Any smps guru's in here?

[mr moose]

I'm trying to fix a power supply for a Behringer 32ch mixing console.  The PSU has a +5v with it's own gnd for the digital circuits, a 48V and a dual rail +/-15V for the audio circuits.  The feedback is taken from the +15v line.   The resistance on the output for the -15v rail is a solid 1.4Mohm, while the +15v rail takes quite a while to settle on 5Kohm.  I have compared capacitance and resistance measurements for every component on both 15v rails and they are all the same (including the transformer windings).   I did this on the board so I know the readings are not accurate to the component spec but they are the same for both sides meaning that if a component has failed then it's counter part on the other rail also failed.

 

When powered it only provides 2.8V on each 15V rail and nothing on anything else. 

 

Q1,  is it possible the output resistance measurements are just different due to the existence of the feedback network attached to the +15v rail?

q2,  should the cny17/3 opto isolater measure 1.1V drop in both directions on the LED side?  If not then I need to fix that first.

 

What I have done is checked the hot side and replaced the mosfet (because they are cheap and it had to come out for testing anyway).  I am getting a solid 340V DC at the mosfet and there are no shorts or open diodes etc.  The transformer is giving me continuity, no shorts between coils and 0.3 to 0.4ohms across each coil.

 

 

cheers guys.

 

[mariushm]

post pics of the board as detailed as possible.

 

check voltage on controller , see if theres a aux winding to feed the controller, check diodes on secondary side, the voltage reference if any

[Heats with Nvidia]

If nothing is obviously burned, its almost certainly a capacitor problem. If you can, check the ESR of the capacitors. If not, just replace them. Especially the first filter caps on the outputs.

[mr moose]

Pics probably won't help as my camera is shit and the board looks good.  The controller is not generating vcc,  it is definitely getting a good feed to HV.   Which is a shit because it is the only damned part on this whole device that element14 (farnell) can't get till next month at best.   Edit: and only sells in lots of 2500 , looks like I am creating a new account with a new supplier.

 

18 hours ago, Heats with Nvidia said:

If nothing is obviously burned, its almost certainly a capacitor problem. If you can, check the ESR of the capacitors. If not, just replace them. Especially the first filter caps on the outputs.

I did test them, but only in comparison to each other on the board.  How likely is it that they would all go at the same time?

 

I don't really want to pull components unless I run out of options.   Being a behringer product I just know the more I pull it apart the more likely I am to fuck it completely.

 

Having said that I have to admire the overall design of this product.  If I can't get this supply to work All I have to do is make an external PSU and the board will be up and running again.

[Heats with Nvidia]

3 hours ago, mr moose said:

I did test them, but only in comparison to each other on the board.  How likely is it that they would all go at the same time?

Did you test them for ESR, or something else? And the probability is not that low. In the vast majority of cases, when i had to repair SMPS, they either worked just fine after replacing the capacitors, or the problem was caused by dying capacitors and the controller and mosfet died because of them.

[mr moose]

34 minutes ago, Heats with Nvidia said:

Did you test them for ESR, or something else? And the probability is not that low. In the vast majority of cases, when i had to repair SMPS, they either worked just fine after replacing the capacitors, or the problem was caused by dying capacitors and the controller and mosfet died because of them.

Not for ESR, I just did a basic dmm resistance and cap test.    To test them properly I have to take them out.  Think I will now anyway as I am running out of options. 

 

Cheers.  

[mariushm]

Try to take some pictures, even if they're shitty.  Can't help if  I / we can't point out what could be wrong. 

 

Capacitors can be bad even if they look fine and measure OK, your multimeter doesn't measure at 100+ kHz and the properties of a capacitor change  at high switching frequencies.  Especially starter caps (small 10-47uF ones usually) don't even have enough electrolyte to swell or leak, they can just dry out. 

 

Also it's difficult to measure things in circuit because often capacitors are in parallel, and/or other components in parallel will affect measurements.

We may be able to spot things you don't or you're not aware they're issues like cold solder joints, cracks in components etc etc 

 

 

[Heats with Nvidia]

10 hours ago, mr moose said:

Not for ESR, I just did a basic dmm resistance and cap test.    To test them properly I have to take them out.  Think I will now anyway as I am running out of options. 

 

Cheers.  

The DMM cap test basically only tells you if they have a short orsome capacitance left. For capacitors in SMPS that is basically useless. They will measure OK, even if they don`t work at all in that application since their ESR got too high.

If you want, i can ship you a small bucket full of "mostly OK" capacitors that failed in such power supplies. The vast majority of them will measure totally fine with your DMM.

