Say, has anyone seen this circuit topology:
http://www.valvediy.com/simplexpg1.html
http://www.valvediy.com/simplexreduxpg1.html
http://www.plitron.com/PDF/polisois1957.pdf
I only just stumbled upon these articles yesterday. I find the circuit intriguing in the way it addresses the problem of coupling driver to output tube, and rids the output tube of the dreaded cathode resistor (!) ... very elegant. I do not know what type of output transformer the author, Ari Polisois, uses in the 6C33C version, but his output distortion specs look *linear* if a little high:
http://www.valvediy.com/simplexpg7.html
And this with unregulated supplies. Ari says he's built >10 amplifiers and that this amplifier circuit bests them all.
I suspect tying output cathode to driver power supply is the weak link of the circuit as any driver supply variations will modulate a very sensitive part of the circuit. Makes me wonder how the circuit would perform with rock-solid regulation.
My hat off to Ari Polisois for an interesting circuit innovation.
http://www.valvediy.com/simplexpg1.html
http://www.valvediy.com/simplexreduxpg1.html
http://www.plitron.com/PDF/polisois1957.pdf
I only just stumbled upon these articles yesterday. I find the circuit intriguing in the way it addresses the problem of coupling driver to output tube, and rids the output tube of the dreaded cathode resistor (!) ... very elegant. I do not know what type of output transformer the author, Ari Polisois, uses in the 6C33C version, but his output distortion specs look *linear* if a little high:
http://www.valvediy.com/simplexpg7.html
And this with unregulated supplies. Ari says he's built >10 amplifiers and that this amplifier circuit bests them all.
I suspect tying output cathode to driver power supply is the weak link of the circuit as any driver supply variations will modulate a very sensitive part of the circuit. Makes me wonder how the circuit would perform with rock-solid regulation.
My hat off to Ari Polisois for an interesting circuit innovation.
DCMB Schematic
Here's the schematic from the links above.
The SIMPLEX Single-ended Audio Amplifier
by Ari Polisois
©2002 A. Polisois
Here's the schematic from the links above.
The SIMPLEX Single-ended Audio Amplifier
by Ari Polisois
©2002 A. Polisois
An externally hosted image should be here but it was not working when we last tested it.
serengetiplains said:
(continuing this conversation with myself) Actually, after further thought, driver supply variations always necessarily appear on the output grid. Coupling those variations to the cathode will therefore reduce their overall influence on the output signal ... the cathode coupling, if you will, providing a form of feedback relative to grid. Very elegant!
Just so's you don't have to talk to yourself...
The coupling and twin power supply idea is quite clever, and because a 6C33 needs a lower HT than its driver, makes a great deal of sense.
Unfortunately, your last post made me think about power supply noise, and I realised that taking the voltage between power supply and anode is not a good thing for a triode with low ra. The undecoupled cathode resistor raises ra, which helps, but without values, it's hard to see what the effect will be. Purely from the point of view of power supply noise, a pentode would be best in this particular application.
The coupling and twin power supply idea is quite clever, and because a 6C33 needs a lower HT than its driver, makes a great deal of sense.
Unfortunately, your last post made me think about power supply noise, and I realised that taking the voltage between power supply and anode is not a good thing for a triode with low ra. The undecoupled cathode resistor raises ra, which helps, but without values, it's hard to see what the effect will be. Purely from the point of view of power supply noise, a pentode would be best in this particular application.
I just started looking at this circuit too. And I am looking for parts. Some changes I am doing is a 10H choke input to replace the resistor in the driver power supply. A larger 10-20H choke input in the output tube power supply. By choke input I mean bridge rectifier then choke then caps. I have had good results with chokes in solid state. DC filiments for driver tubes. I also think its easier to build in dual mono. Mostly because you have more room to fit parts. Also separate channels are better. But, more costly as you now have two separate power supplies on each chassis. Finally passive input pot. Not sure if this will work well but it's easy to add.
gnomus said:
Hi Paul,
Separate power supplies pay big dividends soundwise. Also, as per my previous post, I suspect this circuit enhances power supply noise rejection of the driver supply by coupling similar amplitudes of that noise in phase to the output grid + cathode. I suspect adding a third power supply for the input tube would open the sound of this circuit even more than what the designer claims for it. My comments are conjecture because I have not built this amplifier.
