I wish to make a unity gain tube buffer, preferably cascaded two stage, which will handle up to 100Vpp input/output.
To keep input impedance high, it would be better to use feedback injection from the output of the second stage to the cathode of the first stage .
If OLG of the two stages were around 100 (40dB), then a feedback ratio of this size would give vanishingly low Zout and distortion. However, devising an input stage which will handle 100Vpp is no slouch.
My thinking would be that a differential input using medium mu tubes followed by a larger tube running say 15mA for output might be best.
Does anyone have any experience/thoughts on this topology?
Cheers,
Hugh
To keep input impedance high, it would be better to use feedback injection from the output of the second stage to the cathode of the first stage .
If OLG of the two stages were around 100 (40dB), then a feedback ratio of this size would give vanishingly low Zout and distortion. However, devising an input stage which will handle 100Vpp is no slouch.
My thinking would be that a differential input using medium mu tubes followed by a larger tube running say 15mA for output might be best.
Does anyone have any experience/thoughts on this topology?
Cheers,
Hugh
Thanks Guys,
Frank, thank you for your thoughts. But a 6DJ8/6BX7 would not handle feedback down to unity gain, I don't believe so anyway.
Tim, no, I don't want a Cathode Follower, because the feedback is local, and I need something with interstage feedback, unless....
Could an anode follower do it? Anyone know how it might be configured with say a 300V B+ to input and output 100V with feedback around the stage?
Of course, this might mean a pentode like the 12GN7, but it might just be possible if it could be configured to avoid input overload. That's the tough one; input overload.
Cheers,
Hugh
Frank, thank you for your thoughts. But a 6DJ8/6BX7 would not handle feedback down to unity gain, I don't believe so anyway.
Tim, no, I don't want a Cathode Follower, because the feedback is local, and I need something with interstage feedback, unless....
Could an anode follower do it? Anyone know how it might be configured with say a 300V B+ to input and output 100V with feedback around the stage?
Of course, this might mean a pentode like the 12GN7, but it might just be possible if it could be configured to avoid input overload. That's the tough one; input overload.
Cheers,
Hugh
Tim, Tim, Tim. It's much cooler if you use a complicated circuit with a fancy name, preferably one originally developed for a 1942 vintage table radio. Why do something simple when there's a hard way to do it? Besides which, a fancier circuit will have a few extra poles in it, which can lead to many happy hours trying to stabilize it. A CF is just no fun.
OK, OK........
Let me explain why! I want to use this stage to drive a PP SS output stage, and need the global feedback loop to smooth off the rough edges, like output stage non-linearities and crossover distortion. You can't do this with a CF, of course, and yes, SY, I'm know of the difficulties of poles and associated stability issues.
Of course, it's possible this can't be done with tubes.....
Now, with ignorance and stupidity on my part discounted somewhat, does anyone have any ideas?
Cheers,
Hugh
Let me explain why! I want to use this stage to drive a PP SS output stage, and need the global feedback loop to smooth off the rough edges, like output stage non-linearities and crossover distortion. You can't do this with a CF, of course, and yes, SY, I'm know of the difficulties of poles and associated stability issues.
Of course, it's possible this can't be done with tubes.....
Now, with ignorance and stupidity on my part discounted somewhat, does anyone have any ideas?
Cheers,
Hugh
and no, I don't have anything useful to say, sorry...
Hugh, why will a CF not drive a SS output stage? As long as the source impedance is low and there is sufficient drive current, it ought to work just fine. I've driven O/P stage screens with a 6SN7 CF, and that's a more nonlinear and difficult load than most SS stages.
My own inclination, were this a simple complementary bipolar emitter follower meant to drive a lot of current, would be to insert a MOS source follower between the tubes and the sand. The small gate to drain capacitance is a piece of cake as a load.
My own inclination, were this a simple complementary bipolar emitter follower meant to drive a lot of current, would be to insert a MOS source follower between the tubes and the sand. The small gate to drain capacitance is a piece of cake as a load.
Hi SY,
Absolutely, a CF will drive a double emitter follower SS output stage with ease, no question.
However, to round out the non-linearities, we need some global feedback based on voltage gain. Let's say 40dB tops. This is, of course, a voltage gain in open loop of 100 times, which is a snap with two common cathode triodes in cascade.
But there's a problem. If we apply 40dB to this animal to calm the savage SS breast, we must inject the full output swing of say 80Vpp into the first tube's grid.
Is their such a beast? I know of this one which looks pretty clever: http://www.geocities.com/tokyo/5913/hybrid.html
I should clearly state my interest so those in the know have the option of not responding. I'm interested in this for academic reasons at this stage, but if it works, and I expect a LOT of work to make it happen, it could end up a commercial product for, to risk that ugly, malignant word, PROFIT. I'm reminded of a Yes, Minister program years ago where the Banker, seeking permission to erect a huge office block in London, says, 'If you let us put on another ten stories, Minister, we can deliver more rateable properties and really clean up'. Minister Hacker replies: 'Is that all you ever think about, just profits?'.
