If a crazy man had a bifilar wound OPT to play with... And I don't, so not to worry...
We know from Mac theory that pentode mode translates to UL. But cross coupled triode strap turns to pure pentode mode??? Whats goofier here, screen currents cancel selves completely.
Also messing with flying heaters and a negative main supply. No advantage found yet. Just food for thought experiments...
And a shunt regulated cathode bias, draws constant current from the power supply regardless AB. Maybe this is worthy?
This many Henrys would have made about 5K Plate to Plate had all primary windings been on the plate side. Just sayin for comparison sake to something "normal"...
We know from Mac theory that pentode mode translates to UL. But cross coupled triode strap turns to pure pentode mode??? Whats goofier here, screen currents cancel selves completely.
Also messing with flying heaters and a negative main supply. No advantage found yet. Just food for thought experiments...
And a shunt regulated cathode bias, draws constant current from the power supply regardless AB. Maybe this is worthy?
This many Henrys would have made about 5K Plate to Plate had all primary windings been on the plate side. Just sayin for comparison sake to something "normal"...
Attachments
If it was a class B2 that might work but you could run the filaments from the grid current and bypass the cathode with the Pr shunt to SG. That would unload the OPT and let more GNFB tame the hot Mu. Mu cancels the bias at the grid with the heaters. Duhhh! Mu, Mu, and more Mu! Meeeeuuuuuww....
Cathode to screen voltage is held almost a constant by the cross couple. There is little or no screen feedback. I know what it "loooks" like. But you have to account that cathodes and plate are moving equal and opposite. The opposite plate swings the same as the local cathode. So go figure?
As drawn, sim is delivering 55WRMS into 16 virtual ohms in AB1. And draws constant current from B- supply due to the weird cathode fixed bias shunt. A third leg that runs up the middle and does nothing, except waste power.
When the amp is idle, the shunt is dumping a LOT of power... Not gonna win efficiency contest. Compare shunt to ClassA for inefficiency, except output tubes aren't taking brunt of the wasted heat.
Pic: Cathode and screen moving exactly together = Pentode mode...
Attachments
Last edited:
Mac kept screens at fixed voltage while cathode and plates moved. This bizarre relativity experiment was exactly equivalent to 50% UL.
I've tried that configuration too, but can't get more than 35WRMS without violating AB1. UL + cathode NFB = far too much total NFB. Would require driving G1 further than +0V to realize the full 55W. At least with 6L6 at 450V, this seems too much feedback...
Feedback from cathodes alone, is almost but not quite too much. Can keep it all in AB1, full power out, no big G1 currents to drive. Abuse the same relativity to make Pentode mode instead... With this much cathode feedback, did Mac actually ever need the UL?
Mac screen currents would have appeared in the cathode windings, but not in the plate windings. I'm not sure the fancy Mac winding scheme could extended balancing benefits to screen currents that do not exist in the other half winding? But this crosswire triode to pentode mode has complete balance. Cancel to the dot anyway.
I've tried that configuration too, but can't get more than 35WRMS without violating AB1. UL + cathode NFB = far too much total NFB. Would require driving G1 further than +0V to realize the full 55W. At least with 6L6 at 450V, this seems too much feedback...
Feedback from cathodes alone, is almost but not quite too much. Can keep it all in AB1, full power out, no big G1 currents to drive. Abuse the same relativity to make Pentode mode instead... With this much cathode feedback, did Mac actually ever need the UL?
Mac screen currents would have appeared in the cathode windings, but not in the plate windings. I'm not sure the fancy Mac winding scheme could extended balancing benefits to screen currents that do not exist in the other half winding? But this crosswire triode to pentode mode has complete balance. Cancel to the dot anyway.
Last edited:
So, no takers? I'm not too sure the shunt thing a grand idea as-is. But could be crude beginnings for an autobias of some sort? Third leg shunt maximum current corresponds inversely to AB crossing minimum current. So is an easily detectible/comparible/correctible minima. And could be low passed and fed back to the gate.
