Hi,
I decided to start a new thread on this subject because I've become intersted in meeting the challenge of achieving good quality sound with pentodes and I would like to discuss people's experience in this area. It rarely gets discussed, possibly because it's considered to be a relatively poor solution for "hi-fi" and better suited to guitar amps. However, Gary Pimm seems to have achieved success with it.
I guess most of us have heard about:
* the excessive amount of high odd-order distortion from pentodes;
* the sonic superiority of ultralinear and triode modes over pentode mode (I have an EL34 triode-strapped PP amp that sounds pretty good);
* the difficulty of getting a good damping factor, due to the pentode's high plate impedance;
* how the only way to achieve acceptable sound (good damping, low distortion) with pentodes is through the application of heavy NFB, which is not in favour these days;
* the difficulty of achieving stability with so much NFB;
* etc; etc; etc.
The fact remains that you can get more power from the same tubes in pentode mode than you can in the other modes. You are also able to experiment with a much wider range of operating conditions than with either UL (unless you use a separate tertiary winding for the screens) or triode modes.
The EL34 is my personal choice. It has been praised, in some circles, as the best audio power pentode ever made. And it was definitely intended (initially, at least) to be used as a pentode; if you read Mullard's advertising literature carefully, it's clear that they developed the EL34 before they were aware of the possibility of using it in an ultralinear circuit. The fact that the maximum plate voltage is 800v and the maximum screen voltage is around 400v tends to support this.
I have some ideas of my own about how to achieve success using EL34s in pentode mode. They don't involve my own personal favourite all-balanced design approach, which I have found to be too gain-limited to be able to support sufficient NFB for pentodes. I have tested them, as far as I can, with LTSpice and the results were very encouraging. I'd be the first to admit that it's a big step from Spce simulation to actual building and testing, but that's all the more alluring to me.
Anyway, I'd be really interested in what others have managed to achieve with pentode PP and what advice they may be able to offer.
I decided to start a new thread on this subject because I've become intersted in meeting the challenge of achieving good quality sound with pentodes and I would like to discuss people's experience in this area. It rarely gets discussed, possibly because it's considered to be a relatively poor solution for "hi-fi" and better suited to guitar amps. However, Gary Pimm seems to have achieved success with it.
I guess most of us have heard about:
* the excessive amount of high odd-order distortion from pentodes;
* the sonic superiority of ultralinear and triode modes over pentode mode (I have an EL34 triode-strapped PP amp that sounds pretty good);
* the difficulty of getting a good damping factor, due to the pentode's high plate impedance;
* how the only way to achieve acceptable sound (good damping, low distortion) with pentodes is through the application of heavy NFB, which is not in favour these days;
* the difficulty of achieving stability with so much NFB;
* etc; etc; etc.
The fact remains that you can get more power from the same tubes in pentode mode than you can in the other modes. You are also able to experiment with a much wider range of operating conditions than with either UL (unless you use a separate tertiary winding for the screens) or triode modes.
The EL34 is my personal choice. It has been praised, in some circles, as the best audio power pentode ever made. And it was definitely intended (initially, at least) to be used as a pentode; if you read Mullard's advertising literature carefully, it's clear that they developed the EL34 before they were aware of the possibility of using it in an ultralinear circuit. The fact that the maximum plate voltage is 800v and the maximum screen voltage is around 400v tends to support this.
I have some ideas of my own about how to achieve success using EL34s in pentode mode. They don't involve my own personal favourite all-balanced design approach, which I have found to be too gain-limited to be able to support sufficient NFB for pentodes. I have tested them, as far as I can, with LTSpice and the results were very encouraging. I'd be the first to admit that it's a big step from Spce simulation to actual building and testing, but that's all the more alluring to me.
Anyway, I'd be really interested in what others have managed to achieve with pentode PP and what advice they may be able to offer.
I've built four pentode push-pull amps so far, all using 12-22dB of NFB, and I feel like they are even more "high fidelity" sounding than the triode amps I've built. There is a control, and an ease that triodes just don't seem to have; a lack of congestion.
To address some of your other points:
Pentodes do not always have more higher order distortion than triodes. That is a myth. It is totally dependant on loading, the circuit particulars, and the average drive to the output stage.
Also, stabilizing a feedback amp is not difficult. The techniques are all well documented. What is difficult is finding coherent sources for those techniques, and applying them correctly to your particular circuit/output transformer combination. But the reward is worth the time spent, IMO.
