Sure. Note that I have not tested this buffer yet. I have been operating the main circuit into the F4, which has a JFET buffer as input. The main thing to consider is the dissipation. 80V into 8-10 mA divided equally over the two JFETs means they are operating right at the limit. How well it works and longevity is to be tested. It should work but use it with caution and at your own peril 
Of course, we are part of the FAB gang, so we are not afraid.
You can ignore the jumpers, they are for my testing using the hat. Also, R111 may not be needed. Q106 is the CCS fet and R110 controls the current through the buffer. Lower values of R110 result in more current. Q105 should be selected such that at 8-10 mA, Vgs is about -0.25V - 0V.
You will note that the buffer is direct coupled. When the Vds of Q102 is adjusted to be 40V (using RV1), then the gate of Q105 will be at 40V, and its source will be at 40V minus the Vgs value, which is about 0.25V. So, you would get about 39-40V at the C102. I will note that the Vdg will be 40V, which is higher than the spec, so again, it remains to be seen if it works and if it lasts
The main circuit is mostly unchanged from before except the devices themselves. You could even use the old PCB and swap out the devices. I am calling it the test circuit because the final might change.
Test Circuit
Of course, we are part of the FAB gang, so we are not afraid.You can ignore the jumpers, they are for my testing using the hat. Also, R111 may not be needed. Q106 is the CCS fet and R110 controls the current through the buffer. Lower values of R110 result in more current. Q105 should be selected such that at 8-10 mA, Vgs is about -0.25V - 0V.
You will note that the buffer is direct coupled. When the Vds of Q102 is adjusted to be 40V (using RV1), then the gate of Q105 will be at 40V, and its source will be at 40V minus the Vgs value, which is about 0.25V. So, you would get about 39-40V at the C102. I will note that the Vdg will be 40V, which is higher than the spec, so again, it remains to be seen if it works and if it lasts
The main circuit is mostly unchanged from before except the devices themselves. You could even use the old PCB and swap out the devices. I am calling it the test circuit because the final might change.
Test Circuit
Rick, I am not completely sure and I will probably end up trying a couple.
I am thinking of a follower type, because I think it mates better with bi-amping (gain control, distortion profile/tailoring), and choke loading. Most probably Ben Mah's Fokin amp or your (ra7's) TDVs...
One thing I would not dislike is to integrate an input buffer to the amp. This will give more flexibility on the input side, drive the SIT better and perhaps offer some options on the bias too.
I am thinking of a follower type, because I think it mates better with bi-amping (gain control, distortion profile/tailoring), and choke loading. Most probably Ben Mah's Fokin amp or your (ra7's) TDVs...
One thing I would not dislike is to integrate an input buffer to the amp. This will give more flexibility on the input side, drive the SIT better and perhaps offer some options on the bias too.
I will be getting fets now, do you have an idea of the number to buy (all types) to be able to build two pairs?
I am doing the same thing - an amp for each driver.
Have you chosen a choke? I have found the Lundahl 2733 does a good job - I have not tried any other choke - but that one seemed to be the best easily available for the TDV circuit. I gilded the orchid using the negative supply on the drain per Ben Mah. ra7 was kind enough to draw it out - you can find this on the thread - for folks like me who would be sure to do it wrong without a map. I run them without any additional resistance since I wanted to try higher current. I think most of the improvement I heard was getting those power resistors out of the circuit.
The main thing is to have an abundance of heat sink and a fan to insure longevity. Oversize for consideration of summer temperatures.
Installed a prasi (pcb) sourced LT4320 rectifier on one amp. I had to find out. It does allow almost 10% more voltage (37.6 versus 34.7). I put small heatsinks on the mosfets just to be safe. They are barely warm after two hours of playing.
Two stth6110 had been installed initially. This is a very fine rectifier.
Need to try these on the other amplifiers- seems very promising.
As always, my gratitude to ra7.
Have you chosen a choke? I have found the Lundahl 2733 does a good job - I have not tried any other choke - but that one seemed to be the best easily available for the TDV circuit. I gilded the orchid using the negative supply on the drain per Ben Mah. ra7 was kind enough to draw it out - you can find this on the thread - for folks like me who would be sure to do it wrong without a map. I run them without any additional resistance since I wanted to try higher current. I think most of the improvement I heard was getting those power resistors out of the circuit.
The main thing is to have an abundance of heat sink and a fan to insure longevity. Oversize for consideration of summer temperatures.
Installed a prasi (pcb) sourced LT4320 rectifier on one amp. I had to find out. It does allow almost 10% more voltage (37.6 versus 34.7). I put small heatsinks on the mosfets just to be safe. They are barely warm after two hours of playing.
Two stth6110 had been installed initially. This is a very fine rectifier.
Need to try these on the other amplifiers- seems very promising.
As always, my gratitude to ra7.
I am working on that right now. The design is not too sensitive to the spread in jFET parameters. More on this coming soon.
Yeah, it was added to ensure the drain voltage of the gain device can be adjusted. In the previous design, the follower mosfet Vgs plus the gain mosfet Vgs put the drain voltage in the right region. With JFETs, it needs a bit of voltage on the gate to put it in the right place. But, by doing so, the j175 operates with a Vds of 10V or so, i.e., it is in the linear region. The combination of j175 and j111 allows for this and it keeps the design from requiring really high voltage supply.
Still working on refining the buffer. Almost close to being done but I want to try one more configuration. The latest iteration, while intended for driving line level loads, drives my 55 ohm AKGs just fine. More refinements to come. I promise this will be worth the wait.
Made a few measurements today.
41 volts peak to peak (clipping the F4): 0.44% THD with a nice spectrum. I was measuring clean output to about 60V peak to peak. It could do even more but this is more than enough for a preamp. Headroom is good.
About 11Wpeak into 8 ohms (13.2Vp-p):
At about 1W, it does near 0.05x% THD.
All of this is without the buffer into a 10k load provided by the soundcard (RME ADI-2 Pro). The buffer is limiting the peak output, so I am still working it out.
The whole premise of this design was solid state triodes. The performance (and sound) bears this out. So, I am happy.
41 volts peak to peak (clipping the F4): 0.44% THD with a nice spectrum. I was measuring clean output to about 60V peak to peak. It could do even more but this is more than enough for a preamp. Headroom is good.
About 11Wpeak into 8 ohms (13.2Vp-p):
At about 1W, it does near 0.05x% THD.
All of this is without the buffer into a 10k load provided by the soundcard (RME ADI-2 Pro). The buffer is limiting the peak output, so I am still working it out.
The whole premise of this design was solid state triodes. The performance (and sound) bears this out. So, I am happy.
