DC Offset or bias instability isn't something I've ever encountered with the Circlotron. Its been very common to find that after 6 months of operation, the bias and DC Offset haven't drifted (unless a tube has failed). So I suspect this issue you encountered was not related to the Circlotron itself.
Suppose we build a circlotron and decide to use an interstage transformer as a phase splitter and coupling element: shouldn't it be possible to get some (6dB?) negative feedback "for free" by cross-coupling the cold end of each secondary winding to the opposite tube?
Possibly, yes.
It was a first and quick evaluation test that proved by and large inconclusive compared to the Futterman original patent circuit, that's why I went no further on the Circlotron path at that time.
But I should return the Circlotron way now - and more seriously ! - to check if I can meet the same performance - or maybe better ? - than my U-OTL, and I include simplicity, reliability and compactness in that performance evaluation.
T
For the moment @benchbaron, I have no setup available to measure THD, but given the look of the square waves at 10WRMS output 16R, the THD should be low, possibly less than 1% by my experience...
Below the sine waves at 10WRMS/16R, that show no visible deformations, another clue of a low THD :
I thought I had taken screens of the triangle waves tests, but no 😕 ... I should ! 🙄
T
OTLs have always been very good at waveform tests. THD is a matter of load impedance (lower with higher impedances), how well the tubes are set up and whether there's feedback and how much. Futterman used up to 60dB in some of his designs, but at a price: it was very easy to exceed the phase margin of the amp, and distortion did rise with frequency because the Gain Bandwidth Product was lacking. But he did get some impressive THD values.
I'm of the opinion that getting a flat distortion vs frequency curve over the audio band is more important than the actual THD. Any zero feedback amp will have the flat part down no worries since there's no issue with the GBP. But we've been able to get THD down to 1% at full power by careful setup as well.
I'm of the opinion that getting a flat distortion vs frequency curve over the audio band is more important than the actual THD. Any zero feedback amp will have the flat part down no worries since there's no issue with the GBP. But we've been able to get THD down to 1% at full power by careful setup as well.
A Focusrite Solo USB audio interface is about $100 (3rd gen) and resolves to 0.0003%/-110dB THD.
IVX's COSMOS is at least an order of magnitude better at 2-3x the cost.
I see no good reason not to use any of these.
I do have all what is required : modern softwares and soundcards ! Just my DIY interface box needs to be modded from my old softs and PCs, I think. But I am lazy...
Yes, indeed. That's my point of view too. An even-order distortion, with a regular harmonic amplitude decay, reaching 2-3% is easily accepted...
Now I remember that a member at diyaudio.com here (I think it's @Brice, but maybe I'm confusing) who made a simulation of my circuit and found a THD inferior to 0.8% at 10WRMS/16R on the 5Hz-100kHz bandwidth, IIRC - Oh, I could not retrieve his messages...
T
One of the attractions of the Circlotron output is you can have a fully differential balanced circuit throughout. One of the advantages of this is its inherently lower distortion owing to even ordered cancellation at every stage throughout the amp; distortion is compounded less from stage to stage. This results in a dominant 3rd harmonic (cubic non-linearity as opposed to quadratic of SETs) which is at slightly less than the percentage seen in an SET, were both amps at full power (so an order of magnitude lower distortion in many cases, but a lot more power too). The 3rd is treated much the same as the 2nd by the ear (and is the only odd ordered harmonic that is musical) and is easily able to mask higher orders (FWIW, properly functioning tape machines also make a 3rd harmonic).
I'm about 90% through my stereo M-60 build. I finished the cabinet today. I need to be able to assemble the multiple chassis parts to begin testing.
I have done a lot of "hot rodding" so it is probably not going to be plug & play. Some of the mods will probably be changed or discarded entirely before it is finished. So no sense getting into them now.
In the schematics posted about 50 pages ago, both the M-60 original and the DIY version have XLR plugs, which is great. At the moment, I am stuck with RCA cables so I want to include that with the XLRs on the back. Shouldn't pin 3 jumper to pin1 if using the RCA connection?
I have done a lot of "hot rodding" so it is probably not going to be plug & play. Some of the mods will probably be changed or discarded entirely before it is finished. So no sense getting into them now.
In the schematics posted about 50 pages ago, both the M-60 original and the DIY version have XLR plugs, which is great. At the moment, I am stuck with RCA cables so I want to include that with the XLRs on the back. Shouldn't pin 3 jumper to pin1 if using the RCA connection?
I have also identified a possible failure mode related to the 6AS7GA and over-heating the grid. I would also ask if this condition has been noted in M-60 or Futterman OTLs That use DC on the filaments. I am listening to a Futterman (6 tube/channel) amp running GE 6AS7GAs - the worst version - with no issues what so ever. It is also operating at 160V on the power rails.
Still hoping someone has a comment on the RCA input vs XLR.
Still hoping someone has a comment on the RCA input vs XLR.
Yes.
Yes it has! For this reason we've recommended people stay away from this variant when using our amps. I'm skeptical of the DC filament thing but would love to be wrong about it.
Ok, first to answer the original post, from my prospective. All versions of the 6AS7 will work, but, the worst of them is the GE manufactured GA version. It is easily identifiable because the large structures above the top mica are missing. If you haven't seen them they look like the plate material, just in the wrong place. In testing, these tubes also seem to have the most imbalance between triodes (pair). Some if not all ST bottle versions have these structures below the bottom mica.
So that is the tube I am referring to and the operation of the grid and it's temperature. Also - it doesn't matter what is printed on the tube. Other vendors sold GE manufactured tubes, even RCA.
