Enlight HPC-250-101 is aging

[momaka]

Re: Enlight HPC-250-101 is aging

but I suppose I already made up my mind when it came to the replacement value resistor prior to that second go-around with sanding.

($0.24 metal oxide Yageo OK? Haha)

When it comes to resistors, brand doesn't matter that much - even the cheap Chinese ones tend to be just fine.

In any case, I also suggest grabbing a 150 Ohm resistor, just in case R25 really was the low resistance that it showed initially. That way, if the PSU doesn't work with the 1.5 KOhm resistor, you have an alternative to try. For less than $0.25, it shouldn't break the bank to have an extra resistor. If you end up not using the 150 Ohm resistor, it can always come in handy for later PSU projects - particularly cheap PSUs that have very big load resistors on the output that cook the caps.

Awesome. This influenced my choices a bit - made things easier.

Attached my spreadsheet of previous caps and new caps in the BOM at digikey in case people need a reference in the future (or if you have any suggestions. I probably won't commit it until later in the day on Friday the 13th.)

Looks alright, but I do have some suggestions:

For the Teapo SEK 10uF/50V, instead of going with Nichicon VZ (which is a general purpose cap), you can go with Rubycon YXJ (Digikey part number 1189-2322-ND). For the small caps, matching ripple current and ESR hardly matters, so it's just better to go with low ESR caps with higher ripple current. That said, Rubycon YXJ is just entry level low ESR, so it's perfect. And if offers much longer endurance life over Nichicon VZ.

Likewise, I see a Teapo SEK 470uF/16V on your list getting replaced with a Rubycon PX. Again, instead of going with a general purpose 105C rated cap, you can easily go with the Nichicon PW 470uF/25V choice that you have as a replacement for Teapo SX.

Then there's the G-Luxon SM 1uF/50V... no need to go with Chemicon KMG. Nichicon PW (Digikey part number 493-1887-ND) should be better.

And finally, the Teapo SH 1000uF/16V cap - looks like that one was on the 12V rail on the PSU. I would definitely suggest to increase that capacitance to 1200 uF or even 1500 uF. A lot of older low-end PSUs tend to cheap out on the 12V rail caps and use just one 1000 uF cap, which is quite minimalistic. And again, there's little reason to go with Nichicon VZ here, as PSU outputs are usually better suited towards low ESR capacitors than general purpose. So Here is a list of alternative cap choices you can go with:
Rubycon YXJ 1000uF/16V, Digikey P/N: 1189-1545-1-ND
Panasonic FC 1200uF/16V, Digikey P/N: P11204-ND
Chemicon LXZ 1200uF/16V, Digikey P/N: 565-1934-ND
Chemicon KY 1200uF/16V, Digikey P/N: 565-1529-ND
Rubycon YXG 1200uF/16V, Digikey P/N: 1189-1136-ND
Nichicon PW 1500uF/16V, Digikey P/N: 493-1793-ND
Chemicon KY 1500uF/16V, Digikey P/N: 565-1530-ND

Also, no need to replace that Rubycon YXF that was next to the PWM controller, as it is probably okay. Doesn't seem like it got heat-abused or anything like that. But if you don't have an ESR or capacitance meter, then I suppose you should replace it jut for the peace of mind.

[ckrtech]

Re: Enlight HPC-250-101 is aging

Awesome. Thanks. I went with your suggestions. Upped the Teapo SH 1000uf/16V to a Nichicon PW 1500uf.

It also appears that for the G-Luxon 1uf -> Nichicon UPW replacement, 2.2uf is the lowest you can get. For whatever function those 1uf (Qty: 6) provide, I assume the difference between 1uf and 2.2uf is negligible.

[stj]

Re: Enlight HPC-250-101 is aging

the only decent 1uf i can get now is rubycon YXJ
2.2uf are being phased out of a few panasonic ranges too.

[ckrtech]

Re: Enlight HPC-250-101 is aging

Recapped. De-glued.

Used the little PSU tester for a smoke test. Nothing blew.

The +5VSB is 5.0 and PG is now 100 (was previously 80). I've had questionable results for using this thing for -12V testing (LL showing), but I won't depend on it.

So now I need to rig up something to load 5VSB. I do have several 12V fans sitting around (~0.15A each). I literally have three that have been sitting on the bench for awhile. If I use them, I just gotta figure out how I want to wire it all together and into the ATX connector.

