Come on guys, why start this up again, perhaps less harsh adjectives could be used to tone down the rhetoric?! So what if Mr. Kitic has unorthodox views on the designs than most, RH amplifiers work as its designer intended and no amount of simulation or measurement data is going to get either side to change their minds, it all comes down to personal preference... And I am pretty sure that Mr. Kitic does not need anyone's help to fight his battles for him ;-)
Peace,
Jaz
Peace,
Jaz
I feel your pain.
As I was reading this post I would glance up at my "quick and dirty" RH84 with empathy. Poor thing.
It always amazes me when these discussions get out of hand with the old sound vs distortion debate.
It becomes evident very quickly, at least in my humble opinion that measurements are not the be all and end all when it comes to audio.
Audio is, or can be an interesting art form that still has to abide by technical laws, that being said, parts can have additional characteristics that are grossly different then there white-paper counterparts.
And that is where the two fronts meet.
I joined because this forum seems to have the most well rounded naysayers, which isn't always a bad thing. I will likely need some guidance on my 2A3 winter project and I like the fact that most of these threads don't wind up with 2 strands of Kimber tcss vs the world.
The reality is until the simpler devices (Triodes) are fully understood or appreciated, I don't see the need for creating hyperbole around more complex devices.
I suppose that is why I find valve amplifiers fascinating. Technology marches on, being improved light years in just about every other technological field, comparatively speaking. Yet these "simple" century old devices are the most organic (sounding) thing we have to reproduce primarily organic sounds. It really surprises me when this eludes people that I would consider much more knowledgeable then me in electronics.
I know I am getting way off topic here but a measurment is only useful in the proper context.
Example: You have two identical amplifiers, however one has slightly worse measurements due to a manufacturing error/variance that can also be easily heard. In this case adjusting/fixing that amplifier by lowering it's distortion would yield a better sonic result. This is a specific situation. This is not attaching a specific known quantity to any sort of quality scale, if such a thing existed.
That is why it is is very counterproductive to try and apply a quality attribute while arguing different topology.
What is "better": A garbage truck, race car or a horse?
Horses for courses in my opinion.
As I was reading this post I would glance up at my "quick and dirty" RH84 with empathy. Poor thing.
It always amazes me when these discussions get out of hand with the old sound vs distortion debate.
It becomes evident very quickly, at least in my humble opinion that measurements are not the be all and end all when it comes to audio.
Audio is, or can be an interesting art form that still has to abide by technical laws, that being said, parts can have additional characteristics that are grossly different then there white-paper counterparts.
And that is where the two fronts meet.
I joined because this forum seems to have the most well rounded naysayers, which isn't always a bad thing. I will likely need some guidance on my 2A3 winter project and I like the fact that most of these threads don't wind up with 2 strands of Kimber tcss vs the world.
The reality is until the simpler devices (Triodes) are fully understood or appreciated, I don't see the need for creating hyperbole around more complex devices.
I suppose that is why I find valve amplifiers fascinating. Technology marches on, being improved light years in just about every other technological field, comparatively speaking. Yet these "simple" century old devices are the most organic (sounding) thing we have to reproduce primarily organic sounds. It really surprises me when this eludes people that I would consider much more knowledgeable then me in electronics.
I know I am getting way off topic here but a measurment is only useful in the proper context.
Example: You have two identical amplifiers, however one has slightly worse measurements due to a manufacturing error/variance that can also be easily heard. In this case adjusting/fixing that amplifier by lowering it's distortion would yield a better sonic result. This is a specific situation. This is not attaching a specific known quantity to any sort of quality scale, if such a thing existed.
That is why it is is very counterproductive to try and apply a quality attribute while arguing different topology.
What is "better": A garbage truck, race car or a horse?
Horses for courses in my opinion.
I feel your pain.
As I was reading this post I would glance up at my "quick and dirty" RH84 with empathy. Poor thing.
It always amazes me when these discussions get out of hand with the old sound vs distortion debate.
