pmos
Interesting! How will you handle the biasing? With jfets, it's simple; mosfets, being enhancement mode, are trickier. If you've got something cool and clever, I'd love to hear about it.
I never did like AR's way of doing the cross-coupled circuit, by inserting an anode follower in one leg. There's just something esthetically dissatisfying about that.
EDIT: I see your edits. Never mind!
Interesting! How will you handle the biasing? With jfets, it's simple; mosfets, being enhancement mode, are trickier. If you've got something cool and clever, I'd love to hear about it.
I never did like AR's way of doing the cross-coupled circuit, by inserting an anode follower in one leg. There's just something esthetically dissatisfying about that.
EDIT: I see your edits. Never mind!
Maybe you are right SY and I finally will end up with a pair of J-Fets. But finding a good matched pair of P-channel J-Fets is not easy nowadays. Attached my first intended circuit. Bias current through R5/R6 will be set to 2x to 3x the current through the tubes. That way the Fets will not starve off in case of overload and it will keep V_gs to safe values in case the low supply voltage is not up (very unlikely ). Maybe some improvement is possible by replacing R5/R6 by CCS, but I doubt.
). Maybe some improvement is possible by replacing R5/R6 by CCS, but I doubt.Attachments
Circuit looks good! The nice thing about using MOSFETs in source follower configurations is that the Cgs becomes much less of a problem. But the biasing (and parts count) for a JFET circuit is still probably easier.
P-JFETs are pretty easy to get, at least here in the US; I bought 100 2N5462s for $15 from Mouser or Digikey (I forgot which one!), then set up a matching jig and got two well-matched pairs in about 5 minutes. And that left me with enough FETs to build 48 more channels
It's nice that there's another cross-coupled circuit geek besides me; true push-pull has so many advantages, but it's a bit out of fashion these days.
P-JFETs are pretty easy to get, at least here in the US; I bought 100 2N5462s for $15 from Mouser or Digikey (I forgot which one!), then set up a matching jig and got two well-matched pairs in about 5 minutes. And that left me with enough FETs to build 48 more channels
It's nice that there's another cross-coupled circuit geek besides me; true push-pull has so many advantages, but it's a bit out of fashion these days.
SY said:
But they are rebounding. It is just a lot harder to get a PP amp working right, SE is so simple.
Have you read Allen Wright's PP paper?
http://www.vacuumstate.com/pow_announce_2003.htm
I'm guessing he is using a differential CCS with a rule
I1 + I2 = K
dave
An externally hosted image should be here but it was not working when we last tested it.
Totally SYclotronic, dude!
Could be. Could be a simple current mirror. Or even a plain vanilla CCS with a nice hyperbolic ribbon tied around it. But it's magic. (Someone needs to teach the copywriter the difference between "it's" and "its.")
I decided that my circuit will never catch on until I give it a fancy and mysterious name. I channelled the spirit of Bob Carver and have dubbed my rather simple cross-coupled inverter variant "The SYclotron." When used with a Circlotron output stage, the holistic, circular nature of fine music is dramatically revealed! Jungian synchronicity! Rudolph Steiner! Nikolai Tesla!
Could be. Could be a simple current mirror. Or even a plain vanilla CCS with a nice hyperbolic ribbon tied around it. But it's magic. (Someone needs to teach the copywriter the difference between "it's" and "its.")
I decided that my circuit will never catch on until I give it a fancy and mysterious name. I channelled the spirit of Bob Carver and have dubbed my rather simple cross-coupled inverter variant "The SYclotron." When used with a Circlotron output stage, the holistic, circular nature of fine music is dramatically revealed! Jungian synchronicity! Rudolph Steiner! Nikolai Tesla!
Hi,
And why not call it by its proper name?
The Van Scoyoc phase splitter? Wouldn't that be kosher?
BTW, greetz from Benny Glass...we spent the afternoon at Fallafel in Antwerp discussing the birth of a brand new magazine...
I'll keep you all informed...it might take some time to get to press though...
Cheers,
And why not call it by its proper name?
The Van Scoyoc phase splitter? Wouldn't that be kosher?
BTW, greetz from Benny Glass...we spent the afternoon at Fallafel in Antwerp discussing the birth of a brand new magazine...
I'll keep you all informed...it might take some time to get to press though...
Cheers,
Hi,
Don't do it for me, once in Europe you owe it to yourself to visit Belgium, we have the best of everything and I'm not kidding you in any way.
Chateau Meerdael for hors d'oeuvre...Tempted?...
Oh, we Belgians are numero uno when it comes to cooking you know!!! Not France or Italy.
Cheers and we're not chauvinistic either,
Don't do it for me, once in Europe you owe it to yourself to visit Belgium, we have the best of everything and I'm not kidding you in any way.
Chateau Meerdael for hors d'oeuvre...
Tempted?... Oh, we Belgians are numero uno when it comes to cooking you know!!! Not France or Italy.
