A PP "Unset" configuration with 6AQ5s would work well with these. The source impedance is low and you don't get the power loss you would with triode connection. The attached schematic gives 8W in Class A to below 30Hz with 2 x series connected AN-0115. Output impedance ~7 ohms with no global nfb. You'll have to come up with the front end yourself.
Nice!
Question: Isn't that a cascode?
- The signal goes to MOSFET gate.
- Grid of the 6AQ5 triode acts as 'screen grid' to the MOSFET.
- I'm trying to get my head around how a *pentode* works perched on top of a MOSFET like that. It's sort of like a cascode, since the signal is applied to the MOSFET gate and then the pentode is cathode coupled. I guess from there the screen should be well-stabilized (or regulated) as the plate current varies with signal...
Output impedance of 7 ohms is pretty high, so NFB will need to be employed. Do you think this would react well to plate-grid ("Schade") feedback scheme? I figure it should. Apply the NFB voltage divider from the pentode screen grids? From the plates?
Wow... You guys keep me trying new circuits in LTspice constantly! Learning a lot from this...
--
Last edited:
Your right Pcan Its done in a precision manner in the factory many people cant even hold a metal cutting saw properly let along saw a precision channel , pretty low design standards.
If you’re willing to throw away a few laminations you can usually get a transformer core apart without resorting to cutting. Usually you only lose two or three - I’ve managed to save all but one in several power trafo rewinds (the one I have to mangle to get the process started). You DO want an air gap here so missing two or three laminations isn’t the end of the world.
No it's a sort of cathode-coupled amp. Notice the FETs are P-type. The FET basically acts as a source follower, feeding into the cathode of the tube. This isolates the audio from the local feedback around the tube.
You would get essentially the same end result from an ordinary output stage (no MOSFET) with the same local feedback around the tube, but you'd want to increase the resistor values to keep the input impedance high, and add a blocking cap somewhere, which would deter some folks.
I had a ‘Tubelab’ moment the other day going thru the Antek site (again) and found one that stuck out as different. AN104115. Intended as a step up/down isolation trafo with boost windings. Apparently four 115/120 windings, with the same # of turns on each. Plus a 6 and a 12. If the four 120’s were interleaved that would make a primary that could stand off +/-330 V peak with a single trafo. Now how could this be maximally utilized? Put a 45 volt secondary on it and call it 4 ohms. The existing 18 volts could make the 8 ohm tap - 63 volts). That would be good for 500 watts. A very low impedance primary to be sure (475 ohms a-a) but that’s what sweep tubes are good for. Need 135 watts of dissipation capability and 2.8 amps of current. That’s driveable with 50 bucks worth of sweep tubes, and a B+ of around 400V under load.
If you connect the primaries interleaved like with two cores, won't all flux cancel? even AC? Like connecting the two secondaries out of phase - you get heat but no sound.
I guess it depends on the phasing... Be interesting to try. Except such a low Ra-a is not useful. I don't want to run 8 tubes/channel (EL519, triode).
Also, I don't know how a 200VA coil will be good for 500W? At least not sine drive... I use a PAIR of 250VA on the amps that put out 112W RMS sine 30Hz. I also rate the power transformer like that. That 112W monobloc is powered by a 572VA B+ transformer. 300VA will work if you can stand dropping 10-20V and running the coil hot...
I guess it depends on the phasing... Be interesting to try. Except such a low Ra-a is not useful. I don't want to run 8 tubes/channel (EL519, triode).
Also, I don't know how a 200VA coil will be good for 500W? At least not sine drive... I use a PAIR of 250VA on the amps that put out 112W RMS sine 30Hz. I also rate the power transformer like that. That 112W monobloc is powered by a 572VA B+ transformer. 300VA will work if you can stand dropping 10-20V and running the coil hot...
Last edited:
Yeah, you would have to drive the grid resistor with a follower of some kind. Could be direct coupled if you used the power drive circuit, moving the blocking cap to a high impedance node. Tube k-follower would be possible too, since you would drive a k ohm or two. Driving that cathode directly requires a *very* low source impedance, so the p-channel makes the most sense. It’s just hard to get as good a p-channel as an N, and I dont think there’s a driver tube out there that’s up to the job so staying all tube may not be possible. I’m not really sure what the ultimate limitations are running them in push pull. It was originally developed for single ended class A where the lower mosfet always stays in the linear region. If mosfets stay on all the time they are well behaved. When they go in and out of cutoff, the input and output capacitances have crazy nonlinearities which I’m afraid would introduce the same kinds of crossover distortion products that we’re trying to eliminate by using tubes in the first place.
What I meant was pri-sec-pri-sec, all in phase. Half of the “primary” and half the “secondary” on each side of the “center tap” to enforce DCR and leakage inductance balance. Then trafo I was looking at *is* a 1kVA. Similar idea to the 200 VA unit, but thinking “bigger”. In all these cases, the limiting factor is flux - how much voltage at 50 Hz you can put on the winding before it starts drawing unwanted current. Everything else just turns ratio and how much current you can muster. I wouldn’t want to run 8 EL519’s either. But 8 21LG6’s or even xLW6 is certainly a possibility. I’m just considering options for building the next phase, breaking the 300 watt barrier of the Hammond 1650W, using off the shelf components. I’m not, just not, paying over $1000 for a custom wound transformer to break that barrier.
