Enermax PRO 82+ 525W repair

[homeless_cz]

I´m trying to repair this PSU.
One day I powered on my PC and I heard a blow.
So I disassembled the PSU and found burned out thermistor and 1 MosFET in PFC circuit.
So I replaced both FETs K20J60T to K20J60U but I didn´t know what thermistor was there, so I put SCK055. And new fuse of course.
Now the PSU does strange things. +5V SB works normally, but when I power on(green wire to GND), output voltages goes up for half second. And then nothing.
If I unplug AC cord and then power on, output voltages are OK for 3 seconds until primary capacitors discharge.
PFC is driven by CM6802, tried to replace too, no change.
Any ideas what to check? Thanks.

[stj]

Re: Enermax PRO 82+ 525W repair

first check the output diodes and caps for shorts,
and consider replacing the small electrolytics atleast.

[homeless_cz]

Re: Enermax PRO 82+ 525W repair

Output power diodes OK, all other diodes OK, caps no shorted, good capacity.
I can´t understand why it works when I unplug AC cord only. If I leave AC cord connected and just switch off the main switch, which brakes the phase wire, it doesn´t work.
I have to unplug AC completely and then power on by green wire, then it works until primary caps discharge. WTF?

[stj]

Re: Enermax PRO 82+ 525W repair

then check the primary side snubber parts and replace any zener diodes.

[momaka]

Re: Enermax PRO 82+ 525W repair

WOW!!! The timing of this thread is disturbing.
I just disassembled an Enermax Pro 82+ PSU yesterday that I picked up about a year ago for free due to not working.

In my case, the PSU only has a blown thermistor (SCK 054) that reads about 80 Ohms cold. I haven't done any other testing or work to it, but no other primary parts appear to be blown at this point.

and consider replacing the small electrolytics atleast.

All caps in this PSU are United Chemicon KY and KZE, and none of them appear to be located in hot spots. So I have very little reason to believe the caps are an issue here.

That said, in my unit, one of the two main caps (2x Panasonic UQ, 400V, 180 uF) might be bad. I haven't pulled it out yet, and it doesn't appear bulged at all. However, running my finger on top of its vent, it does feel slightly domed compared to the other. There's no plastic cap, so nothing that is fooling me. I guess I'll desolder it and find out.

Output power diodes OK, all other diodes OK, caps no shorted, good capacity.
I can´t understand why it works when I unplug AC cord only. If I leave AC cord connected and just switch off the main switch, which brakes the phase wire, it doesn´t work.
I have to unplug AC completely and then power on by green wire, then it works until primary caps discharge. WTF?

The CM6802 has a mains detection circuit, which from what I understand operates the APFC according to the input sine wave. So by removing the mains (both Live and Neutral), I suspect that is somehow temporarily disabling whatever issue the CM6802 is seeing with the APFC and mains, thus enabling it to work until the input caps discharge. Of course, just because the power switch on the PSU only disconnects the Live doesn't mean that this is happening... and reason I say this is because you are in Europe and you probably use standard (unpolarized) Schuko plugs, which means the Live and Neutral can be swapped, depending on which way you plug the power cable in the wall.

That aside, I also suggest you check ALL of the small resistors and ceramic caps on the bottom of the PSU, particularly the ones under the CM6802 chip. Also check the ESR and capacitance of the input primary caps. Not uncommon for badly-designed APFC circuits to blow/over-stress these caps.

Meanwhile, I will see what I can find with my unit and post back here. I think we are troubleshooting the same issue. Only difference is yours appears to have blown more parts than mine. I'm also considering disabling the APFC and running the PSU with a voltage-doubler arrangement, if that's even possible.

[homeless_cz]

Re: Enermax PRO 82+ 525W repair

Thanks for suggestion guys. I had a suspicion that primary part works for 1 half wave only, I mean 1 switching cycle. I was checking what happens after power on and I found out that secondary monitoring IC PS231S turns off PFC IC CM6802 after a while. So for the 3rd time I checked all primary parts and found 1 dead transistor which I missed before. CM6802 controls via this transistor the gate of 1 PFC FET. So this confirmed my suspicion.
After replace of this transistor PSU started to work normally, but I didn't have exact replacement. At this moment all voltages are good except +5V which is 4,4V , so I can't get Power good signal because of this. I hope if I will get the same transistor PN2907A the PSU will be OK.
In my case were blown parts > fuse > thermistor > MosFET > resistor 2R2 to MosFET's gate, and a driving transistor for MosFET's gate.