Cheers and we're not chauvinistic either,
Ran a few quick tests with 100u bypass caps across the source-cathode resistors. Never actually tried that before! Anyway, the gain came up some, but not as much as I expected- only about 25. With 500V supply for B+, we end up with 300V on the plate and about 240V p-p max output before onset of clipping. With a square wave, 5 us rise time for 0 to 90%, and an f3 of 100kHz. Balance is still better than I can measure. Bandwidth is identical on the two halves, both on high and low end. This cross-coupling works as advertised, and this circuit has a decently low parts count compared to the "standard" x-couple or the ARC variant.
If I can figure out the logistics of getting the circuit and my computer in the same room, I'll check out distortion.
I'll probably do a second build with some 12BZ7s I have on hand- their mu is higher and plate resistance is lower than the 5691, so I should be able to swing a bit more, get a bit more bandwidth, and a bit more gain. Stay tuned.
If I can figure out the logistics of getting the circuit and my computer in the same room, I'll check out distortion.
I'll probably do a second build with some 12BZ7s I have on hand- their mu is higher and plate resistance is lower than the 5691, so I should be able to swing a bit more, get a bit more bandwidth, and a bit more gain. Stay tuned.
No, I've been browsing my junk box. I played around with the 12BZ7s back in the late '70s, and still had a couple. If memory serves, they're basically the same as two 12AX7s in parallel- same mu, about half the rp, and double the gm. Should match the 2N5462s nicely.
I've already got the sleeper tube, but you won't drag that one out of me- they're damned hard to find. And it's bad enough that you've been blabbing about the 6528!
Guido: Yes, I know what the copy says, but you always have to assume that some, ah, liberties might be taken with the reality.
I've already got the sleeper tube, but you won't drag that one out of me- they're damned hard to find. And it's bad enough that you've been blabbing about the 6528!
Guido: Yes, I know what the copy says, but you always have to assume that some, ah, liberties might be taken with the reality.
> you've been blabbing about the 6528!
6528 is well known, not widely produced, but readily ordered New from PentaLabs for US$85.20 http://www.pentalaboratories.com/product.asp?pf_id=6528&dept_id=1381 or NOS from VacuumTubesInc for US$60.00 http://www.vacuumtubesinc.com/new5.html
Spilling all your beans: it is like a 6AS7/6080 twin power triode, except everything is double-size, more like a 6336. 6.3V5A heater, 30W per plate, and MUCH higher Mu than 6080 or 6336: Mu is 9. Rp is around 250Ω, Gm is of course around 37,000 µMoho or 1/27Ω The high Mu means it can be driven with a resistance-coupled driver working on the same power supply as the output stage, which doesn't work so well when Mu is 2 or 2.7 as in the pass tubes. If you don't mind the heat, it should make a dandy 20 watt push-pull triode power amp.
6528 is well known, not widely produced, but readily ordered New from PentaLabs for US$85.20 http://www.pentalaboratories.com/product.asp?pf_id=6528&dept_id=1381 or NOS from VacuumTubesInc for US$60.00 http://www.vacuumtubesinc.com/new5.html
Spilling all your beans: it is like a 6AS7/6080 twin power triode, except everything is double-size, more like a 6336. 6.3V5A heater, 30W per plate, and MUCH higher Mu than 6080 or 6336: Mu is 9. Rp is around 250Ω, Gm is of course around 37,000 µMoho or 1/27Ω The high Mu means it can be driven with a resistance-coupled driver working on the same power supply as the output stage, which doesn't work so well when Mu is 2 or 2.7 as in the pass tubes. If you don't mind the heat, it should make a dandy 20 watt push-pull triode power amp.
Bring out your dead! (thread, that is)
I finally got around to measuring distortion. Unfortunately, I'm not set up yet to do this differentially, so this is measured at each output. I'm using an HP3581A waveform analyzer- it's good to about .01-.02%- generously on loan from John Curl.
At 1 kHz, 4V peak (8V peak-to-peak) output, the 2nd harmonic is about .06%, third and higher below the measurement limit. At 35V peak output (enough to clip most EL34-type output stages), 2nd rises to 0.3%, third and higher are still below residual.
At some point, I'll rig up a notch filter so I can check this at higher swings.
This is low enough distortion that I'm sorely tempted to go back to the callow philosophy of my youth and build up a zero-loop-feedback amp.
I finally got around to measuring distortion. Unfortunately, I'm not set up yet to do this differentially, so this is measured at each output. I'm using an HP3581A waveform analyzer- it's good to about .01-.02%- generously on loan from John Curl.
At 1 kHz, 4V peak (8V peak-to-peak) output, the 2nd harmonic is about .06%, third and higher below the measurement limit. At 35V peak output (enough to clip most EL34-type output stages), 2nd rises to 0.3%, third and higher are still below residual.
At some point, I'll rig up a notch filter so I can check this at higher swings.
This is low enough distortion that I'm sorely tempted to go back to the callow philosophy of my youth and build up a zero-loop-feedback amp.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.