The vast majority, though certainly not all, of amplifiers use something called a follower output stage. The term follower and the associated concept can be used with all active devices: tubes, bipolar transistors, JFETs, and MOSFETs.
All gain devices have three pins of interest. One is where the current goes in. Another is where the current comes out. The third pin is the one that controls the flow of electrons through the other two. Tubes seemingly confuse the issue by adding more control pins for tetrodes and pentodes, and of course the pins to power the filament, but the basic principle remains the same.
The Circlotron output stage has been around for the better part of a century, having been invented back during the tube days. The configuration of a Circlotron is different from a normal follower output in that the components are arranged in sort of a daisy chain. Interestingly, you have your choice of two different places in the daisy chain where you can tap into the current flow, but nearly everyone chooses the lower impedance nodes at the cathodes (if we're talking about tubes), emitters (bipolars), or Sources (JFETs or MOSFETs).
If you choose to attach the load--meaning the speaker for present purposes--at the Sources, then the output stage is still a follower, even though it's arranged differently.
Followers are popular as amplifier outputs because they have a low output impedance. That means that from the perspective of the speaker looking back towards the output of the amplifier you see a very low resistance. In the real world, there are also small amounts of inductance and capacitance and the sum of the resistance, inductance, and capacitance is known as impedance. The fact that the impedance of a follower is low translates to a high damping factor, which helps the amplifier control the woofer.
The other useful trait of a follower is that it supplies current. You can get as much current out of the other end of the device but the impedance is much, much higher and that's inconvenient if you want to drive a speaker. So the follower--in this case a Source follower since we're talking about MOSFETs in the output stage--allows you to have a low output impedance and lots of current controlled by the pin where the signal goes in.
If you were to complain about the follower, the thing you'd most likely pick to complain about is that it has no voltage gain. In fact, in the purely mathematical sense, it loses a small amount. Fortunately, that amount is small enough that most people just disregard it and say that the follower is a "unity gain" circuit, at least in terms of voltage. By that they mean that they assume the voltage will remain constant (close enough, anyway), but a lot more current will be available.
So the idea that the Thorens output circuit has no voltage gain is unsurprising. The same thing is true of most amplifiers on the market. (If you want an exception, look no further than Nelson's Aleph amplifiers, but they're not hybrids so we can skip that for the time being.)
As for the claim that the follower output in the Thorens allows the tube contribution to dominate the sound...that's where the conversation begins the subtle transition from technical to sales talk. There are assumptions being made; assumptions that he's not making clear to you as he glides past them. Please don't misinterpret what I'm saying. I'm not criticizing him. He made a set of design choices that made sense to him and they will to some extent reflect his personal philosophy. And at the end of the day, the amount of money he takes home to pay bills and buy food with is directly or indirectly linked to the success of this amplifier in the marketplace. So it's natural that a certain amount of sales will enter the equation. Having been in audio myself a while back, I know what it's like to have to make a living selling luxury goods. It's not always easy.
He's also making the assumption that only voltage gain can influence the sound; not current gain. That's a very questionable assumption.
If you look at what he said under a microscope, you'll see that he glossed over the interface between the tubes and the transistors, which is my concern. Imagine yourself to be somewhere tectonically stable, like the middle of the North American continent. You look around and see stability and normality as far as the eye can see. Now imagine yourself to be on the fault line between two tectonic plates. Most times, things are quiet and stable, but when things get rough, they get really rough, fast and without warning. Now pretend that tubes are one continent and solid state another. As long as you are well within the borders of either plate, things are fine, but if you go to the boundary between the two, you'll see the potential for quite a bit of turmoil where the edges of the two plates rub together.
My definition of classic doesn't include the number of units manufactured. There can be lots, or a few, or somewhere in between. They may have been accepted when they were introduced to the market, or not. Only time will tell whether something is truly deserving of the label classic. For example, lots of movies go completely flat at the box office only to come into their own later--Blade Runner might be advanced as an example. It did terribly on release, but its following grew over the years to the point where a lot of people now regard it as a true classic. So much so that they simply assume that it was always so. (I remember when it had Harrison Ford's narration--actually, I preferred it that way.) Is it fair that deserving products sometimes fail commercially? No. But we can't change that. The world is what it is. To award something a "consolation prize" by calling it a classic cheapens the meaning of the term in the long run.
Again, mere production doesn't impress me. Mike Elliot designed unreliable electronics under the name Counterpoint (I sold Counterpoint back when I was in audio--as for the reliability of the product, I have only one thing to say: Ugh!) and the company eventually went under. Now he has a new name, but I don't see the world beating a path to his doorstep.