What I observed - sorry, is very specific, but should be repeatable. I was using my Hickok 539A to pair tubes up for installation . The 539 is capable of about 190V on the plate. Note that the grid bias is set at 40 (high as it goes). The scale setting is "D". this places a 2.5V AC signal on the grid.
The tubes tested fine, but I noted very hot spots on the grid - - SPOTS? Yes, about every 5th or 6th wrap of the grid wire was very hot. The rest looked normal. I can see this as being too hot particularly for the GA. I reduced the drive to .5V or less and have to recalculate the mhos. This reduced the hot spots to what appeared normal.
Since not all of the grid was cherry red, just every 6th turn, this seems to be a time-based issue possibly caused by peaks in the signal used by the Hickok. The AC filaments may have nothing to do with it - but, the Futterman I described earlier and pictured several pages back, has DC filaments and does not seem to have over heating grids. Music peaks may be too low or of inadequate duration on this unit to display heating. Note too that the "every 6th wrap" probably is not valid at frequencies other than 60hz. Since my Stereo M-60 is about a week away, my question should be: Is there or would there be, visible overheating of the grid, particularly at specific intervals, on GA tubes used in the M-60 RUNNING CLASS A only? Does the AC filament voltage have anything to do with it? I can see how this would be unlikely. For those running AB - nevermind.
I am now testing on a Hickok533. For this particular tube it is probably better. And being able to use the "life" test weeds out non-NOS tubes fast.
The GE 6AS7GA tubes are cheap, about 20% of the ST bottle tubes. It is nice to be able to use them.
So that is the tube I am referring to and the operation of the grid and it's temperature. Also - it doesn't matter what is printed on the tube. Other vendors sold GE manufactured tubes, even RCA.
What I observed - sorry, is very specific, but should be repeatable. I was using my Hickok 539A to pair tubes up for installation . The 539 is capable of about 190V on the plate. Note that the grid bias is set at 40 (high as it goes). The scale setting is "D". this places a 2.5V AC signal on the grid.
The tubes tested fine, but I noted very hot spots on the grid - - SPOTS? Yes, about every 5th or 6th wrap of the grid wire was very hot. The rest looked normal. I can see this as being too hot particularly for the GA. I reduced the drive to .5V or less and have to recalculate the mhos. This reduced the hot spots to what appeared normal.
Since not all of the grid was cherry red, just every 6th turn, this seems to be a time-based issue possibly caused by peaks in the signal used by the Hickok. The AC filaments may have nothing to do with it - but, the Futterman I described earlier and pictured several pages back, has DC filaments and does not seem to have over heating grids. Music peaks may be too low or of inadequate duration on this unit to display heating. Note too that the "every 6th wrap" probably is not valid at frequencies other than 60hz. Since my Stereo M-60 is about a week away, my question should be: Is there or would there be, visible overheating of the grid, particularly at specific intervals, on GA tubes used in the M-60 RUNNING CLASS A only? Does the AC filament voltage have anything to do with it? I can see how this would be unlikely. For those running AB - nevermind.
I am now testing on a Hickok533. For this particular tube it is probably better. And being able to use the "life" test weeds out non-NOS tubes fast.
The GE 6AS7GA tubes are cheap, about 20% of the ST bottle tubes. It is nice to be able to use them.
Last edited:
We've had this idea that the grid heatsink of that tube is insufficient for class A2 or class AB2 (depending on the speaker load) operation. So the grid structure warps and arcs. So this limits the use of the tube to either lower power where grid current is not invoked, lower plate Voltage or significantly higher cathode current limiting resistor values.
Thanks Ralph,
I am going to stick to A2 at 8 ohms, but I will be running 160V 0n the power rails, 2 inverter tubes per channel and a little different bias adjustment. will probably be able to report back in a week or two.
I am going to stick to A2 at 8 ohms, but I will be running 160V 0n the power rails, 2 inverter tubes per channel and a little different bias adjustment. will probably be able to report back in a week or two.
With 160V rails you will need a higher load impedance to be at A2 (otherwise will be AB2 and you'll want to run lower bias current on the power tubes). If A2 is important, drop the plate Voltage about 15V. You can do that with a bucking transformer on the primary of the output section B+ transformer.
There are 2 transformers to get the isolation needed for stereo. Each of them have an unused 12V and 6V tap At 6A. I had considered trying to buck one with the other using those taps at 180 deg out of phase - but that was one of the things I had considered. It would seem to me that might work dropping 6, 12 or 18 volts on the primary, but I have never tried using a winding on an active trans like that. Hooking up one like that would probably "buck" both though not to the same voltage. what do you think? That could be switched giving the option of A2 or AB2, though I still don't need AB2.
I guess I could use them to buck the 120V secondary too. That might be safer.
I guess I could use them to buck the 120V secondary too. That might be safer.
Last edited:
You can use windings on an existing transformer to buck Voltages as long as the winding can handle the primary current.
And now that I think about it, bucking the secondary would probably be easier than trying to buck the primary. The 6 and 12V filament taps are rated for 2 - 3 times the current of the B+ taps (about 2.5A compared to 6A).
Except, I would need 4 similar taps to do the secondarys and I only have 2 of each, 6V and 12V. So bucking the primary 18V is probably the way to go - the current on the primary is roughly equal to twice the secondary B+ anyway (with unused filament taps). This would still be adequate.
Except, I would need 4 similar taps to do the secondarys and I only have 2 of each, 6V and 12V. So bucking the primary 18V is probably the way to go - the current on the primary is roughly equal to twice the secondary B+ anyway (with unused filament taps). This would still be adequate.
Last edited: