This thread comes about as a blunder on my part mentioned in another thread.
I had built a circlotron and was testing out the amp stages, but using only one pair of floating independant 200V supplies.
Not really thinking at the time that what I was actually doing by connecting the supplies up to the two channels I was in fact paralleling out the outputs....
Ive been musing on this and wonder if it isnt possible to use only two supplies, or failing that, a transformer with only two 145V windings...
Would the action of two bridge-recs (One for each supply on a single Tx winding) give isolation?
Or use chokes on the DC side to isolate each supply....?😱
I had built a circlotron and was testing out the amp stages, but using only one pair of floating independant 200V supplies.
Not really thinking at the time that what I was actually doing by connecting the supplies up to the two channels I was in fact paralleling out the outputs....
Ive been musing on this and wonder if it isnt possible to use only two supplies, or failing that, a transformer with only two 145V windings...
Would the action of two bridge-recs (One for each supply on a single Tx winding) give isolation?
Or use chokes on the DC side to isolate each supply....?😱
Why use two supplies when you only need one?
http://www.diyaudio.com/forums/tubes-valves/19624-elliptron-schematic.html
Then there is the Crowhurst Twin Coupled amplifier setup.
http://www.audioxpress.com/magsdirx/ax/addenda/media/1960crowhurst.pdf
http://www.diyaudio.com/forums/tubes-valves/19624-elliptron-schematic.html
Then there is the Crowhurst Twin Coupled amplifier setup.
http://www.audioxpress.com/magsdirx/ax/addenda/media/1960crowhurst.pdf
Last edited:
Smoking Amp, did you ever build either the dual supply "Elliptron" or the single supply one with CT choke?
If you did try them how did they turn out?
Cheers Matt.
If you did try them how did they turn out?
Cheers Matt.
Thanks for the links...
While they are quite interesting and alternative, I cant see how I could adapt those ideas to use with a conventional OTL of Circlotron type--Unless I was to add a Transformer.
--Then again I'm not very bright!🙄
While they are quite interesting and alternative, I cant see how I could adapt those ideas to use with a conventional OTL of Circlotron type--Unless I was to add a Transformer.
--Then again I'm not very bright!🙄
"Smoking Amp, did you ever build either the dual supply "Elliptron" or the single supply one with CT choke?"
I'm afraid I haven't yet. I have all the parts rounded up to do either the floating supply Elliptron version (dual bobbin pwr xfmrs) or the single supply design (an extra OT used for the center tapped inductor). I haven't found any good source of inexpensive quality wound center tapped inductors though (hence having to use another OT), so I haven't gotten too excited about it so far. If a good CT inductor were available cheap (and it should be cheap, since it only requires a contiguous layer wind) my excitement would hit critical level.
Another issue with the single supply version is the resistance of the center tapped inductor forms a time constant with the floating electrolytics. For class AB, the power draw varies thru the inductor, so the resonance can get excited. Either you have to use a very low resistance inductor or a servo circuit to control the voltage on the electrolytics.
I do have that servo worked out, it's not too hard to do. Just put resistors from each of the two tube cathodes down to the gate of a Mosfet variable current source (another resistor from gate to source to form a V divider then). The Mosfet current source then pulls its drain current from the center tap of the inductor to keep the electrolytics charged to a constant voltage via the feedback network. Only a limited voltage variation required across the Mosfet, so not a big power waste. And you get regulated voltages to boot.
I mainly got sidetracked by the Schade plate feedback schemes which do the same thing effectively as the CFB. But lately I have seen an Achilles heel in Schade plate feedbacks, due to one outer 60% (for a 40% UL tapped) winding on a typical low cost OT having a very large leakage inductance. This causes the plate feedback on one side to kill the gain (or worse) on one side of the P-P at high frequency. Causing frequency droop in power output and distortion in the waveform.