Joel
To address some of your other points:
Pentodes do not always have more higher order distortion than triodes. That is a myth. It is totally dependant on loading, the circuit particulars, and the average drive to the output stage.
Also, stabilizing a feedback amp is not difficult. The techniques are all well documented. What is difficult is finding coherent sources for those techniques, and applying them correctly to your particular circuit/output transformer combination. But the reward is worth the time spent, IMO.
Joel
Joel,
My experience is the more feedback means, the greater the damping factor hence you mention better "control " I asume you mean control of the speaker cone. Most visitors here do not like this type of sound, too much control hence "harden the sound output". Some prefer a more airy and open sound where the damping factor can be a low as 2-3(ie less feedback). With correct partnership of speaker it can still sound quite good, while other type it may sound sloppy, esp those that dip in its impedance. an example, there is a much acclaimed amp.,based on a single-ended KT88 with output impedance of 3.8ohms, however it has a special feature ie an ability to adjust its output impedance to match the speaker impedance(I don't how it achieve this).
Just a thought.
My experience is the more feedback means, the greater the damping factor hence you mention better "control " I asume you mean control of the speaker cone. Most visitors here do not like this type of sound, too much control hence "harden the sound output". Some prefer a more airy and open sound where the damping factor can be a low as 2-3(ie less feedback). With correct partnership of speaker it can still sound quite good, while other type it may sound sloppy, esp those that dip in its impedance. an example, there is a much acclaimed amp.,based on a single-ended KT88 with output impedance of 3.8ohms, however it has a special feature ie an ability to adjust its output impedance to match the speaker impedance(I don't how it achieve this).
Just a thought.
It has little to do with damping factor, IMO. It is much more related to being able to deliver real power, at or below rated distortion, at the frequency extremes (and having a smooth response rolloff beyond those points as well).
This is, of course, not the sole providence of pentode w/ feedback amps - but they are usually much better performers in this regard. Yes, this is a generalization - the output transformer is vitally important. However, it's utterly astounding how poorly the typical triode no-feedback amp does at 20Hz and 20kHz, in terms of THD. This is what I mean by a "congested" sound.
I honestly can't hear damping factor. Maybe you can though.
Joel
This is, of course, not the sole providence of pentode w/ feedback amps - but they are usually much better performers in this regard. Yes, this is a generalization - the output transformer is vitally important. However, it's utterly astounding how poorly the typical triode no-feedback amp does at 20Hz and 20kHz, in terms of THD. This is what I mean by a "congested" sound.
I honestly can't hear damping factor. Maybe you can though.
Joel
"The EL34 is my personal choice. It has been praised, in some circles, as the best audio power pentode ever made."
Not mine. I favor the 807 instead. This one has a specified Q-Point that gives a THD= 1.8%. I haven't seen any other type that gives such a low THD figure.
As for how I used them: stiff screen supply. For this, I used an active regulator. The active regulator not only nails down the screen voltage (0.8V drop with the DC rail going down some 40V) it also has a very low AC impedance. Both factors go a long way towards improving linearity.
I included two feedback loops. The inner one taking input from the 807 plates back to the grids of the cathode follower drivers. According to an app report, 6.9db(V) of feedback for this loop is recommended. They certainly knew their stuff since that worked just fine. Getting the output impedance of the finals below r(p) with feedback helps with both speaker damping, and low frequency performance since the xfmr's Xl(pri) becomes a larger percentage of output impedance. That local feedback also helped to tame the excessive "brightness" of the high end.
Next, added ~12.0db(V) of gNFB. This outer feedback loop from xfmr secondary to first preamp cleaned up the mids and highs, and also gave the bass plenty of "authority" without any of the under damped "sloppiness" that can occur with VT amps.
Use fixed bias (better linearity and more power since you aren't losing any across cathode bias resistors). Include a decent driver that can source enough current at the high end to charge up the stray and input capacitance to avoid the slew rate problem, and can drive into Class AB(2) transparantly. I favor a DC coupled cathode follower for a driver. That way, there is no coupling capacitor to accumulate excess bias voltage when the signal forward biases the control grid/cathode parasitic diode. In my 807 amp, the OPT actually saturates before the finals clip.
Sonically, beats hell out of a commercial MOSFET amp from the Big Box store.
Not mine. I favor the 807 instead. This one has a specified Q-Point that gives a THD= 1.8%. I haven't seen any other type that gives such a low THD figure.