If I had been thinking, I would have perhaps ordered some load resistors and maybe try to cannibalize something with a mobo-side ATX plug - something to make a clean connection.

How much of a load does a mobo actually put on 5VSB? I may have some "its OK if this thing dies" mobos around here...

btw - some of the caps are listed on the digikey invoice as having a COO: Malaysia. Interesting. Others are Japan.

Pictures included show the PSU tester smoke test results, a full shot of the whole thing, the 1.5Kohm resistor and glue cleaning around the PWM, some glue cleaning around a regulator (I removed two regulators with the small heatsinks bolted on in order to clean the glue), and finally a glamor shot of the 2200uf Nichicon's.

[ckrtech]

Re: Enlight HPC-250-101 is aging

With that said, if the PSU seems to work fine, definitely recap the 5VSB circuit output and check the voltage on the sec. aux. rail, especially with a 0.5-1.5 Amp load on the 5VSB rail. As long as the voltage is well below 30V, the PWM controller should be safe. As for R25, that could have started burning either due to over-voltage on the sec. aux. rail or if it wasn't selected properly (or if the PSU isn't working, the two small TO-92 driver transistors could have shorted or the two glass diodes).

I grabbed a spare motherboard with a processor and RAM, added an ATI Mach 64 PCI video card, and plugged in the PSU (to the board and the wall).

Not sure what sort of load the mobo would put on 5VSB, but I got about 13.5V from the sec. aux. rail before powering up the computer.

After powering up and entering the BIOS, it looks like I was getting around 33.5V from the same position. 1.35kV at the top right corner Trigger display is kinda weird.

Powering off the computer dropped sec. aux. rail back to 13.5V. Flipping off the surge protector switch - the PSU held that 13.5V output for a good 4 or 5 seconds and then dropped out. Fan output was extremely cool and quiet the whole time - but I was maybe up and running for only two minutes.

Anything out of the ordinary?

[momaka]

Re: Enlight HPC-250-101 is aging

How much of a load does a mobo actually put on 5VSB?

Depends on what the motherboard has for options in the CMOS. With Wake-On-LAN (WOL), maybe 0.5 Amps. Without it, I'd guesstimate 100-300 mA. With a motherboard that has S3 (Pentium 3 and newer or AMD Athlon 64 and newer), going into StandBy could pull around 1-1.5 Amp, depending on mobo design and number of memory modules.

btw - some of the caps are listed on the digikey invoice as having a COO: Malaysia. Interesting. Others are Japan.

Yup.
New Nichicon and Chemicon caps often have non-Japanese origins now, even though they are a Japanese brand. I think Rubycon is still the only one made in Japan.

Pictures included show the PSU tester smoke test results, a full shot of the whole thing, the 1.5Kohm resistor and glue cleaning around the PWM, some glue cleaning around a regulator (I removed two regulators with the small heatsinks bolted on in order to clean the glue), and finally a glamor shot of the 2200uf Nichicon's.

Nice work! Looking good.

I think you did miss a spot with the glue, though. See that snubber resistor on the primary side (next to the big main transformer)? It is just South of the small capacitors C5 and C6 in this shot:
https://www.badcaps.net/forum/attach...1&d=1508465837
...
-Make sure to remove that conductive glue there. If anything, it's more critical to remove the tan glue from the primary side than the secondary.

Other than that, it looks like a decent little 250-Watter PSU.

I grabbed a spare motherboard with a processor and RAM, added an ATI Mach 64 PCI video card, and plugged in the PSU (to the board and the wall).

Not sure what sort of load the mobo would put on 5VSB, but I got about 13.5V from the sec. aux. rail before powering up the computer.

After powering up and entering the BIOS, it looks like I was getting around 33.5V from the same position.

Wow 33.5V? That's a bit on the high side, especially considering a 35V Rubycon was filtering it (a bit too close to the cap's voltage rating). Still in spec, though. I wonder why it goes so high, though. Perhaps the motherboard is loading the 5VSB as the PC gets turned on (though typically the reverse happens, and the 5VSB becomes less loaded on most motherboard I've seen.)

In that case definitely try giving the PSU a measurable load on the 5VSB (like those three fans).

1.35kV at the top right corner Trigger display is kinda weird.

You might be picking up too much high-frequency noise. That's probably why the trace for the aux. sec. looks so blurry/"soft" too. If I remember correctly, you have to add a ceramic bypass capacitor between the probes to get rid of the excess noise. I don't remember the exact details, though. But look up SMPS measurement techniques with an o-scope to see what you need to do.

