Zalman ZM600-HP blowing PFC's

[Khron]

Ok, i give up, i'm stumped...

Case in point: an FSP-made Epsilon(?) platform 600w psu (some details can be found HERE.

I got it for nearly nothing, non-working. Had a shorted PFC FET (out of three), and as it turned out, also a shorted PFC diode.

First step was a re-cap (i'm "allergic" to C(r)apXon, and i wouldn't bet my life on Teapo either).

Now, with the PFC heatsink (and associated components) out of the circuit, the PSU works fine. A couple weeks ago, i decided to solder the PFC back in. Lightbulb test went as expected - bulb lit up.

Then i got "brave" and decided to skip the bulb. Result - tripped a breaker in the electric panel

Last night, after taking posession of a triplet of replacement PFC FETs, i tried that again. Result: same deal all over again Once again, one out of three PFC FETs shorted, plus the diode.

Even the NPN/PNP totem-pole pair driving the PFC FETs checks out fine. The primary cap (OST SPS 390u/420v) shows ~349uF on my Tenma multimeter, which is in spec, and it's not bloated.
____________

While typing this up, i got the bright idea of poking around the circuit a bit more, and ended up measuring the "bypass" diode as well (anode before the PFC inductor, cathode joined with the PFC diode's cathode, as can be seen HERE. Guess what - that's shorted as well. Good thing i have some spare UF5408's from fixing the Tagan BZ800 a couple weeks ago.

[Behemot]

Re: Zalman ZM600-HP blowing PFC's

Isn't the PFC choke bad, making the silicon overload? Seems FSP uses worst chokes at all as this would not be the first time I'd see this. I personally replaced one which was fortunatelly still "working", though overheating to 150 °C, in Blue Storm II.

[Khron]

Re: Zalman ZM600-HP blowing PFC's

I sorta doubt it, there are no burn marks or scorch marks or any signs of damage whatsoever.

PSU (still) works fine with the PFC out of the circuit. It "even" worked with that shorted "bypass" diode, and works just as well now, after replacing it with a UF5408.

I'll need to order some more PFC FETs for a couple FSP Everest 1010's i have here anyway, so as soon as i get the new ones, i'll report back with news.

[Khron]

Re: Zalman ZM600-HP blowing PFC's

Slight bit of news: i thought i'd also measure the source resistor from the PFC section. "Eye-metrically", it's a 3-5watt resistor, heatshrinked.

Now, my multimeter "can't decide" between 0.06ohm (60mohm) and 0.1ohm (100mohm).

Odd thing is, i peeled off some of the heatshrink, and i was greeted with a non-standard colour code: green-gold-green-orange-silver.

I'm thinking silver is the 10% tolerance indicator, and i've got a strong hunch the orange band is supposed to indicate the 3watt power rating. If anyone else has a better idea, i'm more than willing to listen

I'm starting to be almost certain that this resistor failed shorted, and the PFC kept the FETs open to try to reach the tripping voltage value across that sense resistor until one of the FETs went bye-bye.

EDIT:
An additional discovery: as mentioned, i've got a couple FSP Everest 1010w that are structurally nearly identical to this 600W Zalman. Now, the PFC current-sense resistor isn't heatshrinked, so i got a real good look at it: same size as the one in the Zalman, but this one's colour code is a tad more conventional: black-red-red-gold-brown, which "should" translate into something like 0.22ohm, if i counted correctly. The thing is, my multimeter has the same opinion about this, as with the Zalman resistor - "waving" between 0.02ohm and 0.06ohm. Coincidence or not, the FSP also has a popped PFC fet (out of three)...

[everell]

Re: Zalman ZM600-HP blowing PFC's

On some of those FSP.......the FETs have nonconductive case where attached to the heat sink. If you use a regular FET you MUST use an isolator between the FET and the heat sink....or it shorts to ground. The FETs with the nonconductive case have less power capability. So when working on this type psu, I use a regular FET with higher power rating and use a nonconductive insulator between heat sink and FET.

Next, the booster coil can be very deceiving. I have fixed several psu with bad APFC by replacing the coil. If everything works well without the APFC circuit, and the APFC circuit is blowing a FET, replacing the coil fixed it.

Because the booster coil is a common problem I have come to the conclusion that manufacturers are using cheap cores not rated high enough for the problems encountered. The end result is a coil that runs hot and fails.

Along with the blue esr meter I purchased, I also purchased a ring tester from the same company. I have found that using the ring tester with the coil still in circuit, it shows a bad coil as shorted and a good coil with a good ring. Takes the guesswork out of fixing this circuit. It was surprising to me how good a booster coil can look yet still be giving problems. Once you get a APFC circuit working, use a laser beam temperature gun to measure the temperature of the coil in several places. The heat is NOT evenly distributed. A coil running hot should be replaced with a coil having a higher quality core.

