I did.
1.7MHz is as relevant as the TMC loop gain analysis without breaking the connection to the output (hiding the fact that TMC is second order). Just another smoked mirror.
Here you feed the amp at +/-50V and push the output at not more than +/-27Vpeak. No wonder there's virtually no THD20 (and PIM). If you bias the output at +/-27V by applying a DC component at the input, the ULGF of 6MHz won't move much, so not much PIM.
Now push the output at say +/-45V (just before clipping) and the THD20 (and PIM) will show it's ugly head.
You may want to try modulating a very low frequency sine (e.g. 1Hz) with a 20KHz small sine, then watch the 20KHz harmonics as they move around. Requires a very long simulation time, though.
That s not representative of real life conditions , 5V below
the rails is quite excessive , 10V seems more adequate as
in Dadod s sims , 50V rails is not enough for whom wants
45V peaks outputs.
Whatever it takes, just before clipping, I haven't see his distortion results at 40V (or more) peak output. But as much as 23V under the 50V rails (as in Dadod's simulation) is completely irrelevant for an apple to apple distortion performance comparison.
Last edited:
I've updated my CFA/VFA write up on my site and clarified a few points. The table describing the differences between the two topologies is also updated.
Bottom line: Above the -3dB closed loop bandwidth, in practical amplifiers of either pursuasion, the loop gain responses are very similar. Reason: You cannot escape the output stage pole and is the primary constraining factor in applying more feedback in both topologies. The good news is that you can apply advanced comp techniques to CFA as well to get extremely low distortion if thats what rings your bell.
CFA vs. VFA: A short primer for the uninitiated – UPDATED
😎
Bottom line: Above the -3dB closed loop bandwidth, in practical amplifiers of either pursuasion, the loop gain responses are very similar. Reason: You cannot escape the output stage pole and is the primary constraining factor in applying more feedback in both topologies. The good news is that you can apply advanced comp techniques to CFA as well to get extremely low distortion if thats what rings your bell.
CFA vs. VFA: A short primer for the uninitiated – UPDATED
😎
Last edited:
OPS local feedback implies multiple loops so you have to be very careful when you talk about "the" ULGF, because there are several.
In many practical implementations it is possible to find a Return Ratio that defines the stability of the entire amplifier and hence sensible to discuss "the" ULGF.
This is practically never the ULGF of the overall loop.
So your low overall amp ULGF here is almost irrelevant, unfortunately.
At least it provides a probable lower bound to the relevant ULGF, so not totally irrelevant.😉
As Waly correctly points out, this also applies to Damir's 1.7 MHz number.
Best wishes
David
Last edited:
This is of considerable interest, more details would be appreciated.
What was the approximate calculation?
What typical PCB trace inductance and capacitance did you arrive at?
Best wishes
David
Last edited:
Member
Joined 2009
Paid Member
Agreed - that's exactly what my simulations showed on my TGM7 comparing VFA and CFA - as the OLG's are the same, the distortion profiles are pretty much the same. But I do like the practical advantage that for pcb layout, the low impedance feedback node is less susceptible to parasitic pick-up.
Black's 1st Feedforward patent is different from the later Negative Feedback in basic principle
Feedforward distortion cancellation has 2 power output stage paths to the the amp output, requires a power level combiner that ideally isolates the reverse/cross conduction - possible with RF "hybrid" but not easily approximated at audio
Quad's version uses a power passive XO approach and achieves Feedforward Cancellation in the crossover region
at very high frequency it transitions essentially to negative feedback around the correction amp
First it was 28V (50W/8).
Here are simulations of THD1k and THD20k with 44V peak output and 120W in 8 ohm, is that good enough?
I like your criticism as that push me to investigate more, but sometime good positive comment would be nice.
Attachments
Just a little remark, why do-you limit CFA to one gain stage in your table comparizon? There is no reason why you cannot add some gain at a CFA input stage, as do L.C in his VSSA. Output levels of this stage are very low, this does not add too much distortion to this stage, and even reduce total one once feedback is closed.
Just i want you to know that i find your work on this subject amazing and *full of interest*. About positive attitude, i believe than, in cars of some countries, it is the positive battery plug witch is grounded. Some should have the same spirit orientation🙂
Last edited:
🙂😎
Two thumbs up ! Dadod.
