SMPS troubleshooting made easy

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Svart

Well-known member
Joined
Jun 4, 2004
Messages
5,134
Location
Atlanta GA USA
OK, the SwitchMode Power Supply only has a few bad reputations for noise, regulation, failure, cost, and general complexity but we can look beyond those trivial reps for it's good points...

When someone thinks of some, let me know.


:shock:



:green:

Ok so I'm only partially kidding( or partially serious), these little devices are handy to use and easy to upgrade and repair if you know the basics of electronics.

I mentioned knowing the basics of electronics.. These devices are full of voodoo, not quite digital, not quite analog, but full of interesting but rather easy to understand concepts that span the whole mixed signal world. I'll go over some background so that you understand where I am coming from when I talk about a specific item, but more on that later.

I know you've read how these things work 100 times but I'll do it once more.

How it works:

The run-of-the-mill SMPS takes in AC, rectfies it to a higher DC voltage via bridge and cap then feeds it to a switching device, usually a N-channel MOSFET which chops it at a high frequency into the primary windings of a transformer. There is some voodoo here, between the interaction of the high frequency signal and the inductance/reactance of the HV transformer but we can go into that later, we are here to fix stuff right now! The pulse widths of the chop determine the average current/voltage coming out of the transformer secondaries. From here it's rectified and sometimes regulated via active and/or passive methods. The chop is simply determined by feedback from the output to the MOSFET driver circuit. Too much voltage and the driver IC makes the pulses more narrow to reduce the average voltage, too little and the opposite happens. Simple eh?

For those of you without humor, please scroll to the end for a short list of items to test/replace. For the rest of you who want to learn a little bit along the way:

Ok let's fix the damn thing already.. With a little humor to boot.

1. No input = no output.

A lot of folks tear into their gear and start taking things apart without checking the basics. Check your fuse. Ohm it out. Sometimes they look ok but they are still broken. Check your cable, sometimes someone ran over your cable with something heavy and it finally broke internally. It's a longshot but it does happen. Many people have wasted a lot of time troubleshooting a powersupply issue only to find out that their power source was taking the dirt nap.

2. Switches are great, you can bring home someone that you would normally need 6 beers to find attractive after only 2 beers + turning the lights off. This saves you money on beer and electricity! However for manufacturers the cost of a "good" switch is often too high for their budgets and they end up using something cheap. This can lead to disaster if your device uses cheap power switches that end up arcing and burning out.

Check the power switch.

3. Bridges are great. The let you drive over rivers and through the woods to grandmother's house you go. What does this have to do with the SMPS? Not a thing.

Check your bridge rectifier. And while you are at it, check the big bulk cap after it. If it looks puffy, change it. If it reads high ESR, replace it. If it's missing but the little legs are still in the holes.. yup. replace it. Above all else, you should probably replace it. Just make sure that the replacement is rated the same or better. I usually increase the uF value by a couple hundred more. If the cap was shorted or exploded, expect that the bridge is dead too.

4. I can't think of anything funny about MOSFETs.

This is one of the parts of the SMPS that die a lot. They are hard to test in circuit because of their gate biasing and other attributes and usually need to be taken off the board to test. Sometimes there are more than one as well. Using the diode test on your DMM, you should NOT see any type of continuance between the gate and any other lead. If you do then the gate oxide is blown and the part is dead, however you may see a diode between the other two legs which is usually normal if the diode only goes one way. Replace this guy with something of higher current and voltage ratings, making sure that they have the same pinout as the orignal. Also be aware of the insulation of the part. A lot of MOSFETs have plastic tabs which mean that they are isolated, most if not all of the metal tabbed versions are NOT isolated. In case you don't know if your FET is isolated, it's best if you go ahead and isolate it.

5. The High Voltage transformer.

I have NEVER seen one of these die. ever. really. never.

6. Transformer output diode. Not much funny here either.

Never seen one of these go bad either, but who knows. First you think that it's a good little diode and it will never give you any trouble. The next thing you know the cops are there asking you questions and you find out it was really a cannibal and kept human corpses in your freezer. Anyway, it's actually usually a dual diode, common anode, that rectifies that bastardized pulsating square wave coming out of the HV tranny.

7. Output filtering.

The manufacturers tend to cheap out on certain areas of the design and this is definately one of the areas. Change the 'lytic caps out for higher temp, higher uF value and same or higher voltage values. Just be careful of the size of the caps, replacements of the same size can differ wildly from model to model and from manufacturer to manufacturer. I recommend the Nichicon HE series, they are usually the smallest in diameter and have nice low ESR.