The designer of this circuit told me very recently the circuit is covered by a French patent. Just so you know.
Tom
Fine with me. Like an unchoked monkey but with a stepped supply instead of wasting incredible gobs of power through an absurd cathode resistor.
Power supply hum is an issue in any triode amplifier (with Rl typically 3xRp, 3/4 of the hum is transformed by the OPT, 1/4 of it is lost in the tube), the only differences between this and an 845 or 211 SE amp are the current and voltage levels, which necessitate an increase in PSU capacitance. With a choke, especially one in the 10-20H range as shown (which will be as large or larger than any other iron on the chassis, given the current a 6C33C needs; I'd go with 1H) this won't be necessary, at least in the stupifyingly brute force ways the SSatanic guys do it.
Tim
Power supply hum is an issue in any triode amplifier (with Rl typically 3xRp, 3/4 of the hum is transformed by the OPT, 1/4 of it is lost in the tube), the only differences between this and an 845 or 211 SE amp are the current and voltage levels, which necessitate an increase in PSU capacitance. With a choke, especially one in the 10-20H range as shown (which will be as large or larger than any other iron on the chassis, given the current a 6C33C needs; I'd go with 1H) this won't be necessary, at least in the stupifyingly brute force ways the SSatanic guys do it.
Tim
Re: Re: Direct Coupling Modulated Bias (SET)
Actually, there is no feedback. Rearrange it as two power supplies, with the output supply on top of the driver supply. You can see the 6C33C's grid-cathode is across the driver plate resistor. Any driver variations will directly affect the output tube, making stable bias in such a high-Gm tube a problem.
Also, I never liked the topology because tubes never cut off. This means the output tube will never be fully utilized because for its grid to reach 0V (with respect to cathode), the driver plate resistor must have zero current through it. This takes an exponential(?) amount of additional negative driver grid voltage to obtain. As a result, distortion will be very high as you approach maximum output (taken as clipping of the output stage).
Tim
serengetiplains said:
Actually, there is no feedback. Rearrange it as two power supplies, with the output supply on top of the driver supply. You can see the 6C33C's grid-cathode is across the driver plate resistor. Any driver variations will directly affect the output tube, making stable bias in such a high-Gm tube a problem.
Also, I never liked the topology because tubes never cut off. This means the output tube will never be fully utilized because for its grid to reach 0V (with respect to cathode), the driver plate resistor must have zero current through it. This takes an exponential(?) amount of additional negative driver grid voltage to obtain. As a result, distortion will be very high as you approach maximum output (taken as clipping of the output stage).
Tim
I have considered the 845, 211 or 805. I looked at a lot of designs and did not find one direct coupled. The other major problem is that NOS tubes sounds the best. But, they are rare and expensive. The 6C33B triode I have heard in many designs. From OTL to SET. It's performance was always good. I preferred it in a SET over OTL. And I think this has to do with a SET being class A. In fact I only like class A even in solid state. I Do not understand why everyone raves about class A pre amps and then mate it with a class AB amp. Yuck.
Driver supply noise reduction
Hi Tim,
Are you saying driver current variations will appear on the output tube grid? This seems true because driver current variations create a varying voltage drop across the driver anode resistor, which voltage appears on the output grid. This problem, however, is not unique to the DCMB circuit, though perhaps is of greater effect, as you said, in a circuit using a high Gm output tube.
Assuming the DCMB topology is, in this regard, no greater culprit than other topologies, the DCMB topology still seems to offer the benefit of driver supply noise feedback in the output stage. To clarify what I mean, let me again describe what I call driver supply noise feedback (the subtle components of which, in my universe, might well be measurablle by only the most sensitive of measuring means). Driver power supply voltage variations, which I define as any voltage deviation from DC whatever---noise, for short---will in the DCMB topology appear in equal phase across the output cathode and grid, though at reduced level on grid relative to cathode (any driver voltage passes through the driver anode resistor, which resistor divides that voltage relative to remaining resistance to ground, not considering alternative pathways to ground). Compare this situation to that of the typical output configuration where driver supply variations appear unreduced on the output grid to there modulate the output at what seems to me a necessarily higher level. The DCMB circuit, from this vantage point, would seem to offer the benefit of quieter operation because of this particular form of feedback. I'm not greatly experienced in hypothesizing operations of a circuit from viewing a schematic, so I appreciate if anyone can see anything I've missed?
