Attached is a schematic and distortion measurements at 1W and 10W of a shunt feedback output stage KT88 amp that I built for my brother. Just thought I'd share.
Attachments
Lucky Brother !
I like the use of the heavy perf for the top plate. Great ventilation and freedom to mount boards pretty well anywhere . . . .
Looking at the graphs I have a question. Are the peaks at the bottom end 60, 120 etc. from the test gear power supply ? I see that they don't move much with power output but otherwise I don't know how to interpret them.
I'm planning and getting parts together to build a 6BX7 or BL7 > 6336 differential amp and lately, looking for ideas and details, I've been reading threads you've posted on but this is the first time I've seen one of your creations. Thanks for sharing it!
I like the use of the heavy perf for the top plate. Great ventilation and freedom to mount boards pretty well anywhere . . . .
Looking at the graphs I have a question. Are the peaks at the bottom end 60, 120 etc. from the test gear power supply ? I see that they don't move much with power output but otherwise I don't know how to interpret them.
I'm planning and getting parts together to build a 6BX7 or BL7 > 6336 differential amp and lately, looking for ideas and details, I've been reading threads you've posted on but this is the first time I've seen one of your creations. Thanks for sharing it!
I'm not sure. I was really rushed when I took these measurements as I was trying to get this done before Christmas so I could deliver the present. I would have liked to get to the bottom of it. I would like to say that it was not the amp, but I used some 800uF caps for the power supplies that I had laying around and I fear that may have been on the large size and some of the large charging current surges may be coupling. 60Hz is hard to explain, though. I don't know. I would have liked to play some more with it to get to the bottom of that but didn't have any time. Also, the amp had a buzzing that I could only hear with my ear about an inch from the speaker cone. I was hoping it would be totally silent with no signal, but some of that is getting out.
It doesn't need to be so high, I'm sure you could get by with much less. I made a symmetrical +-500V raw supply for the regulators and then had a couple of negative regulators for the negative voltages and a few positive regulators for the positive voltages. The power tubes got their own power transformer.
The LEDs are a string of 8 blue LEDs to get the 27V necessary for the 6BL7 bias. I didn't pick anything special for these.
I put the gnfb on the schematic but I didn't connect it, so this amp is pentode-mode without gnfb. That sounds crazy at first, but the shunt feedback reduces the gain of the pentodes to about 5 and gives the resulting circuit far more linear (and lower rp) curves than a 300B, a poor-man's 300B killer if you will. Converting the KT88s to triode mode would give a slight decrease in distortion (probably negligible) and would limit power output because now you wouldn't be able to reach saturation without grid current, which the shunt feedback resistive divider prevents. So you would be giving up a lot of power output for a tiny increase in performance.
I measured onset of clipping at roughly 40W (just by measuring peak voltage on a scope and calculating).
The 27V on the 6BL7 cathode is referenced to the -300V supply. I know that is bad form on my part and confusing, but it was easier for me at the moment I did it and I forgot that I had done that when I posted it. I will correct that.
I'm not sure how I would make this work with an N-channel fet. Do you want to draw up what you have in mind and we can comment on it?
The failures you are talking about were a concern to me as well. I have a microprocessor that is in control of the power via SSRs and one of its inputs is a comparator. Basically, if any of the output tubes draw excessive current, the amp will shut down immediately.
The grid leak is present in the amplifier, I just left it out of this simplified schematic. The full schematic is pretty hideously complex and doesn't actually exist except as collection of a bunch of board schematics.
I made a board for the input stage, a board for the driver CCS (cascode depletion-mode fets) and a board for the output stage. The output stage board includes the p-channel fet and an n-channel fet source follower regulator for the output tube screen with a current-limit. The screen voltage will sag if the tube is driven into the screen-current danger zone
.
Thanks for the comments, guys. I haven't been around for a while and it is nice to come back, especially good to see you here, SY.
If you want to see any more schematics or have any more questions, let me know. This was a really fun project because none of the components were all that expensive and it came out sounding really good. I'm also in the process of building a unity-coupled amp for myself with some Plitron transformers, so I'll be posting on that in another thread that I started ages ago and will have to resurrect.
If you want to see any more schematics or have any more questions, let me know. This was a really fun project because none of the components were all that expensive and it came out sounding really good. I'm also in the process of building a unity-coupled amp for myself with some Plitron transformers, so I'll be posting on that in another thread that I started ages ago and will have to resurrect.
How would you get the current to flow from the source of an n-channel fet up through the shunt feedback divider to the node at the tube's anode?
I'd gladly use an N-channel part as there are much better(higher voltage) parts available, but I don't think you can make it work without making things more complex and at least adding one capacitor. It would work but it wouldn't be as elegant, in my opinion.
I'd gladly use an N-channel part as there are much better(higher voltage) parts available, but I don't think you can make it work without making things more complex and at least adding one capacitor. It would work but it wouldn't be as elegant, in my opinion.
Well, you have one end of the 10k resistor at AC ground (the source with no signal) which doesn't change with an n-channel device. You have the DC from the plate through the feedback resistors and into the source, ditto. Maybe I'm missing something (possible- it was a particularly fun New Years last night!), but it seems to me that this would just be a matter of setting the gate voltage correctly.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.