Cooler Master 1200W Silent Pro Gold - Bad power switch

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Ok - this is where we disagree (nothing wrong with that, I do respect your opinion of course).

I doubt that a Cooler Master power supply is designed in a way so that the mains switch fails on first use.

So, for now, I'd like to think the thermistor has got weak and gets to low resistance too quickly. I've got a replacement coming.

If that doesn't fix the problem, then I'll think of alternatives. Because yes, I am convinced that the switch getting stuck is indeed a problem

Of course you could be right. Maybe the switch got stuck on day one and nobody noticed. I use the switch on other PSUs regularly for various reasons and never have I seen a stuck one. This is a large PSU indeed but of a good brand so I don't immediately assume that this is by design.

[momaka]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Guess I'll chime in for quick second here...

There are two current paths on the primary side for the rectified positive rail after the bridge rectifier before it goes to the positive of the two primary caps. These paths are:
1) Through the diode you circled in blue and then through the thermistor.
2) Through the APFC inductor and then the diode circled in yellow

Path 1) is used only when the PSU is plugged in and sitting in standby.
Path 2) is used when the main PS is turned on and the APFC circuit is active. At this point, path 1) is no longer used. Therefore the thermistor does not need to be rated for the full power (and thus current draw) of the PSU.

Admittedly, I too do not recall to have seen the primary of an ATX PSU (with APFC) designed like this before. But I agree with Per that it is pretty clever, as is doesn't require a relay to bypass the thermistor. Most PSUs I've seen, either the thermistor is always active, or there is a relay to bypass is when the APFC is On.

What I don't understand about this particular design is how current won't flow through both paths 1) and 2) when the PSU is plugged in. I'm sure CM must have though of this... but who knows? For now, I can only assume the APFC inductor has a large enough inductance that it presents a higher impedance path compared to the thermistor, at least on a cold plug in with the bulk caps discharged. Therefore, only path 1) can be used on a cold plug-in. But maybe not? If so, that could be why the switch contacts are getting fused. However, I doubt this could have happened to every one of these PSUs without anyone noticing. Given that these are decent quality PSUs, Q.C. at the factory would surely have caught this.

On the other hand, I wonder what will happen if the thermistor is open-circuited. Perhaps then this might somehow cause the switch contacts to fuse, as only path 2) will be used and there won't be a "soft start" circuit. The APFC inductor might have a higher impedance compared to the thermistor for the first few ms of the rectified wave, but then a much lower impedance shortly after. If so, and if the thermistor is open-circuited, then perhaps this is one way to make the PSU draw a much higher inrush current?

You mentioned your thermistor measure OK, though... so I'm skeptical this is the issue here.

If anything, I'm more inclined to think the replacement switches may be the issue here. Where did you get your replacements? If ebay, Amazon, Ali, or some other China-mall place online, consider their ratings bogus. You'd think the cheap China factories can't get it wrong with making a simple switch... but in fact, they CAN (and I have seen it multiple times myself with cheap switches failing after number of years vs. quality switches working reliably for a long time.)

That's all I can think of at this point.

[Per Hansson]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

On the other hand, I wonder what will happen if the thermistor is open-circuited. Perhaps then this might somehow cause the switch contacts to fuse, as only path 2) will be used and there won't be a "soft start" circuit. The APFC inductor might have a higher impedance compared to the thermistor for the first few ms of the rectified wave, but then a much lower impedance shortly after. If so, and if the thermistor is open-circuited, then perhaps this is one way to make the PSU draw a much higher inrush current?

You mentioned your thermistor measure OK, though... so I'm skeptical this is the issue here.

In my opinion this path should not exist because when the PSU is soft off the MOSFET's for the APFC are off so there is no path through it to the capacitors.
But if one of the MOSFET's failed shorted that would surely be a problem and cause this inrush issue.
tony359: Did you verify that you have ca 340VDC on the primary cap with the PSU is standby and around 400v when you turn on the main power supply?
P.S: To test the theory you could plug in the unit with the thermistor removed (if it still is) and see if it works anyway, that would confirm one of the MOSFET's are shorted IMO.

If anything, I'm more inclined to think the replacement switches may be the issue here. Where did you get your replacements? If ebay, Amazon, Ali, or some other China-mall place online, consider their ratings bogus. You'd think the cheap China factories can't get it wrong with making a simple switch... but in fact, they CAN (and I have seen it multiple times myself with cheap switches failing after number of years vs. quality switches working reliably for a long time.)

That's all I can think of at this point.

