Bob Cordell said:
While that is true, the HF distortion is generally negated by the relatively high localised negative feedback of the unity gain BJT-BJT CFP.
In my experience, the CFP has always provided better high frequency linearity and bandwidth over the same driver+output transistor pair configured as a Darlington.
I think Douglas Self has written about this at length, but I can’t quote a reference because I haven’t read any of his books.
Cheers,
Glen
Bob said:
Interestingly, probably this highlights different work environments.
I can relate with Bob's experience perhaps within the medium - large corporation employee situation, where mistakes may be paid dearly or bosses decide what and how to do.
For the independent engineer on the other hand, while he must be cautious anyway, there is probable more room to pursue sometimes not strictly rational choices.
Rodolfo
Rodolfo (snip)
"...this issue is addressed by never turning completely off either half of the output stage."
This is the most important issue, especially for low distortion at high frequencies (Bob`s "dynamic crossover distortion"), especially with large source/ emitter R´s which are needed for shure current sharing.
With CFP the steering circuit is simpler than for EF stages. I use both, in the last years more CFP´s.
Regards
Heinz!
"...this issue is addressed by never turning completely off either half of the output stage."
This is the most important issue, especially for low distortion at high frequencies (Bob`s "dynamic crossover distortion"), especially with large source/ emitter R´s which are needed for shure current sharing.
With CFP the steering circuit is simpler than for EF stages. I use both, in the last years more CFP´s.
Regards
Heinz!
lumanauw said:Hi, Glen,
This is the CCT to make the sound of CFP more obvious. R2 makes T1 saturated more quickly than normal, but some people just likes this sound
David,
It seems this trades quicker saturation of the driver for later / less saturation of the (slower) output device. The result may well be more pleasing sound-wise.
Jan Didden
Upupa Epops said:
Excuse the late retard, you might like this too :
www.analogaudio.info/audio.pdf
1995, Page 100, Amplifiers and Constructions
What do you think about this variant of CFP? Unfortunately it can only be made for CFP with gain, but one interesting feature that it don't need VBE multiplier to bias it. The input needs only goes to 1 point (A), high impedance because it only drives base(s).
T3 and T4 always biased by R2 and R3.
T3 and T4 always biased by R2 and R3.
Bob, I first used the CFP topology with a very powerful (2000W) amp that I made at Ampex Research in 1969. I used it, because it held the most 'promise' , but I found it problematic because the feedback loop that is formed by each pair does not easily have a dominant pole. I found that emitter followers were actually better in this regard and I have used them ever since.
john curl said:
I have not really pursued this issue, but could it be in the case of a bipolar driver - MOS output there is a natural dominant pole formed by the gate capacitance and driver output impedance?
Admittedly, the gate capacitance is not constant along the large gate-drain voltage swings.
Rodolfo
john curl said:
1969……geezzz John, where’s your sense of adventure?
I can’t really comment on the state of the art of bipolar transistors in 1969 (I didn't arrive until 1977 - into the world, not the audio scene), but modern transistors most likely have much stricter manufacturing tolerances and lesser spreads of parameter variations between devices with the same basic specifications.
The CFP definitely is inherently more prone to instability than a Darlington emitter follower and that’s probably the best grounds I’ve seen for avoiding them. However, I’ve employed them in high power amp designs, which were both inherently stable and repeatable.
I think that, properly applied, the CFP does have desirable characteristics over the standard emitter follower topology and has a place in many audio output applications, but is often not utilized or taken advantage of due to the reluctance of many designers to look at them.
MOSFETs can be difficult to tame too, and that hasn’t stopped audio designers from using them!
Cheers,
Glen
lumanauw said:Hi, Glen,
This is the CCT to make the sound of CFP more obvious. R2 makes T1 saturated more quickly than normal, but some people just likes this sound
OK, cool. Your original comments were general, not directed at that particular derivation though. That aside I really can't see what anyone would like about the sound produced by that circuit in any HIFI application. Was this in a guitar amp or something??
G.Kleinschmidt said:
"Blessed who believeth" (C)
Back in 1977 I selected transistors for my amp and got angry because PNP transistors had enormous spread of parameters when NPN ones were much closer to each other, especially beta and break-down voltages. There were Soviet prototypes of KT818 and KT819, It was in one of laboratories of Institute of Semiconductor Devices.
Recently, I again selected matched transistors and saw the same picture, this time there were already long time in production 2SC5200 and 2SA1943 from Toshiba.
But the major point is, it is an impossible art to produce complementary pairs.
Wavebourn said:
"Blessed who believeth" (C)
Back in 1977 I selected transistors for my amp and got angry because PNP transistors had enormous spread of parameters when NPN ones were much closer to each other, especially beta and break-down voltages. There were Soviet prototypes of KT818 and KT819, It was in one of laboratories of Institute of Semiconductor Devices.
Recently, I again selected matched transistors and saw the same picture, this time there were already long time in production 2SC5200 and 2SA1943 from Toshiba.
But the major point is, it is an impossible art to produce complementary pairs.
I must be blessed then. I wouldn't argue against the fact that there are variations in parameters between devices and some of these variations may be greater for PNP devices than for NPN devices and visa versa, but what is your point?
Are you trying to tell me that semiconductor technology hasn't made significant gains in this regard since 1969?