So now I am back to considering the Elliptron but with a new twist. The typical low cost UL OTs should have good coupling on the inner 40% windings around the B+ center tap due to the windup of the coils. (and the fact that UL mode IS feedback and needs a good coupling to the output also) (testing a limited number of OTs leakage Ls has been consistant so far) The Elliptron configuration makes excellent use of the good coupling of the inner 40% windings, and even uses both at the same time for each tube. (this further cuts the leakage L in half for each tube, lemons into wine)
The new twist is that the "new" "magic" Schade feedback technique cuts crossover distortion dramatically for class AB. This just requires equal winding resistance in the OT to sample the currents during crossover, and a differential driver stage to subtract the feedbacks. By using the CFB (40% UL taps) taps for the differential driver feedbacks (instead of the usual plates), we avoid the badly behaving outer 60% winding in the feedback. ( one of the outer 60% windings will likely be flapping around at HF due to its high leakage L, but this doesn't really matter since, it is driven by the high Z plate side of the tube, which is looking like a current source. A current source drive doesn't care about some little leakage L in the way, it goes right thru to the load with just a bit more voltage swing at the plate) So this techniue can potentially rival the very best OT in class A while using a cheap OT in class AB. Not bad!!
I also have another topological variation in the works that gets rid of the center tapped inductor, but requires dual Mosfet current sources for the cathodes. Also, some effort yet needed to clean up switching distortion in an OT, mainly just some fine touches to the drive signals. Then time to build this finally.
I'm afraid I haven't yet. I have all the parts rounded up to do either the floating supply Elliptron version (dual bobbin pwr xfmrs) or the single supply design (an extra OT used for the center tapped inductor). I haven't found any good source of inexpensive quality wound center tapped inductors though (hence having to use another OT), so I haven't gotten too excited about it so far. If a good CT inductor were available cheap (and it should be cheap, since it only requires a contiguous layer wind) my excitement would hit critical level.
Another issue with the single supply version is the resistance of the center tapped inductor forms a time constant with the floating electrolytics. For class AB, the power draw varies thru the inductor, so the resonance can get excited. Either you have to use a very low resistance inductor or a servo circuit to control the voltage on the electrolytics.
I do have that servo worked out, it's not too hard to do. Just put resistors from each of the two tube cathodes down to the gate of a Mosfet variable current source (another resistor from gate to source to form a V divider then). The Mosfet current source then pulls its drain current from the center tap of the inductor to keep the electrolytics charged to a constant voltage via the feedback network. Only a limited voltage variation required across the Mosfet, so not a big power waste. And you get regulated voltages to boot.
I mainly got sidetracked by the Schade plate feedback schemes which do the same thing effectively as the CFB. But lately I have seen an Achilles heel in Schade plate feedbacks, due to one outer 60% (for a 40% UL tapped) winding on a typical low cost OT having a very large leakage inductance. This causes the plate feedback on one side to kill the gain (or worse) on one side of the P-P at high frequency. Causing frequency droop in power output and distortion in the waveform.
So now I am back to considering the Elliptron but with a new twist. The typical low cost UL OTs should have good coupling on the inner 40% windings around the B+ center tap due to the windup of the coils. (and the fact that UL mode IS feedback and needs a good coupling to the output also) (testing a limited number of OTs leakage Ls has been consistant so far) The Elliptron configuration makes excellent use of the good coupling of the inner 40% windings, and even uses both at the same time for each tube. (this further cuts the leakage L in half for each tube, lemons into wine)
The new twist is that the "new" "magic" Schade feedback technique cuts crossover distortion dramatically for class AB. This just requires equal winding resistance in the OT to sample the currents during crossover, and a differential driver stage to subtract the feedbacks. By using the CFB (40% UL taps) taps for the differential driver feedbacks (instead of the usual plates), we avoid the badly behaving outer 60% winding in the feedback. ( one of the outer 60% windings will likely be flapping around at HF due to its high leakage L, but this doesn't really matter since, it is driven by the high Z plate side of the tube, which is looking like a current source. A current source drive doesn't care about some little leakage L in the way, it goes right thru to the load with just a bit more voltage swing at the plate) So this techniue can potentially rival the very best OT in class A while using a cheap OT in class AB. Not bad!!
I also have another topological variation in the works that gets rid of the center tapped inductor, but requires dual Mosfet current sources for the cathodes. Also, some effort yet needed to clean up switching distortion in an OT, mainly just some fine touches to the drive signals. Then time to build this finally.
Last edited:
- Status
- Not open for further replies.