[momaka]

Re: Enermax PRO 82+ 525W repair

After replace of this transistor PSU started to work normally, but I didn't have exact replacement. At this moment all voltages are good except +5V which is 4,4V , so I can't get Power good signal because of this.

Make sure you are testing the PSU with a proper load - i.e. a motherboard with a 4-pin 12V CPU connector... and preferably one you don't care that much about, should something go bad. The 5V rail should be back to normal then. If not, I suspect your PSU may have other issues as well. This PSU uses a mag-amp regulator for both the 5V rail and the 3.3V rail. So it's a fully independent design pretty much. But with a low 12V load, I think the 5V and 3.3V rails may not regulate properly.

I hope if I will get the same transistor PN2907A the PSU will be OK.

That 2907 is just a standard general purpose PNP transistor. Just match the voltage and current close enough. I think 2N4403, MPSA55, and MPSA56 should probably also work here.

In my case were blown parts > fuse > thermistor > MosFET > resistor 2R2 to MosFET's gate, and a driving transistor for MosFET's gate.

But the question is, why did the APFC blow up?

I investigated my unit more thoroughly and I have to say this about the design: thermistor TH1 is installed in absolutely the most stupid spot. Why? - because if it goes open-circuit or high-resistance for whatever reason (which seems to have been the case with both your and my PSUs) that energy stored in the APFC coil after the APFC MOSFETs charge it has absolutely *nowhere* to go. As a result, you get a blown APFC MOSFET and fuse.

There's more stupidity, but I'll save it for the next post later / tomorrow. I'm still running through some calculations and tests with my unit. I think I found some other circuit design flaws on the APFC. Stay tuned.

[petehall347]

Re: Enermax PRO 82+ 525W repair

search out the fuses

[homeless_cz]

Re: Enermax PRO 82+ 525W repair

Make sure you are testing the PSU with a proper load - i.e. a motherboard with a 4-pin 12V CPU connector... and preferably one you don't care that much about, should something go bad. The 5V rail should be back to normal then. If not, I suspect your PSU may have other issues as well. This PSU uses a mag-amp regulator for both the 5V rail and the 3.3V rail. So it's a fully independent design pretty much. But with a low 12V load, I think the 5V and 3.3V rails may not regulate properly.

You are right, first I tried it with no load, voltages were OK, then I tried to add some old HDDs which made too low load for proper regulation. Today I tried to add 22 Ohm power resistors to each rail and everything looks to be OK.

That 2907 is just a standard general purpose PNP transistor. Just match the voltage and current close enough. I think 2N4403, MPSA55, and MPSA56 should probably also work here.

The only similar spare I had is 2N5401 with lower hFE, but this shouldn't be a problem.

But the question is, why did the APFC blow up?

I'm asking myself the same question.

I investigated my unit more thoroughly and I have to say this about the design: thermistor TH1 is installed in absolutely the most stupid spot. Why? - because if it goes open-circuit or high-resistance for whatever reason (which seems to have been the case with both your and my PSUs) that energy stored in the APFC coil after the APFC MOSFETs charge it has absolutely *nowhere* to go. As a result, you get a blown APFC MOSFET and fuse.

I think you are absolutely right and in my case the destruction started with defective TH1. PCB under TH1 is burned.

There's more stupidity, but I'll save it for the next post later / tomorrow. I'm still running through some calculations and tests with my unit. I think I found some other circuit design flaws on the APFC. Stay tuned.

I'm looking forward

[momaka]

Re: Enermax PRO 82+ 525W repair

I guess this took a bit longer to get to than I expected (my laptop is still down for the time being - good thing I keep backups!)

Part 1: APFC Circuit Analysis

So as I mentioned in my previous post above, apparently thermistor THR1 in these PSUs likes to go open-circuit or high resistance, if we are to judge by the two units in this thread.


And as mentioned, THR1 is located probably in the worst possible spot, between the APFC boost coil (L1) and APFC diode (D4). To get a better idea of why, here is a quickly drawn schematic of the APFC’s power section:


Like any APFC circuit, this one boosts the voltage when MOSFETs Q3 & Q4 (drawn as a single MOSFET for simplicity, along with the Gate-drive resistors) turn on and charge the APFC coil (L1). When the APFC MOSFETs turn off / open, the energy from the charged APFC coil goes through thermistor THR1, diode D4, and gets dumped in the large electrolytic capacitors on the primary C1 & C2.

