SMPS Primary Side Short?

[OpenCircuit]

Am I supposed to have .38 ohms from the + terminal to the  - of the rectifier bridge? 

If I sever the jumper at the red X, the circuit (+ to - of rectifier bridge) goes OL.

Suspecting CM6800 ("Vertical Board").

My light bulb 100W (in series) is glowing bright. There is a definite short on primary side after the rectifier bridge.

The IN5406 was shorted. Q2 was shorted (lower MOSFET on the right). Fuse was also blown. Replaced all of these and still have dead short.

You will have to look past my errors in the schematic I made below. If your hunch is a misreport on my part, please excuse. This is of course a very simplified version of what I am  working on (Corsair SMPS Power Supply).

[peteb2]

...my only input would be go look at the .pdf for the CM6800 device and back engineer from there to see what the designer of your SMPS was thinking. I'm too confused on your schema sorry...

http://www.champion-micro.com/datasheet/Analog%20Device/CM6800.pdf

[james_s]

A short on the input almost always means the chopper transistor(s) are bad, occasionally the rectifier.

[Armadillo]

It's really really very very rare to see a shorted transformer. I've never seen one so far. I've seen transformers with milliohms of primary resistance, so I guess it's possible to get 0.38 Ohms without anything shorted.

not at the + and - terminals of the bridge rectifier.

[rob77]

It's really really very very rare to see a shorted transformer. I've never seen one so far. I've seen transformers with milliohms of primary resistance, so I guess it's possible to get 0.38 Ohms without anything shorted.

not at the + and - terminals of the bridge rectifier.

unless you have a shorted mosfet

[OpenCircuit]

A short on the input almost always means the chopper transistor(s) are bad, occasionally the rectifier.

Q2 was shorted (right side of my diagram).

It's really really very very rare to see a shorted transformer. I've never seen one so far. I've seen transformers with milliohms of primary resistance, so I guess it's possible to get 0.38 Ohms without anything shorted.

I completely removed all transformers on the entire PCB and short was still present. Very likely true in my case it seems.

unless you have a shorted mosfet

Had.

Replaced Q2 (MOSFET) and short is still present. There are only two ways out of the + terminal of the rectifier bridge:
1. Small trace that leads into the CM6800 controller IC.
2. Through the inductor coil where circuit splits
       A. Through  inductor coil then into the gate controller (Q2).
       B. The other way is through a low resistance ceramic resistor that leads into the negative side of the large cap (270uF/400V). If I break the circuit going into from the rectifier bridge +terminal the short is gone. This seems to eliminate the CM6800 being the case as this controller remains connected.


Waiting on desoldering wick...... 

[james_s]

Waiting on desoldering wick...... 

I piece of copper braid from a scrap of coax, dipped in some liquid flux makes a reasonably good desoldering wick.

Vacuum desoldering gun is the way to go if you do much of this though.

[dicky96]

Are you sure that diagram is correct because you seem to have a coil between the positive and negative of the bridge rectifier via the connection where it says .14

That can not be correct as this would produce a DC short or near as damn it.

Seeing as you read .38 ohms so there is a bit of resistance somewhere which means passing current should warm something up, Assuming you have a variable bench power supply, connect it across the + and - of the bridge, set it to maximum voltage and set the current limit to 0A

Now start increasing the current to say 1 amp and feel around for something getting warm.  If nothing warms up turn the current up to 2A 3A whatever your supply will go to, and hopefully something will warm up by now.  There is your short. 

Rich

[OpenCircuit]

Vacuum desoldering gun is the way to go if you do much of this though.

Been gawking at this for a few weeks now, but I know what "cheap" gets "you": http://www.ebay.com/itm/331086297520?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

The spring loaded ones are too big for smaller pin configurations.

[james_s]

Hard to say with stuff like that, it might be ok, it might be junk, I'm leaning toward the latter. I have a Hakko 808 which is a wonderful piece of gear that I've been very happy with but it was not cheap. I suspect there are some reasonable copies of Hakko or Weller gear that might work ok for you and be more affordable for occasional use.

[dicky96]

@Opencircuit
If you refer to the example application circuit for CM8400 you will see the is a PWM controller and also a PFC controller

Look at the Voltage mode Circuit - The two most likely places you will find your short and around Q3/Q4 MosFETS in the PFC section of the power supply or around Q3 in the Switch mode section.  Confusing the application circuit shows two Q3s!  The one in the switch mode section is connected to Q6/Q7 via R28

The 1N5406 will almost definitely be D6 in the application circuit.  The fact that this diode was short circuit suggests your problem is/was short circuit Q3 in the switch mode section.  However the fact that you replaced the 1N5406 and you still see a short across the rectifier says the short must be on the anode side of D6 which suggests you have short circuit  Q3/Q4 in the PFC section.  You don't need to look further than the 1N5406 for now.  Hope that helps

[OpenCircuit]

Are you sure that diagram is correct because you seem to have a coil between the positive and negative of the bridge rectifier via the connection where it says .14

That can not be correct as this would produce a DC short or near as damn it.