If you take commercial availability as the threshold criterion for "classic" status, then every single product ever made is a classic. It takes only a moment's reflection to see that can't be right. Or as they put it in the movie The Incredibles, "When everybody's special, nobody is."
Grey
All gain devices have three pins of interest. One is where the current goes in. Another is where the current comes out. The third pin is the one that controls the flow of electrons through the other two. Tubes seemingly confuse the issue by adding more control pins for tetrodes and pentodes, and of course the pins to power the filament, but the basic principle remains the same.
The Circlotron output stage has been around for the better part of a century, having been invented back during the tube days. The configuration of a Circlotron is different from a normal follower output in that the components are arranged in sort of a daisy chain. Interestingly, you have your choice of two different places in the daisy chain where you can tap into the current flow, but nearly everyone chooses the lower impedance nodes at the cathodes (if we're talking about tubes), emitters (bipolars), or Sources (JFETs or MOSFETs).
If you choose to attach the load--meaning the speaker for present purposes--at the Sources, then the output stage is still a follower, even though it's arranged differently.
Followers are popular as amplifier outputs because they have a low output impedance. That means that from the perspective of the speaker looking back towards the output of the amplifier you see a very low resistance. In the real world, there are also small amounts of inductance and capacitance and the sum of the resistance, inductance, and capacitance is known as impedance. The fact that the impedance of a follower is low translates to a high damping factor, which helps the amplifier control the woofer.
The other useful trait of a follower is that it supplies current. You can get as much current out of the other end of the device but the impedance is much, much higher and that's inconvenient if you want to drive a speaker. So the follower--in this case a Source follower since we're talking about MOSFETs in the output stage--allows you to have a low output impedance and lots of current controlled by the pin where the signal goes in.
If you were to complain about the follower, the thing you'd most likely pick to complain about is that it has no voltage gain. In fact, in the purely mathematical sense, it loses a small amount. Fortunately, that amount is small enough that most people just disregard it and say that the follower is a "unity gain" circuit, at least in terms of voltage. By that they mean that they assume the voltage will remain constant (close enough, anyway), but a lot more current will be available.
So the idea that the Thorens output circuit has no voltage gain is unsurprising. The same thing is true of most amplifiers on the market. (If you want an exception, look no further than Nelson's Aleph amplifiers, but they're not hybrids so we can skip that for the time being.)
As for the claim that the follower output in the Thorens allows the tube contribution to dominate the sound...that's where the conversation begins the subtle transition from technical to sales talk. There are assumptions being made; assumptions that he's not making clear to you as he glides past them. Please don't misinterpret what I'm saying. I'm not criticizing him. He made a set of design choices that made sense to him and they will to some extent reflect his personal philosophy. And at the end of the day, the amount of money he takes home to pay bills and buy food with is directly or indirectly linked to the success of this amplifier in the marketplace. So it's natural that a certain amount of sales will enter the equation. Having been in audio myself a while back, I know what it's like to have to make a living selling luxury goods. It's not always easy.
He's also making the assumption that only voltage gain can influence the sound; not current gain. That's a very questionable assumption.
If you look at what he said under a microscope, you'll see that he glossed over the interface between the tubes and the transistors, which is my concern. Imagine yourself to be somewhere tectonically stable, like the middle of the North American continent. You look around and see stability and normality as far as the eye can see. Now imagine yourself to be on the fault line between two tectonic plates. Most times, things are quiet and stable, but when things get rough, they get really rough, fast and without warning. Now pretend that tubes are one continent and solid state another. As long as you are well within the borders of either plate, things are fine, but if you go to the boundary between the two, you'll see the potential for quite a bit of turmoil where the edges of the two plates rub together.
My definition of classic doesn't include the number of units manufactured. There can be lots, or a few, or somewhere in between. They may have been accepted when they were introduced to the market, or not. Only time will tell whether something is truly deserving of the label classic. For example, lots of movies go completely flat at the box office only to come into their own later--Blade Runner might be advanced as an example. It did terribly on release, but its following grew over the years to the point where a lot of people now regard it as a true classic. So much so that they simply assume that it was always so. (I remember when it had Harrison Ford's narration--actually, I preferred it that way.) Is it fair that deserving products sometimes fail commercially? No. But we can't change that. The world is what it is. To award something a "consolation prize" by calling it a classic cheapens the meaning of the term in the long run.
Again, mere production doesn't impress me. Mike Elliot designed unreliable electronics under the name Counterpoint (I sold Counterpoint back when I was in audio--as for the reliability of the product, I have only one thing to say: Ugh!) and the company eventually went under. Now he has a new name, but I don't see the world beating a path to his doorstep.
If you take commercial availability as the threshold criterion for "classic" status, then every single product ever made is a classic. It takes only a moment's reflection to see that can't be right. Or as they put it in the movie The Incredibles, "When everybody's special, nobody is."
