Sounds like the uncontrolled bias regulator would work great for an amp that never drives a load. Once a load is connected and a signal is applied, the voltage drop across the emitter resistor doesn't tell you the bias level anymore, it changes with output current delivery.
Is the issue with amp classes because it is uncomfortable having the set of Class B amps be an empty set (conduction never exactly 180 degrees)? Why would that matter, its just a border between Class AB and C.
Is the issue with amp classes because it is uncomfortable having the set of Class B amps be an empty set (conduction never exactly 180 degrees)? Why would that matter, its just a border between Class AB and C.
That's exactly why in most class AB or class B bias spreaders, Iq is inferred from temperature rather than directly sensing across say Re. If you try to directly control Iq.Re under dynamic conditions, it can get complicated very quickly for not much added benefit. Class A is a lot easier. You either go SEPP (good examples of the genre from Nelson Pass), a class AB or B type spreader with inferred control (the most common method) or you control it directly which is what I did on my sx-Amp.
That's the problem nobody recognizes! Try it, then let us know 😀
The voltage across an Re is NOT an AC signal you can just low-pass!
(Except for full class A).
jan
Indeed. There is a rather straightforward way to sense Ibias from Vre; anybody knows how?
In my bias measuring test set that I used to track thermal bias transient changes, I faced the same problem. Read the article and all will be revealed 🙂
jan
not true.
ClassA is when the transistor, or transistors, control the output current over the full 360degree cycle of the AC wave form.
Agreed. Most ClassA amplifier manufacturers are not lying about their ClassA output capability.
Not true.
A +-40Vdc ClassA amplifier with 2A of bias through the output stage would have just short of 4Apk of ClassA output current.
The maximum output voltage would be ~35Vpk and certainly a long way short of 40Vpk.
The output stage dissipation is (40+40) * 2A = 160W
The maximum ClassA output power is ~35 * 4 / 2 = ~70W
Note that I did not need to refer to the load impedance. That would be determined by using the maximum output voltage combined with the maximum ClassA output current. In this example it would be: required load impedance = Vpk/Ipk = 35/4 = 8r75
Last edited:
This is not well appreciated. I don't mean to sully things with a tube reference, but see the autobias circuits in Morgan Jones's Valve Amplifiers and Menno Vanderveen's High End Valve Amplifiers 2 where that exact issue (which is inherent in tube push pull stages) is dealt with. I wonder if an adaption of their opamp circuits could be useful in a solid state power amp?
Yes I think conceptually the problem in PP tube amps is exactly the same as in SS PP amps with Rk posing for Re (and v v).
I know Guido Tent has developed an antibias circuit for tubes with has a lot of non-linear processing and valley detectors IIRC.
SY do you have a chapter etc ref for MJ?
Jan
Joe Curcio did one for dynacos some years ago. Somewhat lower parts count so not sure if he was either being clever or cutting corners! CURCIO AUDIO STEREO-70 AMPLIFIER MODIFICATION
Neither take HT variations with wall voltage into account which might matter if you like to push the envelope and have lousy rural mains.
Neither take HT variations with wall voltage into account which might matter if you like to push the envelope and have lousy rural mains.
Most commercial class A power amps are/were not biased to the max peak output level in 8 ohm, but max continuous level at best.
Audioanalyse A9, brochure says 50W Pure class A : 1.2A quiescent current
(35Vdc rails, 160VA toroid per channel, 65W continuous in 8 ohm)
Forté Audio model 1A by Nelson Pass, 50W class A : biased to 1.25A
Accuphase P102, 50W class A : 1.25A bias
(bridged design, 20Vac secondaries, would imply ~25Vdc rails)
http://www.accuphase.com/cat/p-102en.pdf
Need a dozen more examples ?
I have a design somewhere with an individual driver per output device, which enables to bias each power device separately without loading the Vas.
I understand this with the idle measured over Re not are not rearly the same is one also must consider the integrated contribution of the signal over time, this complicates things a lot. We're back where gain and effort are not really in sync
I measured the Vre at the output current zero crossing. At that point, Vre is from Ibias only.
But there are some complications. For instance, if you want say 10% accuracy, the output current should be less than 10% of Ibias at the instant you measure. That means that the zero crossing 'window' can be very short - too short to accurately measure Vre until Iout again starts to rise in amplitude.
Jan
But there are some complications. For instance, if you want say 10% accuracy, the output current should be less than 10% of Ibias at the instant you measure. That means that the zero crossing 'window' can be very short - too short to accurately measure Vre until Iout again starts to rise in amplitude.
Jan
Some bias systems use a common mode loop around the voltage drop in the emitter resistors, I think.
Last edited:
VBE, may be the best analogue computer we have, would consider thermal effects on the OPS not being a major factor in performance (unless idle dynamically drops signifanlty during transients) my feeling is that transients will raise idle and the VBE with its thermal lag will even it out (unless the ops is grossly over compensated). It's also under feedback control as the taking point is after Re.
Last edited:
Those are false advertizement then. 1.25A translates to 12.5W of Class A for 8ohm.
BUT you can have class A in push pull. Just because these are false advertizement does not imply it's not possible. Class A is a definition, nothing to do with what is on the market. You have big enough heatsink and willing to have good air conditioning, you can have high power class A push pull.