Fortunately we have our Great Papa Guru here, so he is teaching skills that were once lost and forgotten...
Thank you Nelson!!
It doesn't require an exponential slope - just a slope that
rises with Plate voltage. You will notice the the Pentode
has a more linear characteristic to the slope than a triode,
and this is true of many of the solid state devices. Having
a degenerating resistor on the Cathode, Source or Plate
will help to get a better cancellation on this. For that
matter, if you have degeneration on a triode, it might be
helpful to use some Plate resistance.
Nelson, degeneration on a triode means: The slope is not so fast rising as an exponential - correct?
But what I don't understand: If one slope is exponential, it requires another exponential to cancel, otherwiese the fourier transform would give differing terms that could not cancel... So pentodes give no perfect cancelation, only some forier terms are extinquished.
Wouldn't it be better to use a non resistive load for a pentode like element that provides fourierer terms in its behavior to cancel pentode terms???
BTW: Very instructive article. I always thougt, only 2 stages, one after the other could provide distortion cancellation...
But what I don't understand: If one slope is exponential, it requires another exponential to cancel, otherwiese the fourier transform would give differing terms that could not cancel... So pentodes give no perfect cancelation, only some forier terms are extinquished.
Wouldn't it be better to use a non resistive load for a pentode like element that provides fourierer terms in its behavior to cancel pentode terms???
BTW: Very instructive article. I always thougt, only 2 stages, one after the other could provide distortion cancellation...
Degeneration on a triode means that you would be adding
a linear term to the curve, and it would not rise so fast.
There is no such thing as perfect cancellation - you take
advantage of what you are given. As long as the slopes
are in the same direction, you can get some cancellation,
it only remains to tweak the parameters to maximize this
(if you want to).
You can always load a gain device with another gain device
to try for distortion reduction. A mu follower would be a possible
example.
Lastly, you can indeed try for a sweet spot between successive
stages.
a linear term to the curve, and it would not rise so fast.
There is no such thing as perfect cancellation - you take
advantage of what you are given. As long as the slopes
are in the same direction, you can get some cancellation,
it only remains to tweak the parameters to maximize this
(if you want to).
You can always load a gain device with another gain device
to try for distortion reduction. A mu follower would be a possible
example.
Lastly, you can indeed try for a sweet spot between successive
stages.
After reading the J2 thread and "sweet spot", I ask myself, if the 2 resistors in the J2 (where output signal is drawn) are choosen for lowest distortion. So in the end it's optimised for a special drive impedance - if we look for perfects results, we should flatten the loadspeakers impedance or add reactiove elements additionally to the resistive ones in the output stage to allow low distortion all over the frequency band...
Is impedance flattening of the speaker a "MUST" for the J2 output stage???
(Sorry for my silly questions, it's so damn interesting...)
Is impedance flattening of the speaker a "MUST" for the J2 output stage???
(Sorry for my silly questions, it's so damn interesting...)
I imagine the circuit is optimized so the FETs run in the sweet spot for speakers which are "tube friendly" as specified in the manual. I guess a very difficult load may make the output stage run out of the sweet spot.
The J2 derives its output impedance from feedback, so
flattening the speaker's impedance is not essential. The mu
follower is not optimized against impedance, it is optimized
for a current ratio between the gain device on the negative
side and the current source on the top side.
The sweet spot article is not about the J2 per se. The J2 is
not designed around low impedance or low efficiency
speakers, and prefers the kind of speakers you would
normally associate with a 25 watt amplifier.
Thanks for the article Mr. Pass.
Instead of my habbit of buying more expensive/exotic parts I should look at tweaking the values to take things to the next level. Can those with more experience recommend newb/budget gear? I did a search on ebay for distortion analyzer and oscilloscope and turned up a bewildering amount of items. What features should I be looking for?
Thanks
Garrett
Instead of my habbit of buying more expensive/exotic parts I should look at tweaking the values to take things to the next level. Can those with more experience recommend newb/budget gear? I did a search on ebay for distortion analyzer and oscilloscope and turned up a bewildering amount of items. What features should I be looking for?
Thanks
Garrett
Just a quick thank you (I think) to Mr. Pass for another great article. Of course, with him showing once again that things aren't as easy as "this one is the best" the whole thing becomes ever more complicated.
What you mean it not just right or wrong?! There's shades of gray in between!!!
Hopefully when it's done, there's more than shades of gray, one hopes the image is in technicolor!
Peace,
Dave
What you mean it not just right or wrong?! There's shades of gray in between!!!
Hopefully when it's done, there's more than shades of gray, one hopes the image is in technicolor!
Peace,
Dave
Enjoyed the article, thank you.
Had a thought, how about plotting the 'Sweet spot' for the common devices used, VGS against ID, that way we could start with building blocks near enough optimized, especially useful for those of us that don't own a distortion/spectrum analyzer. It could be called the NP curve/trace.
Had a thought, how about plotting the 'Sweet spot' for the common devices used, VGS against ID, that way we could start with building blocks near enough optimized, especially useful for those of us that don't own a distortion/spectrum analyzer. It could be called the NP curve/trace.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.