Tom
Sch3mat1c said:
Hi Tim,
Are you saying driver current variations will appear on the output tube grid? This seems true because driver current variations create a varying voltage drop across the driver anode resistor, which voltage appears on the output grid. This problem, however, is not unique to the DCMB circuit, though perhaps is of greater effect, as you said, in a circuit using a high Gm output tube.
Assuming the DCMB topology is, in this regard, no greater culprit than other topologies, the DCMB topology still seems to offer the benefit of driver supply noise feedback in the output stage. To clarify what I mean, let me again describe what I call driver supply noise feedback (the subtle components of which, in my universe, might well be measurablle by only the most sensitive of measuring means). Driver power supply voltage variations, which I define as any voltage deviation from DC whatever---noise, for short---will in the DCMB topology appear in equal phase across the output cathode and grid, though at reduced level on grid relative to cathode (any driver voltage passes through the driver anode resistor, which resistor divides that voltage relative to remaining resistance to ground, not considering alternative pathways to ground). Compare this situation to that of the typical output configuration where driver supply variations appear unreduced on the output grid to there modulate the output at what seems to me a necessarily higher level. The DCMB circuit, from this vantage point, would seem to offer the benefit of quieter operation because of this particular form of feedback. I'm not greatly experienced in hypothesizing operations of a circuit from viewing a schematic, so I appreciate if anyone can see anything I've missed?
Tom
You mean noise in the HV driver supply? Oh, it'll be there. With driver Rp circa 1/5 the value of the respective plate resistor, from B+ to plate to driver GND it acts as a voltage divider to the hum and noise. Thus, 5/6ths of the PSU noise (if Rp and RL are as above) appears across the resistor.
But you're supposed to be filtering supplies better than this anyway.
Tim
But you're supposed to be filtering supplies better than this anyway.
Tim
Further thoughts
I was recently considering the resistor loaded stacked-PSU design in this thread against Bob Hoekstra's inductor loaded Axiom design posted in another thread. I thought I would share my thoughts for comment of anyone interested. The main downsides of inductor (including transformer) loading, IMHRUO, include unwanted phase shifts, increased IM distortion, and restricted HF bandwidth. The upside of inductor loading is high AC reactance. A 100H inductor, for instance, generates ~630Kohm reactance at 1000Hz, much higher than the resistance of a plate load resistor in a comparable circuit. This reactance constitutes a near-constant current source to the tube, lowering THD of the tube, and the more so the higher the frequency. The reactance also acts as a filter for unwanted power supply artifacts, and again more so the higher the frequency the artifact in question, reducing the level of those artifacts, on a typical LC or interstage transformer coupled circuit, on the sensitive grid of the next stage (driver, output, whatever).
I've taken to guessing that PSU artifact reduction perhaps accounts for most of the appeal of inductor and transformer loading, for those who prefer inductor loading; THD reduction probably is also a factor, but would seem less important.
Now consider a resistor-loaded, stacked-supply (DCMB) circuit. In that circuit, artifacts in the power supply of Stage 1 fed to the grid of Stage 2 are reduced by coupling the Stage 2 cathode to the Stage 1 supply, effectively rendering Stage 2, so far as supply variations are concerned, a one-device almost-differential amp (elegant). PSU artifact reduction is not complete in this setup but is proportional to the overall percent the load resistor represents to the total impedance of the tube circuit (resistor + tube + cathode). That proportion, for resistor loading, is high enough to matter, but is practically insignificant for inductor loading, given the high AC reactance of plate load inductors. The Axiom amp, for instance, loads the input tube with 200H = 12.6Mohm reactance at 10KHz. Thus in using inductors in the present circuit, one gets the worst of inductive loading (phase shift etc ... oh, + the not-to-be-dismissed sonic detriment of plastic coated wire) while losing most of the benefit (PSU artifact reduction). Lower THD, a relatively small benefit, is the only overall benefit I can see.
Just as a note, by PSU "artifact" I mean anything but pure DC on the PSU rail. The many sources of artifacts are noise on the AC line (which are only ever partly filtered no matter how stiff the filter), capacitor and inductor wonkiness in the DC filter (dielectric absorption anyone?), induced EMI, insufficient use of Shakti Stones, etc.
