AndrewT said:
In my view, an amplifier is operating in class A when the top and bottom transitors are contributing substantially the same transconductance to the output stage over the whole cycle.
Output stages that merely keep one side from turning off, and often put that transistor into a quasi constant current mode for that part of the cycle, are definitely not class A. They may in some sense be described as non-switching, but that is not class A.
Cheers,
Bob
AndrewT said:
Isn't class A akin to an engine going full tilt and you only use what you need, as opposed to class AB where the engine doesn't shut off but you have a throttle, to rev up the output as needed?
That's how I look at it, right or wrong. This puts the Krill in AB territory.
In Class A the output stage is biased at a high enough current that the output transistors don't turn off. In Class AB the output stage is biased at a lower current and the transistors do switch on and off as more output power is needed.
Johnloudb said:
How would you then call an output stage where the transistors don't switch off but contribute appreciably to the output current for only say 200 degrees? By definition that would be class AB.
Jan Didden
ostripper said:
This is interesting. Your first plot seems to show some non-linear charge storage effects, not surprising with high current lowish Ft devices. Could someone please post just the bias voltage across the output darlington vs. output current with just a DC sweep, if you already have this working in your simulator?
janneman said:
As a general abstraction a complimentary EF at constant voltage bias has the product of the two currents constant so technically neither turns "off". The degeneration messes that up badly though.
MJL, IF the current is only a momentary DC current, then the device is technically on, but it is NOT contributing to making the complete waveform, and is considered, for all practical purposes, class AB. This can be argued, but the final result is the same. Back in the old days, 30 years ago, before high speed output devices, this mode of operation was very valuable, and was implemented by a number of designers, but perhaps in a different way than is described here.
MJL21193 said:
Hi John,
Here's another example: Suppose we have an OPS with ideal BJTs and zero RE (also Ree is zero, of course) and it is biased with a constant voltage across the bases such that the collector currents (Ic1 and Ic2) are 10mA. Clearly this is a class-B OPS, right? However, if we put this OPS to work then Ic1 * Ic2 = 10mA^2 = 0.0001. IOW, the output devices are never turned off completely. Nevertheless, no one would call this class-A or class-AB operation.
Regards,
Edmond.
G.Kleinschmidt said:
This clearly switches but it is a case of what is switched where. As I read it this looks like a variation of Peter Blomley's "new approach to class B" published in WW back in Feb/March 1971. It was described then as "non-switching" but other analyses of it (including in Ian Sinclair's "Audio and HiFi Handbook") suggest that this is a misnomer becasue the switching takes place in the grounded base transistors.
alansawyer said:
Hence my question, if the commutation is switched to high Ft transistors operating at low currents there could be a net improvement.
john curl said:
Hi John,
Thanks for the explanation.
Again, this is how I'm seeing it:
A push pull AB has all devices "on" but they are only conducting (at idle) a portion of the required current to amplify the signal, whereas the class A will be in the same state but will be conducting all of the current that is needed to amplify the signal. The devices in the AB amp get alternately hot and cool down due to the varying current, while the class A stays hot, as the current and dissipation do not change (as much as the AB). This is my engine analogy - like a generator with a demand governor (AB) and one without (A). Class B is start the engine from cold when needed.
I know, clear as mud...🙂
janneman said:
My boo boo. 🙄 You're right, turning on and off would be class B. In class AB there is always a bias current going through both transistors.
Hey. guys, could we not get into the interminable discussion about the definition of class A - it's been done to death so many times on this forum. Does it really matter? The important issue is that non switching OPS avoid some of the Xover distortion mechanisms.
I'm more interested in Steve's 0.005% thd @ 400W NFB amplifer and the techniques! There are some people here claiming this is untrue or infeasible - let the games begin!
I'm more interested in Steve's 0.005% thd @ 400W NFB amplifer and the techniques! There are some people here claiming this is untrue or infeasible - let the games begin!
jkeny said:
0.005% thd @ 20KHz @ 400W/4ohm, GNFB-less! Yes, I am myself challenging this performance and I'd be curious about any further details, beyond blank statements.
Perhaps a new thread would be a good idea.
jkeny said:
OK, but the point is that only pure class A completely avoids cross over distortion by NOT switching at all. Some may not appreciate the distinction. Examination of BJT curves should make it obvious anyway because the point is that they are expected to operate in a linear mode by passing max current at all times they achieve it. Other ways of moving switching elsewhere and AB etc are not the same and will not achieve the same thing as "proper" class A does, done to death or not.
Joshua_G said:
I'm willing to give it a try, but that thread was hard to read with almost daily corrections to the schematic and they have gotten away from the basic theory.
- Status
- Not open for further replies.