Grey
From what you posted above, I understood you to mean that the designer had made the claim.
There's no question that tubes sound different than solid state. I happen to prefer tubes, but grew weary of the cost and aggravation of changing them. To say that the Thorens amplifier has some flavor of tubes about it isn't surprising to me. It would be more surprising if it didn't.
Grey
There's no question that tubes sound different than solid state. I happen to prefer tubes, but grew weary of the cost and aggravation of changing them. To say that the Thorens amplifier has some flavor of tubes about it isn't surprising to me. It would be more surprising if it didn't.
Grey
Hhm, Grey,
I think you are circumventing your own words, which where:
"Assume 1Vrms to full power
Assume a 47k input impedance
This works out to .02mArms to drive this hypothetical amplifier to full power. However, you're looking at (and here I'll use the GR-25 simply because the numbers are fresh in my mind) 1.77Arms at the output...and that's for a mere 25W into 8 Ohms. That's nearly 90,000 times gain for the current."
That was your statement about "current gain" which IMO is not leading to anywhere.
The other things -- where and how to set the optimum interface point between tubes and SS -- are very valid, though.
For the circlotron, even when drawn as a "follower" (load between the sources) it still can be common source circuit (true current output) and its gain (and DF) set by feedback only. And you can have anything in between, it depends at which points along the load you choose to inject the drive signals, on opposite sides==voltage output (follower, with or without feedback at will, ie the Thorens is with feedback), on same sides==current output (transconductance stage, dto). That's one of the unique circlotron features (I'm currently dealing with these intensively). To me its one of the most natural concepts for true symmetrical and balanced amplifier design, especially for outputs of course.
Klaus
I think you are circumventing your own words, which where:
"Assume 1Vrms to full power
Assume a 47k input impedance
This works out to .02mArms to drive this hypothetical amplifier to full power. However, you're looking at (and here I'll use the GR-25 simply because the numbers are fresh in my mind) 1.77Arms at the output...and that's for a mere 25W into 8 Ohms. That's nearly 90,000 times gain for the current."
That was your statement about "current gain" which IMO is not leading to anywhere.
The other things -- where and how to set the optimum interface point between tubes and SS -- are very valid, though.
For the circlotron, even when drawn as a "follower" (load between the sources) it still can be common source circuit (true current output) and its gain (and DF) set by feedback only. And you can have anything in between, it depends at which points along the load you choose to inject the drive signals, on opposite sides==voltage output (follower, with or without feedback at will, ie the Thorens is with feedback), on same sides==current output (transconductance stage, dto). That's one of the unique circlotron features (I'm currently dealing with these intensively). To me its one of the most natural concepts for true symmetrical and balanced amplifier design, especially for outputs of course.
Klaus
Back down the old path
Some timess ideas spring from places where people did not know "We already talked about that and it can not be done, because of the following reasons......".
If we dwell on the reasons why something can not be done we are not being creative or stretching our skills. It is important to recognize that there is a difference between barriers and challenges.
Creativity is the willingness to try!, throw ideas on the table and work them through.
Point - counter point becomes boring and serves no useful growth purpose. These types of discussions easily turn into unproductive emotion based positional flame wars and some one ends up in Texas.
Another way may be more productive.
Think creativity and positive contribution to allow ideas to grow and devlop. Create a fertile environment for the unbiased newbie to put an idea on the table. Some ideas fall by the way side some ideas rise to their potential.
A toxic acid environment does not spawn creativity. Leave that path way for other forums.
I too think the concepts surrounding the Zenkido have great potential. I do not have the skills to go beyond that but I do think there is something there.
Nuff said for now.
Some timess ideas spring from places where people did not know "We already talked about that and it can not be done, because of the following reasons......".
If we dwell on the reasons why something can not be done we are not being creative or stretching our skills. It is important to recognize that there is a difference between barriers and challenges.
Creativity is the willingness to try!, throw ideas on the table and work them through.
Point - counter point becomes boring and serves no useful growth purpose. These types of discussions easily turn into unproductive emotion based positional flame wars and some one ends up in Texas.
Another way may be more productive.
Think creativity and positive contribution to allow ideas to grow and devlop. Create a fertile environment for the unbiased newbie to put an idea on the table. Some ideas fall by the way side some ideas rise to their potential.
A toxic acid environment does not spawn creativity. Leave that path way for other forums.
I too think the concepts surrounding the Zenkido have great potential. I do not have the skills to go beyond that but I do think there is something there.
Nuff said for now.
Re: Back down the old path
I already did. And explained my reasons for thinking that they were worth trying. What no one has done is give me 48 hour days in which to try all the ideas that I come up with.