That also might actually lower the voltage you measured on the aux. sec. when the PC is on. The "33.5V" you measured may be lower. I suggest actually using a multimeter for that.

Fan output was extremely cool and quiet the whole time - but I was maybe up and running for only two minutes.

Yeah, 2 minutes is not much.
Typically you need at least 30 minutes in a warm/hot room.

Anything out of the ordinary?

Other than putting a bypass cap on your scope's probes to (possibly) get a more accurate and less noisy measurement, everything else looks fine.

[ckrtech]

Re: Enlight HPC-250-101 is aging

I think you did miss a spot with the glue, though. See that snubber resistor on the primary side (next to the big main transformer)? It is just South of the small capacitors C5 and C6 in this shot:
https://www.badcaps.net/forum/attach...1&d=1508465837
...
-Make sure to remove that conductive glue there. If anything, it's more critical to remove the tan glue from the primary side than the secondary.

Oh. Good eye!

I removed a few components so I could get that glue out - and it is out now - and saw that the larger resistor that had a "foot" in that glue has some corrosion on the longer leg. Seems rather weird. Any ideas?

But look up SMPS measurement techniques with an o-scope to see what you need to do.

That also might actually lower the voltage you measured on the aux. sec. when the PC is on. The "33.5V" you measured may be lower. I suggest actually using a multimeter for that.

Will do. This guy is pretty close to being screwed back into the P3 case. Thanks for your help!

[ckrtech]

Re: Enlight HPC-250-101 is aging

I did a bit of sanding on the resistor's leg to clean off the corrosion. The glue (or whatever it was) did seem to go up the leg of that resistor, so the corrosion matched the footprint.

With it and the other components I removed put back in circuit, I hooked it up to my test mobo and powered up with a multimeter clipped to the sec aux rail side of the new 1.5kohm resistor (easy to clip the top of that guy).

The multimeter showed the same results as the o-scope. Hit 33V after power up. I tried a different motherboard, and it had about the same result - maybe closer to 32V this time. Haven't used a ceramic cap, though! Just a multimeter rather than the o-scope.

I don't really have much extra hooked up to the motherboards as far as loading down the other rails - no hard drives, case fans, sound card, etc. I am sitting in the BIOS when I get those readings, and letting it POST makes no difference. If the board is up and running, I am getting 32V-33V.

I haven't rigged up a way to connect the fans to +5VSB yet. What are we hoping to see from that vs. the "real world" motherboard hook-ups?

[momaka]

Re: Enlight HPC-250-101 is aging

The multimeter showed the same results as the o-scope. Hit 33V after power up. I tried a different motherboard, and it had about the same result - maybe closer to 32V this time. Haven't used a ceramic cap, though! Just a multimeter rather than the o-scope.

I don't really have much extra hooked up to the motherboards as far as loading down the other rails - no hard drives, case fans, sound card, etc. I am sitting in the BIOS when I get those readings, and letting it POST makes no difference. If the board is up and running, I am getting 32V-33V.

Interesting that the multimeter shows the same thing. In that case, the scope is showing the correct voltages (though when scoping the outputs of the PSU, you still need the bypass cap on the o-scope probes). Most likely the PSU was just designed this way, then. Probably too many winding on the 5VSB aux. sec. transformer. And maybe that's why the original resistor resistor looked so burned up, as it was dissipating too much heat for its own good.

That said, the 5VSB aux. sec. rail is independent of the main power supply, so it doesn't really matter if you put a bigger load on PSU with your motherboard. That said, the aux. sec. rail voltage should increase with a higher load on the 5VSB and decrease as you lower the load on 5VSB.

I haven't rigged up a way to connect the fans to +5VSB yet. What are we hoping to see from that vs. the "real world" motherboard hook-ups?

How high the aux. sec. rail will go. The more load you put on the 5VSB, the higher the aux. sec. rail will go. If you have a way to load down the 5VSB to its maximum rating (1A? 2A?), then that will show you how high the aux. sec. rail can go up. If it goes beyond 35V, you might want to change that 35V Rubycon cap that was filtering it for a 50V cap. And if the aux. sec. goes beyond 37-38V, note at what load it does that, and perhaps avoid using the PSU on newer motherboards that might put too much load on the 5VSB.

Alternatively, you could put a resistor in series with the diode that rectifies the aux. sec. rail to lower the voltage. I imagine something in the range of 220-680 Ohms could work well for that.

