Hi all,
I'm playing with a new amplifier stage; it is a cascode input stage, with two halves of two different 6N6P tubes, fed by a CCS; the input stage drives a SE output stage, made with two 6AS7 tubes.
THIS is the schematic I would like to test.
Actually only the input cascode stage, without any sort of feedback, is built and "working".
My idea is the following: usually the upper grid of the cascode stage is held at a constant voltage (in my case at about 150V with a regulated PS); what about fed the grid with a part of the output signal, directly from the upper tube plate (R22 and C1, local feedback)?
I think it should work fine.
The idea sound so sweet to me that I added local feedback in the same flavour also to the output tube (R16, C11); moreover, due to the fact the 6AS7G tube is not linear at all, I'm thinking to give it a large amount of feedback, almost 100% (lower or remove completely R16).
Your comment?
Thanks in advance.
Ciao,
Giovanni
I'm playing with a new amplifier stage; it is a cascode input stage, with two halves of two different 6N6P tubes, fed by a CCS; the input stage drives a SE output stage, made with two 6AS7 tubes.
THIS is the schematic I would like to test.
Actually only the input cascode stage, without any sort of feedback, is built and "working".
My idea is the following: usually the upper grid of the cascode stage is held at a constant voltage (in my case at about 150V with a regulated PS); what about fed the grid with a part of the output signal, directly from the upper tube plate (R22 and C1, local feedback)?
I think it should work fine.
The idea sound so sweet to me that I added local feedback in the same flavour also to the output tube (R16, C11); moreover, due to the fact the 6AS7G tube is not linear at all, I'm thinking to give it a large amount of feedback, almost 100% (lower or remove completely R16).
Your comment?
Thanks in advance.
Ciao,
Giovanni
Have not tried feedback to grid of top triode in a cascode but I don't see why it would not work. If uou want to linearize the whole cascode, you could just as well feed back to the grid of the bottom tube.
regarding 6AS7G, plate to grid feedback in it's case will not be very usable, as the tube has a mu of only about 2. It is difficult to get the proper swing to drive it to begin with, and with local feedback it gets only worse. besides, the improvement in linearity is only as much as the effective 'mu' of the whole stage is reduced from the mu without feedback, which will be only a little, given that the mu of the 6AS7 starts at 2.
regarding 6AS7G, plate to grid feedback in it's case will not be very usable, as the tube has a mu of only about 2. It is difficult to get the proper swing to drive it to begin with, and with local feedback it gets only worse. besides, the improvement in linearity is only as much as the effective 'mu' of the whole stage is reduced from the mu without feedback, which will be only a little, given that the mu of the 6AS7 starts at 2.
I have been using a similar technique with pentodes. It works great if the top "cascode" (screen input, plate output in my case) plate is only loaded by a CCS, due to the large loop gain available. It has very low distortion and the subsequent follower's gain signature is effectively removed from the resulting output signal. See the figure at bottom of this post:
http://www.diyaudio.com/forums/showthread.php?postid=1172104#post1172104
But for an output loaded cascode, you won't have any gain in the loop to speak of unless you use a follower/buffer like I did to isolate the load.
Nelson Pass also has a SS amplifier, the "First Watt model F4" which uses a similar technique (but with a big honking JFET in resistive mode on the bottom), but with current feedback instead. He calls it "modulated cascode" I think.
Edit: I just noticed the R16,C11 on your schematic. Is just using normal plate-grid feedback on the output stage. Not enough gain from the 6AS7, as mentioned earlier, to bother I think.
Don
http://www.diyaudio.com/forums/showthread.php?postid=1172104#post1172104
But for an output loaded cascode, you won't have any gain in the loop to speak of unless you use a follower/buffer like I did to isolate the load.
Nelson Pass also has a SS amplifier, the "First Watt model F4" which uses a similar technique (but with a big honking JFET in resistive mode on the bottom), but with current feedback instead. He calls it "modulated cascode" I think.
Edit: I just noticed the R16,C11 on your schematic. Is just using normal plate-grid feedback on the output stage. Not enough gain from the 6AS7, as mentioned earlier, to bother I think.
Don
The TubeCad Ultralinear Cascode article:
http://www.tubecad.com/january2000/page5.html
While the Ultralinear case does LOOK similar, the operation with a CCS on the top plate makes for a rather different mode of operation. The Ultralinear case typically uses some percentage of the plate voltage fed back to the screen or cascode grid, and drives the actual output with a varying plate current. Preferably with a low output Rp to drive the load.
The CCS loaded cascode or cas-pentode operates at constant plate current and mandates that the feedback grid operate at exactly the bottom tube's (or screen grid 2 to grid 1) Mu to maintain that constant current.
A high Rp for the top tube or pentode plate is desired to obtain the high loop gain required to enforce the accuracy to Mu. You can expect low distortion from the cascode case if the top plate is not loaded down.
Practically speaking, the cas-pentode is easier to impliment than the true cascode, since it does not require a floating filament supply nor a pull down resistor for the feedback grid.
