For class A the effective anode impedance is 1/2 of plate-to-plate. In class B it becomes 1/4 because when one tube cuts off half of the primary "is not there" anymore, effective turns are 1/2 and impedance becomes 1/4 as it goes with the square of turn ratio...
And yes I think Rongon is quite right. At 40-50 mA I would expect something in the 800R range at least for plate resistance of the EL86. In reality one should figure this out better by drawing the composite tube.
In terms of Zout if the amp goes into class B Zout goes up quite a bit. The Class A amp has better damping factor in other words, not just because of the transformer ratio.
And yes I think Rongon is quite right. At 40-50 mA I would expect something in the 800R range at least for plate resistance of the EL86. In reality one should figure this out better by drawing the composite tube.
In terms of Zout if the amp goes into class B Zout goes up quite a bit. The Class A amp has better damping factor in other words, not just because of the transformer ratio.
Last edited:
However, the Zout alone (or damping factor) is not sufficient to explain the drive capability of the amp. The bias method also has its importance. For example the EL34 triode class A amp with fixed bias can cope better than the output stage with cathode bias and higher quiescent current. It is less susceptible to load variation when the impedance seen through the transformer goes down (lower distortion) despite the self bias amp has better damping factor (because it is normally biased at higher current). On the other end if the impedance see goes up they are equivalent with very good approx. In conclusion, the class A amp with fixed bias behaves better respect to load variation despite its damping factor is a bit worse....tedrodes/pentodes ouput stages are far worse regardless of feedback (GND feedback, I mean). With cathode feedback (or equivalent forms of local fbk) is totally different story and can be better than triode. The triode itself with local fbk improves, of course....
A good approach if cathode bias is used could be to design for the minimum impedance one wants to drive....without exaggeration.
A good approach if cathode bias is used could be to design for the minimum impedance one wants to drive....without exaggeration.
Last edited:
One important thing I forgot to say about the EL34 amp I posted earlier is that the dynamic balance of the amplifier for minimum distortion can be achieved with both the T-shaped resistive network of the splitter and the fixed bias. It is a great advantage to have 2 possibilities because it allows to balance the amp in a wider frequency range. The bias adjustment for dynamic balance also means that a small DC unbalance is likely. So a gapless core is not a good idea, regardless of the primary load.
I know some guru's don't like the para-phase but reality is different, starting from the fact that tubes are never identical, and can be as good or even better than other splitters that should be more ideal in theory.
I know some guru's don't like the para-phase but reality is different, starting from the fact that tubes are never identical, and can be as good or even better than other splitters that should be more ideal in theory.
The UNSET configuration was a long slow development effort that took place sporadically over almost 10 years. Many here know that I like big TV sweep tubes. I had nearly 200 watts flowing from a pair of 6LW6's on about 700 volts about 15 years ago. The low screen voltage requirement made triode mode operation difficult or unreliable. I spent some time experimenting with screen drive, eventually abandoning that effort due to high output impedance (not good for cheap OPT's) and tube failure under adverse conditions (the Dumm Blonde with a guitar preamp set on "Jimi"). Some experiments with dual drive (also called crazy drive) led to the same failures.
At this point I backed up and created a breadboard with a big sweep tube and a pair of mosfets connected to every electrode except the plate. I could put any DC voltage and either phase of drive on any electrode of the tube. I spent several years eliminating possibilities before arriving at something I called the Compound Electron Device, the topology used in the UNSET and a couple of push pull amps. This affords triode like curves from a pentode and eliminates the need for a negative voltage source.
The mosfet is a follower in the cathode circuit of a tube. "Schade" type feedback IS already applied due to the voltage divider from the cathode directly to the screen grid. The ratio of the two resistors used in the divider determines the "triodness" of the tube, and thus it's output impedance / plate resistance. Pick the values that suit your application. Unlike traditional resistive cathode bias the voltage drop across the mosfet (considerable at idle) is not "wasted" as the mosfet drop is near zero at the peak of conduction.
Since the output tube is typically a pentode, a somewhat stiff voltage on the screen is needed. A Zener diode with a mosfet follower has proven sufficient in all cases so far. The output tube can be a triode, with the feedback resistors optimized to improve it's characteristics, IE make a high Mu tube into a low Mu tube with lower Rp.
In my current UNSET amp board design I also use a similar CED stage for the driver. It uses a high Gm TV IF amp tube that is operating as a medium Mu triode.