But what happens when thermistor THR1 opens completely?
-You end up with a charged APFC coil that has nowhere to dump its energy. As a result, the voltage on the right end of L1 rises very high - enough to go over the breakdown voltage of MOSFETs Q3 & Q4. When that happens, damage to either or both of these MOSFETs is very likely (typically fail shorted.) After that, of course the fuse blows (and if you are very unlucky, maybe the bridge rectifier too… but that’s rare). In my case, I was lucky that THR1 went high resistance but did not open-up completely and was still able to dump energy in the primary input caps. (Or was it? Read on.) Thus, thermistor THR1 is clearly installed in a bad spot in the circuit. Ideally, it should be moved somewhere after the fuse or perhaps after the bridge rectifier.

Now the next logical question is why does THR1 go bad? In this PSU, it is a SCK 054 part. Perhaps it’s under spec-d in terms of current? The ThermalTake Smart M850W [SP-850AH3CCB] I have in this thread, appears to have its thermistor connected in the same way as this Enermax PSU, but it didn’t blow – most likely because the thermistor is indeed bigger? On the other hand, that ThermalTake PSU blew its primary cap: a 400V, 470 uF Panasonic HD.

This is where I think I am starting to see a pattern. The very so slight bulge on one of the Panasonic UQ, 400V, 180 uF caps in my Enermax PSU made me suspicious of it. So I pulled them both out and woke up my ESR meter for some work.



No folks, this is no mistake! Both of my caps read only several hundred PicoFarads. I checked them four times each, just to make sure (if not five.) The reading made me so suspicious that I even took out my box of new caps and checked a few of those as well, just to make sure my ESR meter was functioning correctly. Alas, the meter was fine – these primary caps are bad. One then may conclude that perhaps Panasonic primary caps are not good. But I will tell you this: between the writing of this post and the one for the ThermalTake M850W thread, I have seen at least several PSU posts on badcaps.net with blown APFC sections and open primary input caps – some of them with good quality Japanese caps and some of them with inferior quality Chinese/Taiwanese caps. (I can try and dig up those posts/threads, if some of you are really interested.) So the issue is clearly related to APFC and not necessarily the quality of the primary caps themselves… or rather, the design of the APFC, as not all APFC PSUs seem to have issues.

That said, there is one thoughtful provision on the primary of this PSU: ceramic capacitor C47 (0.1 uF, 1 kV). With the primary electrolytic caps blown, this was the only cap that could take energy from the APFC coil. That, combined with my THR1 not going open circuit, is what I think saved my APFC MOSFETs, main PS MOSFETs, and 5VSB offline switch IC from betting blown up. Otherwise, open primary caps are well-known to make the PSU blow more primary components. So for anyone troubleshooting the APFC circuit of this PSU: check your primary caps! If no ESR meter, replace them.

Digging in deeper… you can also see that I drew a quick voltage divider on the bottom right of my diagram. This divider goes back to pin 15 (V_FB) of the CM6802 PWM/PFC chip. It provides feedback for the voltage regulation of the APFC – in other words, to what voltage the primary caps charge. The CM6802 data sheet says that when the APFC is functioning normally, the voltage on this pin should be maintained around 2.5V. And for this PSU, resistors R1, R26, and R103 are 1 MOhm, 1 MOhm, and 13.3 KOhms, respectively. So this means the voltage across the primary input caps is going to be set around DC+ = (V_FB / R103) * (R103 + R1 + R26)
… so DC+ = (2.5 / 13300) * (13300 + 1000000 + 1000000) = 378.44V

However, looking at the electrical characteristics section in the data sheet, the V_FB voltage can vary from chip to chip between 2.45V and 2.55V. This means that the boost voltage across the primary caps, DC+, can vary too. With the above numbers, we get…
DC+ = 370.87V when V_FB is 2.45V and
DC+ = 386V when V_FB is 2.55V.

So it looks like all is well and the maximum boost voltage is not higher than what the primary caps are rated for. But this is in steady-state (running) conditions. What if the APFC boosts harder than it needs to? In other words, is there an over-voltage protection?
- Yes, the CM6802 has a provision for that too. Its data sheet states that once the voltage on the V_FB pin reaches about 2.77V, then the APFC boost drive section will turn off temporarily until the boosted voltage across the primary caps goes down.

