I am have 4 x EL34 and a pair of PP OPTs with no UL taps on the primary winding and only one 0-4-8 ohm secondary winding. This effectively rules out UL/distributed load configuration. The impedance is 3.5k P-P and power handling is nominally 100w but is probably nearer 50w. This gives me the options of either pentode mode or triode mode, if I use an 8 ohm speaker on the 4 ohm secondary.
Up to now, I've been using PP triode-mode but the ratio I get with my 8 ohm speaker on the 4 ohm tap gives me a rather sub-optimal load of 7k P-P instead of the more ideal 5k. I'd like to try pentode-mode PP, to get more output power, but I realise I'd need a lot more voltage gain and NFB to get acceptable damping and distortion.
In fact, the amount of NFB called for with pentodes (at least 26dB) would seem to be much more than some people would advocate these days for an open, detailed sound. I'm confused, because some high-NFB amps (e.g. Citation II with 36dB of NFB) are spoken of with reverence. Or are these different people who are commenting?
I'd try it for myself, except that I think I'd need a quite different topology to change from triode to pentode mode, because of the extra gain required for high NFB. So I'd rather hear what others have to say before going to any great effort to try pentodes.
Up to now, I've been using PP triode-mode but the ratio I get with my 8 ohm speaker on the 4 ohm tap gives me a rather sub-optimal load of 7k P-P instead of the more ideal 5k. I'd like to try pentode-mode PP, to get more output power, but I realise I'd need a lot more voltage gain and NFB to get acceptable damping and distortion.
In fact, the amount of NFB called for with pentodes (at least 26dB) would seem to be much more than some people would advocate these days for an open, detailed sound. I'm confused, because some high-NFB amps (e.g. Citation II with 36dB of NFB) are spoken of with reverence. Or are these different people who are commenting?
I'd try it for myself, except that I think I'd need a quite different topology to change from triode to pentode mode, because of the extra gain required for high NFB. So I'd rather hear what others have to say before going to any great effort to try pentodes.
It all depends on your musical priorities and taste. And your speakers. No matter how much i like open loop triode amplifiers they fall short on at least 50% of the music i listen to. NFB pentodes may not be able to compete in directness, soundstaging and unrestricted dynamics but complex music is better controlled and more listenable. And most speakers expect a low driving impedance.
Pay particular attention to a stable and clean G2 source - it makes a huge difference subjectively.
Pay particular attention to a stable and clean G2 source - it makes a huge difference subjectively.
High NFB need not necesarily mean that you have to realise extra high gain and then use loop feedback from the secondary of the tranformer. Read John Broskie's article on "Partial Feedback": http://www.tubecad.com/march2001/2001_03.pdf , then read it again. And then read it tomorrow. It's quite complicated and takes some time to sink in. He uses a 300B as an example, but the concept is arguably of more use to pentodes as they have little input capacitance in comparison to triodes and have more need for feedback due to high output impedance and higher distortion, but the calculations in the triode example are equally applicable. The reduced output impedance of the output stage also makes life easier for the output transformer - it need not have as much primary inductance for a given cutoff frequency.ray_moth said:
In short, you can reduce the both output impedance and the distortion produced by the output stage by approximately a factor of its gain + 1 (thus realising 100% feedback) without enormous drive voltages. A humble small signal pentode will suffice.
hey-Hey!!!,
Look at the early RCA pentode amp design. No loop FB requird.
http://mysite.du.edu/~etuttle/electron/elect35.htm
That one I've heard( and even in its PPP version the SP20); best off-the-rack amp yet. I'm contemplating it for use with sweep pentode finals.
I've got a Citation II and it is a good amp. I can't get over the feeling that its VA/PI circuitry is a bit much. My E-Linear W6m sound much better and have nearly the same Iron. Before I give up on the Citation, I'll put its OPT in place of the stock ones and see how we do.
cheers,
Douglas
Look at the early RCA pentode amp design. No loop FB requird.
http://mysite.du.edu/~etuttle/electron/elect35.htm
That one I've heard( and even in its PPP version the SP20); best off-the-rack amp yet. I'm contemplating it for use with sweep pentode finals.
I've got a Citation II and it is a good amp. I can't get over the feeling that its VA/PI circuitry is a bit much. My E-Linear W6m sound much better and have nearly the same Iron. Before I give up on the Citation, I'll put its OPT in place of the stock ones and see how we do.
cheers,
Douglas
Ray,
As the happy owner of a Cit. 2, I have considerable respect for the doings of Harman and Hegeman, in that time frame. You don't have to get involved with the complex circuitry of the "Duece", when EL34 O/P tubes are being used. Instead, model the circuitry of the simpler Cit. 5, which uses 6L6 family O/P types. Look here.
Hegeman relied on bandwidth to make lots of NFB work very well. Don't use his circuits in combination with mediocre O/P "iron".