I was recently considering the resistor loaded stacked-PSU design in this thread against Bob Hoekstra's inductor loaded Axiom design posted in another thread. I thought I would share my thoughts for comment of anyone interested. The main downsides of inductor (including transformer) loading, IMHRUO, include unwanted phase shifts, increased IM distortion, and restricted HF bandwidth. The upside of inductor loading is high AC reactance. A 100H inductor, for instance, generates ~630Kohm reactance at 1000Hz, much higher than the resistance of a plate load resistor in a comparable circuit. This reactance constitutes a near-constant current source to the tube, lowering THD of the tube, and the more so the higher the frequency. The reactance also acts as a filter for unwanted power supply artifacts, and again more so the higher the frequency the artifact in question, reducing the level of those artifacts, on a typical LC or interstage transformer coupled circuit, on the sensitive grid of the next stage (driver, output, whatever).
I've taken to guessing that PSU artifact reduction perhaps accounts for most of the appeal of inductor and transformer loading, for those who prefer inductor loading; THD reduction probably is also a factor, but would seem less important.
Now consider a resistor-loaded, stacked-supply (DCMB) circuit. In that circuit, artifacts in the power supply of Stage 1 fed to the grid of Stage 2 are reduced by coupling the Stage 2 cathode to the Stage 1 supply, effectively rendering Stage 2, so far as supply variations are concerned, a one-device almost-differential amp (elegant). PSU artifact reduction is not complete in this setup but is proportional to the overall percent the load resistor represents to the total impedance of the tube circuit (resistor + tube + cathode). That proportion, for resistor loading, is high enough to matter, but is practically insignificant for inductor loading, given the high AC reactance of plate load inductors. The Axiom amp, for instance, loads the input tube with 200H = 12.6Mohm reactance at 10KHz. Thus in using inductors in the present circuit, one gets the worst of inductive loading (phase shift etc ... oh, + the not-to-be-dismissed sonic detriment of plastic coated wire) while losing most of the benefit (PSU artifact reduction). Lower THD, a relatively small benefit, is the only overall benefit I can see.
Just as a note, by PSU "artifact" I mean anything but pure DC on the PSU rail. The many sources of artifacts are noise on the AC line (which are only ever partly filtered no matter how stiff the filter), capacitor and inductor wonkiness in the DC filter (dielectric absorption anyone?), induced EMI, insufficient use of Shakti Stones, etc.
Hi,
The restricted bandwidth is strictly dependent on how well the inductor is built.
In an absolute sense almost any passive component will deviate from the ideal linear behaviour at some point but , just as with OPTs, the inductors being used as a plate load should be as wideband as possible.
That comes at a considerable cost as you can well imagine. It's therefore important to distinguish between "ordinary" PS filter chokes AKA swinging chokes_some are rather "universal" in use, others not at all_and plate load chokes.
As for PS junk rejection, well, if filter chokes are used in the PS already then you shouldn't have to rely on the plate choke to "reject" the crud on the HT line.
You really shouldn't as you can't have it both ways: rejecting HF crud inevitably implies limited BW from the plate choke which will have its impact on HF response of the circuit one way or another.
A good compromise is to use chokes for PS filtering, they're really excellent for that, then use as high a B+ as you care to and use high value plate resistors where apropriate.
This often gives excellent results which can easily outperform plate loaded topologies all else kept equal IME.
Notably more economical too assuming you wanted the best possible plate chokes in the first place.
Cheers,
The restricted bandwidth is strictly dependent on how well the inductor is built.
In an absolute sense almost any passive component will deviate from the ideal linear behaviour at some point but , just as with OPTs, the inductors being used as a plate load should be as wideband as possible.
That comes at a considerable cost as you can well imagine. It's therefore important to distinguish between "ordinary" PS filter chokes AKA swinging chokes_some are rather "universal" in use, others not at all_and plate load chokes.
As for PS junk rejection, well, if filter chokes are used in the PS already then you shouldn't have to rely on the plate choke to "reject" the crud on the HT line.
You really shouldn't as you can't have it both ways: rejecting HF crud inevitably implies limited BW from the plate choke which will have its impact on HF response of the circuit one way or another.
A good compromise is to use chokes for PS filtering, they're really excellent for that, then use as high a B+ as you care to and use high value plate resistors where apropriate.