It becomes evident very quickly, at least in my humble opinion that measurements are not the be all and end all when it comes to audio.
Audio is, or can be an interesting art form that still has to abide by technical laws, that being said, parts can have additional characteristics that are grossly different then there white-paper counterparts.
And that is where the two fronts meet.
I joined because this forum seems to have the most well rounded naysayers, which isn't always a bad thing. I will likely need some guidance on my 2A3 winter project and I like the fact that most of these threads don't wind up with 2 strands of Kimber tcss vs the world.
The reality is until the simpler devices (Triodes) are fully understood or appreciated, I don't see the need for creating hyperbole around more complex devices.
I suppose that is why I find valve amplifiers fascinating. Technology marches on, being improved light years in just about every other technological field, comparatively speaking. Yet these "simple" century old devices are the most organic (sounding) thing we have to reproduce primarily organic sounds. It really surprises me when this eludes people that I would consider much more knowledgeable then me in electronics.
I know I am getting way off topic here but a measurment is only useful in the proper context.
Example: You have two identical amplifiers, however one has slightly worse measurements due to a manufacturing error/variance that can also be easily heard. In this case adjusting/fixing that amplifier by lowering it's distortion would yield a better sonic result. This is a specific situation. This is not attaching a specific known quantity to any sort of quality scale, if such a thing existed.
That is why it is is very counterproductive to try and apply a quality attribute while arguing different topology.
What is "better": A garbage truck, race car or a horse?
Horses for courses in my opinion.
HE,
the problem as I see it is that there are many ways to claim "understanding". The operation of a triode is well understood, what that means for MUSIC reproduction lacks sorely in the definition department. It doesn't even end there, the triode (or pentode in the case of an RH-Amp) is plugged into something else at both ends that modifies its behavior. Until we turn off the agenda and start listening like at a concert hall, we can talk ourselves into believing all sorts of stuff.
I play trumpet professionally and come home sometimes and just turn the stereo on in "chill mode". If I notice "nothing" the reproduction is fine. When I have SETs hooked up, I notice things that aren't on stage with me. That may be a problem with the recording, amps, room, speakers or what happens to my ears during a symphonic performance. Funny enough, I lose a bit of "transparency" with a PP UL amp, but the other "problems" are far less noticeable (honey smeared on the violin strings, contrabasses in the bathroom). I have >95dB (not backloaded or fullrange) speakers so power is not the issue. I am looking for methodology to reconcile the problems, not ******* contests about resistors from "experts" that never mention "sound".
I am going to try a RH307A Super Amp - exactly as it was designed. Then I will know for sure. I have read what Alex Kitic has published and think that he is on to something. Should know by Christmas...............
As far as what is better: a garbage truck when trash is an issue, a race car when time is an issue and a horse to keep from getting lonely when others do not share your views on garbage or time.
People can get as indignant as much as they like but it won't change the reality of a fundamentally flawed design.
All of the problems reside with the driver stage and the reasons are quite straightforward.
A triode works most linearly when loaded with a constant high impedance load which produces an almost flat load line. This is why it is common practice to load a preamplifier triode with a constant current source.
A pentode on the other hand produces roughly the same output signal regardless of the load it is seeing - it is a constant current device in of itself. It will put the same signal into a low impedance load as it does into a high impedance load.
What does plate to grid feedback do to the grid of a pentode ?
Typically the grid of any valve is a high fixed impedance load which is easy for either a triode or a pentode to drive with maximum linearity. However, plate to grid feedback makes the output signal part of the grid load with the effect that the load varies in impedance and will be typically very low (in the order of Kilohms rather than its usual megohms range). This represents a tortous load for a triode - everything it hates. For a pentode it is exactly what it likes to see.
The effect of using a triode driver to drive a plate to grid feedback output stage is to significantly increase distortion and this has been measured and documented by members of this forum. The RH designs can produce barely a few clean watts of power before the distortion climbs significantly. The claimed for power output of a RH design is with such high levels of distortion that it really doesn't compare favourably with an old table top radio.
