Mine was built to drive a pair of EL84. With care I was able to find an OK operating point for both. I ll post the circuit in the appropriate thread. Oh, and it is an LTP in my case, sorry I forgot to mention that. 🙂
Bloodmist:
Why would you want forced equal biasing? Some 10% mismatch in quiescent current between the pair means nothing; it won't age the tube any faster or cause any distortion in normal operation. It will only lower peak power an insignificant amount.
With true forced balance my amps measure and sound just as good with slightly (5 to 10%) mismatched tubes than with carefully matched tubes. Bias balance was important before MOSFET CCS tails were available, and if peak power is paramount.
If you want to build a project based on nostalgia or historical aesthetics, that's one thing.
For maximum fidelity use as linear tubes as you can find (4P1L, 6E5P, 2A3, 45, 71A, 6C45P just for example) with horizontal load lines and forced balance, and no NFB.
EL34 with the topology I presented will do good under ½W, ok under 1W and will clip around 2.4W. That's more than plenty for people with normal rooms and normal speakers. For full power range good / excellent performance look at 6E5P in indirectly heated tubes. Or for smaller power maybe 6N3P.
As per OT;
No, it will work. Connect other end of primary to other MOSFET source, and the other end to the other side MOSFET source. Cap in between one side, or not. (I recommend using the cap.)
Don't connect the center tap to anything.
Not counting the air gap, the only difference between a SE OT and a PP OT is the center tap. If you don't connect the center tap, it's just a regular SE OT (without an air gap).
The primary halves are not somehow wound antiphase, they are (in circuit) antiphase because the tubes anodes have signals that are antiphase.
Why would you want forced equal biasing? Some 10% mismatch in quiescent current between the pair means nothing; it won't age the tube any faster or cause any distortion in normal operation. It will only lower peak power an insignificant amount.
With true forced balance my amps measure and sound just as good with slightly (5 to 10%) mismatched tubes than with carefully matched tubes. Bias balance was important before MOSFET CCS tails were available, and if peak power is paramount.
If you want to build a project based on nostalgia or historical aesthetics, that's one thing.
For maximum fidelity use as linear tubes as you can find (4P1L, 6E5P, 2A3, 45, 71A, 6C45P just for example) with horizontal load lines and forced balance, and no NFB.
EL34 with the topology I presented will do good under ½W, ok under 1W and will clip around 2.4W. That's more than plenty for people with normal rooms and normal speakers. For full power range good / excellent performance look at 6E5P in indirectly heated tubes. Or for smaller power maybe 6N3P.
As per OT;
No, it will work. Connect other end of primary to other MOSFET source, and the other end to the other side MOSFET source. Cap in between one side, or not. (I recommend using the cap.)
Don't connect the center tap to anything.
Not counting the air gap, the only difference between a SE OT and a PP OT is the center tap. If you don't connect the center tap, it's just a regular SE OT (without an air gap).
The primary halves are not somehow wound antiphase, they are (in circuit) antiphase because the tubes anodes have signals that are antiphase.
Yes in my former EL84 amp i immitated the Leak 20 stereo except i modified it to work in pentode mode and changed the V1 to ECC81 and direct coupled the LTP. Worked like a charm in some respects but too much gain.
Guess that one explains why i was never able to direct couple the ECC83 with the ECC88 PI with good results. The phase shifting was much lower but at the price of earlier distortion...
Guess that one explains why i was never able to direct couple the ECC83 with the ECC88 PI with good results. The phase shifting was much lower but at the price of earlier distortion...
If you want to test the topology, and also find out how much power you really need (I'm guessing you'll be surprised at the result), try that topology out with some small tubes you have at hand.
It's great with 6SN7 or 2C51 or ECC88. Yes, you'll drive your speakers very well with ECC88 output stage, forced balance gyrator plate loaded parafeed. 😉
IRF840 is fine, I've also used it. I use MPSA13 for the darlington pair in the CCS.
It won't be an effect machine for nostalgia, it'll be very transparent and lively.
It's great with 6SN7 or 2C51 or ECC88. Yes, you'll drive your speakers very well with ECC88 output stage, forced balance gyrator plate loaded parafeed. 😉
IRF840 is fine, I've also used it. I use MPSA13 for the darlington pair in the CCS.
