Same design as the Revox, I was given two channels stripped from a amp with damaged case. At the moment they are lying in the scrapheap in my backyard as they are awful sounding amps. Some really bad revievs on them around the net too which I concur with. They only sounded decent when biased to class A.
The VFB design shown doesnt need to be in class AB to outperform CFB designs although they are more complex. Marantz is using that VFB design in some of their top of the range.
Good example is lm6171 circuit.
No. And the price was not the same neither. And i confirm, the Revox amp was not so good. Where symmetrical input, good power supply and selected paired parts makes all the difference.
Is was the same for the tape recorders, they where not playing in the same playground.
Btw: What are-you trying to do systematically arguing against all i can write ?
Last edited:
Here is some more information on the Studer A68:
Das STUDER und ReVox Infoportal | STUDER A68 - Analoge Studioendstufe
Here is one studio that uses it:
Sear Sound NYC, NY
Das STUDER und ReVox Infoportal | STUDER A68 - Analoge Studioendstufe
Here is one studio that uses it:
Sear Sound NYC, NY
Some can claim indefinitely that VFB is better than CFB. And even provide bad designed examples to demonstrate-it. Or examples looking like current feedback, but not real CFB.
I said that CFB are more sensible to power supply quality. This is a major point where a bad supply can ruin all the sonic quality you can expect (specially in respect to dynamic behavior).
As far i'm concerned, and during 4 decades of work, CFB topology, when it was possible, had always given to me better results, both on measurements and in subjective listening arguments. That is the only reason why i appreciate this topology.
I have provided a link with two examples you can simulate. The VFB example is yet damn fast, and not degraded to twist the demonstration. The Current feedback mod was tuned for optimal result as well, with the exact same components. The result is clear:
VFB slewrate: 222V/µs CFB slew rate: 1200V/µs
VFB bandwidth: 300Khz, CFB bandwidth: 5Mhz
VFB THD : 0.0023% CFB THD: 0.000204%
IM is divided by 10 too when CFB.
And in real world, the improvement was subtle, yes, but positive, no doubt about that.
Is that talkative enough, homemodder ? Why don't you play with those simul, and try to understand, or, better try build an SSA, for example, to enjoy in real how it can sound GOOD, as reported by all who had build one ? Instead of posting useless opinions in forums ?
Where is the interest to publish every where and each time someone is speaking about CFB that VFB "can be better", with no technical argument ?
You can explore Error correction too, an other interesting topology, with nice results...
If you want to enjoy electronic for audio and improve your skill, do the same that i've done during all my (long) life: Learn, experiment, as objectively as you can, improve, imagine new solutions, break your certitudes all mornings and experiment again. And never stay stuck or believe anything you have read here or here.
Make your mind by yourself.
I said that CFB are more sensible to power supply quality. This is a major point where a bad supply can ruin all the sonic quality you can expect (specially in respect to dynamic behavior).
As far i'm concerned, and during 4 decades of work, CFB topology, when it was possible, had always given to me better results, both on measurements and in subjective listening arguments. That is the only reason why i appreciate this topology.
I have provided a link with two examples you can simulate. The VFB example is yet damn fast, and not degraded to twist the demonstration. The Current feedback mod was tuned for optimal result as well, with the exact same components. The result is clear:
VFB slewrate: 222V/µs CFB slew rate: 1200V/µs
VFB bandwidth: 300Khz, CFB bandwidth: 5Mhz
VFB THD : 0.0023% CFB THD: 0.000204%
IM is divided by 10 too when CFB.
And in real world, the improvement was subtle, yes, but positive, no doubt about that.
Is that talkative enough, homemodder ? Why don't you play with those simul, and try to understand, or, better try build an SSA, for example, to enjoy in real how it can sound GOOD, as reported by all who had build one ? Instead of posting useless opinions in forums ?
Where is the interest to publish every where and each time someone is speaking about CFB that VFB "can be better", with no technical argument ?
You can explore Error correction too, an other interesting topology, with nice results...
If you want to enjoy electronic for audio and improve your skill, do the same that i've done during all my (long) life: Learn, experiment, as objectively as you can, improve, imagine new solutions, break your certitudes all mornings and experiment again. And never stay stuck or believe anything you have read here or here.
Make your mind by yourself.
Oh, my God, i hope they had changed the electrochemical Caps several times during all those decades !
CFB does appear to have two excellent conveniences. It has higher gain capacity, convenient for medium size power amplifiers. With orthodox measuring, you're measuring either strengths of VFB or the weaknesses of CFB (conveniently reduces reliance on listening tests). The severity of the divergence depends on the amount of gain. And that gives me an idea:
I guess that further comparisons would be more interesting with unity gain buffers? The wipes out the gain question apples versus oranges debates. What do you think of that? I'm in need of a buffer. I have a computer. Any normal computer (not equipped with an audiophile card) needs a buffer for assuring level frequency response (even if it is only just to find out if you needed it--you'll certainly find out), so does VFB or CFB do better for a buffer?
If buffer was a dumb question, then howabout comparing low gain preamplifiers run from superreg power? That would be the next stage after the buffer anyway.