Several people with a positive spirit have done an over-the-top effort to help everyone better understand circuits and CFA in particular.
I certainly feel like I got my monies worth 🙂
-Thx to all - Richard Marsh
Last edited:
Maybe. A bit of distortion can be interpreted by the brain in many different ways, and even as giving a more defined sound, I experienced it. This may explained some mysterious listenable but unmeasurable facts about signal amplification.
Called
"quadratic class A" in France by Hephaistos / Gérard Perrot about thirty years ago
"curve linear class A" by Ian Hegglun.
Ian Hegglun published circuits using this kind of output stage in Elecronics World + WW, Sep 1995 p754-756, "Square law rules".
Linear Audio volume 1, April 2011, p6-27 "Square law class A"
(that all the joy of Linear Audio, the authors seem not to have a limited number of pages to explain their investigations and circuits).
There was also an interesting circuit proposed by Ian Hickman as an evolution of Pass's Zen in Electronics World, Aug 1999, p682, "New Zen".
Is "square class A" able to solve the main problems of output stages ?
What Bob Cordell said may apply to it :
"Just a little remark, why do-you limit CFA to one gain stage in your table comparizon? "{
The idea was not to explore all the variants - the document is aimed simply at comparing the two classic topologies, what makes them VFA or CFA and understanding the limitations. There are dozens of derivatives of each - we can see that on the forum. Like I said before, if you are focused on distortion the primary indicator of performance then VFA is probably a bit better (low single digit PPM difference, assume 1~3 MHz ULGF); if you are more interested in SR and BW, then CFA delivers.
With a single gain stage in a CFA, you get >70 dB open loop gain and low distortion - that's where the simplicity comes in. If you want more loop gain, then go for VFA. Simple decision.
I noticed there were many questions from people on this thread 'what constitutes a CFA'. Table 1 in the document should answer that.
But, its easy to build very good sounding amplifiers with either topology, so clearly this topology is either not well understood, or people have just not been exposed to it.
I've built and listened to both, so I speak from experience 😉
The idea was not to explore all the variants - the document is aimed simply at comparing the two classic topologies, what makes them VFA or CFA and understanding the limitations. There are dozens of derivatives of each - we can see that on the forum. Like I said before, if you are focused on distortion the primary indicator of performance then VFA is probably a bit better (low single digit PPM difference, assume 1~3 MHz ULGF); if you are more interested in SR and BW, then CFA delivers.
With a single gain stage in a CFA, you get >70 dB open loop gain and low distortion - that's where the simplicity comes in. If you want more loop gain, then go for VFA. Simple decision.
I noticed there were many questions from people on this thread 'what constitutes a CFA'. Table 1 in the document should answer that.
But, its easy to build very good sounding amplifiers with either topology, so clearly this topology is either not well understood, or people have just not been exposed to it.
I've built and listened to both, so I speak from experience 😉
Last edited:
Do you know 1994 Stochino's circuit
http://www.google.fr/url?sa=t&rct=j...ESP2uMVxaW02VDKEszMyTdw&bvm=bv.57155469,d.bGQ
?
For the use of the vertical MOSFET has a good article:
Stewart Edmond, Autobias for mosfet audio out stage, Electronics World 12/2003 p. 17-20
which shows that the field of vertical transistors tranzmstorov most complementarity Hitachi type 2SK1530/2SJ201 (2SK1529/2SJ200).
This is confirmed by research firm and National Semiconduktor AN-1645
I also conducted research with different FETs in the amplifier:
http://zalil.ru/34828511
and found that the popular and cheap type transistors IRFP240/IRFP9240 give a bad sound, although the parameters of the amplifier measured at work on resistive load high enough.
regards
Petr
Stewart Edmond, Autobias for mosfet audio out stage, Electronics World 12/2003 p. 17-20
which shows that the field of vertical transistors tranzmstorov most complementarity Hitachi type 2SK1530/2SJ201 (2SK1529/2SJ200).
This is confirmed by research firm and National Semiconduktor AN-1645
I also conducted research with different FETs in the amplifier:
http://zalil.ru/34828511
and found that the popular and cheap type transistors IRFP240/IRFP9240 give a bad sound, although the parameters of the amplifier measured at work on resistive load high enough.
regards
Petr