Of course, if you are fixing the PSU just for your Pentium 3 PC, it should be fine to leave everything as is. As I recall, most motherboards from that era had a requirement for the 5VSB for only 1 Amp. So if your PSU can provide that without its aux. sec. rail going past 37-38V, it should be okay.

[ckrtech]

Re: Enlight HPC-250-101 is aging

Awesome.

Thank you once again, momaka. All of your feedback makes me want to start work on another PSU!

(And boy, do I have some goodies...)

[momaka]

Re: Enlight HPC-250-101 is aging

You're welcome.

It's addicting, fixing PSU, isn't it?

Looks like both of the PSUs you show in those pictures are AT type. And the Win Power may be a Wintech, but who knows. Be careful when testing those, by the way - AT power supplies often need a good minimum load to work correctly, especially on the 5V rail (like, at least 1 A). Some have been known to fry themselves if turned ON without a load.

[goodpsusearch]

Re: Enlight HPC-250-101 is aging

Old Enlight usually use delta as the oem.
and old delta, somehow, is always low in 12v when tested with no load.

The OEM is High Power, not Delta.

This psu is suitable for old P3 / Athlon cpus powered from 5V. Try it on a P4 system and you get some serious oscillation.

The Teapo SX 2200uF caps I see in the secondary section are very good caps. I have 6-7 Enlight 250W psus like that and I have removed the Teapo SX caps and then reused them in more modern psus. None of them failed till now.

If you are talking about the 12V rectifier, that may not be a good idea. Old PSUs like this usually use a fast-recovery rectifier on the 12V rail with a typical voltage drop of at least 0.8V. Replacing it with a Schottky (typical voltage drop of 0.5V or less) can sometimes make the 12V rail go way too high, especially with a high 5V load for which the PSU is made.

Already done that, without a problem. 10A rectifier is too low for 12V, I replaced it with 20A MBR20100CTL.

https://cdn.badcaps-static.com/pdfs/...d8ef430be4.pdf

[goodpsusearch]

Re: Enlight HPC-250-101 is aging

I never replace capacitor if it's not necessary, even in 10+ years old psu.
Well, esr meter is very useful here.
As long as it fit the criteria, then it's will be fine.

Yes, capacitor, will degrade. but how much it will affect your system?
That depend on your load.
If it low enough, then a non bulged caps is still good to go.
If it will run at full capacity, then fresh caps is must

This is wrong on so many levels! First of all, a cap could be dead without bulging or gone too high esr. Also, some caps get voltage leaky. If you have a capacitor that can't hold its charge, it puts extra load on the circuit, acting like a resistor. In such situation, even connecting a good cap in parallel wouldn't solve the problem. You need to remove the bad cap.

What you are doing there is quite risky.

First off when dealing with old PSUs, always check if they have a 2-transistor 5VSB circuit. 2-transistor 5VSB circuits do NOT play well with bad caps on the 5VSB output. I've seen burned PWM controllers many many times due to bad output caps on the 5VSB rail.

Furthermore, 2-transistor 5VSB circuits with a "critical" cap (typically 10-47 uF cap on the primary side next to the 5VSB transformer) CAN damage not only the PSU, but also even your motherboard and attached peripherals. Look up Bestec ATX-250-12E if would like to see a good example. A typical failure mode of such circuit is extreme over-voltage on the 5VSB rail (8-15 Volts not uncommon at all).

+1

momaka - I am late for thanking you for your response. Thank you.

I am currently removing/documenting caps as well as making a cap map. On that subject, the large TEAPO output caps (2200uf, SX series) are (according to SX Datasheet) ripple current: 1450mA /100KHz and ESR: 0.095 (Max 25C /100KHz).

I would leave the Teapo SX alone. I have never seen a failed SX.

You should also consider that this platform is very prone to oscillations. It's better not to replace caps with caps with lower esr. One more reason to keep the original caps.

More info here:
https://www.badcaps.net/forum/showthread.php?t=23996
https://www.badcaps.net/forum/showthread.php?t=24101
https://www.badcaps.net/forum/showpo...&postcount=288
https://www.badcaps.net/forum/showthread.php?t=61773

[Wester547]

Re: Enlight HPC-250-101 is aging

The Teapo SX 2200uF caps I see in the secondary section are very good caps. I have 6-7 Enlight 250W psus like that and I have removed the Teapo SX caps and then reused them in more modern psus. None of them failed till now.

