap or the modifications done shouldn't have caused the PSU to blow up...

EDIT: Actually Vcc is ramping like that at 100Hz frequency, not 1kHz. So that comes from rectified mains frequency. But the question still remains that why was that visible from output? It doesn't explain why it blew up, either.

It seems that this capacitor has simply failed due to age.
It must be a polypropylene capacitor (something like a "CBB22 1 µF × 400–630 VDC"); PET ones don’t last that long.
But usually, the metallisation on the plates degrades slowly and gradually, leading to a reduction in capacitance without any visible signs.
One possible explanation is that, following the replacement of the rectifier capacitors in the secondary circuits, the current pulses became narrower, higher in amplitude and sharper.
This was the ‘last straw that broke the camel’s back’.
Only the this camel capacitor was already very unwell and exhausted.

As for the Vcc, that’s more or less how it should look. This is because it is generated up to the PSU’s regulation point and mirrors the rectified mains voltage, taking into account the effect of the load. The main thing is that there are no high-frequency components in the TL494’s power supply.
Next, the internal Vref regulator will filter out all the low-frequency noise and produce the voltage against which the PSU’s output voltage will be compared.

shevalier wrote on 2026-10-02, 05:31:

It seems that this capacitor has simply failed due to age. It must be a polypropylene capacitor (something like a "CBB22 1 µF × […]
Show full quote

It seems that this capacitor has simply failed due to age.
It must be a polypropylene capacitor (something like a "CBB22 1 µF × 400–630 VDC"); PET ones don’t last that long.
But usually, the metallisation on the plates degrades slowly and gradually, leading to a reduction in capacitance without any visible signs.
One possible explanation is that, following the replacement of the rectifier capacitors in the secondary circuits, the current pulses became narrower, higher in amplitude and sharper.
This was the ‘last straw that broke the camel’s back’.
Only the this camel capacitor was already very unwell and exhausted.

As for the Vcc, that’s more or less how it should look. This is because it is generated up to the PSU’s regulation point and mirrors the rectified mains voltage, taking into account the effect of the load. The main thing is that there are no high-frequency components in the TL494’s power supply.
Next, the internal Vref regulator will filter out all the low-frequency noise and produce the voltage against which the PSU’s output voltage will be compared.

Btw, I just realized where the whine was coming from. If you look at the oscilloscope pictures, you can calculate that the switching frequency is about 3.5kHz, and clearly it shouldn't be that low. I feel dumb...

It might have also stressed the components more, even though the load was very light. Also, I was wrong about one of the BJTs having got shorted. It only looked like that at first due to it being connected to those transformers, but after removing it from board they seemed to be fine.

Mike_ wrote on 2026-10-02, 08:08:

shevalier wrote on 2026-10-02, 05:31:

It seems that this capacitor has simply failed due to age. It must be a polypropylene capacitor (something like a "CBB22 1 µF × […]
Show full quote

It seems that this capacitor has simply failed due to age.
It must be a polypropylene capacitor (something like a "CBB22 1 µF × 400–630 VDC"); PET ones don’t last that long.
But usually, the metallisation on the plates degrades slowly and gradually, leading to a reduction in capacitance without any visible signs.
One possible explanation is that, following the replacement of the rectifier capacitors in the secondary circuits, the current pulses became narrower, higher in amplitude and sharper.
This was the ‘last straw that broke the camel’s back’.
Only the this camel capacitor was already very unwell and exhausted.

As for the Vcc, that’s more or less how it should look. This is because it is generated up to the PSU’s regulation point and mirrors the rectified mains voltage, taking into account the effect of the load. The main thing is that there are no high-frequency components in the TL494’s power supply.
Next, the internal Vref regulator will filter out all the low-frequency noise and produce the voltage against which the PSU’s output voltage will be compared.

Btw, I just realized where the whine was coming from. If you look at the oscilloscope pictures, you can calculate that the switching frequency is about 3.5kHz, and clearly it shouldn't be that low. I feel dumb...

It might have also stressed the components more, even though the load was very light. Also, I was wrong about one of the BJTs having got shorted. It only looked like that at first due to it being connected to those transformers, but after removing it from board they seemed to be fine.

Look, it wouldn't even cross my mind to measure the PSU switching frequency.
It's just a dumb RC circuit — there's really nothing there to go wrong. 🙁

shevalier wrote on 2026-10-02, 10:04:

Look, it wouldn't even cross my mind to measure the PSU switching frequency.
It's just a dumb RC circuit — there's really nothing there to go wrong. 🙁

Yeah, it doesn't make any sense. The only thing I can think of is if the controller has got damaged somehow, for example by ESD.