For those of you who anticipated and plugged this OVP threshold voltage in the above formula, this is what we get:
DC+ = (2.77 / 13300) * (13300 + 1000000 + 1000000) = 419.31V
Now do you see the potential trouble there? (And my pun? ) The OVP voltage is higher than what the primary caps are rated for. And while data sheets of electrolytic capacitors usually state that they can handle a small over-voltage for a short period of time (for 400V caps, this is usually 450V), it is NOT advisable to do this. I remember reading a Japanese manufacturer’s capacitor guidelines data sheet once, and it mentioned that prolonged use of capacitors near their maximum rated voltage will shorten their lifespan and is also likely to make the caps fail open-circuit. To make matters worse, the CM6802 data sheet states that the OVP threshold voltage can be as high as 2.85V with chip tolerances/variations. This sets the worst-case over-voltage trip point for DC+ even higher: 431.42V

With that said, the pattern I’m seeing is that I have two different APFC power supplies now with CM6802 chips and blown primary caps rated for 400V. It’s possible the issue may not be unique to PSUs with 400V input caps only. But nevertheless, I strongly believe the voltage rating of the primary caps may have at least a little bit to do with these failures. As mentioned, the boost voltage should be around 371-386V when the APFC is up and running in steady-state. But during startup and transients, perhaps it can peak higher than that, possibly until OVP is tripped. And even with the “best-case” chip variations, the primary caps could still see at least about 409V.

In conclusion, I think 400V-rated caps are not adequate for the APFC circuit of this PSU. If anyone needs to replace the caps on their Enermax Pro 82+ PSU because the old ones are blown, I suggest using 450V parts. They will be more expensive and bigger. However, they should be able to handle any APFC over-voltage better. Moreover, being bigger means they will have a higher ripple current rating.

On that note, I think it’s also possible that APFC circuits blow the primary caps due to high stress. You see, the primary caps have a high-speed switching circuit on their output (i.e. the main PS section of the PSU). And on the input, they have high-speed switching current coming in from the APFC boost coil. Both of these, I think, cause a lot more stress on the primary caps than a simple line-connected (50-60 Hz) voltage booster/divider that we see in classic non-APFC PSUs with two 200V caps in series.

[momaka]

Re: Enermax PRO 82+ 525W repair

Part 2: Disabling the APFC Circuit and Some Testing

Ever since I saw more than a few PSUs with blown APFC circuits, I always wondered if there is a way to disable the APFC section. Well folks, today I will say that I was finally able to do this successfully.

As per the CM6802 data sheet (attached at bottom of post), on page 10 we see it mentions that when the V_FB pin goes above approximately 2.77V, the APFC section will turn Off and stop boosting. However, the PWM section for the main PS will continue to function regardless of this. Once the voltage across the primary input capacitor(s) goes low enough that V_FB drops to at least 2.50V, the CM6802 will start the PFC boosting again.

So… how do we disable the APFC section?
- We trick the CM6802 controller by playing with the feedback network made up of resistors R1, R26, and R103.

In my case, I used an LED in place of R103 so that there would always be a constant voltage drop on the V_FB pin. A Zener diode would serve equally well here. In fact, that may be the better option, but I did not have any Zener diodes rated for less than 5.1V.

Looking at the CM6802 data sheet, it says that the maximum expected voltage at V_FB is 3V. This is not to be confused with the Absolute Maximum rated voltage of this pin, which is, according to page 4, V_REF + 0.3V…. or 7.5 + 0.3 = 7.8V.

I tested the voltage drop of several different color LEDs in my parts boxes with a low-current supply. In particular, I first used two 470 KOhm resistors is series to limit current going to the LED. I then connected this to the primary side of one of my regular non-APFC ATX PSUs, where the voltage across the two primary caps is around 340V DC.

For the LED colors, I tested blue, white, pink (yes, really! ), and purple/UV. The reason for these colors is because these generally have a higher voltage drop. In my case, the blue, white, and pink LEDs all yielded around 2.6V to 2.75V drop across them, which wasn’t quite enough. This wasn’t a surprise, because both white and pink LEDs are based on a blue LED die with a phosphor converter. My purple/UV LED on the other hand, had about 2.9V constant drop across it with the above resistor arrangement. Perfect! Moreover, this voltage drop stayed that way until the voltage across the primary caps on my test ATX PSU fell to almost half (when I unplugged it from the wall). Excellent!

Next, I tried replacing those 470 KOhm resistors with 330 KOhm, simply because the 330k ones were smaller and already had their leads cut so that they fit perfectly on the Enermax PSU PCB. Testing with the 330k resistor didn’t yield significantly different results and the LED voltage drop was almost the same.