Compare this to a beefy pentode driver which will simply keep on putting out a clean signal for most of its range (an inadequate pentode will have its own distortion issues in the form of variable gm with signal).
Remember plate to grid feedback can only correct for errors in the output tube and can do nothing to correct distortion introduced by an overloaded driver stage !!
So why are people claiming such great things for the RH designs ? Well most people will be using them at low power levels for most of the time, here the distortion will mostly be second harmonic (the audibly nice type) and the small amount of time when they stray into gross distortion is the time when human hearing fails to register it as distortion. So whats the problem then ? The problem is that by simply substituting a pentode for a triode, the level of overall distortion will be significantly lower and the available clean power will be at least doubled.
The reality seems to be that people like the type of distortion that the RH designs bring to the table, which is fine by me. However it is not a good example of a partial feedback design, and serves to discredit the benefits which partial feedback brings to amplifier design. If you want to see how to use partial feedback to make an excellent amplifier then I would point you to Gary Pimms Tabor amplifier or Peter Millers Mighty Midget amplifier. If you want to understand what is going on within the circuit I would point you to the Tubecad article (a difficult read to say the least, but worth it).
I started out building a RH design, I have since built about a half a dozen of my own designs using partial feedback as the main feedback mechanism - each getting slightly better than the last. I note that recently Alex has started designing his RH amps correctly with a pentode driver, so I suppose the message has finally sunk in and I have no doubt that these will be far superior to the originals, good on you Alex.
However I also note that he has started to use LM317 bias blocks in his SE amps, a practice I have used extensively in my PP designs, but this seems nothing more than a fashion statement in the RH amps since there is absolutely no technical advantage to their use in an SE design. This is just another example of why I believe that Alex is just a chancer who really hasn't got much of an understanding of what he is talking about - and for him to become something of a guru on that basis brings the tube field into disrepute. The fact is that way back when these amplifiers first came out and Alex had these issues raised with him by formum members here, he became so offensive in his attitude that he was perminently banned from this forum. This should tell you something. The other fact which has rankled with people is that Alex claims to have invented the partial feedback concept, when in fact it has been known since Schade described it over 60 years ago - such an ego.
Built what you like, enjoy it, but please make an attempt to understand what you have done and why it behaves in the way that it does. I know what I am talking about because I took the trouble to understand Partial Feedback and apply the knowledge to amplifier design, I make no special claims for my designs - other than they are a technically correct application of Schades principles.
Shoog
All of the problems reside with the driver stage and the reasons are quite straightforward.
A triode works most linearly when loaded with a constant high impedance load which produces an almost flat load line. This is why it is common practice to load a preamplifier triode with a constant current source.
A pentode on the other hand produces roughly the same output signal regardless of the load it is seeing - it is a constant current device in of itself. It will put the same signal into a low impedance load as it does into a high impedance load.
What does plate to grid feedback do to the grid of a pentode ?
Typically the grid of any valve is a high fixed impedance load which is easy for either a triode or a pentode to drive with maximum linearity. However, plate to grid feedback makes the output signal part of the grid load with the effect that the load varies in impedance and will be typically very low (in the order of Kilohms rather than its usual megohms range). This represents a tortous load for a triode - everything it hates. For a pentode it is exactly what it likes to see.
The effect of using a triode driver to drive a plate to grid feedback output stage is to significantly increase distortion and this has been measured and documented by members of this forum. The RH designs can produce barely a few clean watts of power before the distortion climbs significantly. The claimed for power output of a RH design is with such high levels of distortion that it really doesn't compare favourably with an old table top radio.
Compare this to a beefy pentode driver which will simply keep on putting out a clean signal for most of its range (an inadequate pentode will have its own distortion issues in the form of variable gm with signal).
Remember plate to grid feedback can only correct for errors in the output tube and can do nothing to correct distortion introduced by an overloaded driver stage !!