Mike_ wrote on 2026-10-02, 11:03:

shevalier wrote on 2026-10-02, 10:04:

Look, it wouldn't even cross my mind to measure the PSU switching frequency.
It's just a dumb RC circuit — there's really nothing there to go wrong. 🙁

Yeah, it doesn't make any sense. The only thing I can think of is if the controller has got damaged somehow, for example by ESD.

Well, it’s worth checking the capacitor and resistor.

In later models of the ATX, where mounting glue was used to secure the components, this could, over time, cause the contact cups on the output resistor to come loose.
Or perhaps they simply got the wrong resistor at the factory.

It should be something like 1 nF + 15–20 kΩ.

shevalier wrote on 2026-10-02, 12:31:

Well, it’s worth checking the capacitor and resistor. […]
Show full quote

Well, it’s worth checking the capacitor and resistor.

In later models of the ATX, where mounting glue was used to secure the components, this could, over time, cause the contact cups on the output resistor to come loose.
Or perhaps they simply got the wrong resistor at the factory.

It should be something like 1 nF + 15–20 kΩ.

They are fine, cap is 1nF based on text, and resistor is 27kΩ and measures close to that.

I have a few film caps that could be suitable as a replacement, so I gave this one a try - the original wasn't across both high-voltage caps, it was connected to middle of them so 250V should be enough. And guess what, after soldering in the cap and the transistors, the whine was still there but output voltages measure fine. So it's not completely toast yet.

Next I measured KA7500B's output 1 (pin 8 ) to check whether it really is running at 3.5kHz. And apparently it isn't, it looks like it's about 22kHz instead... so it has to be something else. Maybe there's some oscillation in feedback loop, or something? It also starts to look like the previous film cap blowing up might be unrelated to this...

Mike_ wrote on 2026-10-02, 14:58:

They are fine, cap is 1nF based on text, and resistor is 27kΩ and measures close to that. […]
Show full quote

shevalier wrote on 2026-10-02, 12:31:

Well, it’s worth checking the capacitor and resistor. […]
Show full quote

Well, it’s worth checking the capacitor and resistor.

In later models of the ATX, where mounting glue was used to secure the components, this could, over time, cause the contact cups on the output resistor to come loose.
Or perhaps they simply got the wrong resistor at the factory.

It should be something like 1 nF + 15–20 kΩ.

They are fine, cap is 1nF based on text, and resistor is 27kΩ and measures close to that.

I have a few film caps that could be suitable as a replacement, so I gave this one a try - the original wasn't across both high-voltage caps, it was connected to middle of them so 250V should be enough. And guess what, after soldering in the cap and the transistors, the whine was still there but output voltages measure fine. So it's not completely toast yet.

The attachment replacement_cap.jpg is no longer available

Next I measured KA7500B's output 1 (pin 8 ) to check whether it really is running at 3.5kHz. And apparently it isn't, it looks like it's about 22kHz instead... so it has to be something else. Maybe there's some oscillation in feedback loop, or something? It also starts to look like the previous film cap blowing up might be unrelated to this...

The attachment output1.jpg is no longer available

And what load do you place on the power supply line when taking measurements?
Could it be that the power supply is simply underloaded?
The load acts as a damping element for oscillations due to its low resistance.

Regarding compensation.
A typical circuit is a voltage divider fron Vref formed by two resistors connected to pin 2.
Usually, two identical 4.7 kOhm resistors.
And between pins 2 and 3 – a series RC circuit of a 10 nF capacitor and a resistor 10 times the value of the divider resistors.
Something like 33–47 kOhm, more often 47 kOhm, if the divider resistors are 4.7 kOhm.
Here, too, a film capacitor is highly recommended.
Moreover, these values are copied almost verbatim across all available circuits.

But it seems more likely that the +12V simply isn’t feeding back into the regulation circuit.
In AT power supplies, this was often done to ensure that the +5V was regulated very precisely.
In all sorts of 286/386 setups, the chipset and everything else were powered directly from the +5V.
And the +12V was only supplied to the ‘a certain hard drive motor’.

shevalier wrote on 2026-10-02, 15:07:

And what load do you place on the power supply line when taking measurements? Could it be that the power supply is simply underl […]
Show full quote

And what load do you place on the power supply line when taking measurements?
Could it be that the power supply is simply underloaded?
The load acts as a damping element for oscillations due to its low resistance.