As to how I came to pick the value of these resistors above: I chose to set the current through the LED to around 0.5 mA to 1 mA. Current like that is usually enough to make the LED voltage drop close to what it will be even with maximum rated current. Not only this, but you also have to keep in mind the power dissipation of the resistors. Imagining the LED is not in circuit and the two series resistors above are directly connected between + and – leads on the primary cap (i.e. seeing 325-340V DC), then we use the P = (V^2) / R formula to determine power dissipation.
That is, P = (340^2) / (330 000 + 330 000) = 0.1752 Watts total, or 0.0876 Watts across each resistor. A 1/8 Watt resistor can handle 0.125 Watts, so this is slightly less than ¾ of its power rating, which is good. I don’t suggest going lower than two series 150 KOhm resistors (i.e. 300k total), because the power dissipation then would require ½ -Watt resistors or bigger. With two 150 KOhm resistors, you can use ¼ -Watt resistors. Two 150 KOhm resistors will also allow approximately around 1.1 mA of current through the LED. In my case, I used 1/8 Watt resistors, and my LED current was about 0.5 mA.

Last but not least, I disabled the APFC MOSFETs by disconnecting jumper J17. This is very important! Even though the CM6802 IC should never technically run the APFC section with this mod, this is under normal line conditions. However, during a brown-out or low voltage line, it may try to run.

Speaking of line voltage, this mod *ONLY* works if you have 230V or 240V supply line. This mod will NOT work with 115/120V lines.

All that being said, here is how my modified APFC circuit looked on the PSU board:


And because the primary caps were totally open-circuited, I had to do this:

That big cap there is a Nichicon LQ rated for 450V and 500 uF. Nice upgrade! Unfortunately, it doesn’t fit on the top side of the PCB, so I had to do it on the bottom. (I also put a paper insulator sheet between it and the PCB afterwards, just in case.) That is also the only large 400-450V cap I had in stock. It came from some smashed industrial control board that I picked from a dumpster (actually, there were 4 of these caps on it.)

Now some of you may be wondering how I can run this PSU, since I am located in the USA where we have 120V line voltage (as mentioned, this mod won’t work directly on this line voltage). Easy: the 115<->230V step-up/down transformer is an old friend of mine. Here is the entire (messy) test setup on the floor:


Starting from right to left:
* 120V US cord with exposed ends
* 450 Watt heating element (instead of series incandescent bulb trick)
* 115V to 230V step-up transformer
* PSU
* Incandescent and halogen bulbs (12V), along with resistors serving as a minimum load for testing the power supply.

The black multimeter was for measuring voltage across the primary cap. The gray (Radio Shack) meter measured the voltage drop across the LED (making sure voltage drop is within what it should be – and it was). And finally the red multimeter was for measuring the output voltage of the PSU to make sure it is regulating properly. Oh, and a 40W incandescent globe bulb for discharging the primary cap (it would go down to 25V after unplugging the PSU, which is safe to touch, but not if you want to measure resistance on the PSU.)

The results:
This mod worked fine with the low-load test shown above. Because of the 450W heating element in series with the 115V side of the step-up transformer, I was only getting about 280V across the primary cap with the PSU and loads running. Adding any more load on the PSU would make the primary-side voltage drop even lower. I think I saw the PSU turn off under about 260V DC across the primary cap. Also, with the PSU not running but only 5VSB section running, I had about 325V across the primary cap.

This being said, even if I did remove the series heating element from the step-up transformer and got proper 330-340V DC across the PSU’s primary cap, that’s still much lower than the nominal boosted 380V from the APFC. I’m not sure how much the lowered primary voltage (compared to when there is APFC running) would affect the load capability of this PSU. With a lower primary voltage, the PWM section of the main PS will have to crank harder to maintain the same voltage on the output. All this because I’m pretty sure the main PS transformer was designed with 370-380V primary voltage in mind. So I imagine this could potentially limit this PSU to roughly 350-400 Watts… or maybe 450 Watts with any luck. But I definitely don’t suggest pushing it that high with this mod.

So will I keep the PSU like this or fix it up properly and enable APFC again?
- Most likely, I will enable the APFC again, once I buy some 450V primary caps for it. But that will be with the next Digikey or Mouser shipment, which will be who knows when. In the mean time, I might try putting two 200V caps in a voltage-doubler setup and actually try to use the PSU with a load.

In any case, if you do disable your APFC with this mod here, please know that it is at your own risk. I have yet to do more experimenting with it (if I even do.) I will keep this thread updated if I do, though. In the mean time, have fun!