Rich

Thanks Rich, I totally screwed up.    Fixed and updated the first post (also below).

Squiggly line is the inductor coil.

Yellow wire is a sensor wire that solders to the copper on the protective tape around the inductor coil. Unwrapped and verified this is not the short, BUT resistance is .22 across  Q2's D and S.

2 green dots are on a very small transformer. Resistance is .36 across these two dots which seems expected when the wire is simply wound around a magnet. However, From either green dot to the two question marks on the same small transformer I have .5~.8ohms.

Pulled Q3 and Q4 and they both bench tested fine with 12v. Used a small .2 amp fan.

Variable power supply in 12v, 5v and 3.3v  Considering  a cheap chinese one that you unplug when you leave the room.

When I plug PS into my 100w light bulb safety device the bulb glows like I have stuck a nail across the outlet terminals.

[dicky96]

So Q2 is short circuit D-S then, this will be a FET in the PFC section of the power supply.  Pull Q2 and check it for shorts

[tautech]

Vacuum desoldering gun is the way to go if you do much of this though.

Been gawking at this for a few weeks now, but I know what "cheap" gets "you": http://www.ebay.com/itm/331086297520?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

The spring loaded ones are too big for smaller pin configurations.

You would be surprised...........especially if you've never owned one.
Never had a component I couldn't remove with just a spring loaded sucker and a little bag of tricks when it isn't straight forward. Use mine on SMD too, just got to watch those littles ones. 

[dicky96]

Vacuum desoldering gun is the way to go if you do much of this though.

Been gawking at this for a few weeks now, but I know what "cheap" gets "you": http://www.ebay.com/itm/331086297520?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

The spring loaded ones are too big for smaller pin configurations.

I wouldn't be too sure of the usefulness of that desoldering pump as it is only 30W and that seems a bit low especially for PCBs with a heavy ground plane. 

There is this model that seems to be all over ebay http://www.ebay.co.uk/itm/100W-220V-S-993A-Electric-Vacuum-Desoldering-Pump-Solder-Sucker-Gun-350-c-450/111365837937?_trksid=p2047675.c100005.m1851&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D2%26asc%3D43782%26meid%3De54490e192124202bb0a3609f177155a%26pid%3D100005%26rk%3D3%26rkt%3D6%26sd%3D251640793299  which is 90W but I've never actually used one so couldn't say if it is actually useful or not.  When I used to work for ICL back in the 80s we had Pace desoldering stations which you activated with a foot pedal and they were great, even for 5 or more layer through hole boards we were working on at the time  I can't remember what model they were but think the base stations were blue in colour.

[OpenCircuit]

So Q2 is short circuit D-S then, this will be a FET in the PFC section of the power supply.  Pull Q2 and check it for shorts

Thanks for pointing me there. Another lesson for me....Short resolved.

[OpenCircuit]

You would be surprised...........especially if you've never owned one.
Never had a component I couldn't remove with just a spring loaded sucker and a little bag of tricks when it isn't straight forward. Use mine on SMD too, just got to watch those littles ones. 

I have  a few of them, but would sure like to get something better.

[tautech]

You would be surprised...........especially if you've never owned one.
Never had a component I couldn't remove with just a spring loaded sucker and a little bag of tricks when it isn't straight forward. Use mine on SMD too, just got to watch those littles ones. 

I have  a few of them, but would sure like to get something better.

I might be lucky that the cheap one I got yeeeeeears ago is better that many available these days so I've never felt the need to upgrade it or go to a desoldering iron.
I was advised waaaaay back to steer clear of them but that was in the day when the only units available were Pace and the like whose cost for consumables was excessive but things are different now.
There's threads that come up from time to time of what's available now and some look quite decent but remember that when you NEED something like this you want plenty of wattage for multi layer PCB's. A good 50W station will solder most things but for short dwell time in a desolder iron I'd be looking for much more than 50W.
As you are now past the hill on this repair you are in the position to watch and wait for good advice from those that know more than you and I about recent additions to the marketplace.

[OpenCircuit]

I got some life out of it...although I don't think it is charging the main cap. as it should. It sits around 160 until unplugged from the wall. Ordered some CM6800 and PS229 controller ICs a week ago per another user's suggestion.