I guess that further comparisons would be more interesting with unity gain buffers? The wipes out the gain question apples versus oranges debates. What do you think of that? I'm in need of a buffer. I have a computer. Any normal computer (not equipped with an audiophile card) needs a buffer for assuring level frequency response (even if it is only just to find out if you needed it--you'll certainly find out), so does VFB or CFB do better for a buffer?
If buffer was a dumb question, then howabout comparing low gain preamplifiers run from superreg power? That would be the next stage after the buffer anyway.
2ppm THD without non switching circuitry and error correction
is just plainly impossible moreover if you re talking about your modded
Crescendo that use a classical two amplifying stages topology.
Wahab, you don't understand? It is simulation. To evaluate. Abstract ! Provided for you can verify.
In real life, the distortion is, of course, higher, but i am not able to provide CFB real numbers, HD and IM are under my (old) measurements instruments capability. Just i assure-you they are reduced in a proportion looking more or less like in the model.
Last edited:
One thing is for sure: if you use a separate transistor (inverting input) to carry the feedback, you will add its own distortion.
At gain 1 i don't see any advantages concerning bandwidth. . The difference will be in the slew rate ?
Last edited:
We use the feedback signal to cancel (partially) errors. If you add error (distortion) to this signal, this error will be subtracted from the original signal, means added in opposite phase.
Too, the delay added by the speed limitation of this extra transistor will reduce error cancellation at high frequency , because you subtract a signal to an other witch in not exactly in phase opposition That explain several phenomenas, like reduced damping factor and increased distortion at high frequencies..
You are attenuating local negative feedback to that 1st emitter.
Increased gain, increased impedance, and increased distortion
(unless a global loop is closed elsewhere, then it doesn't matter),
and less sharpness to the knee of the local pole.
I've used this before to attenuate feedback to the cathode of a
Coincertina phase splitter. Giving +Mu/2 and -Mu/2 gains, and
true equal impedance phase splitting.
Increased gain, increased impedance, and increased distortion
(unless a global loop is closed elsewhere, then it doesn't matter),
and less sharpness to the knee of the local pole.
I've used this before to attenuate feedback to the cathode of a
Coincertina phase splitter. Giving +Mu/2 and -Mu/2 gains, and
true equal impedance phase splitting.
Last edited:
I downloaded the Ltspice - missing zener definitions
http://www.esperado.fr/images/stories/SSA-Crescendo/SSA-crescendo.zip
used diode models from: www.ece.utah.edu/~ece2280/diode.lib
CFA sim doesn't bias correctly, VAS is cutoff
I'm traveling for the next week so may not debug the sim any further - care to correct it for the curious here?
quickly found, fixed input collector R value to web schematic- now the VAS bias is nearly 4x as hot as the VFB amp - is that fair?
I don't understand your problem ( I'm not an expert in simulation software, i work in real world, old style).
I run LTspice at home, and this sims runs ok on my computer. Zener had been defined with the included LTspices zeners models. And currents are ok and as expected.
The story is i had made this mod in *real* first, and it worked fine after tuning. Since this time it runs in my system and i am too lazy to open-it and look at the real values.
I had designed this spice months later to respond at a request, and i'm not sure it reflect the exact real values i used. But i'm sure it is not so far, and it measure quite similar. Distortion apart, i can't tell because they are under the limits of my measuring instruments.
I keep Ltspice updated, maybe they've changed their lib components since you ran it - or you added models to your LTspice directories and forgot
It can't be verified that the sim is "fair" if you aren't going to fix the sim values as you claim they need to be to relect your circuits
I and others find you claim of orders of magnitude better 1kHz distortion fro the CFA topology unreasonable - stepping through the circuit values, sim results will help us learn what is going on
It can't be verified that the sim is "fair" if you aren't going to fix the sim values as you claim they need to be to relect your circuits
I and others find you claim of orders of magnitude better 1kHz distortion fro the CFA topology unreasonable - stepping through the circuit values, sim results will help us learn what is going on
Last edited:
I have showed my simulation of the crescendo in other thread(SSA) and result were good but far from what Esperado claim.
THD20k at 70W/8R 0.013422%, at 8W/8R 0.019876%
THD1k at 70W/8R, 0.000984%, at 8W/8R 0.000565%
Strange that THD20k increased at lower power.
Esperado spice models for the laterals were wrong.
dado
THD20k at 70W/8R 0.013422%, at 8W/8R 0.019876%
THD1k at 70W/8R, 0.000984%, at 8W/8R 0.000565%
Strange that THD20k increased at lower power.
Esperado spice models for the laterals were wrong.
dado
They are models for the original Hex from International Rectifier (My parts, not Lateral).
They where picked in the Bob Cordell site, and i recently tried an another source for the same references with very similar results.
Again, the absolute values of distortions and slew rates, has no interest, as they will not reflect the real disparities. Only the difference between the two schematics, because we are in the same landscape, with the same models. Agree ?
But i don't understand those differences. It just convince-me a little more to never lie on Spice simuls for other things than evaluate raw values. or filters fast calculations.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.