I would leave the Teapo SX alone. I have never seen a failed SX.

https://www.badcaps.net/forum/showpo...&postcount=169

It's happened before. But you are right, SX is one of Teapo's better series.

You should also consider that this platform is very prone to oscillations. It's better not to replace caps with caps with lower esr. One more reason to keep the original caps.

True. It's too late anyway, as this PSU has already been recapped. Personally, I would have gone with parts that matched the ESR of the originals (Chemi-con LXY and LXV, Nichicon PM and PJ, Rubycon YXF) so as not to upset the feedback loop and result in oscillations. But as long as the voltage regulation is good and the ripple voltage is in-spec, I suppose that's what matters.

[goodpsusearch]

Re: Enlight HPC-250-101 is aging

https://www.badcaps.net/forum/showpo...&postcount=169

It’s happened before. But you are right, SX is one of Teapo’s better series.

True. It’s too late anyway, as this PSU has already been recapped. Personally, I would have gone with parts that matched the ESR of the originals (Chemi-con LXY and LXV, Nichicon PM and PJ, Rubycon YXF) so as not to upset the feedback loop and result in oscillations. But as long as the voltage regulation is good and the ripple voltage is in-spec, I suppose that’s what matters.

Judging from the photos, even a good cap would have trouble there. The cap was crammed and nearly touching the heatsink.

When I typed the previous posts I hadn't read page 2. Good thing he recapped it anyway, I would use YXF / FC esr grade caps though, as I had bad luck recapping/upgrading this platform.

[ckrtech]

Re: Enlight HPC-250-101 is aging

Hey thanks for the added thoughts, guys. Sorry I was late checking in.

[momaka]

Re: Enlight HPC-250-101 is aging

The OEM is High Power, not Delta.

You're right, I didn't realize that until now.

Hmm... this is not going to be a very good 12V PSU... or maybe not? My oscillating Sirtech / High Power PSUs have an SG6105 PWM chip. I wonder if that makes a difference, as this one is older and seems to use the TL494/KA7500 PWM.

Already done that, without a problem. 10A rectifier is too low for 12V, I replaced it with 20A MBR20100CTL.

Well, I did something similar on a Macron PSU that had a 16 Amp FR (I replaced it with a 20 Amp Schottky). And now with a heavy 5V-load, my 12V rail goes over 12.6V, which is outside of ATX spec. With the old 16A 200V FR rectifier, I was getting 12.3V at most with the same load. Of course with a heavy 12V load, my 12V rail no longer dips below 12V. So I guess it all depends on what you want to use the PSU for.

[goodpsusearch]

Re: Enlight HPC-250-101 is aging

Well, I did something similar on a Macron PSU that had a 16 Amp FR (I replaced it with a 20 Amp Schottky). And now with a heavy 5V-load, my 12V rail goes over 12.6V, which is outside of ATX spec. With the old 16A 200V FR rectifier, I was getting 12.3V at most with the same load. Of course with a heavy 12V load, my 12V rail no longer dips below 12V. So I guess it all depends on what you want to use the PSU for.

Yeap.

Group regulation has its limits. At least it got more efficient now.

When I was recapping my Macron power 350W power supplies I used to put 2 16A rectifiers in parallel for 12V. I used 2xU1620G in 1 of them and had no problem with voltage regulation. If I remember correct I tested on a 12V machine though.

[Relayer]

Re: Enlight HPC-250-101 is aging

Hello ckrtech,
Please see your modified picture below:



Seems you have a nice fan speed controller PCB.
Regards,
Relayer

[momaka]

Re: Enlight HPC-250-101 is aging

When I was recapping my Macron power 350W power supplies I used to put 2 16A rectifiers in parallel for 12V. I used 2xU1620G in 1 of them and had no problem with voltage regulation. If I remember correct I tested on a 12V machine though.

Yes, two 16A regular fast rectifiers in parallel is okay with these older PSUs. Actually, my Macron MPT-401 (400W) uses that configuration, and it has no problems with 12V rail regulation just like yours. But that's because two FR rectifiers still have a higher voltage drop than a single Schottky. Also, the Vf is more stable with two FR rectifiers in parallel than a single one. That's why you didn't have problems. My Macron MPT-301 was also stable and happy with a 12V system prior to the 12V rail rectifier swap.

So that's why I say to be careful with the 12V rectifiers on old PSUs.