So why are people claiming such great things for the RH designs ? Well most people will be using them at low power levels for most of the time, here the distortion will mostly be second harmonic (the audibly nice type) and the small amount of time when they stray into gross distortion is the time when human hearing fails to register it as distortion. So whats the problem then ? The problem is that by simply substituting a pentode for a triode, the level of overall distortion will be significantly lower and the available clean power will be at least doubled.
The reality seems to be that people like the type of distortion that the RH designs bring to the table, which is fine by me. However it is not a good example of a partial feedback design, and serves to discredit the benefits which partial feedback brings to amplifier design. If you want to see how to use partial feedback to make an excellent amplifier then I would point you to Gary Pimms Tabor amplifier or Peter Millers Mighty Midget amplifier. If you want to understand what is going on within the circuit I would point you to the Tubecad article (a difficult read to say the least, but worth it).
I started out building a RH design, I have since built about a half a dozen of my own designs using partial feedback as the main feedback mechanism - each getting slightly better than the last. I note that recently Alex has started designing his RH amps correctly with a pentode driver, so I suppose the message has finally sunk in and I have no doubt that these will be far superior to the originals, good on you Alex.
However I also note that he has started to use LM317 bias blocks in his SE amps, a practice I have used extensively in my PP designs, but this seems nothing more than a fashion statement in the RH amps since there is absolutely no technical advantage to their use in an SE design. This is just another example of why I believe that Alex is just a chancer who really hasn't got much of an understanding of what he is talking about - and for him to become something of a guru on that basis brings the tube field into disrepute. The fact is that way back when these amplifiers first came out and Alex had these issues raised with him by formum members here, he became so offensive in his attitude that he was perminently banned from this forum. This should tell you something. The other fact which has rankled with people is that Alex claims to have invented the partial feedback concept, when in fact it has been known since Schade described it over 60 years ago - such an ego.
Built what you like, enjoy it, but please make an attempt to understand what you have done and why it behaves in the way that it does. I know what I am talking about because I took the trouble to understand Partial Feedback and apply the knowledge to amplifier design, I make no special claims for my designs - other than they are a technically correct application of Schades principles.
Shoog
Last edited:
It has been speculated that there is distortion cancellation going on between the driver stage and the output stage. Even if this is so - and it results in a benign distortion profile at the final output, it cannot maintain that profile as the power output rises and the triode driver starts to distort badly and far in excess of the distortion of the output stage which has feedback to keep it low.
Whichever way you slice it - a triode driver in a partial feedback amplifier can only perform reasonably well at a fraction of the potential power output of a comparable amplifier using a more conventional forms of feedback. This is not the case with a pentode driver.
Why accept a design which cripples the power capability of your precious valves ?
Shoog
Whichever way you slice it - a triode driver in a partial feedback amplifier can only perform reasonably well at a fraction of the potential power output of a comparable amplifier using a more conventional forms of feedback. This is not the case with a pentode driver.
Why accept a design which cripples the power capability of your precious valves ?
Shoog
I am not sure about this. If you load the 6AU6 with 10K or 100K output signal is not roughly the same for a given input.
If the load changes a little ok but distortion is not necessarily the same. In this respect it is inferior to a linear triode or a triode run in its linear region. The triode can work as well if it has enough current to cope with the load just like the pentode. If you change the load a little bit the gain won't change significantly and distortion as well. If you change the load a lot its distortion can be much less in comparison to most pentodes in general. That's the advantage of giving away some efficiency.
If you just mean the output current ok, it will be roughly the same, but it is not true regarding distortion that can change a lot in pentodes for a varying load and that could make things even in practical application.
Last edited:
Compare it to a beefy triode then. Of course if you compare a "beefy" pentode driver to a small signal triode then the triode is worse.
The only advantage of the pentode driver I can see is that you can do the front-end with one stage. But, except for the complexity, adding a further preamp doesn't mean necessarily worse. As always depends on many things...
