Switch Mode Power Supplies

[goldtop5]

Ok, I have a question. I have not much experience with these, in fact...none. But what is the best way to troubleshoot a switch mode power supply? Before we were having to ship them off to the manufacturer or an authorized repair center, but it's getting too costly. Are there any resources online that can give me a better insight on how these things work and how to fix them? Thanks!

[g1]

Some good links and pdf's in this thread:
http://music-electronics-forum.com/t19215/

and here:
http://music-electronics-forum.com/t23974/

[Jazz P Bass]

The first thing that comes to mind is 'Safety'.
There are two separate sides to the supply.
The Hot Side & the Output Side.

The Hot Side does not use a 'ground' as a voltage reference.
The reference is typically one side of the first filter capacitor.

The Output Side will usually use it's own 'zero volts' (which may or may not be at earth ground) as a reference for voltage measurements.

The two sides are coupled together through an optocoupler as a load signal to the switcher circuit.

SMPS have come a long way.
The ic manufacturers have come up with low powered devices that are all in one package.
The high powered units have separate ic's that do all of the brain work.

The biggest issue that I have is the 'protect' circuitry that a lot of manufacturers design into the supply.
Not being a trained EPowerE I have a tough time figuring out a lot of the protection circuits.

[km6xz]

Although safety has been rightly mentioned, it is essential that you understand how everything on the primary side of the small power transformer is HOT to wall mains. There are 100 ways of killing yourself or at least blowing up your test equipment if you do not know it in detail. Most manufacturers print a heavy ink line around that section and all the components within that line are on the primary side. Use an isolation transformer for any testing or measurements on the primary side of the SMPS. A completely floating, non-ground referenced meter can be used for voltage measurements.
SMPS is becoming the norm, soon, all mains operated gear will use them because they are so efficient, light, cool running and powerful. Many DC operated units are using them because multiple voltages can be derived from one battery supply voltage. Units that mix analog and digital circuits will usually have them because they can generate a number of isolated voltages that do not share the same ground returns so noise into analog signal grounds are less a problem. Even, especially, large high power guitar and bass amplifiers and PA amplifiers use them so getting to know them and equipping your workbench as soon as possible is a good idea. The concept is pretty simple but the first SMPS supplies were very complex so a lot of techs refused to work on them. They have become much easier to work on however thanks to extensive development work on controller IC, more effective designs and faster and faster high power SMD switching transistors, rectifier diodes. Older supplies were not quite as efficient as newer less complex designs and that lower efficiency meant more heat. That means older supplies are subject to a common simple problem of secondary side filter capacitors giving up due to external heat and heat generated from gradually increasing dielectric losses. The filter caps are having to filter the high frequency waveforms in the 250khz to 2Mhz range that is the output of the small power transformer. The diodes used for rectifying that high frequency AC have to be fast, the faster the better for lowering heat. It is not unusual for secondary filter capacitors are the only fault if the supply is not blowing fuses. You probably do not have the equipment to measure the speed of the rectifier diode so to be safe, assume if it is an older supply, the rectifier diodes might be deteriorating so replace them as a standard rebuilt when replacing filter caps.
If the symptom is blowing fuses it is usually something in the primary, hot, side. This is best tested passively with an ohm meter and unplugged. The AC input rectifier is easy to test conventionally since it is a normal high voltage rectifier that can handle the lower frequency mains voltage. If the rectifier is ok check the big switching transistors should be checked for shorts. If they are shorted, the low current gate drive path probably also suffered damage, including the controller IC. More modern versions manage to isolate the damage potential more. The reason for the damage is simple. The big heat sink mounted transistors are switching 330 volts DC at potentially a lot of current since only the fuse and rectifier are in the way to the ac power socket. That high voltage is usually labeled VBulk on schematics. The controller circuit generates a fast rise time high frequency pulse to the gates of the switching transistors to cause the switching transistors to "chop" the Bulk constant 330 volts into high frequency AC across the primary of the small power transformer. It can be small because the bulk of a transformer is determined by the frequency of the AC across the transformer. A 500 watt power transformer designed for 60hz mains is large with a lot of core iron. A transformer for 50hz is even heavier for the same power. A 500 watt transformer for 1Mhz is tiny in comparison. So a SMPS is much cheaper to build, the chassis can be much lighter weight, shipping costs much less, and customers save a lot on medical expenses not lugging 120 lb amps to gigs when the same power can be generated in a 12 lb amplifier that uses SMPS.
When repairing SMPS, the components were selected to be enough but no more for the design so do not freely substitute other component values unless you are very familiar with the engineering compromises acceptable for the circuit. Be mindful of the speed requirements of the secondary rectifiers, and temperature range of the filter caps, which should be 105C or better. Some cheap supplies use 85C.
When testing a SMPS, it is nice to isolate it from the load incase the feedback regulation is not working. If it not, the load can get some pretty high voltages across it. Build up a load bank of power resistors to simulate a load, at least for a few watts of dissipation, it does not have to be full power loads for testing.
They use a feedback style regulation that samples one of the secondary voltages and couples that optically to the controller IC that varies the pulse width of the switching primary to raise and lower the secondary voltage to match a reference voltage generated by either a zener diode(or small ic regulator IC) or within the self contained controller IC. That feedback regulation is very reliable is seldom the fault in newer designs but is often damaged in older designs if the primary switching transistors short.
If you have specific questions post them.