High gm to provide slew limiting resistance is very important, when substantial HF error correction signals are present. With that in mind, replacement of the 6CG7 by an ECC99 is something to consider. Obviously, gm is not an issue with the 12BY7 and its close cousins.
BTW, 1 of the changes McShane makes to the Cit. 5 is the inclusion of a choke in the g2 B+ supply. Surprise (NOT!), better regulation there improves things.
As the happy owner of a Cit. 2, I have considerable respect for the doings of Harman and Hegeman, in that time frame. You don't have to get involved with the complex circuitry of the "Duece", when EL34 O/P tubes are being used. Instead, model the circuitry of the simpler Cit. 5, which uses 6L6 family O/P types. Look here.
Hegeman relied on bandwidth to make lots of NFB work very well. Don't use his circuits in combination with mediocre O/P "iron".
High gm to provide slew limiting resistance is very important, when substantial HF error correction signals are present. With that in mind, replacement of the 6CG7 by an ECC99 is something to consider. Obviously, gm is not an issue with the 12BY7 and its close cousins.
BTW, 1 of the changes McShane makes to the Cit. 5 is the inclusion of a choke in the g2 B+ supply. Surprise (NOT!), better regulation there improves things.
Eli Duttman said:
That circuit has relatively little FB that requires high BW from its Iron. There's plate-to-grid around the LTP stage, and cross-coupled plate FB from the power tubes to the LTP grids. The loop from the secondary is minimal compared to most( like the marginally stable Heathkit W5 ).
cheers,
Douglas
I find myself fascinated with these local feedback techniques in lieu of global FB (or to reduce amount). Makes me wonder why they aren't used more often. Does it require just that much more thought to implement correctly?
😕
😕
JoshK said:
It is Heathkit's TotL 70W monoblock amp. Came with a Peerless 16431 opt very similar in spec to the Citation II. 3k2 to 4, 9, 16 and 64 Ohms. It had simiilar PS too; same HV doubler, but w/o a choke, larger negative rail for a cathode follower stage and a user-adjustable mix of current and voltage NFB. Alledgedly very low production.
cheers,
Douglas
Plate to Plate feedback or Partial feedback does deliver the goods. I have two amps at the moment which use the technique. The sound is clear detailed and punchy with plenty of air. No hint of the usual GNFB ills of which people report.
However the design is very exacting as it is a compromise between enough anode load on the drivers verses enough current through the drivers. You don't get both together. The choice of suitable drivers can be very limited. One of the best drivers is a 6AU6 in pentode mode as this is capable of high output impedance (essential) and good current.
It took me months to work out a suitable set of compromises and in the end it looked just like Gary Pimms Tabor amp. Dc coupled and no Global feedback, it kicks some tight bass alright.
Shoog
However the design is very exacting as it is a compromise between enough anode load on the drivers verses enough current through the drivers. You don't get both together. The choice of suitable drivers can be very limited. One of the best drivers is a 6AU6 in pentode mode as this is capable of high output impedance (essential) and good current.
It took me months to work out a suitable set of compromises and in the end it looked just like Gary Pimms Tabor amp. Dc coupled and no Global feedback, it kicks some tight bass alright.
Shoog
In order to get more power in pentode mode you need a transformer with much higher P-P impedance than 3.5k. My guess is that with this transformer you will be better of with triode mode anyway.
Regarding LFB from output plates (or UL taps) into driver stage cathode (or g2 or plate for that matter) - I've done some simulations of this topology recently. One thing to watch carefully is PSU ripple - the common mode power ripple is fed into driver stage, amplified there before being applied to the output grids. Since the transformer impedance for common mode is low common mode amplification runs pretty much in open loop and the common mode current ripple in the final can get out of control and cause intermod even though you may not hear hum directly since xformer does not pass common mode to the speaker.
This topology is better used with dedicated feedback winding with no connection to B+ (either referenced to gnd or floating). With some trade-offs a speaker winding can be used for symmetric feedback into the driver stage by grounding 4ohm tap and using 0ohm and 16ohm for ground referenced feedback.
Regarding LFB from output plates (or UL taps) into driver stage cathode (or g2 or plate for that matter) - I've done some simulations of this topology recently. One thing to watch carefully is PSU ripple - the common mode power ripple is fed into driver stage, amplified there before being applied to the output grids. Since the transformer impedance for common mode is low common mode amplification runs pretty much in open loop and the common mode current ripple in the final can get out of control and cause intermod even though you may not hear hum directly since xformer does not pass common mode to the speaker.
This topology is better used with dedicated feedback winding with no connection to B+ (either referenced to gnd or floating). With some trade-offs a speaker winding can be used for symmetric feedback into the driver stage by grounding 4ohm tap and using 0ohm and 16ohm for ground referenced feedback.
ray_moth said:
Too bad you could not come to BA this year, you could listen to my pentode output (and pentode input!) amp and guess how deep is GNFB. 😉
It is obvious. A switch from plates to screens in some amps are a typical hoax. You have to switch the whole amps, including output transformers, to get both optimal triode and optimal pentode designs.
vladn said:
It is a big difference where to apply the local feedback to. If to apply it to driver anodes however the output stage will be linearized, it's output resistance lowered, but a driver stage will be loaded on much lower resistances, so it's distortions will go up, and specter will go wider.