[dicky96]

@opencircuit
Are you saying you now have 160V across the main electrolytic? Could you post a photo of the component side of the board  or mark where the 160V across the electrolytic on the track side of the board?

Also what do you measure across the +ve and -ve of the bridge rectifier?

What is the mains voltage in your location  110/220V?

Is your limiter light bulb now off, or dim?

Rich

[OpenCircuit]

@opencircuit
Are you saying you now have 160V across the main electrolytic? Could you post a photo of the component side of the board  or mark where the 160V across the electrolytic on the track side of the board?

Also what do you measure across the +ve and -ve of the bridge rectifier?

What is the mains voltage in your location  110/220V?

Is your limiter light bulb now off, or dim?

Rich

Yes; 160v at main electrolytic.

+ve and -ve: 160v

Mains: 120vAC

Limiter: Does not come on at all. With a normal good working power supply the bulb will start bright for a fraction of a second then go out quickly.

Image:



Q3: Drain connects directly to main cap + terminal. Gate is erratic  (image below). Source is .12

[james_s]

160V on the main electrolytic sounds perfectly normal.

[dicky96]

Yes 160V sounds correct if you are on 110V mains

What happens to the 160V when you disconnect the power?  Does it discharge quickly or slowly?

How were you taking those readings on Q3 gate? I'm not familiar with that type of chart, what does the vertical scale represent? Volts?

Do you have an oscilloscope and if so what type?

Cheers
Rich

[OpenCircuit]

Yes 160V sounds correct if you are on 110V mains

What happens to the 160V when you disconnect the power?  Does it discharge quickly or slowly?

How were you taking those readings on Q3 gate? I'm not familiar with that type of chart, what does the vertical scale represent? Volts?

Do you have an oscilloscope and if so what type?

Cheers
Rich

Right at 77ms AC was removed and discharge began (100ms interval):


Software that came with DMM: Y-axis is volts X-axis is milliseconds.

Do have a DSO138 (toy scope), but volts are limited to 50Vpk-max. DMM has Hz on it.

Thanks for the help everyone.

[OpenCircuit]

.........

[dicky96]

Hmmmm.  First thing that makes me wonder, is gate of Q3 has as much as 13V on it.  If that is 13V more than the source then that does not look right.  Are you measuring Q3 gate with other meter probe on negative of the main smoothing cap?  Remind me which one is Q3, is it the power FET from SMPS transformer primary to 0V?

The fact the 160V discharges quickly on the main cap when you disconnect power says something is drawing current like the the SMPS is actually running. 

The DSO138 is actually what we need here because it is running from a 9V supply so it is isolated from the mains as long as it is not conneccted to anything else like a PC which is grounded.  This means we can safely connect the ground of your scope to the -ve end of the main smoothing capacitor :-)  With a 'proper' scope you would need a mains isolation transformer.

OK so on your toy scope, ground connected to main smoothing capacitor negative, what do you see on gate of Q3?  We already know from your multimeter it is less than 50V.   

Rich

[dicky96]

Actually are you sure the power supply isn't working now?

If you have no load then the light bulbs may not light at all, especially as this type of supply has a PFC (power factor correction) circuit, so there is a large inductor between the bridge rectifier and the main smoothing capacitor, plus the smoothing cap will be a relatively low value.  This would reduce power on surge. 

Have you checked the output voltage?

It's just that the discharge pattern on the 160V capacitor looks correct for a working power supply with little or no load.  Nice quick discharge down to 40V at power off, then it goes to a slower discharge at about 40 Volts - which would be when the smps oscillator cut off.

Rich

[OpenCircuit]

1. Are you measuring Q3 gate with other meter probe on negative of the main smoothing cap? 

2. Remind me which one is Q3, is it the power FET from SMPS transformer primary to 0V?

3. OK so on your toy scope, ground connected to main smoothing capacitor negative, what do you see on gate of Q3? 

4. Actually are you sure the power supply isn't working now? Have you checked the output voltage?

Rich

1. I believe I probed the neg on the main cap and Q3 gate.

2. Q3 drain connects directly to the + terminal of the main cap.

3. Pulled the PWM/PFC controller to see if I could conduct any tests to confirm operation: https://www.eevblog.com/forum/beginners/testing-pfcpwm-controller-(cm6800)-out-of-circuit/msg1212880/#msg1212880 Ordered some a few weeks ago just waiting for the 6800s to arrive. Impossible to do any work with the IC daughter board installed. Not sure the primary side is boosting as it should.