If you have to ask, you can’t afford it. That’s the way it always works. $100 says it WAY north of $1 a watt. The limitations of what you can do has never been about the tubes. It’s always the iron. Same goes for solid state amps. You’re always limited by the iron. You spend less doubling the number of power transistors (tubes) as you do going to the next size transformer, once you start dealing with hundreds of watts.
Back to the original discussion, The whole point is cheap iron. $17.50 to $28 is a whole lot easier to stomach than the typical $90 they want for a basic 20 watt OPT. Especially when you can get cheap $3 tubes to drive them. If it was just about building a 1kW per channel tube amp no matter what, I would throw 50 grand at the budget. That would be just about right - but the truth of the matter is that it’s a year’s living expenses.
Back to the original discussion, The whole point is cheap iron. $17.50 to $28 is a whole lot easier to stomach than the typical $90 they want for a basic 20 watt OPT. Especially when you can get cheap $3 tubes to drive them. If it was just about building a 1kW per channel tube amp no matter what, I would throw 50 grand at the budget. That would be just about right - but the truth of the matter is that it’s a year’s living expenses.
If they don't screw the design up $80 for 80Watts is an absolute steal. For high bandwidth, OT toroidals are intrinsically superior, so if even a passing effort to optimise them for audio use is applied I would predict that they would better any EI at multiples of the price. The only intrinsic limitation of toroidals is their intolerance of DC. All that exposed core surface on EI is a visible signal of the parasitics embodied within their design. Interwinding capacitance is the parasitic of toroidals and if you are aware of it and design for it you can reduce its impact reasonably easily - the same cannot be said of leakage inductance of EI transformers.
Most people are not prepared to design for the lack of an airgap and so they will not accept the fact that toroidals are intrinsically better than EI's.
Shoog
Most people are not prepared to design for the lack of an airgap and so they will not accept the fact that toroidals are intrinsically better than EI's.
Shoog
Last edited:
With EI core I can get 70-80KHz bandwidth easy even for BIG Single-Ended transformers. With a rather simple design 1/2P-S-P-S-P-S-1/2P I can make a 6K transformer for the 845 that rolls-off (-3bB) at over 40KHz. Small and cheap EI transformers can even go beyond 100 KHz....
With PP it's much easier to go beyond.....
The real limitation of FeSi EI cores is not the bandwidth but the fact that FeSi EI laminations are not 100% oriented. They are 2/3 oriented and 1/3 at 90 deg. This is not a problem for performance itself. In order to get top performance they need to run at lower induction respect to C cores or toroidals and that means bigger size for a given task. At some point it's not really practical anymore whereas a C-cores are fine....
Exotic materials that are not grain oriented don't have this issue and in fact they are pretty common in top quality transformers (not output tough)...
A Piltron is an expensive beast, but still is a top end LLundahl, I know which one I would choose any day of the week.
To my knowledge there are only two manufacturers of toroidal OT's which makes them able to charge high prices for their products. Using a power Toroidal is a significant compromise but it can allow anyone to achieve good results for a fraction of the price of using one of the big brand EI transformer manufacturers.
Shoog
To my knowledge there are only two manufacturers of toroidal OT's which makes them able to charge high prices for their products. Using a power Toroidal is a significant compromise but it can allow anyone to achieve good results for a fraction of the price of using one of the big brand EI transformer manufacturers.
Shoog
That's ok but it's true for any core-type against the power toroidal.
With some experience on my shoulders I would pick a Sowter almost invariably in place of Lundahl, especially the bigger ones.
Nothing in the price range of a suitable Toroidal (€20-40) could be matched by a commercial EI OT transformer. If you are prepared to spend on a Piltron then there would be almost no commercial EI's which would match their performance, accepting their design constaints (attention to DC balance).
Shoog
Shoog
Shoog you sound like a broken record. I have already said that it's ok.
That Plitron make the best output transformers that's just your claim. They are not in my opinion. Expensive? yes...
It already has plate-to-grid feedback. in fact the feedback is the only thing driving the grid, the signal is driving the cathode through the PMOS source follower. You can add more or less feedback as desired, although with that schematic it also affects the dc balance so you need to alter the bias voltages to suit.
As shown, the output tubes remain in class A up to clipping even if load drops to 4 ohms, so no worries about mosfets cutting off. Biased to class AB the performance is significanty worse.
My initial point was that you just can't generalize. This thing of people talking bad of EI cores is really nonsense...even for the money!
Manufacturers catalogues are necessarily limited, especially those of the small manufacturers. They just can't offer something good for everything. They have to make choices and the core type is no guarantee. Of course custom products are possible but forget about saving money.
I can bet I can find a better Edcor transformer than the one below for the money:
TTG-KT88SE - Tube output UL transformer [3kOhm] KT88 / 300B SE - Shop Toroidy.pl
It resonates at 17.5KHz so the FR will heavily depend on the source impedance. Forget 56KHz regardless of what you use and you will need lots of "acrobatics" to apply some feedback...