All I get in return is specious arguments about current gain.
Since no one else seems to want to talk about the idea, it will have to wait until I get time to try it myself. Unfortunately, that will be a while as I've got other circuits on my mind at the moment.
By your logic, the voltage across that same resistor is of no importance in calculating the voltage gain of an amplifier. This just gets curiouser and curiouser.
Oh, well.
Grey
SCD said:
I already did. And explained my reasons for thinking that they were worth trying. What no one has done is give me 48 hour days in which to try all the ideas that I come up with.
All I get in return is specious arguments about current gain.
Since no one else seems to want to talk about the idea, it will have to wait until I get time to try it myself. Unfortunately, that will be a while as I've got other circuits on my mind at the moment.
KSTR said:
By your logic, the voltage across that same resistor is of no importance in calculating the voltage gain of an amplifier. This just gets curiouser and curiouser.
Oh, well.
Grey
Grey, you seem to be a true master of hair-splitting and putting things into one's mouth that haven't been said at all.
Of course the voltage across the R is all important, but the current isn't, the amplifier in no way knows about this current nor does it need to know (as long as it is voltage input, what is assumed here). The current is returned to the preamp via the cable, and for the preamp the current may matter.
- Klaus (now definitely EOD)
Of course the voltage across the R is all important, but the current isn't, the amplifier in no way knows about this current nor does it need to know (as long as it is voltage input, what is assumed here). The current is returned to the preamp via the cable, and for the preamp the current may matter.
- Klaus (now definitely EOD)
, I am running Zen V9 with the tube (6922) buffer...The market has spoken rather clearly on the issue over the years, in that there has never been a hybrid amplifier that's been a runaway success. Not one. Some have had decent sales. Many have been dismal failures. But compared to all-tube or all-solid state amplifiers, hybrids have made hardly a blip on the radar. In fact, the species is so rare that the average man on the street is unaware of their existence.
So why is that and what can be done to change things? As I've already said, I love the sound of tubes. Many others feel the same way I do. The idea that I outlined above may or may not be a factor in pulling hybrids out of the weeds, but just for the heck of it I'll leave the beginnings of a trail of breadcrumbs for anyone who might find my hypothesis persuasive.
Hopefully I and/or Srajan Ebaen can, singularly or together, make a case for taking tubes at least as far down the amplification chain as the final stage of the preamp. I use tubes that far, and I gather that he does as well. We're hardly the only ones.
So how might one go about designing a hybrid amplifier?
Given that the Pass Labs forum is heavily skewed towards solid state, lets back up and look at the overall structure of a typical tube amp. Why? Because it's not the same as a solid state amp...and that's where part of the problem resides.
A "normal" solid state amp has three stages. It has an input stage which doesn't normally contribute much to the overall gain. It simply serves as a buffer between the rest of the amplifier and the outside world. It presents a high impedance to whatever drives it and acts as a receiving point for feedback. Next comes the second stage, which many people call the VAS. Those three initials already hint at trouble to come, because they stand for Voltage Amplification Stage. The VAS typically does the lion's share of the voltage amplification for the amplifier, taking a modest one or two volt signal and boosting it to the point where it goes from rail to rail. The VAS then hands off the signal to the output stage which is nearly always a follower. As I said above, a follower provides no voltage gain, but lots of current gain. The exact proportion depends on whether you're talking about bipolars or FETs as output devices, but either way, most of the current comes from the output.
Quick recap: Typical solid state amplifiers have three stages, each with distinctly different personalities. A buffer, a voltage gain stage, and a current gain stage.
Typical tube amps have three stages too, but make no mistake, tube amps are not like solid stage amplifiers! In a tube amp, every stage is a voltage gain stage, including the output stage. All three stages have similar personalities.
(Yes, just as there are a hundred variations on solid state amps that don't map in a 1:1 way to my description above, there are a hundred variations on tube amps that don't map 1:1 to my generalized description of a tube amp. We can nail the variations down later if need be.)
If you built a generic tube amp and compared the three stages to the three stages of a solid state amp, life would get confusing very quickly indeed. In a solid state amp, the follower output generates the majority of the current that drives the speaker. So where the hell does the current come from in a tube amp? The transformer.
Uh, oh...now we've really done it...we've opened a can that is simply packed with worms struggling to get out. Why? Because the transformer is the Achilles Heel of a tube amplifier. It has bandwidth limitations, phase shift problems, distortions of various and sundry kinds, it's expensive...and as if all that's not enough, the blasted things are heavy, too. In short, there's little to recommend an output transformer and we'd rather just forget the stupid thing if it's at all possible.
Which brings us smack dab back where we started--hybrid amplifiers.
