Repairing switch mode power supplies

[Alex R]

OK I had a few problems with SMPSs... like repairs that worked for a bit coming back to be blown again, difficulty sourcing ferrite transformers, etc. But more and more PA and bass gear has SMPSs because of the weight thing. So now I just replace the board if that's possible. With Markbass it's ok, they supply the board and though it isn't cheap the repair is still economical. With the Soundcraft Gigrac it is a different problem...

The Gigrac is quite popular, being portable and powerful. I have now seen three of these with bad power supplies. Soundcraft charge over £300 for the power supply board.

OK so my customer found a firm up north who specialise in power supply repairs - uninterruptibles, SMPS, mostly IT power units I think. But they fixed the Gigrac ok, component-level, and it was a lot less than £300. We will wait and see whether it pops again.

So my questions are:

What special equipment might those guys be using? Is it something I might be able to do, fixing SMPS somehow better than I do at the moment (when I tried to fix them I did it the usual way by testing/replacing shorted mosfets and other stuff that might be strained, etc)?

Is there any point in bringing a SMPS up on a variac, watching the current, in the usual way?

And what are your views on this trend, ie using SMPS to run musical amplifiers with the large transients/sudden load changes they see, the tendency for them to be physically maltreated, given involuntary beer shampoos, etc etc?

[Alex R]

...oh yes and where the hell do you get those transformers :-)

[teemuk]

Is there any point in bringing a SMPS up on a variac, watching the current, in the usual way?

It usually won't work; typical SMPS's have driver chips that often want to see a DC voltage level that is above certain threshold before the switching is turned operational. This to prevent running the SMPS at low mains voltages, something putting a variac there essentially is.

So, reducing voltage with a variac or a light bulb current limiter usually turns the SMPS non-operational below certain input voltage levels, thus using such devices is kinda moot. On top of that, reputedly weaker designs may even fail when such things interfere with pre-designed operation of the SMPS.

[Amp Kat]

You need an isolation tranny to keep from getting shocked so you can troubleshoot it easier. Usually the problems are either the mosfets are the transformers. Heres a time when you miss Enzo because he is pretty good with those and sent me some supplies to tinker with but I never took the time to learn them but I do know the ISO will help.

[teemuk]

Agreed on the isolation transformer. Those things are screaming hazard for newbs: Almost without an exception you'll find a rectifier circuit that is tapped directly to mains. In here where we use 240V mains you commonly also find a high capacitance 400VDC filter capacitor to keep the DC at steady level, and it helds its approximately 340V charge for looooooonng. The common rails also tend to be more or less "floating" so sometimes its quite a lot of searching game before DC levels all over the circuit can be accurately verified and bled off after powering off the unit.

It doesn't help that usually the boards have a thick lacquer coating on both sides that effectively prevents reliable probing. You have to be really violent to penetrate that coating and often in such cases you suddenly notice that the 400VDC cap you thought was not charged (probe read 0VDC) WAS charged after all but your probes just didn't read it through all the lacquer during few first tries to probe stuff.

I usually refuse to touch those things without wearing rubber gloves because I hate having "snake bites" all over my fingers when even after tedious cap bleeding procedures random caps give you nasty jolts.

[Enzo]

I don't mind working on SMPS, I was seeing them in arcade equipment 30 years ago.

No variacs, sorry. And yes, iso always. The "ground" - common really - of the switching circuit is often if not usually the -170vDC rail. Sure don't want to clip your scope ground clip to that.

To me the SMPS is two parts, the drive side, and the secondary side. Many times blown up drive circuits are caused by problems on the secondary. In fact when I go through one, the first thing I check is all the rectifiers on the secondary. While there is usually a relatively simple switching system - transistors switching current through a transformer primary - there are also usually extra parts to snub things, so always make sure you don;t have any open diodes or transient suppressors.

I don't recall ever seeing the main transformer bad. And the ferrite hoops are used as inductors filtering the secondary, I don't think of them as transformers.. Oh the incoming mains often runs through common mode filters too.

And don't always blame the SMPS. I find SMPS in those Crate amps with blown rectifiers in the B+ supply. ALmost always caused by bad 6L6 tubes. No fault of the SMPS. An amp that constantly blows up its SMPS could well have a non-functioning curent limiting in the amplpifier circuit.