[homeless_cz]

Re: Enermax PRO 82+ 525W repair

Well I really appreciate you share these conclusions and spent lot of time with writing these long posts
It is all interesting. I'll consider to change primary caps to 450V ones even if they are OK, but are about 8 yrs old.
You were lucky with a thermistor, mine was shorted and blown up like a grenade, only legs in PBC stayed

[momaka]

Re: Enermax PRO 82+ 525W repair

Well I really appreciate you share these conclusions and spent lot of time with writing these long posts

No problems, and sorry if it seems like I'm overtaking your thread.

But I figured since I have the same exact PSU with almost the same issue, we might as well "combine resource" (or so to speak) and post everything in the same thread.

That said, for anyone else who might be reading this, I'm going to upload some pictures of the solder-side of this PSU. I'm doing this mainly as a reference, in case someone has one (or more) of those tiny SMD resistors on the bottom of the PCB blow up. I think my pictures should be clear enough to tell what most of them are.

See bottom of the post for the pictures. Also, I will be putting up a post on this power supply in the PSU quality pictorial thread, along with summary of the components inside this PSU. Stay tuned.

**EDIT**
And here is the post:
https://www.badcaps.net/forum/showpo...postcount=2960

[momaka]

Re: Enermax PRO 82+ 525W repair

Part 3: Conversion to Oldscool Classic Primary Side Without APFC

As mentioned in the last post, I probably won’t be ordering parts anytime soon. This means no new 450V input caps and no fixing the APFC circuit until who knows when. So rather than have this PSU sit uselessly again (and if you remember, there’s also that big cap on the bottom, so I can’t close it), I decided to do another experiment here.

For starters, I cleared the whole primary side.
https://www.badcaps.net/forum/attach...1&d=1541443598
No input caps, no thermistor, no APFC coil, no APFC sense resistors… nothing unnecessary left - time to start fresh.

And the bottom side… Oh my, looks like I am up to something no good again.
https://www.badcaps.net/forum/attach...1&d=1541443598
Not only is the big Nichicon cap gone, but I also cut a trace in one spot, added a jumper in another… and then there’s that white 14 AWG copper wire on the left side. I wonder what that is for.

Alright, let’s turn it over and see the madness.
https://www.badcaps.net/forum/attach...1&d=1541443598
https://www.badcaps.net/forum/attach...1&d=1541443598
https://www.badcaps.net/forum/attach...1&d=1541443598
Yeah, it ain’t so pretty looking.

What you are seeing is a classic PSU primary input side without APFC – i.e. two 200V caps in series. On the point here, these are Panasonic HC, rated for 200V, 680 uF, and 105°C. I got these brand new close to 10 years ago for a PowMax PSU of mine that I never got around to recapping / rebuilding. I might still do that, but for the time being I figured these caps could be more useful here.

They are wired in series, just like any other regular non-APFC PSU. The white 14 AWG wire I mentioned above is connected to Neutral (on the bottom of the PCB) and goes to the mid-point of the two series caps. Why? - Because, I live in a country with 120V AC lines. Connecting the mid-point of the input series caps to Neutral is the same as flipping the 115/230V voltage selector switch on old PSUs to the 115V position. Essentially, the two caps function as a voltage-doubler circuit. On the negative peak of the line, one cap chargers to ~170V DC (that is, 120V AC * √2). Then on the positive peak, the other cap charges to ~170V DC. The total voltage seen on the primary is then the sum of these two voltages – i.e. around 320-340V DC (depending on load and line variations). Mine was actually fairly stable around 330V, even though my line is typically 121-122V AC.

I also added two 330-KOhm balancing resistors across each of the two 200V caps. These resistors help to balance the voltage across the caps. They also bleed the charge in caps when the PSU is disconnected from the line and the voltage across them falls too low for the 5VSB circuit to discharge them (i.e. without these, the caps may remain charged to some low voltage, which may not be lethal, but could damage a multimeter if you tried to measure something on the primary.)

On that note, I should mention that these 330-KOhm resistors were what I had on the APFC feedback circuit (i.e. R1 and R26). Therefore, I also put new resistors for R1 and R26: a 150-KOhm and a 470-KOhm, respectively (which adds up to 620 KOhms – about the same as the two 330-KOhm resistors.)