Yes I think so too, I also noticed that generally power supplies of this wattage uses the bigger type of power switch but this one uses the same size like all PSU's have used for eons so there's that too...

[momaka]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

In my opinion this path should not exist because when the PSU is soft off the MOSFET's for the APFC are off so there is no path through it to the capacitors.
But if one of the MOSFET's failed shorted that would surely be a problem and cause this inrush issue.

Well, not quite.
The two APFC MOSFETs are not part of the current path from the bridge rectifier (+) lead to the bulk caps (+) leads. Rather, the APFC MOSFETs pulse On and Off when the APFC is activated, shorting the positive rail from the bridge rectifier to ground through the APFC inductor. This "charges" the APFC inductor each time the MOSFETs are On, and then "discharges" through the APFC diode and into the bulk caps when the MOSFETs switch Off.

So if one of the APFC MOSFETs failed shorted, either the fuse or the APFC MOSFETs source resistors would instantly blow.

The circuit is pretty much the same as the one I drew a simplified schematic for a while back for an Enermax Pro 82+ PSU, shown here:
https://www.badcaps.net/forum/attach...1&d=1539360637
In the case of the above schematic, Q3 and Q4 are the APFC MOSFETs with R7 and R8 the respective Source resistors. L1 is the APFC inductor, and D4 is the "diode" (Enermax used the S-D junction of a MOSFET there for whatever reason) corresponding to the diode circled in yellow on Tony's picture. THR1 is the thermistor... which is not where it is in this Cooler Master 12000W PSU. C1 and C2 are the primary bulk electrolytic caps.

Yes I think so too, I also noticed that generally power supplies of this wattage uses the bigger type of power switch but this one uses the same size like all PSU's have used for eons so there's that too...

Yup, that could be the issue too.
If the original switch was of higher quality than the O/P's replacements, then it probably could withstand a few good power cycles and probably how the whole design got past testing / Q.C. and into production. But in reality, it's probably just marginal for the power / inrush current requirements of this PSU, and hence why it failed. And with the replacement switches possibly beling of even lower quality, it wouldn't be hard to imaging why they fuse every time.

[Per Hansson]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Thank you momaka!
I'm sure you are correct, my mistake.
I based it on the following, but I guess that pin I asked about there then must be connected as your schematic shows and not like this:

You're welcome, can you also please check resistance at the point I marked with red text as "Heatsink Pin???" in the same picture, if that leads to the heatsink or if you can see where it leads in the top of the PCB?

I confirm those are the heatsink pins.

But then I fail to see how the thermistor can even help at all?
I mean it will have a high resistance when connected to mains so the current will just go through the APFC inductor to the bulk capacitors then?
Just as you said in your previous post #22
The issue gets even worse if a thermistor with a higher resistance is used like I suggested together with a relay so forget about that then...

tony359: what's the color coding of these two resistors?

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

If anything, I'm more inclined to think the replacement switches may be the issue here. Where did you get your replacements? If ebay, Amazon, Ali, or some other China-mall place online, consider their ratings bogus. You'd think the cheap China factories can't get it wrong with making a simple switch... but in fact, they CAN (and I have seen it multiple times myself with cheap switches failing after number of years vs. quality switches working reliably for a long time.)

Hi Momaka, thanks for chiming in!

I cannot remember where I got those. That said, the original switch "works" until you use it. It gets fused when you switch on the PSU but it gets "unstuck" after toggling it a few times.

As you all said, it could be a marginal design and the switch works just enough to pass QC.

tony359: Did you verify that you have ca 340VDC on the primary cap with the PSU is standby and around 400v when you turn on the main power supply?
P.S: To test the theory you could plug in the unit with the thermistor removed (if it still is) and see if it works anyway, that would confirm one of the MOSFET's are shorted IMO.

Thanks, I'll check that.

tony359: what's the color coding of these two resistors?

attached. Could it be 3.8 Ohm? I'm reading a short there - my probes read 0.5Ohm if I short them and I read 0.5 ohm on those resistors in circuit.

[stj]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

do you have space around the switch to enlarge the hole and fit a bigger one?

for high inrush you need a specific type of switch that has a large gap that opens and closes fast.
this reduces arcing when it's switched
you can tell these switches from how they feel, they have very agressive springs in them.

[stj]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

how did you get 3.8ohm from black(0) yellow(4) silver gold black?

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

I get Black Orange Grey gold - and the last black might be some special tolerance?

I need to check about the switch, I don't think so.
However, that's for a later stage. For now I'd like to find out whether this is a fault or - as speculated - a marginal design which needs improvement.