[km6xz]

Ditto....Transformers used to be a problem with the first ones decades ago but never have seen a bad one in years.
The secondary side is pretty conventional and can be referenced to ground normally but check first. Anything that impacts speed of switching is suspect, when a ultrafast diode is spec'd, do not use a general silicon diode, loss and high temps in the switching transistors is usually due to switching transition timing.
A scope with the proper probe and input amp can make troubleshooting a breeze. A true differential input plug-in if using a lab type scope, they are so cheap now industrial surplus there is no excuse for not having a Tek 7400 series scope for things like this. The differential measurements can disregard grounds(up to some specified maximum potential) so work on the primary side is safe. Before taking any measurement in a switcher, figure what should be there, and how the probing would impact the circuit. A probe of less than 10meg input can cause problems. Forget the variac as others have said, unless testing a operational supply, the voltage range spec'd for it should be checked, dropping out or pulling excess current at the minimum rated mains voltage means you are not done with the repair. If it says 60-250, see if it overheats at 60, usually it won't at 250, in fact it will run cooler. Don't try under the rated mains input, it will pull too much current or shut down protectively.
If you are going to do a lot of them, get set up and get a supply of the original parts and specialize in it, most shops dealing with pro audio are afraid of them and you can build up a wholesale department.
Back in the ADAT era, the supplies were a frequent problem and few shops, nor the factory even tried to repair them so I set up one bench just to do them and rebuilt thousands of them at a flat rate of $50 for Alesis and their service centers. Getting the original parts directly from China, in 500 lots, meant a complete rebuild would be about $6 in parts cost and 20-30minutes labor. We did the same with main processor boards for a few companies. Once you build up a proper specialized bench and a few jigs, specializing in something can be very profitable.
A nice thing about working on a lot of anything is building up a symptom catalog, and expected measured performance. On your own load bank, a quick measurement of mains draw, cut out level, and secondary regulation, you can tell in 5 seconds whether it is working. These are pretty much a go/no-go test, not like a tube amp were everything is a sliding scale of proportions. If it is rated at +/- 15volts at 5amps, it is broken if it can only source 4amps. There is nothing ambiguous.
That is what I always liked about RF systems, the performance tests were enough to unambiguously tell you if it needs work. If rated at 150watts into a 50 ohm load, or a 20 db SINAD at 0.2 microvolts, and it does not do it, something is wrong. Everything in the RF system needs to be working optimally to produce a very reproducible parameter. Audio systems are more nuanced and shades of grey, with more points of user control(eq, gain, loads, side chains, etc). Every amp on my bench would get a few basic quick measurements, such as mains current at idle. After a while the log of those will give you a good clue if one is not within your historic record tolerance. Same with distortion, a lot can be inferred by a simple 5 second single freq distortion measurement at some specific power level, a spectrum display is even more suggestive than a THD measurement, but a single THD figure in the log is easier to compare with the historic record when starting and ending a repair.

[Steve Conner]

I was trying to think of something to say, but looks like Enzo and Stan have covered everything.

A decent-sized modern SMPS has more like three stages. Active PFC, primary driver and the secondary stuff.

Bringing them up on a variac does kind of work. They'll do nothing at first, but should sputter into life when the voltage gets above the minimum rated. And since you do this test with the output unloaded, the draw from the mains should be negligible. If the ammeter pegs at this point, you know you have a shorted diode on the secondary side or similar, turn off quick. But many SMPS chips nowadays will protect themselves against shorted rectifiers: they just shut down.

For really hardcore investigations, you can run the controller chip off a bench power supply and the DC bus off a really big bench power supply. That's what the SMPS designers do in their labs.

If it pegs from the start, even before the switcher has started switching, you know the fault is on the primary side: a blown switching FET or primary rectifier. Usually the onboard fuse will already have failed violently in this case.

You need an isolation transformer. If you've ever worked on the old live chassis TVs, this is no different. You plug the SMPS into it and ground the negative rail of the DC bus, then you can happily scope things on the primary side.

The transformers and inductors don't burn out. I've never seen a bad one either. It's the diodes, capacitors and power FETs that go. The first port of call when fixing one is to test all the power semiconductors. Any bad ones are usually obviously shorted, if not blown to smithereens.