And the Banker says, 'Not just profits, Minister [pauses for effect], Profits...'
Cheers,
Hugh
Absolutely, a CF will drive a double emitter follower SS output stage with ease, no question.
However, to round out the non-linearities, we need some global feedback based on voltage gain. Let's say 40dB tops. This is, of course, a voltage gain in open loop of 100 times, which is a snap with two common cathode triodes in cascade.
But there's a problem. If we apply 40dB to this animal to calm the savage SS breast, we must inject the full output swing of say 80Vpp into the first tube's grid.
Is their such a beast? I know of this one which looks pretty clever: http://www.geocities.com/tokyo/5913/hybrid.html
I should clearly state my interest so those in the know have the option of not responding. I'm interested in this for academic reasons at this stage, but if it works, and I expect a LOT of work to make it happen, it could end up a commercial product for, to risk that ugly, malignant word, PROFIT. I'm reminded of a Yes, Minister program years ago where the Banker, seeking permission to erect a huge office block in London, says, 'If you let us put on another ten stories, Minister, we can deliver more rateable properties and really clean up'. Minister Hacker replies: 'Is that all you ever think about, just profits?'.
And the Banker says, 'Not just profits, Minister [pauses for effect], Profits...'
Cheers,
Hugh
Well, if you want a final gain of 26 dB (a pretty common number for power amps), then you're only injecting 1/20th of the output swing into that first stage grid. Four volts ought to be pretty easy for a competently designed input stage. Even with the CF in the picture, I'll guess you could set a dominant pole at a reasonably high frequency and still stabilize things pretty well.
SY,
I don't want gain in closed loop, as I want to add an input tube stage for the necessary 26dB, perhaps a 6SL7 in plate load.
So, I guess the question remains, can a two triode cascade deliver unity gain to 80Vpp output with 40dB of global feedback?
With a tube bias of say -12V from say a 5687, could it deal with an input of 80Vpp if the feedback was plumbed back into the cathode resistor?
It might be time to buy John Broskie's TubeCAD, perhaps even build the thing.......
Thanks for all the contributions, keep 'em rolling if you have any good ideas!
Cheers,
Hugh
I don't want gain in closed loop, as I want to add an input tube stage for the necessary 26dB, perhaps a 6SL7 in plate load.
So, I guess the question remains, can a two triode cascade deliver unity gain to 80Vpp output with 40dB of global feedback?
With a tube bias of say -12V from say a 5687, could it deal with an input of 80Vpp if the feedback was plumbed back into the cathode resistor?
It might be time to buy John Broskie's TubeCAD, perhaps even build the thing.......
Thanks for all the contributions, keep 'em rolling if you have any good ideas!
Cheers,
Hugh
Hugh, I'm getting a little confused Originally I thought you were asking for a local NFB unity gain buffer that doesn't contain a CF. But, after the last post it appears that you might be wanting to take global NFB from the ouput back into the cathode of the buffer.
Since the latter is fairly standard practice, I'm assuming the former. In which case, it seems to me that a cascaded pair has a few disadvantages: 1) you'll probably need at least one bypass cap on the second stage 2) local NFB into the cathode will require matching the bias of the input stage in the feedback loop which may necessitate capacitative coupling which will add its own features to the transform space.
So, here is another thought just to have something to pick at. It will swing 50Vpp easily. It has on OLG of about 300. I assume you don't mind a SS CCS. There are a few flaws in this, including the need for several voltages, but it may offer some food for thought.
Unfortunately, TubeCad won't help you with this one. I have it, it's a great tool, I use it all the time, but it won't solve this problem.
Originally I thought you were asking for a local NFB unity gain buffer that doesn't contain a CF. But, after the last post it appears that you might be wanting to take global NFB from the ouput back into the cathode of the buffer.Since the latter is fairly standard practice, I'm assuming the former. In which case, it seems to me that a cascaded pair has a few disadvantages: 1) you'll probably need at least one bypass cap on the second stage 2) local NFB into the cathode will require matching the bias of the input stage in the feedback loop which may necessitate capacitative coupling which will add its own features to the transform space.
So, here is another thought just to have something to pick at. It will swing 50Vpp easily. It has on OLG of about 300. I assume you don't mind a SS CCS. There are a few flaws in this, including the need for several voltages, but it may offer some food for thought.
Unfortunately, TubeCad won't help you with this one. I have it, it's a great tool, I use it all the time, but it won't solve this problem.
Attachments
If you're going to feed back across multiple stages, and to say a cathode, then bias doesn't matter. You could feed back the exact output voltage to the first tube's cathode, and then the entire amplifier would act be a voltage follower. So why not feed back a few volts to the more sensitive 6SL7?
Tim
Tim
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.