[goldtop5]

It seems there is a lot going on compared to a non-smps. Which means a lot more can go wrong. I've watched a few youtube vids on smps. I'm kind of wondering....will it be cost effective to work on these? There are lots of sensitive components, so even if I were to repair these, they have a good chance in coming back. Plus, I would lose time and money repairing them under our service warranty.

[km6xz]

Don't videos on Youtube, most of everything posted is bad practice or guesswork. Study how they work so they are not so mysterious and are more effectively diagnosed. The shop in the area which can repair them effectively will be the shop that eventually puts the others out of business. Everything is going to high efficiency power supplies, in some areas they will be required soon, like high efficiency fluorescents and LED lighting is mandated in a growing number of countries.
They are not complicated even though they have more parts than linear supplies. If repaired properly they should be reliable. For some models, replacements are so cheap, like computer supplies, that it does not make much sense to repair them at $30 each but power supplies for MI and pro audio gear is not going to be so cheap for a 300-500 watt supply because they are not made by the 10's of millions.
If a shop learns about them, and has the required test gear, they should have no problem making money by repairing them quickly.

[J M Fahey]

Yes, I know local shops which *only* repair SMPS and nothing else.

Average repair time 5 minutes, they "already know what's wrong before opening them" ... applied experience of course.

[Mick Bailey]

I had a run on commercial SMPS battery chargers from the construction industry. All the same make, same model, same failure; shorted switching MOSFET and blown PWM controller. It took longer to clean the circuit boards than to fix them. I uprated the MOSFET and never had one coming back.

Jestine Yong's book is a really good introduction to SMPS repair.

The principles are straightforward, though some supplies appear complex due to lots of ancilliary components.

[tedmich]

Being able to talk about (and fix) SMPS does raise a tech to mysterious shaman like status with old tube amp heads as well! Like any intrinsically HV circuit, it is easy for Mfg. to save a little money and create a design thats something less than long lived. Witness the older flat panel monitors which depended on a HV CCFL backlight and had so many fail due to the cap plague. Or the two SMPS versions of the really good Yamaha PS series power amps with the P3500S/P2500S units being totally bullet proof and portly with linear PS, while the SMPS powered P5000S/P7000S are a few kg lighter and often show up busted.

[T9c]

I found this link. You may have already been there, but for the thread here it is. Pretty basic, but good for SMPS newbies like myself.

Notes on the Troubleshooting and Repair of Small Switchmode Power Supplies