As for cross triode strappage, thats a sure winner with this OPT. Assuming such an OPT exists separate from a classic McIntosh? I wouldn't suggest desecrating history just to recover the iron.
If its a patenty thing that holds back new production, I can show you an even better (takes same idea even further) winding for the purpose. Most likely never been tried before. Ask me. Dare me to whip it out... Goahead, see what it gets ya?
As for cross triode strappage, thats a sure winner with this OPT. Assuming such an OPT exists separate from a classic McIntosh? I wouldn't suggest desecrating history just to recover the iron.
If its a patenty thing that holds back new production, I can show you an even better (takes same idea even further) winding for the purpose. Most likely never been tried before. Ask me. Dare me to whip it out... Goahead, see what it gets ya?
Last edited:
hey,
a few thoughts...
First as drawn the output is actually an autoformer. Not sure if it was your intention but it seems to make a difference since when wired as an autoformer your output voltage appears at the plate but not at the cathode. Still haven't gotten my head around what that means in reality but wiring the OT as a transformer changes things a bit.
Also,
I'm not sure how mac did it with the bifilar pairs but it appears that you could simply do a quadfilar and be done with it.
dave
a few thoughts...
First as drawn the output is actually an autoformer. Not sure if it was your intention but it seems to make a difference since when wired as an autoformer your output voltage appears at the plate but not at the cathode. Still haven't gotten my head around what that means in reality but wiring the OT as a transformer changes things a bit.
Also,
I'm not sure how mac did it with the bifilar pairs but it appears that you could simply do a quadfilar and be done with it.
dave
As far as I recall, Mac didn't use UL mode. The screens were either bootstrapped to opposite plates, or had a separate CT'd winding on the xfmr to supply isolated screen V+. (Some imitators like AR used the speaker winding for CFB and a simple fixed screen voltage giving UL mode) Some Macs had windings for feedback to the driver stage too.
You can build a Mac clone easily by just using two identical OTs wired up al'a Norman Crowhurst's Twin Coupled amp. He used cross-coupled caps between the opposite cathode and plate windings to sub for the bifilar effect. Looks to me to be completely equivalent. Edcor's CXPP60-MS-2.4K or CXPP100-MS-1.7K or 2.5Kcomes to mind.
Another alternative is the Elliptron configuration, this is half way between Mac/Twin and Circlotron. The tube plates go to the usual plate winding ends and the cathodes connect thru cross coupled caps to the opposite UL taps on the OT. A CT'd inductor supplies -DC to the cathodes then. This gives 1.4*1.4 ~> 2X the primary OT impedance equivalent for 40% UL taps, and gives about 29% CFB. Easier to drive than the Mac 50%. Needs an extra CT winding on the CT'd inductor for screen voltage distribution if you don't want the UL mode. Or use floating screen supplies (zeners, caps).
---------------------
Re: quadrifilar
The Mac used two bifilar sets of windings (each is bifilar between a plate winding and the opposite cathode winding) because those have identical (but opposite between bifilar sets) AC signals on them. Combining those into one quadrifilar set would cause enormous distributed capacitance problems. Also, of note, the Mac bifilar pairs have the full B+ difference between adjacent bifilar wires, this requires extra insulation enamel layers on the wire to avoid arc-over. The Twin avoids this issue. I would build a Twin, NOT a Mac as a first attempt. Although there are some super insulated magnet wire sources available nowadays.
You can build a Mac clone easily by just using two identical OTs wired up al'a Norman Crowhurst's Twin Coupled amp. He used cross-coupled caps between the opposite cathode and plate windings to sub for the bifilar effect. Looks to me to be completely equivalent. Edcor's CXPP60-MS-2.4K or CXPP100-MS-1.7K or 2.5Kcomes to mind.
Another alternative is the Elliptron configuration, this is half way between Mac/Twin and Circlotron. The tube plates go to the usual plate winding ends and the cathodes connect thru cross coupled caps to the opposite UL taps on the OT. A CT'd inductor supplies -DC to the cathodes then. This gives 1.4*1.4 ~> 2X the primary OT impedance equivalent for 40% UL taps, and gives about 29% CFB. Easier to drive than the Mac 50%. Needs an extra CT winding on the CT'd inductor for screen voltage distribution if you don't want the UL mode. Or use floating screen supplies (zeners, caps).