8. Voltage regulators.

Some SMPS use regular old linear regulators for their lower output voltages that don't need a lot of current. This is a topic that we have beat into the ground.. much like how Britney Spears looks these days.

9. Feedback. Something you get when you eat too much and get on a rollercoaster.. you get your feed...wait for it.. back! Ahem..

This is the start of the really fun, interesting, and meaningful part of the SMPS.. Ok I'm lying. This is where the confusion starts to set in. Nine times out of ten a shunt regulator is fed from the highest voltage rail of the system. This creates a lower reference voltage that feeds back to the PWM controller IC. Sometimes this is simply an ON/OFF signal, sometimes this is a signal that tracks the output voltage. Either way, the PWM IC adjusts accordingly and changes the pulse widths of the MOSFET. Feedback is great especially if your wiring shorts somewhere in your system, it will usually shut the PWM down completely and thus shut the whole SMPS down and avoid letting the factory smoke out of your expensive junk. I would actually start troubleshooting the feedback circuit by testing the output of the SMPS. Make sure that your + rails are not shorted to the -/ground rail, which is one of the most common non-failure problems and is usually found in the wiring and/or on the board that the SMPS is feeding.


10. The PWM IC. The heart of the system. This IC does all the thinking for you, however it is not the one who thought they could make it to work, knowing they had diarrhea, only to end up getting stuck in traffic and praying to a higher being that you don't ruin those 50$ silk boxershorts.. It doesn't do that kind of thinking for you. However it does do a mighty fine job translating a linear voltage level into related pulses.

It takes in a voltage signal and adjusts it's output signal accordingly. Amazingly enough these parts are usually less than .25$. In cheap designs, the FET is driven DIRECTLY by the PWM IC. If your MOSFET has a blown gate, expect this part to be bad as well. I have yet to see one handle the 150-400vdc that the FET is exposed to and live to tell about it. In expensive units you will likely see some kind of optoisolator between the PWM IC and the MOSFET. While they protect a .25$ part, expect the 2$ optoisolator to be bad unless you have an O'scope and check it out for a nice square wave. In REALLY well designed units this part might be a Photovoltaic gate driver which needs no power on the secondary side to drive the MOSFET, it generates it's own voltage exactly how solar panels do. The downside is that they don't slew fast and they don't drive with a lot of current. The PWM IC is usually powered by a resistor/zener regulator from the rectified line voltage, just after the bridge. Sometimes this is done for the period during startup until the rails stabilize and take over, sometimes it's always powered from the rectified line. A small army of resistors and caps could also be fried around the PWM IC depending on how the HV decided to attempt to get to ground so beware of those and look for burnt parts.

And that is it for the general overview of the SMPS. It's actually a lot simpler than you would think until my totally excellent and simplistic explaination made it clear for you.

If not, here are the things to check in a nutshell:

1. Line input AC
2. DC out of the bridge rectifier pre transformer
3. DC out of the rectifier post transformer

Things that will likely need to be replaced:

1. PWM IC
2. MOSFET
3. Bulk cap after bridge(pre transformer)
4. Output caps


You'll likely be able to fix 8 out of 10 SMPS by just changing those parts alone if you are the gambling type.
 
I would like to add

If the unit does not power up the output side be careful the Bulk cap(s) can hold a charge and burn a mark in a screwdriver ring etc..

In some of the cheaper small switchers ( like in external hard drive boxes) I have found the bulk cap and transformer leads to have broken solder, Resolder the connections and units worked fine under a active load bank and scope. I guess from movement because of the mass of the parts or just bad soldering?
 
Good info.

I try to stay away from switch mode supplys if i can help it.......

I got one on my bench right now that gives out a loud click at 500ms intervall........its not just that i can see the DC bouncing on the scope......but its actually making a sound.

Its a beatbox.
 
Gus, good info, I have actually seen the solder failure a lot too but it slipped my mind. There are two causes for that, usually the powersupply is too hot inside since switchers are made to be extrememly small and practically shoehorned into tiny little plastic boxes without any airflow. The solder stresses from the heat/cool cycles and eventually fatigues and cracks. It can also be as you state, the heavy transformer soldered to a thin PCB with inadequate support that separates after being bounced around a lot. Gus is again dead on when he mentions those large caps. Well built supplies usually have a bleeder resistor to discharge that cap but the cheap ones tend to leave that off because it's "not necessary" and only adds cost.