Regarding compensation.
A typical circuit is a voltage divider fron Vref formed by two resistors connected to pin 2.
Usually, two identical 4.7 kOhm resistors.
And between pins 2 and 3 – a series RC circuit of a 10 nF capacitor and a resistor 10 times the value of the divider resistors.
Something like 33–47 kOhm, more often 47 kOhm, if the divider resistors are 4.7 kOhm.
Here, too, a film capacitor is highly recommended.
Moreover, these values are copied almost verbatim across all available circuits.

But it seems more likely that the +12V simply isn’t feeding back into the regulation circuit.
In AT power supplies, this was often done to ensure that the +5V was regulated very precisely.
In all sorts of 286/386 setups, the chipset and everything else were powered directly from the +5V.
And the +12V was only supplied to the ‘a certain hard drive motor’.

I'm using a 10Ω 25W resistor for 12V line and 3Ω 50W resistor for 5V line. The latter is perhaps somewhat light load, so I also tried a 1Ω resistor for 5V line. The whine is still there, but at slightly different pitch. Also, this time heatsink for the BJTs got quite warm really fast (I touched it only after removing the wall plug!). I also tried removing Cdt, but that had no effect. 🙁

Now that I think about it, I actually measured this PSU last winter, and it was fine then. I even used same load for testing. So something has broke down since then... But I think I'm fed up with playing with this thing for now, I'll have another look at it if I actually ever need it.

However, using 1Ω load for 5V line made the other PSU behave. There's no longer 1kHz sine wave in 12V line, and ripple for 5V line is also reduced. So at least I got two of the three recapped PSUs to work properly.

Anyways, thanks for help with debugging these! This got a lot more complicated than I expected a recap job to be. 😀

For me the answer is yes,

I am (so far) happily using the Akyga Basic 700W (AK-B1-700) without it blowing up. Powering my Athlon XP 2800+ Barton on EPoX 8K3A.

These are inexpensive PSUs for the 38A they promise to deliver on +5V rail.

I ordered three and got free shipping for extra cheapness. CheapyD approves and much love to Poland.
https://www.csmegastore.de/i/1876828/akyga-ak … siv-300-w-384-w

Kasreyn wrote on 2026-10-27, 10:55:

For me the answer is yes, […]
Show full quote

For me the answer is yes,

I am (so far) happily using the Akyga Basic 700W (AK-B1-700) without it blowing up. Powering my Athlon XP 2800+ Barton on EPoX 8K3A.

These are inexpensive PSUs for the 38A they promise to deliver on +5V rail.

I ordered three and got free shipping for extra cheapness. CheapyD approves and much love to Poland.
https://www.csmegastore.de/i/1876828/akyga-ak … siv-300-w-384-w

What a bizarre PSU. Heavy 5V rail, but when you look at the connectors, it seems like it is designed for pretty modern systems which tend to be 12V heavy instead.

How does the design and component quality look like?

Kasreyn wrote on 2026-10-27, 10:55:

For me the answer is yes, […]
Show full quote

For me the answer is yes,

I am (so far) happily using the Akyga Basic 700W (AK-B1-700) without it blowing up. Powering my Athlon XP 2800+ Barton on EPoX 8K3A.

These are inexpensive PSUs for the 38A they promise to deliver on +5V rail.

I ordered three and got free shipping for extra cheapness. CheapyD approves and much love to Poland.
https://www.csmegastore.de/i/1876828/akyga-ak … siv-300-w-384-w

Does it keep the voltages under load? Because, if it does, I may be interested in picking up a few units (hoping they ship here were I live).

I'm also seeing other units from the same brand on that store, at lower wattages and lower costs. They still declare beefy 3.3 and 5v rails. If the voltages mantains under stress, they may be good.

Other thread that may be relevant here as well:

Project "Revenant": An Idea for an open-source modular PSU for any vintage PC (and anything else that uses same voltages

Mondodimotori wrote on 2026-10-27, 21:08:

Kasreyn wrote on 2026-10-27, 10:55:

For me the answer is yes, […]
Show full quote

For me the answer is yes,

I am (so far) happily using the Akyga Basic 700W (AK-B1-700) without it blowing up. Powering my Athlon XP 2800+ Barton on EPoX 8K3A.

These are inexpensive PSUs for the 38A they promise to deliver on +5V rail.

I ordered three and got free shipping for extra cheapness. CheapyD approves and much love to Poland.
https://www.csmegastore.de/i/1876828/akyga-ak … siv-300-w-384-w

Does it keep the voltages under load? Because, if it does, I may be interested in picking up a few units (hoping they ship here were I live).

I'm also seeing other units from the same brand on that store, at lower wattages and lower costs. They still declare beefy 3.3 and 5v rails. If the voltages mantains under stress, they may be good.

