I’m going through a QSC 1400 power amp to make sure it checksout in all areas and I have taken note that upon a cold start up it measuresabout 160mv DC offset on the left channel and roughy 40mv on the right channeloutput. As I observed my meter over thenext 15 minutes the 160mv slowly dropped all the way down to 7mv and settledthere. The other channel also came downslowly to almost zero. My question isthis:
Am I correct in assuming a likely cause is leaky filter caps?This is my first encounter with pro audio gear and I must say the unit issurprisingly well built. I’d love to getit dialed in and see how close to “audiophile” criteria the sound quality is. I’ve read other people say QSC amps are notas musical as others but this sounds like nonsense to me. An in spec amps just amplifies whatever goodor bad signal they are given without any coloration. Anyway I’m curious if any has tips orcomments on this particular amp. Thanks!
Am I correct in assuming a likely cause is leaky filter caps?This is my first encounter with pro audio gear and I must say the unit issurprisingly well built. I’d love to getit dialed in and see how close to “audiophile” criteria the sound quality is. I’ve read other people say QSC amps are notas musical as others but this sounds like nonsense to me. An in spec amps just amplifies whatever goodor bad signal they are given without any coloration. Anyway I’m curious if any has tips orcomments on this particular amp. Thanks!
Most likely it is 'just the way it is' and what you are observing are the effects of thermal drift from a design that doesn't pay attention to such matters. As always though, having a circuit diagram available would give a better idea.
160mv would be considered high by modern standards for 'hi-fi' and typically an offset of around 50mv or less would be more acceptable with top flight amps achieving under 10mv under all conditions. That said, 160mv wont cause any issues as such.
160mv would be considered high by modern standards for 'hi-fi' and typically an offset of around 50mv or less would be more acceptable with top flight amps achieving under 10mv under all conditions. That said, 160mv wont cause any issues as such.
Ok here is a schematic. Actually another model but the same exact circuit. Speaker out put is unique in that it is directly from the filter caps.
They are a slightly unusual output stage for sure. So it looks like the load is ultimately AC coupled which would explain the slight variable voltage you see. I can't see how the output could support a DC offset tbh, at least not with a load present.
Have you tried measuring the offset with a load in place ?
Have you tried measuring the offset with a load in place ?
Sincerely, I would discard the filter caps, they don't participate in the biasing of the stages. Please, do as follows. Take a hair dryer, and heat slowly in certain zones while measuring the offset. There are few options: an input differential pair, a leaky cap in the NFB loop or a devalued resistor.
In your case, I doubt of D1 and D2 biasing diodes and the two opamps.
Keep us posted. 160mV is too much.
In your case, I doubt of D1 and D2 biasing diodes and the two opamps.
Keep us posted. 160mV is too much.
Last edited:
Good point about a load connected. I can't remember if my 8ohm dummy load was connected or not so I need to double check that later. I was planning a re cap either way so I guess that might be the next step. It seems like the filter caps are usually good on these amps but I should probably ESR check em.
Any difference in leakage current of filter capacitors (serving also as output coupling caps) will be present on ouput as DC ofset (with slow correction via NFB).
I haven't seen a topology like that before. Slowly getting my head around it, can someone point me in the right direction please?
In this strange circuits, you are right, the do participate. In this case, a couple of resistor of the same value across the caps will equalize quickly the middle point. Also check both 47K resistors in the return of the DC loop.
Ok thanks BV. So if I understand correctly...the negative feedback loop basically forces a correction of the offset through a time constant of appx 15 minutes?
I've been testing resistors off the main rails and so far they match but I'm assuming that even a slight difference in resistor values could lead to an unequal middle point. I believe 1% accuracy resistors are used but I better double check for imbalances.
I saw this circuit once in a guitar amp, the live becomes ground and viceversa.
Time constant must be milliseconds, not minutes.
NFB divider (also for DC NFB)is 10k/200R, so about 50x. 200mV at output will result in about 4mV at +In at IC1. Each OA has some DC unsymetry. Here is also 47k/47 network sensing PSU DC symetry, to the same +In input of IC1. And capacitors leakage current will slowly decrease ("re-formating") after switch on. Here are too many variables. It is so simple made as possible..Main advantage is in direct mounted power transistors (colectors without isolation) on common heathsink, so maximal use of heathsink cooling capacity, small DC cares nobody..Costs cut is primary.
Last edited:
Take a conventional Class AB amp such as Doug Selfs 'Blameless' and re-arrange it as follows. The amp has now becoming inverting in configuration.
1/ The conventional output point is now 'grounded'.
2/ The amp runs on a single rail.
3/ The feedback is returned differently and the amp now becomes inverting.
Also, and relevant to the original question is the fact that a DC offset is not possible because the load is AC coupled. Any offset measured should collapse when a load is connected.
Attachments
Audio signal go to the output via the power rails in place of the joint of the emitters of the output transistors. In other words, the supply "ground" becomes the output "live" for the load. A great stupidity, because all PS (Lytics and trafo) becomes live for audio signals. I don't understand why reinvent the wheel.
The amp has a thermal loop which has a time constant.
As the loop settles down to working temp the dc bias moves a bit.
As the loop settles down to working temp the dc bias moves a bit.
Thanks folks, as I mentioned above I was getting it once I realised the caps were output coupling not just smoothing. I agree, daft idea
- Status
- Not open for further replies.