Mr K religiously defended the lack of data measurements on the RH designs for years. Now that there are some measurements, I think the criticisms have been shown to be on the mark. What the real issue simply boils down to is benign distortion profile sounding good to some listeners. At least at low levels.
As far as lower distortion arguments go, I would like to point out that taking the Schade/plate feedback back to the driver cathode is way better than either of the conventional triode or pentode Schade driver approaches. Both stages get cleaned up, and a benign load now for the driver. Unfortunately, vanishingly low distortion just makes it sound "clinical", without the residual 2nd H, 3rd H, 4th H tail to give it "detail".
Either you like tailored distortion or not......... Really nothing more to it than that.
As far as lower distortion arguments go, I would like to point out that taking the Schade/plate feedback back to the driver cathode is way better than either of the conventional triode or pentode Schade driver approaches. Both stages get cleaned up, and a benign load now for the driver. Unfortunately, vanishingly low distortion just makes it sound "clinical", without the residual 2nd H, 3rd H, 4th H tail to give it "detail".
Either you like tailored distortion or not......... Really nothing more to it than that.
He he, they should play with some class A2 or AB2 amps to appreciate that where the load really goes from very high to very low and low source impedance is practically the only solution....
"Could you please direct me me to these measurements."
Here are some particularly appalling measurements:
http://www.diyaudio.com/forums/tubes-valves/236769-why-use-pentode-5.html#post3512545
-----------------------------------------------------------------
"He he, they should play with some class A2 or AB2 amps to appreciate that where the load really goes from very high to very low and low source impedance is practically the only solution.... "
Umm, it takes a Mosfet practically to fix the problems there. You have proven my point. Thankyou.
Certainly one COULD use a very low Rp triode to drive the Schaded grid through a fixed resistor to make it look like a current source. But that is not how the RH is configured.
I myself don't see a huge difference between a triode driver and pentode driver for the RH configuration. Either way the driver needs significant attention to linearizing it. The pentode makes for the easiest fixup by simply operating it at high idle current and some cathode degeneration. The triode needs significant cathode degeneration to increase its Rp, and that makes it difficult to achieve the high current drive for driving the low Z Schade feedback point.
The 12AT7 driver however, seems to be specifically chosen for high 2nd harmonic generation (cancellation purposes), since its data sheet refers to it as a MIXER. (high 2nd H distortion required for this service) And a low Z load will make it into a fountain of 2nd Harmonic.
If you want to fix up the Schaded RH setup, get rid of both triode or pentode approachs, and put the feedback back to the driver cathode. (doesn't care if its a triode or pentode driver)
Here are some particularly appalling measurements:
http://www.diyaudio.com/forums/tubes-valves/236769-why-use-pentode-5.html#post3512545
-----------------------------------------------------------------
"He he, they should play with some class A2 or AB2 amps to appreciate that where the load really goes from very high to very low and low source impedance is practically the only solution.... "
Umm, it takes a Mosfet practically to fix the problems there. You have proven my point. Thankyou.
Certainly one COULD use a very low Rp triode to drive the Schaded grid through a fixed resistor to make it look like a current source. But that is not how the RH is configured.
I myself don't see a huge difference between a triode driver and pentode driver for the RH configuration. Either way the driver needs significant attention to linearizing it. The pentode makes for the easiest fixup by simply operating it at high idle current and some cathode degeneration. The triode needs significant cathode degeneration to increase its Rp, and that makes it difficult to achieve the high current drive for driving the low Z Schade feedback point.
The 12AT7 driver however, seems to be specifically chosen for high 2nd harmonic generation (cancellation purposes), since its data sheet refers to it as a MIXER. (high 2nd H distortion required for this service) And a low Z load will make it into a fountain of 2nd Harmonic.
If you want to fix up the Schaded RH setup, get rid of both triode or pentode approachs, and put the feedback back to the driver cathode. (doesn't care if its a triode or pentode driver)
Last edited:
Don,
I usually don't play with pentodes but prefer cathode fb from the OT (that can be enough especially for UL) or a mix of CFB and low amount of loop fb. CFB works really well.