[Khron]

Re: Zalman ZM600-HP blowing PFC's

Indeed, all the FETs are "fullpak" - with isolated tabs; i guess that's why they're used in triplets. The replacements i got were the same (i figured i'd stick with the original design).

Any idea what value the source / current sense resistor "should" have, though? My meter says they're both suspiciously low AND uncertain...

[Behemot]

Re: Zalman ZM600-HP blowing PFC's

If it is for current sensing, it has to be low, otherwise there would be no current flowing through it into the supply, right

[Khron]

Re: Zalman ZM600-HP blowing PFC's

The CM6800 datasheet does contain a formula for determining the value of the resistor, but... How can i know what values the designers had in mind?

Too damn bad my meter isn't all that accurate on the low end of the scale - it's a bit "iffy" even on a 0.1ohm/2w Panasonic resistor, "wobbles" between 0.06 and 0.1ohms indicated

Oh, and curse all these non-standard colour-code markings too!

[Khron]

Re: Zalman ZM600-HP blowing PFC's

M'kay, today i put together a small low-ohms measuring rig, as per HERE. Didn't have a 12ohm resistor, so i paralelled an 18ohm and a 47ohm, to get 13ohm.

I then measured the black-red-red-gold-green resistor i extracted from the FSP. If my math isn't too messed up, it came out somewhere in the 36-37 milliohm range. How that "fits in" with the coloured markings on it, i'll never know.

The CM6800 datasheet, on page 9, says the following:

"RS =0.8V x Vinpeak/(2x Line Input power)
For example, if the minimum input voltage is 80VAC, and the
maximum input rms power is 200Watt, RS = (0.8V x 80V x
1.414)/(2 x 200) = 0.226 ohm."

Ok, assuming the same 80Vac low-line, and an input RMS power of about 1200W, then the resistor does indeed come out to around 38-39 milliohm.

How stupid would it be to try out a test-run with only two of the three FETs mounted, and a 0R22 (220 milliohm) Rs?

[Behemot]

Re: Zalman ZM600-HP blowing PFC's

Depends whether the transistors are indeed in parallel, or there is some modified circuit and they have to be all present for it to work correctly.

But I would not change the resistor, the circuit is balanced to measure current according to known voltage drop I think.

[ben7]

Re: Zalman ZM600-HP blowing PFC's

First try disconnecting the control chip, and then pulling the MOSFETs LOW using a resistor (before the two transistor totem-pole). Do the bulb test. The PFC should be 'disabled' and seemingly bypassed. If no suspicious activity occurs then the PFC chip is toasted (or locked up for some reason). Just a thought of how to test the circuit though, don't take that for granted

[Khron]

Re: Zalman ZM600-HP blowing PFC's

The circuit's kinda like this:

PFC PWM signal comes from the CM6800 chip, goes into the totem-pole. After this, there's a 10k pull-down resistor, and a 10 ohm on each of the 3 FET gates.

One thing, though: i scoped the output of that gate drive signal (at the output of the totem-pole), and it looked... A bit odd, IMHO. Couldn't snap a photo, 'cause both my hands were busy with the two 'scope probes ("improvised" differential probing). The amplitude seemed way too great, first of all, the frequency seemed a bit too low, and also, my scope might've had some "fake" triggers.

I'll see if i can get a friend to help with snapping a decent screenshot these days

[ben7]

Re: Zalman ZM600-HP blowing PFC's

The circuit's kinda like this:

PFC PWM signal comes from the CM6800 chip, goes into the totem-pole. After this, there's a 10k pull-down resistor, and a 10 ohm on each of the 3 FET gates.

One thing, though: i scoped the output of that gate drive signal (at the output of the totem-pole), and it looked... A bit odd, IMHO. Couldn't snap a photo, 'cause both my hands were busy with the two 'scope probes ("improvised" differential probing). The amplitude seemed way too great, first of all, the frequency seemed a bit too low, and also, my scope might've had some "fake" triggers.

I'll see if i can get a friend to help with snapping a decent screenshot these days

What sort of cap is filtering the Vcc of the CM6800?

Try taking out the totem pole driver resistors and see what the PSU does on the bulb test. (With mosfets and PFC diode in of course!)

[Khron]

Re: Zalman ZM600-HP blowing PFC's

Post-recap, a Rubycon YXF 47u/50v, plus whatever ceramics reside on the daughterboard with the CM6800 itself.

Well, as specified before, with the PFC components removed (3 FETs and a TO-220 diode), but with the 3A diode "bypassing" the coil and TO-220 diode, the PSU works fine, all secondary voltages within ATX spec.