This often gives excellent results which can easily outperform plate loaded topologies all else kept equal IME.
Notably more economical too assuming you wanted the best possible plate chokes in the first place.
Cheers,
Absorbing dielectric absorption
Frank, thanks for your advices, which are always well regarded. One source of power supply crud I'm wanting to eliminate, to any possible degree, is the amusical behaviour of power supply capacitors. Assume a passive CLC supply, the inductor will filter some percentage of nasties created by the first C, but not those of the second C. Those nasties account for the obvious sonic differences between capacitors and in a sense are a non-musical capacitor-sourced AC voltage injected, if you will, onto the rail of an otherwise nicely filtered supply, and injected, in the example of the second C of a CLC supply, just before the plate load resistor or inductor. If I can't have a 100uF foam teflon (or better, air) capacitor, how to rid a supply of particularly those capacitive effects becomes a question I find nicely challenging. This is where the Shakti Stones come in, and fervent prayer. Or perhaps one shouldn't overlook the possibility of using a 10,000VDC supply and tiny air capacitors?
J'digress. The only solution I can think of is placing those capacitive effects differentially across the input of the following stage.
Frank, thanks for your advices, which are always well regarded. One source of power supply crud I'm wanting to eliminate, to any possible degree, is the amusical behaviour of power supply capacitors. Assume a passive CLC supply, the inductor will filter some percentage of nasties created by the first C, but not those of the second C. Those nasties account for the obvious sonic differences between capacitors and in a sense are a non-musical capacitor-sourced AC voltage injected, if you will, onto the rail of an otherwise nicely filtered supply, and injected, in the example of the second C of a CLC supply, just before the plate load resistor or inductor. If I can't have a 100uF foam teflon (or better, air) capacitor, how to rid a supply of particularly those capacitive effects becomes a question I find nicely challenging. This is where the Shakti Stones come in, and fervent prayer. Or perhaps one shouldn't overlook the possibility of using a 10,000VDC supply and tiny air capacitors?
J'digress. The only solution I can think of is placing those capacitive effects differentially across the input of the following stage.
Hi,
Tom,
While I've no idea what exactly it is you have in mind, I can't help but feel that most of the problems you describe can be dealt with rather easily by sticking strictly to class A operation.
At least that would eliminate a big chunk of the moving targets.
You won't have to resort to huge exotic caps either as the current draw is mainly constant and much of the imperfections of the passive components can be canceled out by going PP/differential, at least in theory.
Sticking to PP circuits, fully symmetrical topologies such as the well known circlotron circuit should even do better still in this respect.
The main reason, IMHO, it's not that widely used is one of cost.
Other than that it sure looks appealing.
For SET amps there's no simple solution I know of, the endresult will invariably depend on whatever is in the signal path...
Unfortunately, it all is...
So whichever way you turn it, every single component will have its impact on the final result.
It then all becomes a matter of chosing your poison and finding a recipe where flavour A of component X complements or counters flavour B of component Y at nauseam.
At the end of the day that's what will set a finely tuned design apart from a run of the mill PanaPhlipsky.
A form of art not to be found in any engineering design book but then you won't find the finer secrets of good winemaking in a book either..........
Cheers,
Tom,
While I've no idea what exactly it is you have in mind, I can't help but feel that most of the problems you describe can be dealt with rather easily by sticking strictly to class A operation.
At least that would eliminate a big chunk of the moving targets.
You won't have to resort to huge exotic caps either as the current draw is mainly constant and much of the imperfections of the passive components can be canceled out by going PP/differential, at least in theory.
Sticking to PP circuits, fully symmetrical topologies such as the well known circlotron circuit should even do better still in this respect.
The main reason, IMHO, it's not that widely used is one of cost.
Other than that it sure looks appealing.
For SET amps there's no simple solution I know of, the endresult will invariably depend on whatever is in the signal path...
Unfortunately, it all is...
So whichever way you turn it, every single component will have its impact on the final result.
It then all becomes a matter of chosing your poison and finding a recipe where flavour A of component X complements or counters flavour B of component Y at nauseam.
At the end of the day that's what will set a finely tuned design apart from a run of the mill PanaPhlipsky.
A form of art not to be found in any engineering design book but then you won't find the finer secrets of good winemaking in a book either..........
Cheers,
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.