[mr moose]

6 hours ago, Heats with Nvidia said:

The DMM cap test basically only tells you if they have a short orsome capacitance left. For capacitors in SMPS that is basically useless. They will measure OK, even if they don`t work at all in that application since their ESR got too high.

If you want, i can ship you a small bucket full of "mostly OK" capacitors that failed in such power supplies. The vast majority of them will measure totally fine with your DMM.

 

16 hours ago, mariushm said:

Try to take some pictures, even if they're shitty.  Can't help if  I / we can't point out what could be wrong. 

 

Capacitors can be bad even if they look fine and measure OK, your multimeter doesn't measure at 100+ kHz and the properties of a capacitor change  at high switching frequencies.  Especially starter caps (small 10-47uF ones usually) don't even have enough electrolyte to swell or leak, they can just dry out. 

 

Also it's difficult to measure things in circuit because often capacitors are in parallel, and/or other components in parallel will affect measurements.

We may be able to spot things you don't or you're not aware they're issues like cold solder joints, cracks in components etc etc 

 

 

 

It's all good, I believe you.  I will take them out and try a few new ones.   Although this PSU is only 4 years old it has jamicon caps,  so even if they hadn't failed they are probably going to go in the next few years anyway.     

 

Funny thing is why companies like behringer use 105C rated caps but cheap out on quality.

 

 

 

 

[mariushm]

Jamicon is not that bad, it just depends on where the capacitors are used in the circuit, if they're not abused they can last a long time.

 

The thing with power supplies is that companies often don't make them in house, they just go to an OEM that makes power supplies... companies like FSP (Fortron), Delta, Chicony, Meanwell ... this way they don't have to pay for all kinds of certifications and tests that need to be done for various countries... it's cheaper.  That's the same reason why lots of products simply come with a wallwart adapter or a laptop style brick power supply - these can be made by oem companies that handle all the certifications and safety tests.

 

It's possible Behringer made a contract and bought let's say 100k power supplies and the contract specified a model or minimum performance levels (efficiency, minimum warranty, max failure rate etc) but maybe it didn't go into specifics like "only these 2-3 brands and series of capacitors are permitted into this part of the circuit" and the OEM may have switched to second option capacitors to either save money or because original part couldn't be sourced - it's normal to have alternate parts just in case one is out of stock or has too long delivery time.

[mr moose]

3 hours ago, mariushm said:

Jamicon is not that bad, it just depends on where the capacitors are used in the circuit, if they're not abused they can last a long time.

 

The thing with power supplies is that companies often don't make them in house, they just go to an OEM that makes power supplies... companies like FSP (Fortron), Delta, Chicony, Meanwell ... this way they don't have to pay for all kinds of certifications and tests that need to be done for various countries... it's cheaper.  That's the same reason why lots of products simply come with a wallwart adapter or a laptop style brick power supply - these can be made by oem companies that handle all the certifications and safety tests.

 

It's possible Behringer made a contract and bought let's say 100k power supplies and the contract specified a model or minimum performance levels (efficiency, minimum warranty, max failure rate etc) but maybe it didn't go into specifics like "only these 2-3 brands and series of capacitors are permitted into this part of the circuit" and the OEM may have switched to second option capacitors to either save money or because original part couldn't be sourced - it's normal to have alternate parts just in case one is out of stock or has too long delivery time.

Your last paragraph is most likely the truth.  Given Behringer are close the the cheapest for everything I'd be betting a pretty penny that if they did outsource the PSU that they screwed the price down so far the contractor would be looking to cut costs.   Having said that, it's not outside of Behringers MO to simply take an existing product and copy it completely.  Either way there generally isn't enough meat in the end sale price to fill me with confidence that they have not cut any corners.

 

 

[Heats with Nvidia]

Four years is enough for Behringer. You are out of every warranty period. Also yes, Jamicon caps are good, but in most cases these capacitors have the lowest capacitance possible for this application, so they need to be good to at least survive the warranty period. Even some more expensive manufacturers do it that way. Also, Jamicon also has some cheap capacitor series and it might be one of those.

 

When you replace them, please use 105°C capacitors of the 5000...10000h class and with a low ESR, so they at least have a chance of lasting more than those four years. The most critical ones and the most prone to fail ones, are the first capacitors behind the rectifier on the secondary side.

[mr moose]

I looked up the jamicon parts and apparently they were 10,000 hour rated at 105.    Given this unit has not likely even done 4000 hours in static temp controlled environment, I am going to assume something at least is less quality than it could have been.  Especially if several have gone at once.

 

But I won't know that till I pull them out and test them properly.   I have ordered rubycon and panasonic all with 105 ratings and the longest hour ratings I could source. I think the lowest one is 7000. 

 

Thanks for the help guys.