At this point, I'm 100% sure Wavebourn's solution is far more
relevant to Silas' immediate needs. But I'm gonna post this
anyway. Worked too darn hard on it not to.
Zen bent backwards through a real triode controls the mess.
Aleph with the gyrator thingie goin on is the Anti-Triode.
The JFET cascode up front moves the signal common to the
bottom rail, and improves overall PSRR. No voltage gain here.
Center holds plenty well enough to be both OTL and OCL.
Only thing I don't care for is 1000uf coupling cap, but was
necessary evil to swing Zen and Aleph in phase at 20Hz.
I'll be thinking hard how to get rid of it.
relevant to Silas' immediate needs. But I'm gonna post this
anyway. Worked too darn hard on it not to.
Zen bent backwards through a real triode controls the mess.
Aleph with the gyrator thingie goin on is the Anti-Triode.
The JFET cascode up front moves the signal common to the
bottom rail, and improves overall PSRR. No voltage gain here.
Center holds plenty well enough to be both OTL and OCL.
Only thing I don't care for is 1000uf coupling cap, but was
necessary evil to swing Zen and Aleph in phase at 20Hz.
I'll be thinking hard how to get rid of it.
Attachments
So: D1 + D2 creates the -1.25V servo reference.
Emitters of Q1 + Q2 simply stand on top of that,
And .44V dropped thru R5 when set to idle at 2A.
We are back at GND without too much error.
Q1 and Q2 serve completely different functions,
though I've drawn them sort of symetrical, as
they both need a similar bootstrapping.
The source of the upper mosfet was chosen to
drive both bootstraps, cause it has the lowest
impedance, and least effect upon the output.
The 3K3 in the lower bootstrap is no mistake.
If its the same as the top side, it don't work.
Both straps grant an extra +/-5V to drive with.
Emitters of Q1 + Q2 simply stand on top of that,
And .44V dropped thru R5 when set to idle at 2A.
We are back at GND without too much error.
Q1 and Q2 serve completely different functions,
though I've drawn them sort of symetrical, as
they both need a similar bootstrapping.
The source of the upper mosfet was chosen to
drive both bootstraps, cause it has the lowest
impedance, and least effect upon the output.
The 3K3 in the lower bootstrap is no mistake.
If its the same as the top side, it don't work.
Both straps grant an extra +/-5V to drive with.
Wavebourn said:
Only if swinging 35W SE triode within 1V of either rail.
Actually, in this case, its more like SE UL...
I had problems with grid leak when starving the plate
too much in saturation. And VGS bias when too near
cutoff. So the triode couldn't easily be run full swing.
But that doesn't mean I was ready to abandon it.
I've cheated and abused its plate to become in effect
the "screen" of the lower mosfet. 50% feedback etc...
Triode stays at happy (though awfully low) voltages.
Wavebourn,
Checked Salas circuit in LTspice and lowend gets f-cked up when going down this low. 4,7u gives a Fu of 95Hz. So to reach down to 5Hz the cap has to be 100u!
I also get a peak of +6dB at 150kHz.
Otherwise it sims fine when using proper tubes instead of 12AU7.
Compared to a 100H choke the PSRR is almost on par, but compared to a CCS it is lousy.
I wouldn't sweat LTSpice. Gives goofy answers that might be
frequency dependent anytime the impedances swing too stiffly.
I like to set up a dummy signal at 96KHz just to make sure it
don't attempt too many "smart" calculation shortcuts. Quite
a significant calming effect upon the higher order harmonics.
frequency dependent anytime the impedances swing too stiffly.
I like to set up a dummy signal at 96KHz just to make sure it
don't attempt too many "smart" calculation shortcuts. Quite
a significant calming effect upon the higher order harmonics.
Kenpeter,
So, what does your sims say?
I didn´t care very much about the peak as it might as well be the MOSFET-models. Checked with some other models and it was gone.
But the hf-peak wasn´t the issue, it was the calculation of the 47uF cap with its -3dB at 95Hz. Also a possible bettering of PSRR could be of interest.
In simple terms:
Seems like the gyrator as is about equivalent to a ca 60H/2k choke with a parallell cap of ca 60p. But without the low-end loss of it.
It also seems to have the equivalent resistance of over 300k.
Nice circuit!
So, what does your sims say?
I didn´t care very much about the peak as it might as well be the MOSFET-models. Checked with some other models and it was gone.
But the hf-peak wasn´t the issue, it was the calculation of the 47uF cap with its -3dB at 95Hz. Also a possible bettering of PSRR could be of interest.
In simple terms:
Seems like the gyrator as is about equivalent to a ca 60H/2k choke with a parallell cap of ca 60p. But without the low-end loss of it.
It also seems to have the equivalent resistance of over 300k.
Nice circuit!
Mine? Not currently relevant to Anti-Triode, though could be...revintage said:
All harmonics in this example FFT -85db down or better...
Couldn't best -45db without the 96KHz dummy circuit you
might take notice in the lower left corner.
Try this same LTSpice trick with Wavebourn's Gyrator, may
give better simulated results.
Attachments
revintage said:
I simmed and calculated in mind, fot 6V6 tube. Sorry. Use 47 uF, or 4.7 uF, but ten times higher values of resistors.
Speaking of reference to ground, it is not the cap to ground; it is a voltage regulation in respect to ground. Different topology. But instead of that I would prefer better PS filtering.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.