After all, wouldn't it be great if there were some way to tack a solid state output stage onto a tube front end?
To which Srajan Ebaen and I say in unison, "Well, duh!"
(He feels the goal has already been reached. I remain unconvinced. But let's get back to the top-down design of a hybrid.)
First stage: Well, we've got to have tubes or it won't be a hybrid, right? As as I said earlier, tubes don't do the level shift thing so well, and they don't do current very well either, so we'll need to play to the tube's strengths. They've got as near infinite impedance as you can imagine (think FET levels of Zin, here), low capacitance, and the potential to swing volts in a manner that solid state devices can only dream of. So buffering isn't a problem, but if there's not going to be a VAS, we need to start working on gain sooner rather than later. It's not difficult to get 25-30dB of gain out of a single triode stage, so we'll plan on that. In fact, life would be easy if we wanted to build a no-feedback amp. Just set the first stage for 26dB gain, drop in a tube follower for the second stage, and end with a standard solid state push-pull follower. Ta da! You're done.
But who'd want a no-feedback amp?
(Ahem...a shameless joke at my own expense.)
Okay, back to reality. Most people are going to want to use feedback, so we're going to need more voltage gain in order to burn it off as NFB later on. So now the second stage needs to contribute voltage gain as well. Mind you, in a tube amp this is the natural order of things; for that matter, it's the normal thing in a solid state amp, too, but...
And this is where (I think) things get wonky...
A solid state amp's VAS creates gonzo amounts of voltage gain. I mean, we're talking 50-60dB easy, and can be arranged to deliver a reasonable amount of current as well. But with tubes, you're looking at another 20 or 30dB or so and only modest amounts of current.
Okay, we are now in serious trouble. Let's sit down and talk about the compromises we're facing:
1) We can take the 20 to 30dB of gain, add that to the gain from the first stage, and accept a comparatively low level of negative feedback. In fact, this is exactly what tube amps generally do--they might have an open loop gain of, say, 40dB and use 14dB of that for negative feedback, leaving 26dB of closed loop gain for the user to play his music through.
2) Set your sights on more gain, add a third stage to get another 20-30dB, use something like 40-50dB of NFB, and see where that gets you.
3) Really stir the pot and use P-flavored solid state devices as active loads to juice up each tubes' gain nearly to the theoretical maximum (in tube terms, this is "mu") and use the highest gain tubes we can lay hands on, which would most likely lead you to the 12AX7 (mu=40, if I recall correctly) and try to get away with two gain stages.
All of which leaves us with the unsavory fact that we're generating lots of volts, but virtually no current, and one way or another output stages demand current. Practically speaking, there are only two solid state devices suitable for an output stage--bipolars and MOSFETs. By their very nature, bipolars gobble current at their bases, and MOSFETs, while not devouring electrons whole, require lots of current to charge and discharge the capacitance at their Gates.
Or, in the vernacular, there's no such thing as a free lunch.
Practical solutions are few:
1) Parallel common cathode tubes, taking the signal from the plate. (For those used to MOSFET parlance, this is equivalent to common Source.) The Zout of the stage will drop in inverse relation to the number of tubes you parallel. There are two problems here, one being that the impedance is still pretty high, even with two or three tubes in parallel, the other being that the available drive current isn't rising as quickly as we might wish. If you're happy with 5 or 10mA, then things won't be too bad, but if you're wanting 20-30mA or more, you're going to be building a tube farm.
2) Take the second stage as a voltage gain stage, then use a follower after that to lower your driver stage impedance. The bother here is that although the Zout is much better, the current is still low and you're looking at parallel tubes again.
3) Go to a solid state driver stage, which leads to other problems that I'll skip for the moment.
4) Change the topology to something other than a normal three-stage amp. And no, Circlotrons aren't "different" in this context. They've got the same drive requirements as a "normal" amplifier.
Anyone who has another idea is welcome to add to the list, but I predict that additions will be few. There just aren't a lot of options. I've got a couple of ideas that I'm going to sit on until I get a chance to try them. Out of the options listed above, and assuming that I wanted to use feedback, I'd probably go for two tube voltage gain stages, a tube follower stage (probably paralleled), and a solid state output.
Or, as I said earlier, it just ain't as easy as it looks.
KSTR,
If it's any consolation, I was a little flustered the first time I ran across an analysis of an amplifier in terms of current gain. It's a paradigm shift and it didn't come easily for me.
Grey
So why is that and what can be done to change things? As I've already said, I love the sound of tubes. Many others feel the same way I do. The idea that I outlined above may or may not be a factor in pulling hybrids out of the weeds, but just for the heck of it I'll leave the beginnings of a trail of breadcrumbs for anyone who might find my hypothesis persuasive.