Discharging a cap of (charged to)300v/400uf will dump 300A of instantanious current into a 1 ohm load @ 120kuC. That's a pretty good charge and it WILL burn a hole into your skin. I've seen them burn holes through sheet metal too.
 
The first SMPS I worked on was a crusty old tube car radio. that one used a vibrator. Nowdays I work on Behringer, and a common problem I see is one of the rectifier diodes after the transformer shorts, usually the diode that feeds back.
 
Funny this thread would pop up now. I just finished repairing a Crate Powerblock. It uses a SMPS to power a class D amp and the PWM and MOSFET are combined in the same IC, a KA5M0965Q. It would power up, turn off, power up, turn off, etc. cycling about once a second. Since I didn’t have a KA5M0965Q lying around, I tried changing the 4.7/50 electrolytic at the “soft start” pin first. It was a bit deformed (bulging out at the bottom) and I thought “Ah ha! This has to be it!” But turned out to be the KA5M0965Q instead. :sad: I ordered a free sample from Newark and finally got the thing back together and working. I had $25 in the Powerblock and the part was free, but that thing was a BEAR to take apart and put back together, so I don't know if it was worth it or not. :?
 
great post, Svart    ;D

fun to read, currently, I'm learning about trouble-shooting of PSU/computer.
 
Great post from Svart and great thread. Thank you so much

I would like to ask you guys, what to do when the smps is emmiting an acoustic High Pitch noise?
I have that in a lot of smps, normally is around 10Khz and it's really really anoying.

In all the smps that I have that acoustic noise it comes from one the transformers
 
Mechanical noise from transformers is really not something that can be fixed, I think.

Back in the days of CRT displays, this could be a huge problem in the studio - only thing to do was to carefully selecting by listening to many displays..  :eek:

Jakob E.
 
Potting can help. As well as isolating the transformer or inductor mechanically from the pcb or chassis (rubber, ...).
 
Greetings,

Svart, thanks much for the tutorial! I have a dead switcher in the broke stuff pile from a PC that I swapped out. Was going to trash it, but I'll use your info here to learn if I can fix it. Good work!

Regards,  Jim
 
gyraf said:
Mechanical noise from transformers is really not something that can be fixed, I think.

Well there needs to be a solution because it's a really huge problem with Switch Mode Power Supplies.
Being it the Cheaper walwart SMPS or modules inside units, it's pretty common to have an irritating mechanical sound at around 10khz coming from the first transformer (probably a choke)

I've had and have this in many devices, it's common to go to someone's house and I notice a 10khz frequency coming from somewhere, normally is from the Router/Modem PSU

I've opened some of the problematic ones and took the choke out and put it in varnish. It solved some of them, others attenuated the noise, half of them are still the same.

I'm working an a TC Electronic M2000 PSU that the 10hkz (more or less) mechanical noise is just unbearable


 
volker said:
Potting can help. As well as isolating the transformer or inductor mechanically from the pcb or chassis (rubber, ...).

Thank you so much Volker, I will try isolating it with rubber. Good idea

As  for the Potting, what do you advice to use to pot a small choke or transformer?
 
I've used standard two-part epoxy a couple of times because I have it around anyway for other things. But it's nasty stuff and requires a bit of delicate handling, you wouldn't want to do that on your kitchen table. There is also beeswax and paraffin wax,  the latter which I've used once for potting a guitar pickup I wound. Both have a relatively low melting point, so might not be best for your application in a power supply. I'm sure there are more specialized materials available for the transformer winders (all sorts of epoxies, acryl based, silicone based, ...), but I am not a magnetics guy. Maybe CJ knows more.
 
Thanks Volker

I think Epoxy might work fine.
I will try that

Parafin and beeswax I used that in guitar pickups to tame down microphonics but in this application the transformer or coils vibrate so fast that I don't thin the soft parafin or beeswax will be able to block that vibration.
There's also the melting factor of those materials and it gets hot inside PSUs

I will try a combination of drowning the small transformer in varnish, leak it soak up an dry
Then use Epoxy to pot the varnished transformer and then use rubber to de couple it from PCB and Chassis

I will try on diferentes PSU and let you know the results
 
Looks like MG makes a few epoxies

https://www.mgchemicals.com/products/potting-compounds/epoxy-potting-compounds/



I wonder if dipping  in that Plasti Dip stuff would work for damping....it's pretty tough stuff..... solvent base though I think
 
Back
Top