Don
The circuit is on page 420... that you could not find it right away confirms my idea that the book could be structured somewhat better (or that the subject index could be more extended). But that is also the only criticism!
Broskie also presents and auto-bias circuit for class AB operation.
http://www.tubecad.com/2005/May/blog0046.htm
Erik
The circuit is on page 420... that you could not find it right away confirms my idea that the book could be structured somewhat better (or that the subject index could be more extended). But that is also the only criticism!
Broskie also presents and auto-bias circuit for class AB operation.
http://www.tubecad.com/2005/May/blog0046.htm
Erik
The "servo" presented by Vlad has neither minimum nor maximumsmoking-amp said:
detection that I can determine. Its only function seems to be a low
pass, and therefore no better than CCS bypassed with a huge cap.
The average cathode current in class AB1 will be higher than it is
during the class A crossing. Trying to hold steady this average is
not the right way to maintain a consistant bias as one enters to
class B territory.
I've shown you one possible way to detect and hold the peak or
trough of the crossing (depending weather you might choose to
observe that event from the top or bottom rail).
The fix can be applied to cathodes or grids, but applying fix to
cathodes is positive (pull down) feedback when using those
same cathodes for negative trough sensing. So I deliberately
didn't sense that way, I looked for the positive peak up top.
One could even bypass the cathode resistor with a cap when
sensing from the top, if so doing might prove an advantage.
If I had chosen to fix the grids, my method of applying that to
the grids would look a lot like Vlad's. And I have no problem
with that part of his circuit. But I don't think Vlad has drafted
a proper sense and hold of the crossing minimum. I think he
has so far only given us a circuit that acts upon a long term
average, which is not the optimum metric to be measuring.
It wouldn't take a huge re-arrangement of Vlad's collection of
parts to detect the true minimum, hold it, and act upon that
minimum instead of the average. He is almost there, but not
quite yet.
I don't have Moray's book to see what he's done. Some of
Broskie's soloutions seem acceptable, but others needlessly
complex. He may not have presented all of them with intent
other than to educate, or he'd have cleaned them up better.
Ken,
I think you're missing the essence of the circuit. It doesn't sample
the minimum current but rather compares it to a threshold at each
trough.
The minimum cathode current is detected by having Q1 charge
C1 whenever the turn-on threshold is exceeded. This seems to
rectify and filter the negative going extrema, or troughs, of the
cathode current waveform. Grid DC voltage is increased until
the cathode current troughs cross the threshold. Should work.
The MoJo scheme and the Broskie scheme rely on clipping the
current waveform and filtering the resultant to follow the
zero-crossing voltage. This looks like a perfectly reasonable
approach with some potential advantages. I need to study this
some more.
Michael
I think you're missing the essence of the circuit. It doesn't sample
the minimum current but rather compares it to a threshold at each
trough.
The minimum cathode current is detected by having Q1 charge
C1 whenever the turn-on threshold is exceeded. This seems to
rectify and filter the negative going extrema, or troughs, of the
cathode current waveform. Grid DC voltage is increased until
the cathode current troughs cross the threshold. Should work.
The MoJo scheme and the Broskie scheme rely on clipping the
current waveform and filtering the resultant to follow the
zero-crossing voltage. This looks like a perfectly reasonable
approach with some potential advantages. I need to study this
some more.
Michael
I am familiar with Broskie papers on shifting bias (well at least with some of them). His bias ideas are nice but different. As to "Valve Amplifiers" I do not have this book so I can not comment - I'd appreciate if someone explains what is done there.
The bias circuit here (and in Michael Koster circuit) does not shift bias when going from A to AB. This is strictly an attempt to stabilize fixed bias vs varying plate and screen voltages and tube aging. It does so by detecting the trough of the 2nd harmonic of the signal. Q1 works as a detector (the role of D1 in the first diagram), comparator and pseudo- sample/hold circuit. It is *not* a low pass filter / amplifier, sharp non-linearity and trough following detection is the key. C1 is part of a hold circuit.
In a perfect world C1 voltage is constant regardless of input signal level or operation class. It should only change when plate/g2 supply changes or the tube ages. In my case I have both output devices in a single envelope so I hope that balance can stay more or less the same with the aging. Michael circuit has auto-balance.
The bias circuit here (and in Michael Koster circuit) does not shift bias when going from A to AB. This is strictly an attempt to stabilize fixed bias vs varying plate and screen voltages and tube aging. It does so by detecting the trough of the 2nd harmonic of the signal. Q1 works as a detector (the role of D1 in the first diagram), comparator and pseudo- sample/hold circuit. It is *not* a low pass filter / amplifier, sharp non-linearity and trough following detection is the key. C1 is part of a hold circuit.
In a perfect world C1 voltage is constant regardless of input signal level or operation class. It should only change when plate/g2 supply changes or the tube ages. In my case I have both output devices in a single envelope so I hope that balance can stay more or less the same with the aging. Michael circuit has auto-balance.
Blumlein's Garter could be abused for side to side,
with top sense servo controlling the minimum sum.
Or twin pair of well bypassed CCS. With grounded
grids, and no need for the voltage lost in the lower
half of Blumlein's criss-crossed taps.
I dropped one pole from the low pass up top, as
the bypass caps below offer that same function.
-------------------------------------------------------------
I'm going back through the other autobias circuits
that I earlier dismissed too easily (due to my own
difficulty in understanding them), and try to get a
grip on those alternate theories of operation.
This shouldn't have confused me as much as it did.
I must be slipping hard...
with top sense servo controlling the minimum sum.
Or twin pair of well bypassed CCS. With grounded
grids, and no need for the voltage lost in the lower
half of Blumlein's criss-crossed taps.
I dropped one pole from the low pass up top, as
the bypass caps below offer that same function.
-------------------------------------------------------------
I'm going back through the other autobias circuits
that I earlier dismissed too easily (due to my own
difficulty in understanding them), and try to get a
grip on those alternate theories of operation.
This shouldn't have confused me as much as it did.
I must be slipping hard...
Attachments
For your info:
The Circuit in Morgan Jones Valve Amplifiers basically uses a current sense to monitor the current in the tube. This signal goes to zero as the tube cuts off. To balance that a circuit is added to the sense circuit to clip the signal at 2 times the idle current.
That is the sense circuit works symmetrically around the desired idle point (between 0 and 2 x idle). This clipped signal is then averaged and feeds the servo amp.
From what I can gather the Menno VdV/Tentlabs Servo Board works exactly the same way but instead of clipping at 0 and 2x idle it uses a much narrower clipping window around the desired idle point. It is claimed that this gives superior idle point stability as you are then working over a smaller part of the Ia vs Vg1-k curve of the tube. It is also claimed that this gives better push pull side idle current balance as it is less susceptible to tube variations.
Cheers,
Ian
The Circuit in Morgan Jones Valve Amplifiers basically uses a current sense to monitor the current in the tube. This signal goes to zero as the tube cuts off. To balance that a circuit is added to the sense circuit to clip the signal at 2 times the idle current.
That is the sense circuit works symmetrically around the desired idle point (between 0 and 2 x idle). This clipped signal is then averaged and feeds the servo amp.
From what I can gather the Menno VdV/Tentlabs Servo Board works exactly the same way but instead of clipping at 0 and 2x idle it uses a much narrower clipping window around the desired idle point. It is claimed that this gives superior idle point stability as you are then working over a smaller part of the Ia vs Vg1-k curve of the tube. It is also claimed that this gives better push pull side idle current balance as it is less susceptible to tube variations.
Cheers,
Ian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.