As for how I used them: stiff screen supply. For this, I used an active regulator. The active regulator not only nails down the screen voltage (0.8V drop with the DC rail going down some 40V) it also has a very low AC impedance. Both factors go a long way towards improving linearity.
I included two feedback loops. The inner one taking input from the 807 plates back to the grids of the cathode follower drivers. According to an app report, 6.9db(V) of feedback for this loop is recommended. They certainly knew their stuff since that worked just fine. Getting the output impedance of the finals below r(p) with feedback helps with both speaker damping, and low frequency performance since the xfmr's Xl(pri) becomes a larger percentage of output impedance. That local feedback also helped to tame the excessive "brightness" of the high end.
Next, added ~12.0db(V) of gNFB. This outer feedback loop from xfmr secondary to first preamp cleaned up the mids and highs, and also gave the bass plenty of "authority" without any of the under damped "sloppiness" that can occur with VT amps.
Use fixed bias (better linearity and more power since you aren't losing any across cathode bias resistors). Include a decent driver that can source enough current at the high end to charge up the stray and input capacitance to avoid the slew rate problem, and can drive into Class AB(2) transparantly. I favor a DC coupled cathode follower for a driver. That way, there is no coupling capacitor to accumulate excess bias voltage when the signal forward biases the control grid/cathode parasitic diode. In my 807 amp, the OPT actually saturates before the finals clip.
Sonically, beats hell out of a commercial MOSFET amp from the Big Box store.
I'm starting to think that most of the hype about triode mode is the possibility to drive cheap OTs and get good damping factor out of them... go and get some serious ones and you'll see how good a pentode amp can sound!!!
I'm designing a prototype of an EL36PP amp, I'll have the option of configuring it in triode or pentode mode, maybe I'll post some results... I like the approach of returning error signal from the anode of the drivers to the driver tube.
Miles, could you point me at your 807 design? I have some, I love their look and have heard great things about this beam tetrode.
I'm designing a prototype of an EL36PP amp, I'll have the option of configuring it in triode or pentode mode, maybe I'll post some results... I like the approach of returning error signal from the anode of the drivers to the driver tube.
Miles, could you point me at your 807 design? I have some, I love their look and have heard great things about this beam tetrode.
Giaime said:
Hear, hear!
You can find out more about it Here (under a different handle) or see a pic in the Photo Gallery. They do have a nice "retro" look don't they? "...have heard great things about this beam tetrode." And from them.
I'm using push-pull EL84 in pentode at the moment with terrific results. Distortion spectra look fine (at 1W out, 2nd is dominant at -75dBW, 3rd and 5th are below -85dBW, nothing higher above residual), and the advantages that pentodes bring to the table are considerable. But, yes, it means the output transformer has to be an excellent one.
good for me 
Hi, tetrode lovers !
Just finished that:
http://www.dissident-audio.com/PP_6L6/Page.html
I'm more than satisfied
Yves.
Just finished that:
http://www.dissident-audio.com/PP_6L6/Page.html
I'm more than satisfied
Yves.
Even 802 sound good in pentode mode! I bet you have some that you can't find a use. Go and make a 2A3 SE amp killer...
http://giaime.altervista.org/802.html
I bet you have some that you can't find a use. Go and make a 2A3 SE amp killer... http://giaime.altervista.org/802.html
My contribution:
1. Yes, pentodes have low enough distortion - when working into a well-defined load! Typical distortion performance for the 6L6 is well-documented, e.g. fig. 13.30, p.570, Langford-Smith. The point is that in amplifiers the loudspeaker is an atrocious load, from which one can simply not get away.
2. If one insists on pentodes (as said previously, being able to use different screen and anode voltages is advantageous), local feedback is preferable to multistage (or both).
3. For rather more stable NFB globally or at least including the OPT, there is the option of using a separate unloaded winding for feedback, thus obviating the effect of leakage reactance. This only leaves the loudspeaker winding "unattended", which is not that serious. I have seen a circuit with 32 dB of NFB very stably applied in this way - but it does again require that such a secondary is available.
4. I do not see mention of combined positive and negative feedback. NFB over the whole amplifier with positive FB taken off the stage before the power stage, can render the power stage almost "invisible". In an illustration 10W setup, when IM distortion without feedback was 40%, it decreased with NFB only to 8% and with combined n. and p. feedback to 1,9%, etc.