In the above shots, you can also see why I cut a trace on the bottom of the PSU. Or can you? If not, here’s the reason: relocating the input NTC thermistor. In this case, it’s still on the (+) DC bus. But instead of being between the (removed) APFC coil and primary input caps, I now have it between the bridge rectifier positive (+) terminal and before the (removed) APFC coil. If anyone is fixing this particular Enermax PSU and is keeping the APFC circuit, I think this might be a better spot for the thermistor.

Finally, I moved cap CX1. Currently, it’s connected across the two 200V caps (i.e. the 320-340V bus) and provides additional bulk storage. It was moved mainly because it was in the way of things. But I moved it also because of the long wires with which the two series input caps are connected – after all, long wires means higher ESR. Given that the primary side (main PS and 5VSB) switch at high frequency, this will generate more EMI/RFI. Having CX1 cap there should help a bit.

So after doing all of this, I was ready to test the PSU. However, something crossed my mind about the new thermistor location: while this new spot will work just fine if the APFC circuit was still there, it wouldn’t work so effectively with the two 200V caps in the voltage-doubler circuit. In particular, only one of the two 200V caps will see the NTC inrush limiter and the other one won’t at all. Therefore, depending at which time the PSU is connected to the wall relative to the line’s sine wave, there can be no, little, or very large inrush current. Obviously that wouldn’t be a good thing.

For this reason, I decided to move the NTC thermistor yet again.
https://www.badcaps.net/forum/attach...1&d=1541443598
https://www.badcaps.net/forum/attach...1&d=1541443598
Now it’s connected to that Neutral wire that makes the two 200V primary caps into a voltage doubler circuit. This ensures that the inrush current is limited for both caps. On old PSUs, normally the NTC is placed on the AC-side - either on Live or Neutral after the fuse and plug wires. But there was just no other spot I could put it, simply because of how packed the PSU is. Same goes for the two big 200V caps. You may be wondering why I didn’t just solder them to the board where the original caps were and cut / jumper some traces. Technically, that would have been the neater solution. But I simply didn’t want to mod the PCB traces too much. While moving the thermistor, I also re-routed that mid-point connected 14 AWG wire, so that it is no longer soldered to the bottom of the PSU. Instead, as you can see above, I found an existing hole on the PCB that connects to the Neutral line and simply soldered a copper wire in there. Then I put some insulation on it and bent it to go to the NTC thermistor. Yes, it's ghetto but it will hold just fine. Oh, and the hot glue is there just for extra insulation and to dampen any vibrations. Not really necessary, but I had the hot glue gun hot already.

That said, here is the final look of my conversion /experiment here:


After this, I closed up the PSU and tested it with a simple dummy load again – it worked fine, as expected. The voltage across the primary caps stayed dropped to a little under 300V with my 450W series heating element. But once I removed that, it stayed close to 330V at all times.

Next, I tested on a real PC load. My guinea pig was an HP Pavilion desktop with an Athlon II X2 CPU and no HDD. To my surprise, it just did not want to boot with this PSU. What’s surprising is that all voltages came up and stayed within proper levels (at least on my multimeter… I don’t have a scope to check ripple). Yet, the PC would shut down its fans while still keeping the Enermax PSU turned On. Some measurements showed me that the Power Good (PG) signal remained pulled low. I tried putting additional load on the 5V and 12V rails, thinking this may be the issue. However, that HP PC still refused to boot with this PSU. Changing the PSU to a known-good 300W Delta made the PC work fine, though. I performed these tests three times, just to make sure there wasn’t something else going on with the HP PC. Alas, it was fine and just didn’t like my modded Enermax PSU for whatever reason.

A little depressed after all of this work, I decided to try another computer – this time, my trusty ASUS A8V-MX motherboard. I was expecting the same shenanigans, but once more to my surprise, the ASUS A8V-MX mobo actually booted fine. I went inside the CMOS and set all the options, then saved and reset – no problems!

So the PSU does work with a PC load. However, that’s all the testing I’ve had time to do with my modded Enermax PSU to this point. I still haven’t figured out why it didn’t want to work with the Athlon II X2 HP Pavilion PC. Nonetheless, I do plan to do more tests on a few other PCs to see what happens.

As mentioned in my previous post above, removing the APFC and not having the voltage on the primary side boosted to what it was before (~380V DC with APFC vs. ~330V DC now), I think it’s possible this may be affecting the PSU in some way. Also, this lower primary voltage will likely lower the overall power rating of the PSU. So on that note, I do need to test the PSU with a higher load PC to see what happens. Hopefully nothing will blow up. Keep your fingers crossed for me here.

