I have no idea what you mean by "switch to pure class B at any level". What class do you suppose they do switch to? Either they have the low eficiency of Class A or the high efficiency of Class B at full output. The Rotel doesn't put out 20 watts, so what do you suppose it is doing then?
"switch to" was actually your words, imo not the best description there is no switching as such, I simply followed along. I meant they are still bias to operate as class AB. The bias can be controlled such that the amp is bias in AB class full time, or it can also be controlled to in fact operate in class A at low input signal level, and "switch to class B at some point..." as you previously described.
The point is, they don't have to (but yes they could, if the designer choose to do so) switch to class B, they can, and I believe in most cases, the bias are designed for them to be in AB practically full time. In some cases they could also be designed to operate in class A bias and "switch to" class B at some point, but not always, as they can also be designed to switch from A to AB. As I mentioned before, I do own two amps that operates in class A up to 8W (Halo A21), the other up to 20-25W according the the manufacturer's engineer. If you are curious enough you can email Parasound and ask if their Halo amps would switch to class B at some point. I am sure the answer will be "no".
Please read some technical documents before you post misinformation. This is fundamental stuff, easily researched.
And then we have "not all class AB amps run in Class A bias even at low level."
That is not very nice for you to say I posted misinformation, unless you have proof that I did.
As ski2xblack said in post#12:
Verdinut, class A/B amps, by definition, conduct slightly more than 180 degrees. They still do that, even at high power levels (they don't somehow switch to class B operation). Class B is inappropriate for audio purposes.
http://sound.whsites.net/articles/amp-classes.htm
He said basically what I said with much fewer words
, though I slightly disagree with his point of class B being inappropriate, that's just too black and white.
Thanks, but I read that article before you even linked it. If I remember right, nothing in that article contradicts what I have been saying so far.
The definitions of classes presented in that article were:
- Class A: The amplifiers single output transistor conducts for the full 360o of the cycle of the input waveform.
- Class B: The amplifiers two output transistors only conduct for one-half, that is, 180o of the input waveform.
- Class AB: The amplifiers two output transistors conduct somewhere between 180o and 360o of the input waveform.
According to the definitions from that article, if the bias is controlled such that output transistors conduct for even 181 degrees, it can be correctly defined as class AB. Now, would you really consider such amp class A at low input signals, with just 1 degree overlap? I supposed you may do so technically speaking, but I doubt any such amps would bother claiming to run in class A at any point in practical sense, I meant like by saying their amp runs in class A for up to 10 mW ouput?
By the way, I decided to re-read that article, and I found the following paragraph that actually made basically the same point I have been trying to make.
"The biasing of a Class AB amplifier output stage is generally adjusted to suit a particular amplifier application. The amplifiers quiescent current is adjusted to zero to minimise power consumption, as in Class B operation, or adjusted for a very small quiescent current to flow that minimises crossover distortion producing a true Class AB amplifier operation."
Note the word "true", that is exactly my point.
Let me summarize by saying there are several scenarios in the class AB design:
1) As you described, one that operates in class A up to a certain point, then "switch to" class B above that level, hence the term "class AB". Note: I don't think this is the popular design.
2) Bias to operate with the output devices (push/pull) conducting for more than a half cycle, but very slightly, such as between 182 degrees to 190 degrees (just as an example, the designer will choose), hence class AB by definition, because the conduction angle is > half cycle < full cycle. Note: I believe this is one of the more popular way, I would assume lots of avrs such as those by Yamaha, D&M's class AB design works this way, just my educated guess though.
3) Bias to operate with the output devices conducting for full cycles at low signal levels so that the amp actually operates in class A up to a few watts or even higher, practically enough to be considered as an amp that runs in class A up to a point and then in class AB above that point.
Examples of such amps: Parasound Halo line, Emotiva XPA-1L
How many amplifiers have you designed? Feel free to ask me the same question.
I won't ask you the same question because I prefer just open minded conversations in less challenging tone. To me this is just a hobby and I come here for idea exchange opportunities, fun, and sometimes also opportunities to help each other by sharing knowledge. I am a practicing EE, and I understand the basic theories in the design of audio amplifiers very well but I have not designed one myself, did build a couple long time ago. I have done a lot of reading, naturally including text books (still have them) in university. I have communicated with those who do it as a living as well as a professor friend who does it as a hobby. One can "design" amps, but that does not mean one understands all the theories behind.
If you read my posts with open mind as I did when I read yours, you may see that we just have a very slight disagreement on the way class AB amps operate in the bias. I accepted for the first part of the way you describe class AB in the first sentence of your post#58, but I take partial exception only on the part "switch to" class B at some point...." because you implied it can only be that way. In fact, that is just one way, there are other ways as I listed above, and you too can do more Googling for further details.
For short version, one just have to read ski2xblack's post#12.