---------------------
Re: quadrifilar
The Mac used two bifilar sets of windings (each is bifilar between a plate winding and the opposite cathode winding) because those have identical (but opposite between bifilar sets) AC signals on them. Combining those into one quadrifilar set would cause enormous distributed capacitance problems. Also, of note, the Mac bifilar pairs have the full B+ difference between adjacent bifilar wires, this requires extra insulation enamel layers on the wire to avoid arc-over. The Twin avoids this issue. I would build a Twin, NOT a Mac as a first attempt. Although there are some super insulated magnet wire sources available nowadays.
Last edited:
"For a practical build, couldn't you use two identical normal transformers - one for the plates and one for the cathodes, with their secondaries tied together in parallel? "
Some issues to consider. Using two transformers, you would want to connect their 16 Ohm secondaries in parallel to get an 8 Ohm output. So they need 16 Ohm secondaries in that case. (Two 8 Ohms secs in parallel would give 4 Ohm out) The tubes each see 2X the primary Z of a single OT, so you want low Zpri Ots usually. With two OTs you can now get twice the power out too, so more tube power ability may be sensible. But then the B+ requirement may increase substantially over what the single OT is insulated for. This can be solved by splitting the B+ into B+/2 and B-/2, so each OT sees half the doubled B+ with respect to frame or ground. Also keep in mind that the 50% CFB requires a LOT of driver signal level.
Some issues to consider. Using two transformers, you would want to connect their 16 Ohm secondaries in parallel to get an 8 Ohm output. So they need 16 Ohm secondaries in that case. (Two 8 Ohms secs in parallel would give 4 Ohm out) The tubes each see 2X the primary Z of a single OT, so you want low Zpri Ots usually. With two OTs you can now get twice the power out too, so more tube power ability may be sensible. But then the B+ requirement may increase substantially over what the single OT is insulated for. This can be solved by splitting the B+ into B+/2 and B-/2, so each OT sees half the doubled B+ with respect to frame or ground. Also keep in mind that the 50% CFB requires a LOT of driver signal level.
Last edited:
Have noticed that Plitron one for some time. A little pricey for me using $1 tubes and $15 power xfmrs. Would be nice if Edcor would come out with a CXPP25-MS-2.5K model for a cheap Twin. I have a testbed coming along for trying out an Elliptron (cheaper with only one OT). May try the Twin later. Also the Hammond 1650T looks good for a circlotron. (it's rated a 1.9K primary at 30 Hz, but that would be equivalent to a 950 Ohm primary at a 20 Hz bottom end, using the 16 Ohm output as the 8 Ohm output. With the Circlotron giving 4X the nominal OT Zpri for each tube, that would then look like a 3.8K OT equivalent good for down to 20 Hz).
Waiting for Ken to tune in with his next "Hallucination".
Waiting for Ken to tune in with his next "Hallucination".
Last edited:
A pentode circlotron also does this with a lot less windings on the core, in fact, an autoformer would be the right thing to use. It has the added advantage that you really only have one winding so no problems with imperfect magnetic coupling between several windings, which greatly reduces crossover transients. Of course, you pay for all this with lots more winding on the power transformer...
IMHO something like sweep tubes would be more suited for this sort of thing because lower plate voltages can be used (if straight pentode circlotron is used, it becomes a requirement due to low maximum G2 voltage on sweep tubes), translating into lower drive voltages, and this in turn may substantially simplify the driver circuits.
IMHO something like sweep tubes would be more suited for this sort of thing because lower plate voltages can be used (if straight pentode circlotron is used, it becomes a requirement due to low maximum G2 voltage on sweep tubes), translating into lower drive voltages, and this in turn may substantially simplify the driver circuits.
"Of course, you pay for all this with lots more winding on the power transformer..."