[ChaosLegionnaire]

Re: Enermax PRO 82+ 525W repair

oh no momaka! are u building a chinese junk power supply? whats with all the jumper wires going all over the place?! lol!

anyway, i know enermax is a taiwanese company not chinese. hehe... just messing with ya!

as for that hp pc not playing nice with the psu, im just gonna put it down to oem weirdness. oem shit is typically fussy with the ram, cpu, gpu, everything. wouldnt surprise me if its fussy with the psu too.

[momaka]

Re: Enermax PRO 82+ 525W repair

as for that hp pc not playing nice with the psu, im just gonna put it down to oem weirdness. oem shit is typically fussy with the ram, cpu, gpu, everything. wouldnt surprise me if its fussy with the psu too.

Not all OEM hardware is fussy (actually most is okay, IME). But I suppose this one particular motherboard was for whatever reason. Sure it doesn't make sense why the PSU wouldn't work on that motherboard, but I tested it a few more times and nothing changed. That said, I also tested the Enermax on a very similar OEM HP board as the one above, but this one had a Athlon II X4 CPU. Guess what? - That one worked no problem. I also added a medium-power PCI-E GPU just to give the 12V rail a bit of extra stress (I'd guesstimate about 120-140 Watts load on the 12V rail total). PSU was fine again. So it's possible this PSU may just have problems with very low loads, especially on the 12V rail. Even though it's an independent design of sorts, the 12V rail does need more load than the other rails so that they can reach their target levels.

Anyways, I'm still not done with testing it more, but will update the thread when I do even more/harder load tests.

[sam_sam_sam]

Re: Enermax PRO 82+ 525W repair

Momaka

I have to ask you how much amperage will the heater element let go through in this setup to the switching power supply


I know when you use a light bulb you might get about 1 amp or so or less depending how your setup is

[Drack]

Re: Enermax PRO 82+ 525W repair

Well it's a 450w element so probably close to 3.75 amps but that's just guessing

[momaka]

Re: Enermax PRO 82+ 525W repair

I have to ask you how much amperage will the heater element let go through in this setup to the switching power supply


I know when you use a light bulb you might get about 1 amp or so or less depending how your setup is

Well, if the power supply is to completely short-circuit on the primary side, it would be as Drack pointed out: 3.75 Amps.

However, with the PSU running, you won't get anywhere near that current, because once the input voltage drops below a certain threshold, the PSU will cut out. This goes for any SMPS that's *not* based on flyback topology or a regular line transformer-based. Generally, the cutout voltage is around 100V AC for most 120V-rated SMPSes and I think somewhere around 190V AC for 230V-rated ones.

So with that said, as more power is drawn from the SMPS, it will pull more current on its primary side. And as more current is pulled through the primary side, the voltage drop across the heating element will become larger, making the voltage across the SMPS primary side smaller.

As I found out, the PSU in this thread would cut out at around 270V DC (across its primary) with a light load. That translates to around 95V AC (with voltage doubler). Thus, the heating element was probably dropping around 20-25V across it... which means the current though it was somewhere around 0.78 Amps at most.

Therefore with a standard 100 Watt light bulb (@ 120V AC), you're not going to see anywhere near 1 Amp of current draw with the PSU running (for 230V, the current will be half). I'd guesstimate the current will be around 0.2 Amps max before the SMPS cuts out (again, that's for 120V AC).

There is another (easier) way to get a gross estimate of this, though. When you have two series devices, the maximum power draw of your DUT (device under test) will occur when both devices have equal resistance.. and hence equal power going across them. So for a 100W series light bulb, you should be able to draw 50W at most with your PSU. However, this means the voltage across the PSU will also be half (60V AC for 120V line and 115V for 230V line). As long as the PSU can still work at half the input voltage, you should be able to draw that much power from its primary. But chances are, you won't, unless it's a flyback or a regular linear transformer power supply. Thus, when using a series current limiting device (incandescent light bulb or heating element), I'd say a good gross estimate is to take 1/4 to 1/5 of its rated power as the maximum that your DUT can draw before the input voltage drops too much. So for a 100W light bulb, expect maybe around 25W from the PSU. And for a 450W heating element like mine, it would probably be limited closer to 1/5 power, if not more... so around 90 Watts or less.

Sorry for the wall-of-text for an answer. I hope this clears it up for anyone that experiments with a series current limiting device.

[Syn-fi]

Re: Enermax PRO 82+ 525W repair

hi, what is the value of the NTC designated at THR1?
thank you