[olddawg]

Like others I repaired switching supplies to component level for 25 years almost daily for a multitude of different manufacturers and applications. I would say that before you dive into them really understand how they work. You can't effectively troubleshoot without that. Modular supplies are usually the easiest to repair. Most will run free standing. If the smps is just a corner of a main board it's trickier. Over the years the most numerous failure I saw was bad caps drying out. Now that they use a lot of SMD components on paper/resin boards, the boards can become conductive from heat. Many companies used to offer rebuild kits with all of the necessary components for the modular supplies. Don't assume the odd 500K or 1meg resistor isn't open. If you do it right you will test the modulars with a load jig.

[Alex R]

That's cool, thank you guys. The iso tip makes a whole lot of sense! - I habitually use one, running through a variac and an analogue ammeter so I can watch the current, but the response of the first SMPS I tried to fix was so jumpy when I brought it up on the ammeter I kind of gave up. I know how they work in principle, thanks to a conversation on here a few years ago, but I wasn't understanding how the control chip needs a certain supply voltage to function. It is useful to know they won't draw much at all with no load, that's something to watch for.

So what do the driver chips do - time the mosfets?

[Euthymia]

And what are your views on this trend, ie using SMPS to run musical amplifiers with the large transients/sudden load changes they see, the tendency for them to be physically maltreated, given involuntary beer shampoos, etc etc?

I can't speak to the inherent reliability; on the one hand we do have millions of computers running on the things, on the other hand, the power supply in a PC is a reasonably inexpensive module that is tossed in the bin if it ever goes out.

I have enough clients shaking their heads and telling me that it's getting more and more difficult to find someone to fix their tube amps (even more so the vintage solid state amps I repair), so I can guess that finding someone who both works on audio gear and will troubleshoot switching power supplies to a component level would be next to impossible.

So far I have only had one amp in that needed its switcher fixed, and I was able to fix it, but I also put that type of amp on my "no fly" list of things I must refer the owners of to the company that produced them.

Now if the company made the power supply a separate PCB that could be replaced for a reasonable cost, then it might be worth the time, but the amp I worked on had everything on one big board. Sorry, "call the factory and see who they recommend."

[Mars Amp Repair]

Well I guess I can add my 2 cents, too.
We have only delved into the smaller/simpler switches like the one in the Crate V series mentioned above. Ones of this caliber don't frighten us too much. for one, there is a fairly small number of components to check in the first place. As mentioned by many above it's usually the main switching FET and some of its supportive circuitry that has either failed 'just because' or a result of a leaky/shorted rect diode in the secondary. Agreeing with Enzo, the biggest cause of demise for these is the 400+ volt jolt they get as a result of shorting/arcing power tubes. Not too many semis in those little switchers are rated for that. It actually surprises me that they didn't design in some protection for that eventuality. Of course you'll notice Crate dropped the idea pretty quick & was basically giving them away for ridiculously cheap.

As mentioned, we don't try to sub any of the diodes or components as speed is paramount to keep the FET out of it's active range. We've been able to find the exact replacements thanx to Mouser and places like Digikey.

I built an isolation transformer jig out of a toroidal transformer from this huge power amp that just happens to have 117volt secondaries. Didn't cost me anything & works fine. I do need to outfit my benches with line ammeters. Up til now we have just used the limiter bulbs on non-smps amps & such.

That is some good advice on starting the switchers up at the minimum line voltage they are rated at...if you can find that data.

The SMPS's that seem daunting & basically unapproachable are the super elaborate ones like you see in the Line 6 Vetta. They're chuck full of supportive diodes & transistors as well as little solder in circuit boards. What in the world do you do with something that complicated? I know in the Case of the Vetta, Line 6 has stopped supporting them & usually you end up with a switcher that also takes out the power amp board with all its smd components. the amp really isn't worth repairing with that much wrong with it as well as not knowing if the main processing has been damaged until you get the power supply working. BUT,

in the case of the really complicated switchers like the MarkBass aforementioned with all its supportive circuitry & highly silicone gooped electrolytics?!! I'm thinking as long as the amp sells for $1200.00 & they supply the board for a couple hundred bucks or so, that's the best way to go for now.

We didn't know how good we had it in the days when VCR's came out & the manufacturers went from city to city holding invaluable seminars teaching us soooo much about digital circuits, processors & servos & all that you needed to know to be a great tech on VCR's & camera's & such (esp when a lot of the controller circuits were discrete). Once vcr's & cd players all started coming out in blister packs, that was pretty much the end of that. Glad I was around to glean as much as I did during that era. Ahhh how I loved the complex electro-mechanical married to the electronic controllers of the earlier video recorder days. How much fun was that! But I digress...