But the decision must be difficult. In practice there is simply not denying the advantages of UL compared to the simplicity. My own biggest effort in this respect was a 100W job using 4 x 6L6GC in UL. I can with great difficulty bring myself not to use this topology, to be honest.
But it is an interesting topic, Ray_Moth, one which I will certainly follow to see what is contributed!
Regards.
1. Yes, pentodes have low enough distortion - when working into a well-defined load! Typical distortion performance for the 6L6 is well-documented, e.g. fig. 13.30, p.570, Langford-Smith. The point is that in amplifiers the loudspeaker is an atrocious load, from which one can simply not get away.
2. If one insists on pentodes (as said previously, being able to use different screen and anode voltages is advantageous), local feedback is preferable to multistage (or both).
3. For rather more stable NFB globally or at least including the OPT, there is the option of using a separate unloaded winding for feedback, thus obviating the effect of leakage reactance. This only leaves the loudspeaker winding "unattended", which is not that serious. I have seen a circuit with 32 dB of NFB very stably applied in this way - but it does again require that such a secondary is available.
4. I do not see mention of combined positive and negative feedback. NFB over the whole amplifier with positive FB taken off the stage before the power stage, can render the power stage almost "invisible". In an illustration 10W setup, when IM distortion without feedback was 40%, it decreased with NFB only to 8% and with combined n. and p. feedback to 1,9%, etc.
But the decision must be difficult. In practice there is simply not denying the advantages of UL compared to the simplicity. My own biggest effort in this respect was a 100W job using 4 x 6L6GC in UL. I can with great difficulty bring myself not to use this topology, to be honest.
But it is an interesting topic, Ray_Moth, one which I will certainly follow to see what is contributed!
Regards.
I certainly hope it proves to be! I'm happy that people have already contributed some interesting stuff. By saying "pentode," by the way, I didn't mean to exclude beam tetrodes. To me, they are equivalent in most respects, although I know there are differences in the way they behave.
I am aware of your preference for UL, Johan, and you argue in a very logical way. I must admit that I overlooked one of the pentode's disadvantages that you mention, namely, its sensitivity to plate load. Distortion increases rapidly either side of the optimum and, as you rightly say, a speaker is far from being a constant load. That's one area where the UL arrangement really scores points, I admit, since it affords much greater latitude. Still, I would like to pursue the pentode/tetrode path, because I think it has much to offer.
There are some interesting aspects to the loading questions. Take a look at the RCA datasheet for the 7027A (my favorite medium-power pentode). There's a chart labelled "Operation Characteristics Push-Pull Class AB1". The chart shows changes in power and distortion with load for a pentode stage with 540V B+ and a 400V screen.
Note the distortion curves. There is indeed a distinct and sharp null at about 6700 ohms p-p. Naively, one might thing to run the stage that way, and indeed, it will result in a very high 76 watts output. But... note what happens if you reduce the plate load to 4500 ohms or so. The max power drops to 60W (not significant), the 3rd rises to 4%, but look at the 5th and 7th! They've dropped out almost entirely. And, open loop, that 4% is not too bad. And note that variation in load either have little effect on the third (load lower) or actually take it toward the null (higher). So using a 4500 ohm nominal loading drops the power slightly, but greeatly reduces load sensitivity.
Note the distortion curves. There is indeed a distinct and sharp null at about 6700 ohms p-p. Naively, one might thing to run the stage that way, and indeed, it will result in a very high 76 watts output. But... note what happens if you reduce the plate load to 4500 ohms or so. The max power drops to 60W (not significant), the 3rd rises to 4%, but look at the 5th and 7th! They've dropped out almost entirely. And, open loop, that 4% is not too bad. And note that variation in load either have little effect on the third (load lower) or actually take it toward the null (higher). So using a 4500 ohm nominal loading drops the power slightly, but greeatly reduces load sensitivity.
Another point about speaker impedances is that a 2-way speaker "system" with crossover, does not have the wild impedance rises and dips of a naked voice coil. Yet, this is what people seem to always be referring to when they mention this "problem". Why? I've measured my 16 ohm 848A's, and they present almost exactly 16 ohms across the spectrum.
This is also easily proved by building a speaker voice coil simulator (Aiken's site has one) out of some chokes and caps, and then looking at square waves with it attached, and with your actual home speaker attached. The first looks awful, the second doesn't.