There are two drawbacks to the CED topology:
If the mosfet shorts, bias is lost and the output tube becomes very unhappy. The BIG sweep tubes I'm using can pass AMPS of current when this occurs. I had two tubes shatter, one violently during the early experiments. Fuse the cathode circuit.
The stage does not invert the signal. Making a phase inverter with this topology is not easy.
P channel fets are not as good as their N channel counterparts, but good ones exist. The first amp I built with the CED was a push pull amp. It made 80 WPC from some tiny 6GF5's. I have broken the 300 watt barrier with a pair of 36LW6's although they were rather red faced at the time. I'm aiming at 500 WPC with 4 or 6 of them. Yes, I'm using some pricey OPT's for this. I get them a long time age when they were cheap. Experiments with cheap toroids will follow, including a pair of Toroidy SE OPT's.
More info is here:
UNSET is coming?
Last edited:
The “barrier” has never been about the tubes - its has always been the transformers. Care to speculate on what those big Plitrons of yours would cost to just order them *today*? I’m smart enough to know not to even ask unless you’re prepared to write a blank check. Looking at abusing Antek power trafos as OPTs it seems there are a couple of $200 possibilities to make a tube amp put out the power of a CA12. That I can live with.
The one thing I wonder about the CED is what the crossover distortion is like running them in AB push pull. You would be running that lower mosfet over its last 10 volts producing extremely nonlinear capacitances. In my experience, that’s what limits push pull solid state amps - transistors not going in and out of conduction cleanly. Limited more by capacitance changes vs. voltage than composite gm. Even low biased PP tube amps sound better in the first 10 watts (to me anyway).
I was thinking a way to minimize that effect would be to use a good strong bipolar as the follower. A CFP using a much smaller P channel (using a linear Miller cap to swamp out nonlinear cgd) and a good ol MJL3281 doing the burnt of the work. Their SOA is better than many mosfets eliminating the threat of exploding parts.
The one thing I wonder about the CED is what the crossover distortion is like running them in AB push pull. You would be running that lower mosfet over its last 10 volts producing extremely nonlinear capacitances. In my experience, that’s what limits push pull solid state amps - transistors not going in and out of conduction cleanly. Limited more by capacitance changes vs. voltage than composite gm. Even low biased PP tube amps sound better in the first 10 watts (to me anyway).
I was thinking a way to minimize that effect would be to use a good strong bipolar as the follower. A CFP using a much smaller P channel (using a linear Miller cap to swamp out nonlinear cgd) and a good ol MJL3281 doing the burnt of the work. Their SOA is better than many mosfets eliminating the threat of exploding parts.
I have a good deal of experience using dumpster power toroids for OPT's. I have a guitar amp that runs a 50 VA Antek driven by a pair of UL84's which are EL86/6CW5's with a 45 volt 100 mA heater. I get about 20 watts of power on 340 volts of B+ and 170 volts on the screens. The frequency response was pretty flat out to 20 KHz, but I never ran a complete sweep since it did what I wanted it to......play loud and clean.
I got a bunch of toroidal power transformers from a dumpster that were all from some thermal printers that were being scrapped. There were two vendors for the same transformer with the same part number on it. One made good OPT's, the other sucked. I did use some in HiFi amps being driven by triode wired E130L's, also from scrap, power supplies this time.
As I had stated some time ago I found some isolation transformers with 4 X 120 volt windings and no plastic in the center, so I wired all 4 windings in series, and wound my own secondary through over the outer wrap. Playing with the order of the series connections does affect the HF response and balance. It is also possible to stack two or more such toroids and wind a single secondary through two or more stacked toroids. These were all 160 VA toroids.
I have some 4 X 120 volt winding toroids rated at 600 VA waiting for me to play with. I will once the amp is fully working with the Plitrons. I just got the first batch of UNSET boards back from the PCB house. I will be using two or three boards in parallel to bust the 500 watt barrier again. The first time I pushed 500+ watts through these OPT's was back in 2010 using Pete's big red board. Post #438 here:
Posted new P-P power amp design
Want to do something really crazy, drive each toroid with its own pair of output tubes, taking the speaker output from a single secondary wound through the stacked cores.
Note, I have also done something similar with conventional EI core OPT's. Wire each of 4 OPT's to its own set of output tubes, but wire all secondaries in series - parallel. I was getting over 300 watts from 8 medium sized 13BG5 sweep tubes and 4 OPT's. All were fed from the same driver board, but each tube had it's own bias pot. I set up each output stage individually before wiring them all together to avoid fireworks.