So has anyone been successful repairing the switchers in the higher powered P.A. power amps such as you find in the Crowns? I've just been steering away from them. and then there's the wonderful 'D' class. Thanx, glen

[km6xz]

Switching supplies obviously scare a lot of people but to remain in or to enter into the business of repairing audio gear, one is going to have to accept that they are not going away. More and more people who want to be in the business or have been, are getting backed into a smaller and smaller corner of specialty work that is also populated by a lot of people in the same boat. There are dozens of people who will work on tube guitar amps in probably any town, most refuse to work on any modern electronics. Some very common gear is hard to get repaired for lack of any nearby shops. How many really competent keyboard shops are there? Or the computer based recording gear, both of which sells at very high levels but fewer and fewer repair options are available.
I've been out of the active commercial activity of repair for several years when I gave my shop to my employees and moved here. At that time there was about 1/5 th of the competition for the bulk of new product repair in my area(San Francisco Bay Area) as when I opened the shop in the early 90s. There were at the same time of my leaving, about a 10 times as many one-man shops only working on tube guitar amps. For a digital tape deck, keyboard component level repair, or anything with dense surface mount by 2004 for all of California really had only a dozen or so shops and 3-4 which had the competency to do it on a consistent level.
That means, for a bright ambitious young person who really wants to make a go of a real repair shop, the opportunities are greater now than ever....if he avoids the crowded corner and accepts that all the newest, as well as older, technologies will be the core of his business. The largest available customer base is being under served, while the specialty of tube guitar amps is so overcrowded with techs who are more electro-mechanics, so there is an oversupply and shrinking customer pool.
If a kid in school or before college, prepares himself by devouring every book on things like switching power supplies and processors, and becomes comfortable with working with them, and outfits his bench with the needed gear, in most cases will be the only game in town...or in some cases, the only game in the state.
Power amp design is going through a revolution now and high power, low weight high efficiency designs are only going to snag more and more of the new product development. This will further push the mechanics who never kept up, into a small corner.

[Alex R]

Agreed - though here in the UK there aren't too many people doing tube amp repairs either. I think it is a bad sign if someone simply refuses to look at solid-state - it's not that hard and there's a lot of work out there. Switch-mode and surface-mount are just a few steps further. But if manufacturers stop charging so much for their mass-produced gear, we might get moved into the kind of uneconomical-to-repair area that has largely shut down the TV/VCR/DVD player repair market.

Well, I have accumulated a few bits of gear with blown SMPS shipped to me and abandoned - if I get a bit of free time I'll practice on them. Have to move with the times!

[Mars Amp Repair]

Yep Alex,
The stories are beginning to all sound the same.

As for keyboards, one of the largest keyboard servicers here in the Denver area lost their tech & have not been able to find a tech to replace him, so they have been referring all their keyboard work to us as there pretty much isn't anyone else around to do it. There is one guy who runs a mobile keyboard repair business, but he's out of commission for a while due to surgery and he's just one guy. We're swamped with keyboard work.

We are the largest & busiest servicer for music electronics in the Denver area & our biggest issue is merely finding qualified techs to do the work. We presently are about 3-4weeks just getting to items coming in our doors today.

Now I know that historically businesses thought this is some kind of great guaranty to have a back-log like this, but we haven't slowed down since Summer last year. Personally, I look at it as not so great customer service. I'd love to be able tell customer's 1 - 1 1/2 weeks max.

Our biggest issue is finding more qualified techs to do the work. We have a great core of 4 techs, but you can only get just so much production out of that number.

As for hiring addl techs, out of the resume's we receive when we run an ad we pretty much get laid off rocket scientists (much empathy for them and their families) or wanna-bees who have soldered some connections on their equipment but don't have even close to the basic electronics knowledge/skills necessary to support the training involved to get them up to speed. Then there's the need for the tech to have a clue about talking musician talk with the knowledgeable musician customers.

It's a dilemma. We do O.J.T. of course, but a tech still needs the basic electronics knowledge as a starting point as well as the desire [sometimes I think 'sick' desire ;-] ] to do this kind of work.

I guess it's no secret why the Vocational schools like DeVry & all stopped gearing their programs for the repair field. It was mostly geared for the consumer field which as prior posts indicate just dried up when VCR's & Camcorders became blister-pack items.

ANYONE KNOW ANY TECHS INTERESTED IN LIVING IN COLORFUL COLORADO & WORKING IN THIS FIELD? ;-] glen