1/2 of the tube for each channel?

[Wavebourn]

As I mentioned before, there is another effect: isolation of signal source impedance from input impedance of the amp. If that impedances are non-linear, or reflect load, tubes minimize their impact on sound.

 

[SY]

Really? What's the source impedance of a CD player or iPOD?

 

[Pano]

Low. 100 ohms or less. Volume controls placed after them? Not so low.

 

[SY]

2k5 max at half volume. Lower at all other settings.

Input nonlinearity of the chip amp?

 

[Wavebourn]

We can experiment: take a test signal from anode of power amps' output tube, and feed it to input of a SS amp through a resistor of huge value, to form a voltage divider of this huge resistor and input impedance of the SS amp. I used resistor from 220V power outlet to output of a SS amp to test it's output impedance linearity.

 

[SY]

How about something easier? Set the volume control to -6dB, then measure the distortion.

 

[Pano]

-6dB is not typical. -20dB or more is. It's not good gain structure, but it's very typical.
Might also want to measure freq. response.

 

[SY]

At -20dB, the max source impedance will be 850 ohms or so, reducing the effect (if it even exists) even more.

 

[Pano]

a 50K pot at -20dB would have an output impedance of 850R?

 

[Wavebourn]

How about something easier? Set the volume control to -6dB, then measure the distortion.

The easier to me means to stretch as wide as possible, observe, then extrapolate back to real conditions.

If people report some improvement it must have some rational explanation except boring "they prefer distortions".

 

[Pano]

Most of the tube buffers I've heard seemed to have more "drive" than without. I don't know why this should be so, but I hear it often. Some have no noticeable effect.

 

[SY]

a 50K pot at -20dB would have an output impedance of 850R?

Schematic shows a 10k pot feeding the amp. If you use a 50k pot, the max source Z at -20dB is about 4k, still pretty darn small.

 

[Pano]

Yes, I see about 3.6K for the 50K pot. That's not real high.

 

[Pano]

Looking back at the schematic, wouldn't it have been better to place the volume pot before the tubes? 50K could be used there, instead of 10K after the tubes.

 

[Wavebourn]

Looking back at the schematic, wouldn't it have been better to place the volume pot before the tubes? 50K could be used there, instead of 10K after the tubes.

Probably yes.

And, I am curious, what is to use JFET buffer instead?

 

[SY]

At those voltages, FETs would work better.

 

[wavycertified]

Intresting. I found the original buffer article with analysis, please have a look Good Design Bad Design and DIY: Entry to wonderful world of tube audio

 

[Pano]

Thanks for the link!