For Circlotrons, the dual bobbin industrial control xfmrs 120V/230V like Stancor TGC175-230 (or any of Magnetek-Triad, Hammond etc equivalents) provide a mere 67 pF common mode capacitance from secondary. Two of these will make the floating HV supplies for Sweep tubes. Smaller dual bobbin versions like the TGC43-230 will work for floating screen supplies. I intend to eliminate most of the HV filter inductor requirements by putting a 60 Hz resonant LC filter on the 120 V primary side of all these connected together (ie, only one L does the whole bunch!). The LC filter only passes 60 Hz sine wave current from the power line, but generates square wave voltage to the xfmrs to make continuous current draw into the rectifier/Cap outputs. Downside is that the LC filter gives only RMS DC voltage output like a normal L only DC filter, rather than peak V.
For Circlotrons, the dual bobbin industrial control xfmrs 120V/230V like Stancor TGC175-230 (or any of Magnetek-Triad, Hammond etc equivalents) provide a mere 67 pF common mode capacitance from secondary. Two of these will make the floating HV supplies for Sweep tubes. Smaller dual bobbin versions like the TGC43-230 will work for floating screen supplies. I intend to eliminate most of the HV filter inductor requirements by putting a 60 Hz resonant LC filter on the 120 V primary side of all these connected together (ie, only one L does the whole bunch!). The LC filter only passes 60 Hz sine wave current from the power line, but generates square wave voltage to the xfmrs to make continuous current draw into the rectifier/Cap outputs. Downside is that the LC filter gives only RMS DC voltage output like a normal L only DC filter, rather than peak V.
Last edited:
I contacted Bud Purvine about the possibility of winding some unity-coupled transformers but he said that I would have to purchase a batch of special high voltage insulation wire for which he has no other use. He also said it is a pain to work with, also increasing the cost. In the end, I figured I would just go with the Plitron if I needed to since it would be cheaper.
Don,
You are suggesting a separate power transformer for plate and screen. Is there a reason that you wouldn't use the floating plate supplies to derive that since they are already referenced to the cathode (which I am assuming is what you would be going for).
Don,
You are suggesting a separate power transformer for plate and screen. Is there a reason that you wouldn't use the floating plate supplies to derive that since they are already referenced to the cathode (which I am assuming is what you would be going for).
"You are suggesting a separate power transformer for plate and screen. Is there a reason that you wouldn't use the floating plate supplies to derive that since they are already referenced to the cathode (which I am assuming is what you would be going for). "
Just that the low capacitance dual bobbin xfmrs I mentioned are limited to 120V/240 V (L input giving an RMS DC rather than peak DC voltage B+) without going to a doubler arangement, so I would stack two xfmrs up each side to get 120 V for screen (sweep tubes) and 480 V for plate. Or use a resistor, zener and cap from cathode to the +240 V tap to get maybe +150 V for the screen.
One could I suppose use one xfmr on each side and go for a full wave quadrupler (240 V center tapped drive ). Might be lower ripple that way than a half wave doubler (especially using the 60 HZ LC filter on the primary side to keep it continuous conduction rectification.) Have to try that one out.
Or just go with C input to get 1.4X voltage off one xfmr.
Just that the low capacitance dual bobbin xfmrs I mentioned are limited to 120V/240 V (L input giving an RMS DC rather than peak DC voltage B+) without going to a doubler arangement, so I would stack two xfmrs up each side to get 120 V for screen (sweep tubes) and 480 V for plate. Or use a resistor, zener and cap from cathode to the +240 V tap to get maybe +150 V for the screen.
One could I suppose use one xfmr on each side and go for a full wave quadrupler (240 V center tapped drive ). Might be lower ripple that way than a half wave doubler (especially using the 60 HZ LC filter on the primary side to keep it continuous conduction rectification.) Have to try that one out.
Or just go with C input to get 1.4X voltage off one xfmr.
Last edited:
Huh, whats all this then? Separate power for flying heaters all I was on about... As long as heaters fly, hardly matters if main supply is positive or negative... Well, might matter to front end cathodes? I had yet to consider for them.
