Single or dual differential?

[hitsware]

Originally posted by PMA
Simple is good, but not the best. This amp (post 81) will have pretty high distortion, driving gate-source capacitance of the Mosfets.

I've done something simular (single ended 2sk170 driving Hitachis with no feedback) and got ~0.6% THD 20 to 20k. Not as bad as one would expect. (or get with IR types) IMO .... mike

[Nelson Pass]

Every type of device has it's own sweet spot(s). Don't
substitute devices into the same circuits and imagine that
you have a real comparison.

[hitsware]

Originally posted by Nelson Pass
Every type of device has it's own sweet spot(s). Don't
substitute devices into the same circuits and imagine that
you have a real comparison.

Ya Mon .... Forgot to mention used the Hitachis as followers ...

[hitsware]

Originally posted by john curl
The circuit should work, just as it is. Gate resistors are most probably optional on this design. Why? I don't know why you like gate resistors of jfets. All it does, in this case, is to make the input more noisy.

If'n you use a gate resistor with a jfet (at the input) ..... You get a rolloff that completely negates any kind of TIM .... I use 10K with 2sk170 .......

[Steve Eddy]

Originally posted by hitsware
If'n you use a gate resistor with a jfet (at the input) ..... You get a rolloff that completely negates any kind of TIM .... I use 10K with 2sk170 .......

Hmmmm. But since FET input capacitance is so non-linear, wouldn't such a high value input resistor just acerbate the problem and produce greater amounts of high frequency distortion which could fold back down into the audio band?

A shunt RC network would ultimately be preferable, yes? Assuming of course that one's goal is optimal objective performance.

se

[hitsware]

>But since FET input capacitance is so non-linear, wouldn't such a high value input resistor just acerbate the problem and produce greater amounts of high frequency distortion which could fold back down into the audio band?

It seems that would be the case, but in practice it works. It's a differant ball game than with something like the mundo IR type fets where the output (drain) reflects back to the gate. Using a 10kHz squarewave as referance, I get nice smooth response.

>A shunt RC network would ultimately be preferable, yes? Assuming of course that one's goal is optimal objective performance.

Yes, theorieticly. But sometimes to me super simple elegance in the schematic can outweigh other factors. I must say however I've not aurally A-B 'd the 2 approaches. Usually my squarewave approach is pretty dependable though...........mike

[PMA]

A question is if we consider THD of 0.6% to be low or high. For me that is high. How is the THD at 10 - 20kHz? Did you measure IMD - probably more problems with Mosfet's non-linear input capacitance.

[hitsware]

Originally posted by PMA
A question is if we consider THD of 0.6% to be low or high. For me that is high. How is the THD at 10 - 20kHz? Did you measure IMD - probably more problems with Mosfet's non-linear input capacitance.

For a 0 feedback (global anyways) amp, I consider it slightly high but workable. No I didn't measure IMD (or THD below 20Hz), so that could be bad, but again I've always had good results using carefull scrutiny of squarewave response.

[PMA]

I am sorry I meant 10kHz - 20kHz, as THD is usually worse at higher frequencies.

[Bricolo]

Originally posted by hitsware


For a 0 feedback (global anyways) amp

what scematic are you talking about?