Hopefully I and/or Srajan Ebaen can, singularly or together, make a case for taking tubes at least as far down the amplification chain as the final stage of the preamp. I use tubes that far, and I gather that he does as well. We're hardly the only ones.
So how might one go about designing a hybrid amplifier?
Given that the Pass Labs forum is heavily skewed towards solid state, lets back up and look at the overall structure of a typical tube amp. Why? Because it's not the same as a solid state amp...and that's where part of the problem resides.
A "normal" solid state amp has three stages. It has an input stage which doesn't normally contribute much to the overall gain. It simply serves as a buffer between the rest of the amplifier and the outside world. It presents a high impedance to whatever drives it and acts as a receiving point for feedback. Next comes the second stage, which many people call the VAS. Those three initials already hint at trouble to come, because they stand for Voltage Amplification Stage. The VAS typically does the lion's share of the voltage amplification for the amplifier, taking a modest one or two volt signal and boosting it to the point where it goes from rail to rail. The VAS then hands off the signal to the output stage which is nearly always a follower. As I said above, a follower provides no voltage gain, but lots of current gain. The exact proportion depends on whether you're talking about bipolars or FETs as output devices, but either way, most of the current comes from the output.
Quick recap: Typical solid state amplifiers have three stages, each with distinctly different personalities. A buffer, a voltage gain stage, and a current gain stage.
Typical tube amps have three stages too, but make no mistake, tube amps are not like solid stage amplifiers! In a tube amp, every stage is a voltage gain stage, including the output stage. All three stages have similar personalities.
(Yes, just as there are a hundred variations on solid state amps that don't map in a 1:1 way to my description above, there are a hundred variations on tube amps that don't map 1:1 to my generalized description of a tube amp. We can nail the variations down later if need be.)
If you built a generic tube amp and compared the three stages to the three stages of a solid state amp, life would get confusing very quickly indeed. In a solid state amp, the follower output generates the majority of the current that drives the speaker. So where the hell does the current come from in a tube amp? The transformer.
Uh, oh...now we've really done it...we've opened a can that is simply packed with worms struggling to get out. Why? Because the transformer is the Achilles Heel of a tube amplifier. It has bandwidth limitations, phase shift problems, distortions of various and sundry kinds, it's expensive...and as if all that's not enough, the blasted things are heavy, too. In short, there's little to recommend an output transformer and we'd rather just forget the stupid thing if it's at all possible.
Which brings us smack dab back where we started--hybrid amplifiers.
After all, wouldn't it be great if there were some way to tack a solid state output stage onto a tube front end?
To which Srajan Ebaen and I say in unison, "Well, duh!"
(He feels the goal has already been reached. I remain unconvinced. But let's get back to the top-down design of a hybrid.)
First stage: Well, we've got to have tubes or it won't be a hybrid, right? As as I said earlier, tubes don't do the level shift thing so well, and they don't do current very well either, so we'll need to play to the tube's strengths. They've got as near infinite impedance as you can imagine (think FET levels of Zin, here), low capacitance, and the potential to swing volts in a manner that solid state devices can only dream of. So buffering isn't a problem, but if there's not going to be a VAS, we need to start working on gain sooner rather than later. It's not difficult to get 25-30dB of gain out of a single triode stage, so we'll plan on that. In fact, life would be easy if we wanted to build a no-feedback amp. Just set the first stage for 26dB gain, drop in a tube follower for the second stage, and end with a standard solid state push-pull follower. Ta da! You're done.
But who'd want a no-feedback amp?
(Ahem...a shameless joke at my own expense.)
Okay, back to reality. Most people are going to want to use feedback, so we're going to need more voltage gain in order to burn it off as NFB later on. So now the second stage needs to contribute voltage gain as well. Mind you, in a tube amp this is the natural order of things; for that matter, it's the normal thing in a solid state amp, too, but...
And this is where (I think) things get wonky...
A solid state amp's VAS creates gonzo amounts of voltage gain. I mean, we're talking 50-60dB easy, and can be arranged to deliver a reasonable amount of current as well. But with tubes, you're looking at another 20 or 30dB or so and only modest amounts of current.
Okay, we are now in serious trouble. Let's sit down and talk about the compromises we're facing:
1) We can take the 20 to 30dB of gain, add that to the gain from the first stage, and accept a comparatively low level of negative feedback. In fact, this is exactly what tube amps generally do--they might have an open loop gain of, say, 40dB and use 14dB of that for negative feedback, leaving 26dB of closed loop gain for the user to play his music through.
2) Set your sights on more gain, add a third stage to get another 20-30dB, use something like 40-50dB of NFB, and see where that gets you.
