Memory Distortion - 100dB Feedback Amp on step 8

[Tachikoma]

Memory Distortion Philosophies

I have no idea what he's talking about... but the way his design blatantly goes against all "audiophile design philosophy" is rather intriguing

[nelsonvandal]

I also find this article intriguing and substantial. We all know that THD and harmonic spectrum doesn't say it all. I've tried CFP and cascoded input stages, but not yet the combination advocated by the author. Contradictory to the article I find CFP to be a major improvement compared to cascode. The combination of JFET input and bipolar "slave" gives the best performance with the best of both worlds - warm and full musical sound and low bias input currents from the JFETs and the clean, detailed airy presentation of bipolars. Try this and be impressed!

[Dreadhead]

hmmmmm Drive a transistor with a ultrashort step function and there is distortion in the circuit? WOW. I wonder what would have happened to any speakers attached to said system.....

Seems like quackery to me.

[nelsonvandal]

Fancy cables, connectors, capacitors, volume knobs, fuses, expensive wooden earcups, power cords etc etc is quackery. Adding CFP to the input is not. Cascoding is not. It's very substantial both by measurement and in listening tests. Maybe it's not memory distortion that explains why these tweaks make a better sound, maybe it is. Do the tweaks and find out for yourself if it's all BS or a revelation.

[Dreadhead]

Fancy cables, connectors, capacitors, volume knobs, fuses, expensive wooden earcups, power cords etc etc is quackery. Adding CFP to the input is not. Cascoding is not. It's very substantial both by measurement and in listening tests. Maybe it's not memory distortion that explains why these tweaks make a better sound, maybe it is. Do the tweaks and find out for yourself if it's all BS or a revelation.

I'm not saying that feeback or cascoding are quackery. Listening is not a test it's an opinion (unless it's DBT) I do them too but I at least understand that when I go in (I'm not saying you don't).

What I think is quackery is blaming something you need a testbench and a square wave impulse to measure for something that you can hear. For example you can measure the effect of cables impedance and resistance but after you do some math and show that they do incredibly small things to the signal that are almost certainly below the fidelity you can measure using instruments at the transducers (speakers or headphones) much less hear (unless they inserted a active resistor or other component in the cables).

Many people disagree with me on this and many other things and to be honest as long as you're not trying to force your cables/demagnetizer/etc on me go for it.

It also appears that this distortion is apparently fairly well known in circles where the amps deal with large transients all the time (i.e. not audio).

Btw: I don't think wooden earcups are necessarily quackery if they change the volume/shape to of the air behind the driver, then they can make a difference though how big I don't know.

[nelsonvandal]

Btw: I don't think wooden earcups are necessarily quackery if they change the volume/shape to of the air behind the driver, then they can make a difference though how big I don't know.

OK, not quackery, just fetishism. Sorry, I couldn't help myself. I agree, a wooden cup probably make a difference, but I don't think they could sell an equally good plastic cup for the same price.

About memory distortion, I don't think it's about pushing spikes through a transistor. You're supposed to measure the offset caused by heating a transistor with the amp running at open loop, and sum it up with the dissipated power of the same transistor fed with an arbitrary input signal with the amp running in closed loop to get an "annoyance factor". One would think that a class (A)B output stage would be most affected but it's not. I think this is in harmony with my listening experience - the input stage has a much greater influense over the sound compared to the output stage. Another way to lessen the subjective annoyance is to use better transistors with low Cob for the input-, cascode- and VA-stages.

[aerius]

hmmmmm Drive a transistor with a ultrashort step function and there is distortion in the circuit? WOW. I wonder what would have happened to any speakers attached to said system.....

Seems like quackery to me.

You don't get it do you? It's done to isolate the effects so that they can be seen. It's the same reason a simple 1kHz sine wave is used for distortion spectrum measurements instead of playing music through the amp.

It's basic science, you have a new parameter which needs to be investigated. It has to be isolated in such a way that it can be reliably measured and controlled, and once we know how to control it we go back to the real world and see if we can find correlations in listening tests with actual music. In other words, once we can measure and adjust the amount of "memory distortion" in an amp, we then build a series of amps with differing amounts of MD ranging from zero to lots and see if we can hear a difference and where the threshold is.

Btw: I don't think wooden earcups are necessarily quackery if they change the volume/shape to of the air behind the driver, then they can make a difference though how big I don't know.

They don't need to change the volume or shape of the air behind the driver to make a difference. It's called resonance. Build a speaker out of MDF. Then build an identical one out of Baltic birch ply. They ain't gonna sound the same, and the same is true of headphone cups. Materials matter, and it can be measured.

[slwiser]

This sounds a little like the patent of David Berning and his *-ZOTL designs.

I had one of his amps for a while and it is real nice. What his circuit does is isolate the speaker from the amp such that it the amp sees a constant impedance of about 4K. This results in very high damping factors.