CMA Target number -
Large signal bandwidth at low distortion -
A competitve target number for our CMAmp today would be less than .001% thd and fewer transistors. See what you can do with your circuit here to get to that target level of performance as everyone's goal here.
[PS... that would be as 'measured' distortion. Often not the same as SIM'ed FFT results]
Thx-RNMarsh
Large signal bandwidth at low distortion -
A competitve target number for our CMAmp today would be less than .001% thd and fewer transistors. See what you can do with your circuit here to get to that target level of performance as everyone's goal here.
[PS... that would be as 'measured' distortion. Often not the same as SIM'ed FFT results]
Thx-RNMarsh
Last edited:
Build it with new awareness-
Now, some practical advice about building with CMA topologies.... which can give bad results if unaware --->
Thx-RNMarsh
Now, some practical advice about building with CMA topologies.... which can give bad results if unaware --->
Thx-RNMarsh
Last edited:
Dear Esperado,
Kindly post the .asc file also in each case. This is to help us in learning.
--gannaji
Esperado, it would be much more instructive if the loop gain was plotted.
This way, as the feedback resistance network value is changed, you can see the loop gain bandwidth change - which is a prime feature of CFA.
This way, as the feedback resistance network value is changed, you can see the loop gain bandwidth change - which is a prime feature of CFA.
Your sims are flawed...
The LTP version input stage has 1/10 of the CFA
input stage transconductance...
The LTP version VAS has 1/3 of the CFA version VAS gain.
I expected that this thread is going to be about CFA design not a war between pro CFA and pro VFA.
Esperado here is not the place to insist on this comparison(technically very flawed).
The golden curve were taken at the input of the VAS. Everybody can see the levels at 100 Hz are identical. Around -57db. And the output levels the same too. Mean the VAS have the same gain in the two situations, if i do not make mistakes.
If you simulate the circuits you will see that the currents of each stage are similar too.
Now, as i said, up to everybody to provide their own sims in order to improve the VFA version, as well as the CFA version. I asked this some time ago, because i'm upset by the attitude of few contributors, in my ignore list, who drive some personal war, far from the gentleman attitude we all try to keep in this forum. Nobody had done-it. It seems they are more talented for criticism than to provide any useful inputs.
For the moment, my purpose was just to look at the evolution of the bandwidths. They will not change if we increase in LTP the gain of the input stage. Dadod, i believe on the contrary that the only way to get a better understanding of the two topologies are to compare them with circuits as close as possible. But they are several aspects: impedances and gain, each ones with effects on distortions or bandwidths, hard to conciliate. If you look at my first two versions of LTP, you will understand what i try to show. Just a beginning.
Please, provide your LTP version, as i asked several time.
If you simulate the circuits you will see that the currents of each stage are similar too.
Now, as i said, up to everybody to provide their own sims in order to improve the VFA version, as well as the CFA version. I asked this some time ago, because i'm upset by the attitude of few contributors, in my ignore list, who drive some personal war, far from the gentleman attitude we all try to keep in this forum. Nobody had done-it. It seems they are more talented for criticism than to provide any useful inputs.
For the moment, my purpose was just to look at the evolution of the bandwidths. They will not change if we increase in LTP the gain of the input stage. Dadod, i believe on the contrary that the only way to get a better understanding of the two topologies are to compare them with circuits as close as possible. But they are several aspects: impedances and gain, each ones with effects on distortions or bandwidths, hard to conciliate. If you look at my first two versions of LTP, you will understand what i try to show. Just a beginning.
Please, provide your LTP version, as i asked several time.
Last edited:
Lets start at square 1 with a basic CFA amplifier, understand circuit operation more fully, and then evolve the design to higher performance - I dont see any point in rushing in with complicated designs when it clear there is a lot of learning needed to be done in general on this subject - me included. So here goes!
I've included an overview .pdf and then my .asc files as well covering transient and AC analysis.
I propose we stay focused on the development of a basic design and forget about trying to contrast it with VFA - that argument is tiresome and not productive in my view.
I've included an overview .pdf and then my .asc files as well covering transient and AC analysis.
I propose we stay focused on the development of a basic design and forget about trying to contrast it with VFA - that argument is tiresome and not productive in my view.
Last edited:
VSSA is simpler (4 transistors less), flatter, more stable and measure better (0.01% at 20KHz in sims). I do not see where is the problem to work with it.
Last edited:
Bonsai, I totally don't understand. I do not care about VFA vs CFA war. I'm, and i believe all of us, interested to see the differences and the expected advantages of each of those two topologies.
We expect from VFA less harmonic distortion (because the LTP) and we expect from CFA extended bandwidth and slew rates (and less IM ?).
Are-we right ?
And what are the various ways to improve the weaknesses of each ones ?
And what is their impact on our listening experience ?
None of those two topologies are a religion. From the early beginning, i don't understand this fight.
Last edited:
Esperado,
1. Lets use a classic CFA topology amp (VSSA is a very clever CFA design, but its not classic CFA as you will find in text books or inside op-amps. In fact the design I posted would not have the TIS I showed, but would normally use a current mirror)
2. Focus on CFA - no need to worry about VFA in this thread - just my advice otherwise we get bogged down in endless circles and dont move forward.
3. I dont think this thread is about proving one is better than the other - I use both and am happy to continue doing that - so the fact that we start off with a design (the one I posted) that is 0.04% is ok - the mission is to see how we can evolve a classic CFA to high performance.
1. Lets use a classic CFA topology amp (VSSA is a very clever CFA design, but its not classic CFA as you will find in text books or inside op-amps. In fact the design I posted would not have the TIS I showed, but would normally use a current mirror)
2. Focus on CFA - no need to worry about VFA in this thread - just my advice otherwise we get bogged down in endless circles and dont move forward.
3. I dont think this thread is about proving one is better than the other - I use both and am happy to continue doing that - so the fact that we start off with a design (the one I posted) that is 0.04% is ok - the mission is to see how we can evolve a classic CFA to high performance.
Last edited:
http://www.diyaudio.com/forums/solid-state/240712-cfa-topology-audio-amplifiers-17.html#post3601284
If a fair comparison between CFA and VFA is out of discussion, this thread will lack a great part of its interest.
Thanks Esperado, great work.
There's no VFA with the same parts count to come close to VSSA. The name tells everything Very Simple Symmetrical Amplifier and it is best sounding simple amplifier of them all.
Regards L.C.
The problem is that the people are affraid to evolve the amp from VSSA up, since that would show they are all wrong. As I already explained in this thread the right way would be to transform VSSA to top level amp with super (compound) parts, made out of three to four BJT-s instead of each BJT part now present.
The main idea lies in simple block schematic and that is made out of three blocks only: input modulator (input + feedback signal subtracted in the same part), voltage amplifier (transconductance stage), output buffer (drivers + mosfets). Each of this should strictly follow the basic VSSA topology, with the only difference to implement super compound parts instead.
If VSSA is not classic CFA I really don't know what it is??? 😕
Diamond buffer sounds worse than single modulated part in the input stage, because it has the same problem as LTP, the signal is subtracted through different parts (input transistor and feedback transistor), so each part is playing different role and there you miss coherence of the signal. Loads of unwanted math functions are inserted to the system without any reason.
LTP and diamond buffer in power amps should be avoided miles around!