3) Really stir the pot and use P-flavored solid state devices as active loads to juice up each tubes' gain nearly to the theoretical maximum (in tube terms, this is "mu") and use the highest gain tubes we can lay hands on, which would most likely lead you to the 12AX7 (mu=40, if I recall correctly) and try to get away with two gain stages.
All of which leaves us with the unsavory fact that we're generating lots of volts, but virtually no current, and one way or another output stages demand current. Practically speaking, there are only two solid state devices suitable for an output stage--bipolars and MOSFETs. By their very nature, bipolars gobble current at their bases, and MOSFETs, while not devouring electrons whole, require lots of current to charge and discharge the capacitance at their Gates.
Or, in the vernacular, there's no such thing as a free lunch.
Practical solutions are few:
1) Parallel common cathode tubes, taking the signal from the plate. (For those used to MOSFET parlance, this is equivalent to common Source.) The Zout of the stage will drop in inverse relation to the number of tubes you parallel. There are two problems here, one being that the impedance is still pretty high, even with two or three tubes in parallel, the other being that the available drive current isn't rising as quickly as we might wish. If you're happy with 5 or 10mA, then things won't be too bad, but if you're wanting 20-30mA or more, you're going to be building a tube farm.
2) Take the second stage as a voltage gain stage, then use a follower after that to lower your driver stage impedance. The bother here is that although the Zout is much better, the current is still low and you're looking at parallel tubes again.
3) Go to a solid state driver stage, which leads to other problems that I'll skip for the moment.
4) Change the topology to something other than a normal three-stage amp. And no, Circlotrons aren't "different" in this context. They've got the same drive requirements as a "normal" amplifier.
Anyone who has another idea is welcome to add to the list, but I predict that additions will be few. There just aren't a lot of options. I've got a couple of ideas that I'm going to sit on until I get a chance to try them. Out of the options listed above, and assuming that I wanted to use feedback, I'd probably go for two tube voltage gain stages, a tube follower stage (probably paralleled), and a solid state output.
Or, as I said earlier, it just ain't as easy as it looks.
KSTR,
If it's any consolation, I was a little flustered the first time I ran across an analysis of an amplifier in terms of current gain. It's a paradigm shift and it didn't come easily for me.
Grey
Yuck Foo, is this ever going to stop ?
Or witnessing that (Hudson) Hawk crab flash by again, in the words of Danni Aiello : can you fckg believe it ?
Here's the original of that A50-A60 business, E83CC/6FQ7 (or ECC803S/E80CC) and 4 pairs of the TO3 Hitachi junk, biased at 2.4A.
Probably had to be insane to build stuff like that in those days.
Or witnessing that (Hudson) Hawk crab flash by again, in the words of Danni Aiello : can you fckg believe it ?
Here's the original of that A50-A60 business, E83CC/6FQ7 (or ECC803S/E80CC) and 4 pairs of the TO3 Hitachi junk, biased at 2.4A.
Probably had to be insane to build stuff like that in those days.
Attachments
Actually, "Yuck Foo" was merely meant as a synonym for Yuk, not as a metaphor for the other business, the other part was just a quote from a US MADE movie.
But you are spot on, i'm a retard, profoundly rude, and likely very vulgar.
Must be difficult, living in a culture that is pushing out profanities through all media pores 24/7 without any censorship ?
The choice of words in my previous post was intended as an expression of discontent of two references to semi-commercial hybrid amp designs which are cheap ripp-off's of a truly good one.
Frits Savelkoul's A80, the one i posted, did amount to about $3k to assemble in 1988, properly done. The running mate for that power amp, Frits' hybrid pre-amp design, cost even more to build.
The total assembly ticket is the hurdle for decent hybrid amplifiers, how amplifiers sound like is what most are here for, not words.
But you are spot on, i'm a retard, profoundly rude, and likely very vulgar.
Must be difficult, living in a culture that is pushing out profanities through all media pores 24/7 without any censorship ?
The choice of words in my previous post was intended as an expression of discontent of two references to semi-commercial hybrid amp designs which are cheap ripp-off's of a truly good one.
Frits Savelkoul's A80, the one i posted, did amount to about $3k to assemble in 1988, properly done. The running mate for that power amp, Frits' hybrid pre-amp design, cost even more to build.
The total assembly ticket is the hurdle for decent hybrid amplifiers, how amplifiers sound like is what most are here for, not words.
Well, silly me. I thought that 'yuck foo' was 'f-you' with the first letter of each word reversed to avoid being censored on the forum. Naturally, since that is what I thought, I was a little pissed. But if you say that thats not what you meant, I'll give you the benefit of the doubt.