Re: Re: Pentod PP - too much NFB needed to sound nice?
hey-Hey!!!,
Not exactly so, the loadline changes, but for pentodes it is of great importance to stay away from the knee for best distortion performance. If you're going for maximum output swing for a given B+, you will drive through part of the knee. Pentodes go to nearly all even and low order as the load line steepens. That is what is going on here, and the LTP circuit does away with that.
cheers,
Douglas
Wavebourn said:
hey-Hey!!!,
Not exactly so, the loadline changes, but for pentodes it is of great importance to stay away from the knee for best distortion performance. If you're going for maximum output swing for a given B+, you will drive through part of the knee. Pentodes go to nearly all even and low order as the load line steepens. That is what is going on here, and the LTP circuit does away with that.
cheers,
Douglas
Re: Re: Re: Pentod PP - too much NFB needed to sound nice?
Sure, pentodes are more forgiving to load abuse.
Bandersnatch said:
Sure, pentodes are more forgiving to load abuse.
Re: Re: Re: Re: Pentod PP - too much NFB needed to sound nice?
I wouldn't exactly call it load abuse. Adjust the gain and output Z, not to mention the possible output voltage capability with the load value. In any case, it is quite useful...🙂
cheers,
Douglas
Wavebourn said:
I wouldn't exactly call it load abuse. Adjust the gain and output Z, not to mention the possible output voltage capability with the load value. In any case, it is quite useful...🙂
cheers,
Douglas
Re: Re: Re: Re: Re: Pentod PP - too much NFB needed to sound nice?
It is always useful to use particular tubes in modes they perform the best. My comment was about loading of triode driver stages on very low dynamic resistance of an output stage with parallel feedback. Free lunch may be found in a mouse trap only. The electronic design always means optimization, no such thing as "The best of all bests" 😉
Bandersnatch said:
It is always useful to use particular tubes in modes they perform the best. My comment was about loading of triode driver stages on very low dynamic resistance of an output stage with parallel feedback. Free lunch may be found in a mouse trap only. The electronic design always means optimization, no such thing as "The best of all bests" 😉
Shoog said:
Does the RH84 design fall into this category? The circuit is well documented and easy to find through the forum. I've considering building one to see how it performs.
Ray,
I'm looking at "partial" feedback for a parallel push pull KT88 amp at the moment. There are quite a lot of schematics out there which use feedback from the output tube anodes returned to the driver tube cathodes. However, I'm looking instead at extending my "Baby Huey" scheme instead using something like 6AU6, 6BR7, 6CL6 or EF86 as pentode diffamp front end and then a driver buffer 2nd stage to direct couple to the output tube grids.
The local shunt feedback lowers the output tube rp and so extends the output tranny roll offs at both high and low frequency ends. This then makes application of a little global feedback easier (stability issues are reduced). I have actually found that up to about 5dB of global feedback can help "focus" the stereo image and that amount of global feedback will give nearly a halving of output impedance which is often handy.
Pity the Output Tranny doesn't have 4 Ohm and 16 Ohm taps, then you could also apply some cathode feedback to the output tubes.
Cheers,
Ian
I'm looking at "partial" feedback for a parallel push pull KT88 amp at the moment. There are quite a lot of schematics out there which use feedback from the output tube anodes returned to the driver tube cathodes. However, I'm looking instead at extending my "Baby Huey" scheme instead using something like 6AU6, 6BR7, 6CL6 or EF86 as pentode diffamp front end and then a driver buffer 2nd stage to direct couple to the output tube grids.
The local shunt feedback lowers the output tube rp and so extends the output tranny roll offs at both high and low frequency ends. This then makes application of a little global feedback easier (stability issues are reduced). I have actually found that up to about 5dB of global feedback can help "focus" the stereo image and that amount of global feedback will give nearly a halving of output impedance which is often handy.
Pity the Output Tranny doesn't have 4 Ohm and 16 Ohm taps, then you could also apply some cathode feedback to the output tubes.
Cheers,
Ian
I built a parafeed version of this but never really liked the result to much. Having said that I used suboptimal anode chokes which cobbled it a bit. Others have built it and love the result. Personally I would redesign using a decent driver pentode such as the 6AU6. I would also go with split rails (using a silicone bridge for the neg rail) and direct couple it.
Shoog
Shoog said:
Shoog,
Did you have any trouble with DC stability?
The balance and bias of the output stage is determined by the current through the input valves, and the circuit, if without the output stage CCS, would have the output stage control grids at B+ at turn on, reducing slowly as the input valves warm up and begin conducting. (Admittedly, the latter concern is lessened with slow warming valve rectifiers)
- Status
- Not open for further replies.