Hello all,
I recently have had an interesting issue which I would like to share, actually revisiting what was discussed in post #320 on pg. 16.
(Yes, this thread is getting very long).
It is about the capacitor that (sometimes) is placed around the bias transistor circuit. Values I have seen in amp designs range from 10nF to 47uF.
And still to date I haven't found any textbook or serious article description of the 'raison d'etre' for this capacitor - or how to properly design its value.
Recently I had a beautiful, original RA-810A on the bench to upgrade. After opening the box, I first noted that the left channel had this C613 bypass cap - but the right channel had nothing???
So I decided to put in a 1uF film cap as I usually do (although this time before starting the actual upgrade). And soon the circuit went hot.
Scope on - and oops, there was severe HF oscillation. Removed the 1uF - and the oscillation was gone. Hmm.
Apparently these bias bypass caps are maybe not as innocent as they seem?
Anyway, I then completed the full upgrade successfully and thought little more about it as the amp was rock stable - and sounded fabulously.
But the next 'customer' on the bench was a heavy non-working RB-980BX. The four T4A secondary fuses had been replaced with T7.5A (!) and the bridge rectifiers had blown. Something was seriously wrong with this amp.
While I was waiting for delivery of the new rectifiers, I removed the 7.5A fuses and powered one channel up using lab supplies with current limits set at 250mA.
At first the supply currents looked relatively ok, but when I briefly touched the 10uF bypass cap the limiters immediately set in.
Scope on - and there was a large 1MHz oscillation.
Trusted remedy - two 100pF Miller caps from the base to collector of the VAS trannies immediately stopped it, currents normal - problem solved.
But being an upgrader rather than an repairer, my curiosity had already been triggered and I just had to investigate and experiment a bit.
So I first tried reducing the Millers to 15pF - and the oscillation returned. Then up to 47pF where things went stable.
Eventually remembering the RA-810A - what if I instead just removed all Millers and the 10uF bias bypass cap? So I did - and the amp was stable.
The VASs, the bias tranny and the 10uF cap were then all measured out of the circuit and were fine. All drivers and power trannies were also ok, and so were the current sources for the LTP's. No other visible or obvious damages that may have caused the problem.
I guess that left the LTPs. They actually measured ok (although there was not much hFE matching left) and the thermal compound in the plastic tubes had turned brown from heat. Even so, they worked just fine - once the 10uF bias cap was gone.
If anyone on this thread has had any similar experiences - or even better, a plausible explanation to the above, I am definitely all ears....
And also 'all eyes' if someone has a link to a relevant paper on the issue.
Per
I recently have had an interesting issue which I would like to share, actually revisiting what was discussed in post #320 on pg. 16.
(Yes, this thread is getting very long).
It is about the capacitor that (sometimes) is placed around the bias transistor circuit. Values I have seen in amp designs range from 10nF to 47uF.
And still to date I haven't found any textbook or serious article description of the 'raison d'etre' for this capacitor - or how to properly design its value.
Recently I had a beautiful, original RA-810A on the bench to upgrade. After opening the box, I first noted that the left channel had this C613 bypass cap - but the right channel had nothing???
So I decided to put in a 1uF film cap as I usually do (although this time before starting the actual upgrade). And soon the circuit went hot.
Scope on - and oops, there was severe HF oscillation. Removed the 1uF - and the oscillation was gone. Hmm.
Apparently these bias bypass caps are maybe not as innocent as they seem?
Anyway, I then completed the full upgrade successfully and thought little more about it as the amp was rock stable - and sounded fabulously.
But the next 'customer' on the bench was a heavy non-working RB-980BX. The four T4A secondary fuses had been replaced with T7.5A (!) and the bridge rectifiers had blown. Something was seriously wrong with this amp.
While I was waiting for delivery of the new rectifiers, I removed the 7.5A fuses and powered one channel up using lab supplies with current limits set at 250mA.
At first the supply currents looked relatively ok, but when I briefly touched the 10uF bypass cap the limiters immediately set in.
Scope on - and there was a large 1MHz oscillation.
Trusted remedy - two 100pF Miller caps from the base to collector of the VAS trannies immediately stopped it, currents normal - problem solved.
But being an upgrader rather than an repairer, my curiosity had already been triggered and I just had to investigate and experiment a bit.
So I first tried reducing the Millers to 15pF - and the oscillation returned. Then up to 47pF where things went stable.
Eventually remembering the RA-810A - what if I instead just removed all Millers and the 10uF bias bypass cap? So I did - and the amp was stable.
The VASs, the bias tranny and the 10uF cap were then all measured out of the circuit and were fine. All drivers and power trannies were also ok, and so were the current sources for the LTP's. No other visible or obvious damages that may have caused the problem.
I guess that left the LTPs. They actually measured ok (although there was not much hFE matching left) and the thermal compound in the plastic tubes had turned brown from heat. Even so, they worked just fine - once the 10uF bias cap was gone.
If anyone on this thread has had any similar experiences - or even better, a plausible explanation to the above, I am definitely all ears....
And also 'all eyes' if someone has a link to a relevant paper on the issue.
Per
Hello AngelP,
I recommend to install high-speed ceramic series F for example Siba.
Its purpose is to shunt the alternating current, although low, but the collector-emitter resistance of the bias transistor.
Hi Sam,
Yes, I know that this bias stage bypass capacitor does speed up the turn-off action of the drivers. Doug Self notes that "its use has been questioned because of the possibility of unhelpful charges building up on it during asymmetrical clipping". So, the issue must have been extensively discussed somewhere, but he gives no link to any reference.
How should its value be determined - Empirically? By trial and error? And why/how can it cause oscillations?
Bob Cordell suggests to instead put in an emitter follower after the bias stage which would dramatically reduce its output impedance - across the entire frequency spectrum. But that is not straightforward in an upgrade of an existing amp layout.
I thought of testing whether the 1MHz oscillation was due to parasitic capacitances in the power stage by soldering the 10uF Black Gate back in again and put in a 15pF between one of its legs to ground. Alas, the old black cap had already been sent to recycling, so it is probably in a landfill somewhere.
I may have another used 10u Black Gate lying around which I could try and see if the oscillation returns - and if so, whether a small cap to ground helps.
Anyway, I guess that the rule of thumb is that if you have an amp that has suffered serious and damaging oscillations - you should assume that all electrolytic capacitors have been damaged by heavy reverse polarity voltages. And simply replace them all.
Yes, I know that this bias stage bypass capacitor does speed up the turn-off action of the drivers. Doug Self notes that "its use has been questioned because of the possibility of unhelpful charges building up on it during asymmetrical clipping". So, the issue must have been extensively discussed somewhere, but he gives no link to any reference.
How should its value be determined - Empirically? By trial and error? And why/how can it cause oscillations?
Bob Cordell suggests to instead put in an emitter follower after the bias stage which would dramatically reduce its output impedance - across the entire frequency spectrum.
But that is not straightforward in an upgrade of an existing amp layout.I thought of testing whether the 1MHz oscillation was due to parasitic capacitances in the power stage by soldering the 10uF Black Gate back in again and put in a 15pF between one of its legs to ground. Alas, the old black cap had already been sent to recycling, so it is probably in a landfill somewhere.
I may have another used 10u Black Gate lying around which I could try and see if the oscillation returns - and if so, whether a small cap to ground helps.
Anyway, I guess that the rule of thumb is that if you have an amp that has suffered serious and damaging oscillations - you should assume that all electrolytic capacitors have been damaged by heavy reverse polarity voltages. And simply replace them all.
Hello AngelP,
The amplifier is not stable, either a malfunction or modifications have made it so.
