amp troubleshooting... the dreaded volt or two offset condition

[guitardad]

Some of the most difficult amp conditions I have had to diagnose are always when you have just a few volts offset. Rails are good. Most of the time its several leaky or open semi's, but in my years of troubleshooting and repairing these amps, seems that condition is the most challenging than say full rail offset or definitive shorts.

I wonder if anyone can share their methods of attacking these type circuit conditions. Seems in the past several months most of the ss amps were slight offset issues. From Yorkville box mixers to rack power amps.

Might even help some others like myself reading posts here to have better starting points. I do mostly tube amps so ss can be a pita when i see this kind of condition.

[Mick Bailey]

When I'm getting a few volts offset I look first to see if the amp isn't biasing correctly. Sometimes it will adjust out if there's provision in the circuit and other times it can be the result of a diode biasing string that's shifted over the years and this can go either way - slight crossover or slight offset. Often amps have been run with an offset for years, completely unnoticed by the owner until something happens to cause heavy distortion or no output. If the amp has a capacitor coupled output this can leak DC and is an easy fix, but they'll show a high DC offset with no load connected.

I find that generally offset needs to be measured into a load. With a 10M multimeter sometimes you can get a phantom reading that disappears once the load is connected.

[Steve Conner]

In a properly working solid-state amp, offset and bias are completely separate issues. The global feedback loop should hold the offset constant irrespective of the bias, and a well-designed bias string should hold the bias constant irrespective of the offset. Even a poorly designed bias string should manage fine for small offsets.

The global feedback loop will compensate for any DC offsets inside it. Problems inside the feedback loop will cause distortion before they cause excessive offset. So when troubleshooting offset, look outside the feedback loop. The input stage and its DC block capacitors would be the first place to look. The input stage is usually a differential pair of transistors that should be matched.

Some amps have two separate feedback loops, one to manage distortion at audio frequencies and another that functions as a DC servo to bring the offset to zero. In some designs, like the hybrid SVTs, only the DC servo is present. Obviously in a design like this you would check the DC servo first. If the opamp output is pegged to one of the rails, that should tell you something. An excessive offset has developed somewhere else in the circuit and the DC servo is trying its hardest to correct it and failing.

In solid-state work it really helps to get the control engineer's way of thinking. They are all just feedback loops, and troubleshooting systems with heavy feedback is different to tube amps.

[Enzo]

Bias problems CAN cause excess current to flow in the output stages, even without a load. Offset problems CAN cause excess current to flow when loaded.


A solid state amp is one big fed back loop. it wants to center itself. If you have a split power supply, centered is zero volts. If you have a single 90v power supply, then the output centers on +45v.


If my power supplies are not even, then the circuit will center between them, but not at zero volts.

Keep in mind there may be more than one level of power supply. The outputs may run on +/-50v, but there may also be an op amp running on +/-15v. Those low voltage supplies can come from a power supply circuit, or they might be derived from those 50v rails right in the power amp.

My op amp input stage wants to center as well, and whatever comes out of it is probably what comes out your power amp. So if it is missing one rail, it will center between what it has, and you might then see 7.5v offset (zero and 15)

People automatically start assuming there is a shorted transistor when they see unwanted DC. But an open one can do the same thing. And it doesn't have to be the transistor itself. An open resistor can turn a transistor off.

The feedback loop of an amp allows it to center the voltages. It does that by detecting some point is too high, so it adapts to send it lower. But if it cannot move the voltage that way, it tries within its ability. SO you might have an off voltage because the amp cannot push some other voltage the other way.

[Jazz P Bass]

I have found, in simple designs, it is often the differential pair that causes the slight (or severe) offset.
Ideally they should be 'matched'.
So if just one of them is fluky, you will have offset.
This is in regard to simple designs.
Now the complicated ones, I will admit to pulling out hair.

[guitardad]

Yeah I think we all know a lot about leaky 1n4148's in the bias. In older amps when I have a slight offset, that is the first thing I check or sometimes just replace them if its a mind stumper. I learned to use a Huntron anlayzer for some situations, but these days its about as cost effective to swap out low power semis and take your chances. Those diodes don't seem to like freeze spray either, so I spray a cleaning swab and just touch the casing to see if I get drift.

There's the remove the rail source method I have used a few times. Measuring resistance from one rail to the common line down each section. Good for finding opens or leaky caps when A-B ing channels without having the rail affecting measurements.

We all have our own methods of course. For El caps I love my Creative Electronics ESR meter.

Sometimes I miss the days of Sams Photofact which supplied working voltages. I have had many units where I curse the vendor for not providing operating voltages.

[Mick Bailey]

In a properly working solid-state amp, offset and bias are completely separate issues. The global feedback loop should hold the offset constant irrespective of the bias, and a well-designed bias string should hold the bias constant irrespective of the offset. Even a poorly designed bias string should manage fine for small offsets.

Agreed, but many older amps don't properly centre on 0v if they're over-biased, but can be brought down to reasonable offset levels if they're biased just clear of crossover distortion, but before the quiescent current draw starts to climb. If the offset is more than (say) a volt then there's something else wrong that needs fixing. It's a reasonable arguement to say that any amp where bias can affect offset isn't properly working, but sometimes decades old amps suffer from a less-than-perfect design and the blight of component drift. What's difficult to see nowadays is what was acceptable (say) 40 years ago. I had a transistor Vox which had a 130mV offset. Sounded fine, and the owner hadn't complained about the sound, but what would it have read when brand new?

[km6xz]

I would not worry about a leaky 1N4148 or two, that will just bring the bases closer together and run closer to class B. Open level shifters will cause more concern.

Rail stiffness imbalance with high base current can easily present an offset regardless of what the diff amp is trying to correct. Before swapping parts randomly like suggested take a look at the drop across the collector load resistors to see which stage is showing much different than expected current.
The fusable collector resistors are always suspect for good reason.
High resistance ground returns will generate an offset.
I do not see where shotgunning or hand wringing is needed if the circuit is understood.

[guitardad]

I would not worry about a leaky 1N4148 or two, that will just bring the bases closer together and run closer to class B. Open level shifters will cause more concern.

Rail stiffness imbalance with high base current can easily present an offset regardless of what the diff amp is trying to correct. Before swapping parts randomly like suggested take a look at the drop across the collector load resistors to see which stage is showing much different than expected current.
The fusable collector resistors are always suspect for good reason.
High resistance ground returns will generate an offset.
I do not see where shotgunning or hand wringing is needed if the circuit is understood.

IDK about that. We have seen a bunch of units with leaky 4148's in them.

Shotgunning is kind of a last resort. Sometimes its more cost effective though since you really can't work on units that have limited access. They build new stuff for fast mfg assembly, not service techs. No access plates. limited cable lengths, etc so when you have to tear into these things sometimes it's cheaper to start with known working diff amps, 4148's and so on.