The only capacitance that really matters in a follower is the Crss, as it is directly seen by the driver. A good low impedance driver can deal with a variable capacitance as long as it stays fairly small. I have tried several different mosfets including some pretty big ones ( the 9 amp FQPF9P25) and see little change in the HF response (-3dB is 65 to 70 KHz limited by the OPT) and THD at low levels compared to smaller fets like the FQP3P50.
I have not experimented much with BJT's since I blasted several ON semi parts in half early on. Maybe I'll experiment further on at a later time.
I think any experiments I do with unset will probably *be* single ended. Last SET I played with was triode strapped EL34’s. Six watts. At some point I’ll need to drop $600 on a decent pair of SE iron to try and bump that up a couple of dB. I’ll have a few xLW6’s left over after my kilowatt-ish experiments have run their course, and they’d be good candidates. Cheaper than 300B’s or even KT88’s. The super wattage will likely be just pentodes with a power drive mosfet each. Global NFB doesn’t scare me, and that 200 watt Williamson-SVT-ish thing I built sounds pretty damn good. And doesn’t oscillate. One thing at a time.
The first thing I may try is driving a pair of “junker” 600 VA 22-0-22 interleaved toroids with some cheap $4-each 21LG6’s to get 400 ish watts. Limited by the turns ratio, 4 ohm load, and a 400 volt rail. Two stacked trafos still only lets you hit it with 340 volts peak (at 50 Hz, probably safe to 40). Won’t have to buy anything for the experiment. The better route for higher power would be the AS15T950 as a OPT, as it’s about 1k a-a and would handle over 700 volts peak applied, producing 63 volts RMS on the 120v “primary” all day long down to 25 Hz. Can you say PA subwoofer amp? I sure can. Would go great with my Williamson on the top cabs. Wouldn’t use all the current capabilities of sweep tubes, but they will run with 800-900 V plates, won’t they? Top caps won’t arc over like cheap octal tube sockets will when they get dirty, and I would have 40 watts dissipation each to work with. I’m thinking that the voltage would not really be that insane. I did buy one of those trafos to have on hand some time ago thinking I would be doing it this way when I was hoarding LW6’s.
The first thing I may try is driving a pair of “junker” 600 VA 22-0-22 interleaved toroids with some cheap $4-each 21LG6’s to get 400 ish watts. Limited by the turns ratio, 4 ohm load, and a 400 volt rail. Two stacked trafos still only lets you hit it with 340 volts peak (at 50 Hz, probably safe to 40). Won’t have to buy anything for the experiment. The better route for higher power would be the AS15T950 as a OPT, as it’s about 1k a-a and would handle over 700 volts peak applied, producing 63 volts RMS on the 120v “primary” all day long down to 25 Hz. Can you say PA subwoofer amp? I sure can. Would go great with my Williamson on the top cabs. Wouldn’t use all the current capabilities of sweep tubes, but they will run with 800-900 V plates, won’t they? Top caps won’t arc over like cheap octal tube sockets will when they get dirty, and I would have 40 watts dissipation each to work with. I’m thinking that the voltage would not really be that insane. I did buy one of those trafos to have on hand some time ago thinking I would be doing it this way when I was hoarding LW6’s.
Just dropping in here to mention that I am planning sort-of bigger amplifier, using all toroids. I will be using 12AV5GA in pentode mode, and will regulate the screens to about 150 volts, with an AS1T150 power transformer for around a 370-400 plate voltage at full power. I decided to try the Antek AS-1206 dual 6 volt secondary as outputs, which at an 8 ohm nominal load should be a reflected ~2.9k or so. I'm hoping for 30 watts or so. I think the power transformer should work well with enough capacitance, as I don't plan for extended high output, but worst case I can add a second one later if I need more current handing. My only doubt is to whether or not I should have gone with a higher voltage power transformer, but I can always upgrade later. I already have the iron, just need to start on a chassis.
Last edited:
Just remembered I got measurements RE: the 120:120 primaries of the triad coils I use. Those measurements are for one coil. I figure two in series doubles the inductance. No wonder they pass 2Hz!
I can't speak for other brands, but I know Triad works as OPT in my designs. If it ain't broke, and it's cheaper (for a Canadian) why change?
Output transformers a cheap alternative?
I can't speak for other brands, but I know Triad works as OPT in my designs. If it ain't broke, and it's cheaper (for a Canadian) why change?
Output transformers a cheap alternative?