Question about repair old NEC switching PSU

[Harp]

Question about repair old NEC switching PSU

Hello!

I got old 34v psu (made in Japan) who was find in trash smashed apart. After I fix traces, and do test with unloaded output, I find that it has a problem and it is not secure to use.
The output is galvanic isolated, and katode voltage of regulator is strange behaveour. If the unit is power on/off with enough big delay its seems that everything is good, but if it power on too quickly, before he discharge himself, it pass big current from primary. The problem is rigid and repetatible.
I think that is some problem on output side so I traces circuit. On primary DC it have big STR16006 on heatsink, all component seems ok, only slightly discoloration below power resistor.
I wish to have understand what is doing there and fix if possible. It is only high voltage source I have. To be even clearer what I meen, I record a small video of 2 min below.

38MB

https://filebin.net/5syj61bgjjoc8gh8

direct link

https://filebin.net/5syj61bgjjoc8gh8/nec_sw_psu_34v.mp4

Please help if can!
Thanks!

[Altra]

I seem to remember the old switched mode power supplies needed a load or load resistor on the outputs. I think modern power supplies have one built in. There is nothing connected to your +5 volt output? Search google for "SMPS load resistor", Just a thought.

[Skippy]

I see date codes for 1992. Your electrolytic caps could be due for replacement, especially the three big ones. But check the small ones too, they are easiest to remove, test and replace. There may be a 'soft-start' feature on the regulator, and the electrolytic caps nearby affect this. If they have lost capacity, the soft-start may be too fast, overloading the rest of the supply.

[Harp]

The 5v rail is isolated and regulated via own vreg. No feedback. I now solder a 1k resistor at this rail, but in first video there was noting connected. The 34v rail has onboard 600R resistor.
Electrolitic has old but not bad, smaler electrolite I was replaced, but the big one I do not have. I test it with esr meter and it is pass. I doubt that this electrolite can do presented behaveour?

http://www.fileconvoy.com/dfl.php?id...5a5488e296cdd4

I put the 150R load on 34v rail, and first it work normaly, but if power on before discharge, goes to catastrofic failure, turn 60w fully on and not have neither 5v nor 34v on rails. So, what now?

[Skippy]

If you're unsure about the big electrolytics, and don't have replacements, this is my recommendation:
Add additional capacitors in parallel with the ones on the PCB. Get a 400V capacitor from another SMPSU, even 100 to 200 microfarad is enough, and attach it to the underside of the board. Likewise, add a 40V cap, 470 uF or more across the DC output cap.
If it starts to look like it's working better, then you can investigate more, find out if it's lack of capacitance, high esr, etc.

[Harp]

I was desoldering all three big electrolitic caps, and test it with chinese transistor tester. The smaler two caps in the middle of picture seems to be in relatively good readings, but the biggest one 10000/40v got esr of 0.16R. All three caps are japanise Nitsuka brand. I found a Rubycon 6800/50v, but he has similar esr reading of 0.15R. So I suspect that this tester can not do some measurements on esr with resolution about 0.01R or less, so my guess is that all caps are preety ok and do not doing a problem. Or all caps that I have is similary bad. I will put it back and add Rubycon parallel with original.

In the middletime, can anyone confirm that transistor tester can realy measure 0.01R of esr?
EDIT:
I shorted clips (with and without wire), and got same reading of 0.16R, so I must substract it from measurement. In that way caps are fully ok.

[Harp]

After paralleling Rubycon 6800uF, the bulb first briefly flash and then turn fully on. Voltage goes from 32, 33v then slooowly up. I desolder additional Rubycon and psu work like usual, with same problems.

Then I do schematics of the primary side, and I think to know where the problem might to be. All 4 little electrolitic cap is 100uF, and there are missing on first place in trash. I took replacement guessing that this should be enough, but it seems that are too big, or wrong value. Searching the web hopeing to find some example of str16006 psu circuit, but I cant find.
Assume that this tought is correct, what cap will be suspect and with what value replaced?

I put complete schematics in attach.

[Skippy]

Out of the four 100uF caps, I think the one connected between G and K of the P2M8X thingy is the most likely to be critical to get correct operation, so if you're going to tinker with it, try 47uF then 22uF, then try larger 220uF and see if any changes are detectable.

[Harp]

To eventually speed up I go to extreme point, to see if can gather some information who had difference vs previous operation. I desoldered all 100uF caps and replaced it with 4u7. The operation is stay completly the same: its normaly turn on, then I turn it of, and if turn it on when Vout is 3v or higher it fail, and if turn it on when Vout is 2v or lower it pass.

Thus, not even any change up. So, I replaced caps between G-K to 680uF, and tested again.
Absolutely nothing change. The same behaveour as the first video, as the mentioned above.
Only information for shure about these caps I got from silk screen, where only G-K caps is 5mm diametar, and all other is slightly smaller about 4.5mm.

What cause the latch, and what is going between 2-3v ? Only voltage in that range is referent voltage of c1093 with 2.2v. In paralleling 6800uF to output, I am not expected to fail, why is that happening?

[Skippy]

One of the 100uF caps is clearly the smoothing cap in a half-wave-rectified supply from the transformer. As such, 100uF seems a fair value, BUT, I suggest using a low ESR part intended for high-freq SMPSU applications.

[Harp]

Correction:
A-K (not G-K) caps is 5mm diametar, and all other is slightly smaller about 4.5mm.

Now situation on shematic
C10 - on pin #1 str16006 chip is 4u7
C11 - on pin #4 chip is back 100uF (max to fit silk screen)
C14 - G-K is 4u7
C19 - A-K is 47uF (I put it bigger because silk screen)

The setup is working with same problem. How to troubleshoot the cause of problem? I dont know, but guess it something connected with feedback throw optocoupler, the secondary side? The psu is working just on that demand.

The A-K (sensing?) voltage is 1.14v if it is in ok state, and 3.3v if it latch to fail.

[Skippy]

If you're using the PCB markings of capacitor diameters as any kind of guide, you need to keep in mind this is for 1992 capacitors, and they would likely be smaller capacitance value than modern-spec parts.

[Harp]

I have desoldered a couple component but nothing there.

The transistor c2719 is ok, regulator c1093 I replaced with TL431 but it stay the same, some resistor with suspectialy low reading, but all is ok. Diode is not shorted... nothing burned.
I was changed all small 4 electrolites with 4u7 and work the same way as with 100uF. Later I was traing with 1uF but not worked, the same moment power on the lamp. I suppose that psu is (mean to be) designed to if anything fail, that go off, and not to blow herself. If he wish to blow yourself then must be a very clear and logic reason why is that.
I dont know, and if any of you cant tell me a step forward to troubleshoot, I'll put it a side.