Just as a point of interest, I have used this CCS-cascode-follower approach to make a true SS triode using the "Early effect" of a bipolar transistor for the bottom device.
Call it an "Early triode".
The "Early effect" is responsible for the slope of the collector curves of bipolar SS devices, which makes them all converge back to a hypothetical negative "Early voltage" when extended back to the left on typical collector voltage versus current plots. Selection of a suitable device can make for a nice linear amplifier stage (but with a constant current to voltage transfer ratio). (I use this as the voltage gain, or VAS, stage in a SS Lin topology amplifier which then does not require the usual global feedback of most SS amps.)
Don
http://www.tubecad.com/january2000/page5.html
While the Ultralinear case does LOOK similar, the operation with a CCS on the top plate makes for a rather different mode of operation. The Ultralinear case typically uses some percentage of the plate voltage fed back to the screen or cascode grid, and drives the actual output with a varying plate current. Preferably with a low output Rp to drive the load.
The CCS loaded cascode or cas-pentode operates at constant plate current and mandates that the feedback grid operate at exactly the bottom tube's (or screen grid 2 to grid 1) Mu to maintain that constant current.
A high Rp for the top tube or pentode plate is desired to obtain the high loop gain required to enforce the accuracy to Mu. You can expect low distortion from the cascode case if the top plate is not loaded down.
Practically speaking, the cas-pentode is easier to impliment than the true cascode, since it does not require a floating filament supply nor a pull down resistor for the feedback grid.
Just as a point of interest, I have used this CCS-cascode-follower approach to make a true SS triode using the "Early effect" of a bipolar transistor for the bottom device.
Call it an "Early triode".
The "Early effect" is responsible for the slope of the collector curves of bipolar SS devices, which makes them all converge back to a hypothetical negative "Early voltage" when extended back to the left on typical collector voltage versus current plots. Selection of a suitable device can make for a nice linear amplifier stage (but with a constant current to voltage transfer ratio). (I use this as the voltage gain, or VAS, stage in a SS Lin topology amplifier which then does not require the usual global feedback of most SS amps.)
Don
First of all, thanks to everybody for your answers.
THIS is what is now on my desk, not yet tried it.
I will draw a different shcematic, maybe without the need of an external, floating power supply.
So, it seems that the input cascode stage would work fine, low distorsion and output not influencing it; of course I have now some other questions for you.
1. What about tube drift with age? will the lower 6N6P change its plate voltage (operating point) with age, moving the operating point of the whole cascode stage?
2. If local feedback on the 6AS7 output tube is not efficient, how can I make the output stage more linear?
What about 100% local feedback?
I could also add a driver stage (a CCS loaded ground cathode) with a 6N6P, and thus I can have plenty of gain for the local feedback.
Ciao,
Giovanni
Well, actually the CCS on top is not only a CCS, but a CCS with a "mu" tap; I mean, a cascode mosfet CCS that keeps the load "disconnected" from the lower tube.smoking-amp said:
THIS is what is now on my desk, not yet tried it.
I will draw a different shcematic, maybe without the need of an external, floating power supply.
So, it seems that the input cascode stage would work fine, low distorsion and output not influencing it; of course I have now some other questions for you.
1. What about tube drift with age? will the lower 6N6P change its plate voltage (operating point) with age, moving the operating point of the whole cascode stage?
2. If local feedback on the 6AS7 output tube is not efficient, how can I make the output stage more linear?
What about 100% local feedback?
I could also add a driver stage (a CCS loaded ground cathode) with a 6N6P, and thus I can have plenty of gain for the local feedback.
Ciao,
Giovanni
"1. What about tube drift with age? "
I have not operated this configuration long enough to gain such experience, but here is my guestimate:
The Mu is pretty much a geometrical feature and will stay pretty much constant. The transconductances of the grid and plate on the other hand will both fall as emission drops, but they will maintain there ratio (Mu). So, effectively, the loop feedback gain of the stage will fall with time (but not the closed loop voltage gain Mu), causing an increase in output distortion. Once cathode emission falls below the CCS current however, the plate will suddenly jump to max + voltage in a futile attempt to correct it.
2) 100% local feedback, I assume, means going to cathode follower mode, which will greatly increase the input drive voltage requirement. Otherwise, some kind of gain put in somehow.
Don
I have not operated this configuration long enough to gain such experience, but here is my guestimate:
The Mu is pretty much a geometrical feature and will stay pretty much constant. The transconductances of the grid and plate on the other hand will both fall as emission drops, but they will maintain there ratio (Mu). So, effectively, the loop feedback gain of the stage will fall with time (but not the closed loop voltage gain Mu), causing an increase in output distortion. Once cathode emission falls below the CCS current however, the plate will suddenly jump to max + voltage in a futile attempt to correct it.
2) 100% local feedback, I assume, means going to cathode follower mode, which will greatly increase the input drive voltage requirement. Otherwise, some kind of gain put in somehow.
Don
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.