4. Was getting .6v from  3.3, 5 and 12v rails.

[dicky96]

Hi
I don't think CM6800 is your problem here 

From the discharge pattern on the 160V main smoothing cap It's pretty obvious to me that the SMPS is actually running to some extent, in other words the CM6800 is driving the chopper FET Q3 and you are getting some voltage (albeit a low amount) on the output of the PSU.  To get that the CM6800 has got to be driving PWM to Q3.

Now if we just refer to page 15 of the CM6800 datasheet and the typical application, which is probably very similar to your PSU:

http://www.champion-micro.com/datasheet/Analog%20Device/CM6800.pdf

The top section of the schematic, that is the PFC circuit. This is where we had Q2 short circuit causing your lamp to light full.  Now we have 160V on the main smoothing cap this is sorted.  Actually even if the PFC section was not oscillating you would still get charge to the main smoothing cap via the coil L1 and the PSU would run.  The whole point of PCF is to make the PSU appear as a resistive load and increase efficiency.  You would still have the PSU running without this section working.

Now you said Q3 drain is connected directly to +ve of the main cap.  I don't think this is actually the case, if we are referring to Q3 in the bottom section of the schematic on CM6800 page 15 then you will see the drain connects to V+ of the main cap via a primary winding on the SMPS transformer.  On your DMM this would read near to zero ohms to main cap +ve but if you look at the PCB and am fairly certain you will find Drain goes to V+ via the winding.

I believe you said way back on this thread that Q3 was short circuit and you replaced it?  If that is correct did you fit the same part number as a replacement?

Now here is the interesting thing.... you were saying you had like 12-15V varying voltage on gate of Q3 with respect to -ve of the main smoothing cap.  That had me puzzled for a while as it can't actually happen, unless Q3 was faulty maybe and if the SMPS is basically running and generating some output Q3 must be working.   But.......

If you look again on Page 15 schematic bottom section, you will see Q3 source connects to -ve main smoothing cap via R31.  This will be the current sense resistor and very low value.  It may even be 4 or 8 SMD resistors in a series/parallel arrangement.   

It seems likely that what is happening here is when Q3 was short circuit it blew some of these resistors so they went high value.

This would cause a couple things to happen - first the source of Q3 would probably be floating at some volts above 0V.  Then of course the gate could read 12V or 15V or whatever with respect to 0V. The PowerFET would probably run but the current through the SMPS transformer primary would be very limited and you would get minimal output.  Secondly the CM6800 would think it had an overcurrent fault on the SMPS output and reduce the PWM duty cycle to minimum giving you even less output.  And thirdly your light bulb would not even flicker.  Hmmm sounds a bit familiar?

OK so...
1. Verify Q3 Drain goes to cap +ve via primary of SMPS transformer, not directly.

2. Measure resistance Q3 source to -ve main smoothing cap.  It should be very low maybe 0.1R or less.  If it is high, try to find R31 which may be an array of SMD or a hefty through hole resistor and see what value it should read.

3. With the PSU running measure voltage Source to Gate of Q3 or even better put you scope Ground on Q3 source and probe on Q3 gate.  This is safe to do as your scope runs from a low voltage supply and is not grounded, as long as it is not connected to anything else other than the device under test at the same time.  You should see voltage no more than 3V or pulses about 3V high

Right.... let me know what you find?

Rich

PS there is another scenario could be the problem - that your PSU actually has a high side and low side FET driving the SMPS transformer and Q3 is OK but the other one is not.  But do the above checks first before we worry about that possibility.  Anyway it should be fairly obvious if this is the case as you would have two large power FETS (both same type) around the primary of SMPS transformer

[OpenCircuit]

...did you fit the same part number as a replacement?

This is essential. Same part number, but package type.  FET body isolated v. non-isolated

Q2 was shorted (right-middle; see below) marked "Q2".

"IN5406" was shorted middle of picture.

Q3 (top-left; see below) click on image then when the page loads click the image again. This will get you high-resolution of the traces. When looking at the FETs (markings side) the left leg is the gate, then middle drain and finally source right. Just double-checked this to make sure diagram is correct.

Thinking of running wires for remote mounting of the CM6800 daughter card allowing me to probe the 6800 IC terminals directly while in-circuit.

I will look into the other things you mentioned as soon as I can get this back together. I agree, I think it is running since I was getting .6v on all rails on the secondary side The block diagrams are roughly the same from 6800 block diagram and my application, but certainly operation differences. This is a 600w power supply for a desktop computer. Thanks a ton for all the ideas and information. This is the entire purpose of this project-learning.

[dicky96]

No worries nmate that is what the forum is for and also it is an interesting challenge for me to try and diagnose something via forum posts only.