The factory switch is indeed much more springy than the one I fitted.

The problem with ordering electronics in the UK now is that the respectable sellers (CPC, Farnell, RS) are charging an arm and a leg for shipping small orders. It used to be free over £5, now it's free over £50 and for small order you get charged extra too. So I often end up ordering from Ebay or - even worse - from AliExpress as Ebay has got expensive as well sometimes.

[stj]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

99% of my stuff is from ali now,
only capacitors and specialist stuff comes from u.k. sources.
as you said, they are taking the piss with the minimum order value - although i think cpc is still only £20

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Indeed no handling charge over £20 - and no shipping fees even with small orders? That's new!

The website says "free delivery over £20" but I have simulated a small order and I'm getting free delivery.

Anyways, no extra fees over £20 is a bit more reasonable thanks. I can add some consumables to get there. £50 is more difficult.

Will make things a bit easier thanks for that.

[Per Hansson]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

The resistors should be read with regular color scheme for a standard 4-band resistor.
So it is black-orange-silver for 0.03Ω then gold for 5% tolerance and the final black band is manufacturer specific (could mean non-inductive for example).

I played around in Falstad with a simulation of the circuit (grossly simplified) and it behaves as expected if the inductance is extremely high (the default in Falstad was 1 Henry).
I tried Googling what a reasonable inductor is for a ATX power supply but eventually gave up and desoldered one from an old power supply I had laying around.
It measured 325µH all the way from 100Hz up to 100kHz.
If I input that in Falstad something crazy happens: it starts oscillating, I have no idea why the simulation does that.
Anyway, I was hoping to be able to understand how the thing works but maybe there is a reason we don't see this design in other power supplies?
I mean the cost savings are pretty large, so if there would be no drawbacks I'm sure everyone would be doing it like this?

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Ah, I was looking on the 5-bands and silver as third band doesn't exist there. Thank you.

Also thank you for investing time on this subject, I'll do the tests you mentioned asap.

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

tony359: Did you verify that you have ca 340VDC on the primary cap with the PSU is standby and around 400v when you turn on the main power supply?
P.S: To test the theory you could plug in the unit with the thermistor removed (if it still is) and see if it works anyway, that would confirm one of the MOSFET's are shorted IMO.

Sorry for the delay.

I tested without the thermistor.

I get 344V in standby and 397V when turned on. Yes, it does work without thermistor.

What do we get from that?

I have ordered some replacement switch from CPC - they are from Arcoelectric here in the UK. I could only find 10A switches of that size. They do feel much better than the ones I have so maybe who knows, it was just a dodgy switch.

I've also received the thermistor replacements. I've replaced it but I'd like to add an X capacitor to the switch before installing the switch.

Only difference I think I see is that the capacitor seems to drop from 397V to 340V slowly? Maybe it was happening before and I did not notice?

The mains socket has two X capacitors in parallel. 1x1uF and 1x0.47uF across N and L. Plus two small tantalum (?) across L-E and N-E.

Shall I select something close to the same 1.47uF capacitance used across the mains socket?

[Per Hansson]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Sorry for the delay.

I tested without the thermistor.

I get 344V in standby and 397V when turned on. Yes, it does work without thermistor.

What do we get from that?

It means that the APFC is working as expected, I had a suspicion that it was turning on directly when you applied mains: that would result in a too large inrush but this confirms it is working as intended.
I do have my doubts that the design is sound though, otherwise we would not see power supplies use expensive relays to take the thermistor out of the circuit .
I tried complicating the simulation on Falstad but could not get it to behave as expected with an inductor of expected size (325µH).

I would just install the new switch and see if it works, might be all there is too it.

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Thanks. You're probably right. But I wouldn't mind adding some capacitors to mitigate the sparks before I try, shall I go with the same I see installed on the mains socket? 1x0.47uF + 1x1uF.

[Per Hansson]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

The capacitor would go in parallel with the switch.
But even 1uF would be enough for 20mA of current.
That might leave the PSU constantly on or at least constantly in a loop trying to charge the main cap:
Then switch on 5VSB and drain it, looping like that forever (I've never tried it).

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

oh I see, very interesting. Any recommendations then? Different value? Different cap?

[Per Hansson]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

As I said in my previous post I would just install the new switch and see if it works

[tony359]

Re: Cooler Master 1200W Silent Pro Gold - Bad power switch

Ok - I was just trying to avoid burning another switch, that's it

Shall test and let you know!