It won't be an effect machine for nostalgia, it'll be very transparent and lively.
You're right... i feel a bit stupid now. Must be the fatigue from work...
I must admit i'm a bit curious why your amp clips at 2.4W, could it be the class A operation or the topology? Or perhaps a bit of both.
It is essential to have high supply voltage, rather 400 V than 300 V, when cathodyne is used to produce high level signal or used as DC-coupled.
My own amp clips at 2W (4P1L balanced). I never run it full volume with speakers.
The EL34 circuit I attached in this thread clips at about 2.4W (I didn't simulate the exact clipping point, but at that point the harmonics have creeped up to -50dB or whatever it was) because the circuit runs out of current and/or voltage.
Make the B+ (and plate voltage) higher, and the current higher (my example had 35mA per tube; quite low but very reliable and powerful enough), and you have a higher clipping point.
The gyrator plate loads make sure that even if the circuits swings almost double the quiescent current of one tube (about 60mA in my example) thru the OT, the tubes themselves operate very linearly in nearly horizontal load lines.
Only the inherent unlinearity of the EL34 tubes makes the harmonics creep up with increasing output power. Use more linear tubes to remedy this.
Update: decided to use mixed bias, that will gain the best of both worlds, meaning O.P. stability AND less of the IMD and sag produced by the cathode resistors. Cathode resistors will be only 340 ohm and bypassed with 1000uF each + 10uF film caps. Then the fixed bias will need to be about -20V. The voltage drop on the cathode resistors will be compensated by increasing the plate voltage to 420V rather than 400V. The voltage drop on those resistors while using -20V fixed bias is 15-20V. I'm still aware i will get lower than "datasheet" output power because the transformers, but that will instead increase the damping factor and decrease distortion at lower power. The distortion profile may change a bit as a result but until i hear it playing i won't know if i love it or hate it. But if slight low order distortion is a "feature" for these tubes at least when triode strapped, then my guess is i will absolutely love the sound.
Did you get a power figure yet? I was thinking 11 watts EL 34 triode cathode bias is as good as it gets. I asked JJ if she will make a TR34. It could be slightly better than the EL34 as triode. A PX25 of sorts. If no one does I might have to. I suspect the anode could be slighly improved. If the cost is more than + 20 % a normal pentode I would say we have been ripped off.
I feel rather uncomfortable to risk an opinion now after almost 8 months and much considered advice before. But if allowd a few general points:
1. I feel it a little unfair to blame the Williamson topology simply because a certain execution has been found unstable. (As far as I am concerned much of that was as a result of the low 8µF filter caps available at the time and perhaps poor consideration of coupling cap poles - but let's leave it at that.) It is still one of the best simple topologies in the field, with some 0,05% distortion at full output.
2. I myself have often used the 'Mullard' topology - actually first made polular by H.J. Leak - because of the slight advantage of an input pentode. (I do not prefer triode input, because of the effect of triode Miller capacitance plus undefined feeding impedance on n.f.b. stability, particularly with a 12AX7 in there.)
3. It is erroneous to regard the direct-coupled Schmitt (ltp) phase inverter as fully direct-coupled. The bottom grid earthing capacitor is as much a coupling capacitor, and if not looked after can create serious l.f. phase shift/imbalance in the OPT.
4. One often hears about the 'high nfb' of the Williamson. 20dB high? With ss circuits mostly boasting > 40dB? Most Leak circuits used 26 dB with the famed Mullard 5-20 even higher. I cannot see the problem as long as stability is kosher, something which is easily accomplished with any half-decent design.
5. Although the OP stated preference for triode output, I would not leave UL (I prefer distributed load) topology out of the equation. Inspection will indicate that e.g. in an EL34 stage with maximum triode output = 15W, its UL configuration will yield lower thd at 15W, with still some 10W to go and similar d.f.
I am certainly not disrespecting other suggestions; only I have found that there is little if any audible advantage to be gained by going more complicated than the above.