So, I feel safe throwing out the "I can't use pentodes because they are sensitive to load impedance changes" myth. Do you?
Joel
This is also easily proved by building a speaker voice coil simulator (Aiken's site has one) out of some chokes and caps, and then looking at square waves with it attached, and with your actual home speaker attached. The first looks awful, the second doesn't.
So, I feel safe throwing out the "I can't use pentodes because they are sensitive to load impedance changes" myth. Do you?
Joel
ray_moth said:
"Pentode" means five electrodes, does it not?
1) Cathode
2) Control Grid
3) Screen Grid
4) Beam Formers
5) Plate
Five electrodes: yup, it's a pentode
(Some versions of the 6BQ5 were built with suppressor grids; others with beam formers.)Same problem the solid state people have. Like a synchronous AC motor (which is what a loudspeaker is -- it just goes back and forth instead of 'round and 'round) the actual impedance can vary tremendously in both magnitude and phase. Nothing you can do about that, except deal as best you can.
Strong points made by SY and Joel.
But.......
Joel (yours is easier!) Yes, I have also dabbled a bit in speaker equalisation, though I am by no means an expert. (As an aside, and perhaps because of the latter disqualification, I needed some 3 x L, 6 x C and 3 x R to make a cross-over plus impedance equaliser for a simple 2-way system before I could get an impedance within 6 - 9 ohms and phase angle within +/- 15 degrees.) I thus fear for what is generally on the market. Judging by some of the graphs published for some not unknown loudspeakers, things certainly seem to go quite frenetic, impedance and phase wise. So.
SY,
Yes, somehow I did not look up those characteristics - will rectify that soon. Still, why do I go pentode instead of UL - and this is an honest question, not an attempt to wave the flag for this or that. You mentioned distinct advantages for pentode use, and I will patiently wait for your article.
But here I must confess to going a step further in high-class tube designs, making matters easy for me. I really use the UL configuration of the cathode feedback type (Quad style) of circuit. Immediately special output transformers jump out, but I have the advantage of designing and having my own wound, simply because there is a transformer company close by, and to import any decent transformer is uneconomical with our present balance of economy. (It may not make too much sense to compare, but to have a good 100W C-core output transformer made locally cost me the equivalent of US $88. I cannot import an equivalent Hammond or Lundahl for nearly this.)
If one needs to go to many 100Ws, making separate anode/G2 voltages desirable without the availability of the Quad type of output transformer, then my previously stated position regarding local feedback or +/- will be my route. (You guys are naughty. I have not done a +/- feedback design since decades ago, and am strongly tempted to try that again! But having retired I sadly do not have nice spectrum analysers at my disposal any more - perhaps someone at work still loves me.)
Regards all.
But.......
Joel (yours is easier!) Yes, I have also dabbled a bit in speaker equalisation, though I am by no means an expert. (As an aside, and perhaps because of the latter disqualification, I needed some 3 x L, 6 x C and 3 x R to make a cross-over plus impedance equaliser for a simple 2-way system before I could get an impedance within 6 - 9 ohms and phase angle within +/- 15 degrees.) I thus fear for what is generally on the market. Judging by some of the graphs published for some not unknown loudspeakers, things certainly seem to go quite frenetic, impedance and phase wise. So.
SY,
Yes, somehow I did not look up those characteristics - will rectify that soon. Still, why do I go pentode instead of UL - and this is an honest question, not an attempt to wave the flag for this or that. You mentioned distinct advantages for pentode use, and I will patiently wait for your article.
But here I must confess to going a step further in high-class tube designs, making matters easy for me. I really use the UL configuration of the cathode feedback type (Quad style) of circuit. Immediately special output transformers jump out, but I have the advantage of designing and having my own wound, simply because there is a transformer company close by, and to import any decent transformer is uneconomical with our present balance of economy. (It may not make too much sense to compare, but to have a good 100W C-core output transformer made locally cost me the equivalent of US $88. I cannot import an equivalent Hammond or Lundahl for nearly this.)
If one needs to go to many 100Ws, making separate anode/G2 voltages desirable without the availability of the Quad type of output transformer, then my previously stated position regarding local feedback or +/- will be my route. (You guys are naughty. I have not done a +/- feedback design since decades ago, and am strongly tempted to try that again! But having retired I sadly do not have nice spectrum analysers at my disposal any more - perhaps someone at work still loves me.)
Regards all.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.