For this test, considering the FETs would be off, i don't suppose it'll matter if i just stick the two good ones that are left, back in, right?

[Behemot]

Re: Zalman ZM600-HP blowing PFC's

The circuit's kinda like this:

PFC PWM signal comes from the CM6800 chip, goes into the totem-pole. After this, there's a 10k pull-down resistor, and a 10 ohm on each of the 3 FET gates.

One thing, though: i scoped the output of that gate drive signal (at the output of the totem-pole), and it looked... A bit odd, IMHO. Couldn't snap a photo, 'cause both my hands were busy with the two 'scope probes ("improvised" differential probing). The amplitude seemed way too great, first of all, the frequency seemed a bit too low, and also, my scope might've had some "fake" triggers.

I'll see if i can get a friend to help with snapping a decent screenshot these days

Low frequency could be due to low output power. Many of the CM chips even use different modes of operation on low load and than on higher load.

[Khron]

Re: Zalman ZM600-HP blowing PFC's

Low frequency could be due to low output power. Many of the CM chips even use different modes of operation on low load and than on higher load.

Good point there, this *had* crossed my mind... But i had to switch my scope to 5ms/div, to be able to fit about 2.5 periods of the gate drive signal on the screen - that's *WAY* too low for any power-saving whatever

[Behemot]

Re: Zalman ZM600-HP blowing PFC's

Well nobody said the PFC is efficient on low power, you may also find that some modern CM family chips disable PFC on low power. Also the PF itself is usually pretty bad on low power.

[Khron]

Re: Zalman ZM600-HP blowing PFC's

Ok, i promised some scope shots. The settings stayed the same: 5V/div for both probes, 10ms/div (that's milliseconds), both channels DC-coupled. Channel 2 = permanently hooked up to the negative leg of the mains cap (used as "ground"-reference). PFC components not in circuit

Picture no.1: PSU plugged into the mains (through bulb-tester), but not turned on (ie. only 5Vsb on); both probes hooked up to the negative leg of the big mains capacitor

Picture no.2: Ch1 probe before the gate resistor of one of the PFC FETs

Picture no.3: Ch1 probe on the (tied together) bases of the totem-pole transistors

Picture no.4: Ch1 probe on the Vcc supplying the totem-pole

With both channels switched to Gnd, the "base line" is the first horizontal line above the lower edge of the 'scope's reticle.




EDIT: Out of sheer curiosity, i scoped the 5Vsb line as well (Ch2 on the secondary ground (black ATX lead) and Ch1 in the purple 5Vsb lead), with the same settings as above, and i ended up with the waveform visible in picture no.5 If i scope the 5Vsb "normally" (single channel, ground clip on secondary ground, probe tip on 5Vsb) i get some ~20mV ripple at some tens of kHz (the 5Vsb chip's supposed to be switching at 67kHz).

Something's fishy here.....

[ben7]

Re: Zalman ZM600-HP blowing PFC's

Ok, i promised some scope shots. The settings stayed the same: 5V/div for both probes, 10ms/div (that's milliseconds), both channels DC-coupled. Channel 2 = permanently hooked up to the negative leg of the mains cap (used as "ground"-reference). PFC components not in circuit

Picture no.1: PSU plugged into the mains (through bulb-tester), but not turned on (ie. only 5Vsb on); both probes hooked up to the negative leg of the big mains capacitor

Picture no.2: Ch1 probe before the gate resistor of one of the PFC FETs

Picture no.3: Ch1 probe on the (tied together) bases of the totem-pole transistors

Picture no.4: Ch1 probe on the Vcc supplying the totem-pole

With both channels switched to Gnd, the "base line" is the first horizontal line above the lower edge of the 'scope's reticle.




EDIT: Out of sheer curiosity, i scoped the 5Vsb line as well (Ch2 on the secondary ground (black ATX lead) and Ch1 in the purple 5Vsb lead), with the same settings as above, and i ended up with the waveform visible in picture no.5 If i scope the 5Vsb "normally" (single channel, ground clip on secondary ground, probe tip on 5Vsb) i get some ~20mV ripple at some tens of kHz (the 5Vsb chip's supposed to be switching at 67kHz).

Something's fishy here.....

Do you have an isolation transformer?

I think you should put the PFC parts in, and run the PSU with the lightbulb trick. Then scope each gate pin of the three PFC MOSFETs.

A 1ms period is 1kHz; most switching regulators (PFC, SMPS, etc..) run at 50kHz-100kHz.

[Khron]

Re: Zalman ZM600-HP blowing PFC's

No, i don't have an isolation trafo, but for what it's worth, i've only got an IEC connector soldered to the PSU, and the earth wire isn't connected to anything on the PSU.

Yeah, that was my point, that frequency's way too low for... anything, really