Somewhat OT question to Artosalo: Referring to your schematic in your post #8: With due respect, are the voltage/component figures of the input stage correct? I find it strange that with the screen resistor (R10) as high as 1 meg compared to R9 (plate load) of 47K, you still get Vg2=165V compared to Va=141V, with the other values as given? Not serious, just curious - unless the used pentode has a quite unusual µ(g2 - a).
1. I feel it a little unfair to blame the Williamson topology simply because a certain execution has been found unstable. (As far as I am concerned much of that was as a result of the low 8µF filter caps available at the time and perhaps poor consideration of coupling cap poles - but let's leave it at that.) It is still one of the best simple topologies in the field, with some 0,05% distortion at full output.
2. I myself have often used the 'Mullard' topology - actually first made polular by H.J. Leak - because of the slight advantage of an input pentode. (I do not prefer triode input, because of the effect of triode Miller capacitance plus undefined feeding impedance on n.f.b. stability, particularly with a 12AX7 in there.)
3. It is erroneous to regard the direct-coupled Schmitt (ltp) phase inverter as fully direct-coupled. The bottom grid earthing capacitor is as much a coupling capacitor, and if not looked after can create serious l.f. phase shift/imbalance in the OPT.
4. One often hears about the 'high nfb' of the Williamson. 20dB high? With ss circuits mostly boasting > 40dB? Most Leak circuits used 26 dB with the famed Mullard 5-20 even higher. I cannot see the problem as long as stability is kosher, something which is easily accomplished with any half-decent design.
5. Although the OP stated preference for triode output, I would not leave UL (I prefer distributed load) topology out of the equation. Inspection will indicate that e.g. in an EL34 stage with maximum triode output = 15W, its UL configuration will yield lower thd at 15W, with still some 10W to go and similar d.f.
I am certainly not disrespecting other suggestions; only I have found that there is little if any audible advantage to be gained by going more complicated than the above.
Somewhat OT question to Artosalo: Referring to your schematic in your post #8: With due respect, are the voltage/component figures of the input stage correct? I find it strange that with the screen resistor (R10) as high as 1 meg compared to R9 (plate load) of 47K, you still get Vg2=165V compared to Va=141V, with the other values as given? Not serious, just curious - unless the used pentode has a quite unusual µ(g2 - a).
The one good Williamson stereo pair I heard was very like a good transistor amp. Tight and dry, more so than a Quad 303. The leak TL12 + sounds more like the 303, I prefer it. To my ears the Willaimson outperforms the few Krell amplifers I have heard. The Marantz 9 out performs them all. The Dynaco is like the 303. If I had a Williamson I would try less feedback and make the KT 66 grid leak a bit smaller to stop grid current. If memory is right the Willaimson gets about 1 watt boost this way. The no feedback graphs show it. I have never heard a JLH 10 watt class A, he thought it like a Williamson. A Sony TA5650 in higher bias ( when first sold before blow ups ) much like a TL12+ / Dynaco. Most wonderful valve design I heard was Heathkit with EL 84. No idea what number. Tighter than the Dynaco and very open. Most of these observations were through Quad ESL 57. That is chance and not intention. The Quad 2/22 is not my cup of tea. I suspect it is the control unit output arrangement. The Leak Varislope is better although impractical (anode output ). The transistor amps that sound good seem to have very few transistors in the signal path. Taking the point of view that everything is in the signal path I still count amplifying devices as major devices and current sources as fancy resistors.
I am OT here, but while Nigel referred to the Quad 22 control unit: I need to point to one reason why it might (and probably will) sound lousy; pardon if read before.
For some strange reason 6 styrene capacitors/channel used in the tone control circuits, in time drifted to up to twice their values. One needs not say what that would do to response. These should first be replaced before making any judgment regarding its quality. Its output arrangement (like others of the time) is also suited to feed a Quad II only; for general applications alterations need to be made.
Back to topic.
For some strange reason 6 styrene capacitors/channel used in the tone control circuits, in time drifted to up to twice their values. One needs not say what that would do to response. These should first be replaced before making any judgment regarding its quality. Its output arrangement (like others of the time) is also suited to feed a Quad II only; for general applications alterations need to be made.
Back to topic.
- Status
- Not open for further replies.