---------------
McBizarroFilar-er.. er Whatever.. Got big voltages right next each other, so enamel better be good! McIntosh didn't design a true Balun, nor go beyond the first most obvious step to try and fake it... Thats where I would press to change things up a little.
What I mean its not a true balun? Well the signals go in opposite ends, meet in the middle as they pass each other, and then leave opposite ends. The signals travel in opposite directions. At each end you got a big voltage swing laying against a wire that is not swinging, where is this cancelling?
When I post my next hallucination, we will address this. But I gotta draw pictures first. Else the explanation of it would be lost in too many words. We will take the next step toward faking a transmission line transformer. Fake is the magic keyword.
I suspect Pi winding could be as effective as Bifilar (considering that its never real transmission line transformer with signals travelling the same direction, not even after what I would do with it) without need for extra fancy enamel to lay opposing wires right against each other...
---------------
McBizarroFilar-er.. er Whatever.. Got big voltages right next each other, so enamel better be good! McIntosh didn't design a true Balun, nor go beyond the first most obvious step to try and fake it... Thats where I would press to change things up a little.
What I mean its not a true balun? Well the signals go in opposite ends, meet in the middle as they pass each other, and then leave opposite ends. The signals travel in opposite directions. At each end you got a big voltage swing laying against a wire that is not swinging, where is this cancelling?
When I post my next hallucination, we will address this. But I gotta draw pictures first. Else the explanation of it would be lost in too many words. We will take the next step toward faking a transmission line transformer. Fake is the magic keyword.
I suspect Pi winding could be as effective as Bifilar (considering that its never real transmission line transformer with signals travelling the same direction, not even after what I would do with it) without need for extra fancy enamel to lay opposing wires right against each other...
Last edited:
The Mac bifilar setup solves the class B coupling problem well between tube sides for switchover transients. The main weakness I see is that the two cathode windings are not closely coupled, so any variation in couplings to the secondary will show up as distortion since these are feedback control signals. The Circlotron, UL and conventional xfmr also have the same issue.
Actually, the Twin is the only one I have seen than can solve that issue in theory. Say the two cathode windings have some small coupling difference to their secondary. As long as the two xfmrs making up the Twin design are exactly identical. Due to the close coupling (from the cross coupled caps) between a left cathode winding and a right plate winding and similarly a right cathode windg and a left plate windg. Each tube cathode gets coupled to an xfmr primary left side and a primary right side because of the cap coupling. So an equal coupling average for each cathode results. So the Mac is beaten in theory.
I'm not following your complaint about the ends of the Mac bifilars not tracking. The AC signals on each bifilar are exactly matched at each end. Just a DC difference between pair wires.
Baluns, common mode chokes, and transmission line xfmrs can lead to a number of interesting OT designs. Some have been discussed here before, like the set of transmission lines connected in series at one end and in parallel at the other (each line configured as a common mode choke). Maybe even could throw in some HF switching carrier to reduce the size or wire turns of some function.
Awaiting further "hallucination" inputs.
Actually, the Twin is the only one I have seen than can solve that issue in theory. Say the two cathode windings have some small coupling difference to their secondary. As long as the two xfmrs making up the Twin design are exactly identical. Due to the close coupling (from the cross coupled caps) between a left cathode winding and a right plate winding and similarly a right cathode windg and a left plate windg. Each tube cathode gets coupled to an xfmr primary left side and a primary right side because of the cap coupling. So an equal coupling average for each cathode results. So the Mac is beaten in theory.
I'm not following your complaint about the ends of the Mac bifilars not tracking. The AC signals on each bifilar are exactly matched at each end. Just a DC difference between pair wires.
Baluns, common mode chokes, and transmission line xfmrs can lead to a number of interesting OT designs. Some have been discussed here before, like the set of transmission lines connected in series at one end and in parallel at the other (each line configured as a common mode choke). Maybe even could throw in some HF switching carrier to reduce the size or wire turns of some function.
Awaiting further "hallucination" inputs.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.