Where does the cost come from in the A80? I mean, I've spent about $1500 on my tube amp so far, but I don't understand how one can spend $3000 on an amplifier, especially one with no output transformers.
Also, contributing to the subject of the thread, I have noticed a lot of new Fairchild high voltage, high power mosfets with extremely low reverse transfer capacitances. I would think that these would be quite effective at going from high tube output impedances to large power fet low input impedances.
Where does the cost come from in the A80? I mean, I've spent about $1500 on my tube amp so far, but I don't understand how one can spend $3000 on an amplifier, especially one with no output transformers.
Also, contributing to the subject of the thread, I have noticed a lot of new Fairchild high voltage, high power mosfets with extremely low reverse transfer capacitances. I would think that these would be quite effective at going from high tube output impedances to large power fet low input impedances.
hybrid amps
Grey, I read your technical description of hybrids with interest. Back in my OTL days, 250watt/ch 6AS7 outputs (20 bulbs-40 tube sections per side) I also wanted to experiement with hybrids.
I used to call my OTL's the popcorn amps as tube sections would flame out if exposed to any dc pulse (like dropping the needle on a record after a wine session). When 6AS&'s were a few bucks a peice no worries but not now at $30 each.
I built a 12AX7 Voltage stage using both sections in parallel driving a 6AS7 mu follower driver stage cap coupled into 4 pairs of parallel mosfets in a classic class AB output stage with a passive bias arrangement for the outputs. I also had a dc sense device with a relay to protect my speakers from start up pulses as the amp warmed up. The AC coupling protected the gates of the fets from the tube start up pulses and I also had zener protection to limit the input voltage.
From what I remember the amps were not too bad but of course they really didn't distinquish themselves either from the normal mass of circuits. I used them as spares when my OTL's were having a snit. Finally I used them as sub amps. The driver boards were dc regualted by floating op amp regulators based on the ARSP 10 preamp. I had them for over 10 years without any problems. I finally gave the little tanks away when I moved.
I have found the Pass projects much more enjoyable and eliminating high voltage from my life is even better. I don't have any interest in going the direction of Hybrids but having a tube preamp driving a ss power amp to me is the best of both worlds if you like tubes. I am now using the DEQX system for speaker and room correction as my preamp driving my GR designed XA100 amps built from this forum.
Grey, I read your technical description of hybrids with interest. Back in my OTL days, 250watt/ch 6AS7 outputs (20 bulbs-40 tube sections per side) I also wanted to experiement with hybrids.
I used to call my OTL's the popcorn amps as tube sections would flame out if exposed to any dc pulse (like dropping the needle on a record after a wine session). When 6AS&'s were a few bucks a peice no worries but not now at $30 each.
I built a 12AX7 Voltage stage using both sections in parallel driving a 6AS7 mu follower driver stage cap coupled into 4 pairs of parallel mosfets in a classic class AB output stage with a passive bias arrangement for the outputs. I also had a dc sense device with a relay to protect my speakers from start up pulses as the amp warmed up. The AC coupling protected the gates of the fets from the tube start up pulses and I also had zener protection to limit the input voltage.
From what I remember the amps were not too bad but of course they really didn't distinquish themselves either from the normal mass of circuits. I used them as spares when my OTL's were having a snit. Finally I used them as sub amps. The driver boards were dc regualted by floating op amp regulators based on the ARSP 10 preamp. I had them for over 10 years without any problems. I finally gave the little tanks away when I moved.
I have found the Pass projects much more enjoyable and eliminating high voltage from my life is even better. I don't have any interest in going the direction of Hybrids but having a tube preamp driving a ss power amp to me is the best of both worlds if you like tubes. I am now using the DEQX system for speaker and room correction as my preamp driving my GR designed XA100 amps built from this forum.
hybrid amps
Grey, I forgot, I did use feedback, I can't remember but I think it was around 12db to the cathode of the 12ax7 from the speaker output. I can't remember if I cap coupled the input or not. I never had a problem with gain as I installed a black beauty volume control pot on the input. It was also fan cooled.
dave
Grey, I forgot, I did use feedback, I can't remember but I think it was around 12db to the cathode of the 12ax7 from the speaker output. I can't remember if I cap coupled the input or not. I never had a problem with gain as I installed a black beauty volume control pot on the input. It was also fan cooled.
dave
I don't know, would it be great? You've jumped a bit ahead of yourself, Grey. Before duh-ing, you'd best first answer why you would want a tube front end instead of solid state. What is wrong with a solid state front end and what is right (or at least better) with a tube front end? Only with a firm grasp on the answers to those questions can you decide whether tacking a tube front end to a solid state output stage is a "Well, duh!" in theory before proceeding to the practice of formulating hybrid amp design goals and implementation.GRollins said:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.