Looking quickly at those distortion measurements you linked if those figures are correct one might get lower distortion from a triode strapped KT88 without fb!
In the case of the original Shade a good candidate for the transformer coupled driver could be a 46 triode (65V peak into 10K, no positive grid drive) that, playing just a little bit with its quiescent point, would just give some 2nd H, between 1-2% at max drive I guess, and very little or nothing relevant for the rest. It's a SE to PP configuration so if that 2nd H cancels directly with that of the 6L6's or cancels because of the PP it won't be there anyway.
For positive grid drive one can essentially say that at some point distortion is proportional to the source impedance. Depending on the power device you can drive it in positive grid for a bit also with some source impedance (like that typical of low mu DHT's if they have enough plate voltage and can provide the current), provided there is absolutely no blocking (so transformer coupling, for example), but then grid current kicks in consistently and there you need a really low output impedance beefy driver. With the 45 and 2A3 I find that a cathode follower, despite not having as low output impedance as a mosfet, is able to drive them for a good bit with THD still low. But not any cathode follower. It has to be one able of high peak current.
I usually don't play with pentodes but prefer cathode fb from the OT (that can be enough especially for UL) or a mix of CFB and low amount of loop fb. CFB works really well.
Looking quickly at those distortion measurements you linked if those figures are correct one might get lower distortion from a triode strapped KT88 without fb!
In the case of the original Shade a good candidate for the transformer coupled driver could be a 46 triode (65V peak into 10K, no positive grid drive) that, playing just a little bit with its quiescent point, would just give some 2nd H, between 1-2% at max drive I guess, and very little or nothing relevant for the rest. It's a SE to PP configuration so if that 2nd H cancels directly with that of the 6L6's or cancels because of the PP it won't be there anyway.
For positive grid drive one can essentially say that at some point distortion is proportional to the source impedance. Depending on the power device you can drive it in positive grid for a bit also with some source impedance (like that typical of low mu DHT's if they have enough plate voltage and can provide the current), provided there is absolutely no blocking (so transformer coupling, for example), but then grid current kicks in consistently and there you need a really low output impedance beefy driver. With the 45 and 2A3 I find that a cathode follower, despite not having as low output impedance as a mosfet, is able to drive them for a good bit with THD still low. But not any cathode follower. It has to be one able of high peak current.
Last edited:
Shoog seems to understand finally that his personal "technical aspects" are not shared by (some)others in the know..........
What I do not understand is a crusade against something that does not affect us personally. I have looked everywhere for "proof" that a SET, SEP, Pentode driver, Triode driver or even MOSFET driver is better or worse. My brilliant discovery? Only the implementation matters. There really are multiple paths to decent sound and BEST SOUND is only in the ear of the beholder. Even more significant: there seems to be an inverse relationship to the believability of those in attack mode. Maybe feedback is a good thing for our "public" lives?
What I do not understand is a crusade against something that does not affect us personally. I have looked everywhere for "proof" that a SET, SEP, Pentode driver, Triode driver or even MOSFET driver is better or worse. My brilliant discovery? Only the implementation matters. There really are multiple paths to decent sound and BEST SOUND is only in the ear of the beholder. Even more significant: there seems to be an inverse relationship to the believability of those in attack mode. Maybe feedback is a good thing for our "public" lives?
There are indeed multiple paths, ands we have to learn by the errors of our own experiments, but there is no benefit in starting out on the path with an approach which is going to result in a low power and high distortion design - that seems peverse to me
Shoog
I dunno... My Novar Spud is an input triode and output pentode with Schade feedback. With a pentode output stage, I get about 6 W out (3 % THD) using the 6LR8. 1 W THD is well below 1 %. With the output stage triode strapped, I get a bit over 4 W (3 % THD) and 0.3~0.4 % THD at 1 W, midband.
I happen to like the sound. Especially with the triode strapped output.
~Tom
I happen to like the sound. Especially with the triode strapped output.
~Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.