If it is a 600W PSU it is likely to have high side and low side FETs driving the primary.    Are Q3 and Q4 both the same type of device?  If so pull Q4 and test it with a bench supply and fan for example

Are there just three terminals on the primary side of the SMPS transformer? If this is a single sided board  then that looks to be the case - 3 transformer primary connections to the left of Q3 Q4, and one unsolderd terminal.  In that case, if these three terminals all read pretty much zero ohms to each other I think we have a two FET output stage and a centre tapped primary in which case you have a push-pull output stage (see below).

So we probably need to work out the topology (type) of smps design you have

Take a look
Could be half bridge type design or push pull design
http://www.smps.us/topologies.html

Half Bridge LLC
http://www.infineon.com/dgdl/Infineon-ApplicationNote_600WHalfbridgeLLC_Evaluation_Board_with_600V_CoolMOS_C7-AN-v01_00-EN.pdf?fileId=5546d4624e24005f014e2ef4761132a3
but then we would see MosFETS rather than rectifiers on the secondary side

There is an interesting link to SMPS topology here as well.
http://ww1.microchip.com/downloads/en/DeviceDoc/dsPICSMPS%20AC_DC%20Users%20Guide.pdf

Regards CM6800 daughter board: exending it so you can work on it would be helpful but the only really important thing now is what do you see on your scope on pins 11 and 12 with respect to -ve of main capacitor.  These are probably connected direct to pins where the daughter board fits the main board, or maybe via a resistor so once you trace that out you can test from the daughter board pins with your scope.  I know I have said this before but do not ground your scope - if it is just connected to it's PSU then you will be fine.

[OpenCircuit]

Are Q3 and Q4 both the same type of device?

Are there just three terminals on the primary side of the SMPS transformer? If this is a single sided board  then that looks to be the case - 3 transformer primary connections to the left of Q3 Q4, and one unsolderd terminal.  In that case, if these three terminals all read pretty much zero ohms to each other I think we have a two FET output stage and a centre tapped primary in which case you have a push-pull output stage (see below).

Q3 and Q4 are the same.

Power supply has three transformers, but one really small transformer is situated (physically) between Q3 and Q4. If you see the green marks on the image you can see the pins to that transformer and connectivity to Q3 and Q4. The secondary side (4 pins total) of this transformer is connected to Q3 gate and Q4 gates and appears to controlthem. Removed both Q3 and Q4 and tested them with a fan (12V).

Got the CM6800s in the mail and was tracing (see below) circuitry...found an open 75ohm SMD that runs to the 6800 Isense terminal. Nothing in my scrap bin.  More waiting.
 
I will have to look at the various topologies mentioned, but I think push/pull is correct.

[dicky96]

Q3 and Q4 are the same.

Power supply has three transformers, but one really small transformer is situated (physically) between Q3 and Q4. If you see the green marks on the image you can see the pins to that transformer and connectivity to Q3 and Q4. The secondary side (4 pins total) of this transformer is connected to Q3 gate and Q4 gates and appears to controlthem. Removed both Q3 and Q4 and tested them with a fan (12V).

Got the CM6800s in the mail and was tracing (see below) circuitry...found an open 75ohm SMD that runs to the 6800 Isense terminal. Nothing in my scrap bin.  More waiting.
 
I will have to look at the various topologies mentioned, but I think push/pull is correct.

Hmm o/c resistor in current sense circuit.... well isn't that an interesting thing   Certainly could explain quite a lot seeing as both power FETs test as good. 

The very small transformer between Q3 and Q4 would be part of the gate drive circuit.  Most likely either
1. the source of one of the FETs is connected to the low potential (0V) side of the main SMPS transformer primary and the gate is driven direct by CM8400 and the other FET is on the hot side with Drain to main cap V+ and the gate is driven by a seperate winding on that little transformer thus isolating the CM8400 from the high voltage. 

Or 2. as you describe 4 pins on the secondary side, you have CM8400 driving a primary on the little transformer and each power FET gate drive is on a separate secondary.

You may well have gate drive (and maybe even the entire PSU) something like this schenmatic http://www.smpspowersupply.com/atx-power-supply.html   

This schematic I think is half bridge not push pull as there is no centre tap on main primary. I'm not an expert in any way on these dual FET drive PSU but that sounds right to me, I'll bet someone more knowledgable than me could confirm or refute that.

Anyway we do seem to be getting somewhere, happy fault finding my friend... and this I think is is a journey of discovery and education for both of us 

Rich

[OpenCircuit]

Got the new CM6800 installed and the resistor replaced. No change.