With a quick search, I found these posts:

https://www.elektroda.com/rtvforum/topic34363 … 7.html#17066567
https://www.reddit.com/r/pcmasterrace/comment … nt_buy_a_cheap/

All in all, if the brand hasn’t improved significantly very recently, it doesn’t look good and it more than often applies to these unknown cheap PSUs. This just seems another bottom of the barrel Chinese junk brand. Someone should at least first make some measurements and inspect the unit design and components before this could be recommended option. As it is, I’d say a hard pass and not worth wasting the money.

CharlieFoxtrot wrote on 2026-10-28, 05:39:

With a quick search, I found these posts: […]
Show full quote

With a quick search, I found these posts:

https://www.elektroda.com/rtvforum/topic34363 … 7.html#17066567
https://www.reddit.com/r/pcmasterrace/comment … nt_buy_a_cheap/

All in all, if the brand hasn’t improved significantly very recently, it doesn’t look good and it more than often applies to these unknown cheap PSUs. This just seems another bottom of the barrel Chinese junk brand. Someone should at least first make some measurements and inspect the unit design and components before this could be recommended option. As it is, I’d say a hard pass and not worth wasting the money.

So... No different from that chinesium unit that I bought a few years ago. Even if the PC would post and be stable, voltages in EVEREST would drop to 4.6 on the 5v when under load. I now use it only for test benching untill I can get a new reliable unit from a good brand. (Thanks to memory crysis, PSUs prices are going down hard, since nobody is building anything).

Basic Line, only lower wattage:
https://www.youtube.com/watch?v=nCG-9SJ1nI0
Another Akyga piece:
https://www.tweakpc.de/news/51164/netzteil-mi … zurueckgerufen/
No, thanks.

Ydee wrote on 2026-10-28, 10:13:

Basic Line, only lower wattage: https://www.youtube.com/watch?v=nCG-9SJ1nI0 Another Akyga piece: https://www.tweakpc.de/news/511 […]
Show full quote

Basic Line, only lower wattage:
https://www.youtube.com/watch?v=nCG-9SJ1nI0
Another Akyga piece:
https://www.tweakpc.de/news/51164/netzteil-mi … zurueckgerufen/
No, thanks.

Nice fireworks 🤣

From the translated article:
” Later it turned out that the power supply was equipped with a fake PFC, which was simply filled with metal scrap to simulate a higher weight”

This is a new one for me, a scrap metal PFC!

CharlieFoxtrot wrote on 2026-10-28, 10:26:

Nice fireworks lol […]
Show full quote

Nice fireworks 🤣

From the translated article:
” Later it turned out that the power supply was equipped with a fake PFC, which was simply filled with metal scrap to simulate a higher weight”

This is a new one for me, a scrap metal PFC!

Wouldn't be the first time I handle a crap unit that weighs a lot. I wouldn't be surprised to find ballast in it to make it heavier.

Mondodimotori wrote on 2026-10-28, 10:32:

CharlieFoxtrot wrote on 2026-10-28, 10:26:

Nice fireworks lol […]
Show full quote

Nice fireworks 🤣

From the translated article:
” Later it turned out that the power supply was equipped with a fake PFC, which was simply filled with metal scrap to simulate a higher weight”

This is a new one for me, a scrap metal PFC!

Wouldn't be the first time I handle a crap unit that weighs a lot. I wouldn't be surprised to find ballast in it to make it heavier.

Faking passive PFC choke is something else. It's not like it is an expensive component and even manufacturing that fake one isn't free. Everything on that PSU is probably on a level that it barely functions.

Efficiency of that unit is also atrociously bad, perhaps around 50% or so, even less.

CharlieFoxtrot wrote on 2026-10-28, 14:26:

Mondodimotori wrote on 2026-10-28, 10:32:

CharlieFoxtrot wrote on 2026-10-28, 10:26:

Nice fireworks lol […]
Show full quote

Nice fireworks 🤣

From the translated article:
” Later it turned out that the power supply was equipped with a fake PFC, which was simply filled with metal scrap to simulate a higher weight”

This is a new one for me, a scrap metal PFC!

Wouldn't be the first time I handle a crap unit that weighs a lot. I wouldn't be surprised to find ballast in it to make it heavier.

Faking passive PFC choke is something else. It's not like it is an expensive component and even manufacturing that fake one isn't free. Everything on that PSU is probably on a level that it barely functions.

Efficiency of that unit is also atrociously bad, perhaps around 50% or so, even less.

It's cheaper to get scrap into a metal can than to wind copper onto a proper core. That's exactly why they did that. They did the numbers 😁

And who'll ever notice that the power factor is crap for that one PSU in the household.