If you remove the bypass cap, you can drive from one side (ground or NFB to the other tube) and completely skip a phase inverter! Of course this necessarily burns up a good bit of power in the CCS and output tubes (class A only), so for any efficiency you must use a seperate splitter stage.
It won't balance the bias current because the way it works is 5+5 = 10 = 6+4. It won't stabilize bias, it will only provide a smooth, level base for it to operate on (for bias, a steep valley is preferable for stability: imagine a ball rolling around in a bowl, it wants to stay in the center).
Tim
It won't balance the bias current because the way it works is 5+5 = 10 = 6+4. It won't stabilize bias, it will only provide a smooth, level base for it to operate on (for bias, a steep valley is preferable for stability: imagine a ball rolling around in a bowl, it wants to stay in the center).
Tim
but only class A
A single o/p LTP is possible with one CCS. But will sound crummy if driven into class AB/B when the CCS reaches the limit of it's current-sinking ability: ie tail-current>CCS, shutting down tail 2. Or am i wrong?
I was considering replacing conventional by-passed cathode resistors with by-passed CCS's.
Has anyone seen this done?
A single o/p LTP is possible with one CCS. But will sound crummy if driven into class AB/B when the CCS reaches the limit of it's current-sinking ability: ie tail-current>CCS, shutting down tail 2. Or am i wrong?
I was considering replacing conventional by-passed cathode resistors with by-passed CCS's.
Has anyone seen this done?
> why not use a current-source to bias an output tube?
Using only hollow-state technology, the current source is either a fatter tube than the outputs, or wastes a lot of power.
With a sillycon device, it is quite practical.
This does of course force the stage to ONLY work in Class-A. If one side goes near cut-off, the output just clips. In contrast, most tube "Class-A" stages really do get into Class AB at full power, one side near cut-off and the other side covering the difference.
Distortion is in fact higher unless you bypass. When one side goes near cut-off, its cathode impedance rises. That reduces the effective gain of the other (active) side, causing soft-peaks. If it stayed fully Class-A, the variation of the two cathode impedances nearly cancels. Or with resistor bias and going just a little into cutoff, the resistor tends to be of a value to limit the loss of gain on peaks. Bypassing the resistor is not needed in true A operation but usually helps reduce peak distortion because most "A" amps do swing pretty near cut-off.
My gut reaction is that for audio and loudspeakers, it would be better to let the resistor and capacitor support AB swings. A transient at a speaker low-Z dip may swing into Class AB even if the amp has plenty of 8Ω power in Class A. Distortion rises as it goes into AB but not as bad as the clipping imposed by a constant current source.
However I have seen a guitar amp that in effect used constant current drive. It is certainly practical and may have unexpected advantages.
In a VERY different field, vari-gain circuits for recording limiters, I finally proved to myself that constant current biasing actually increases distortion and/or reduces usable input level, because of loss of Gm near peaks. If the CC source is bypassed with a resistor, it gets better. A good value for the resistor is very nearly the value of bias resistor you would use for a non-vari-gain stage. Which confirms why many P-P amps with common resistor and no bypass sound fine.
Try it. With and without bypass(es). At low level it will work fine unbypassed. At high levels you may (or may not) prefer the sound with a bypass. Don't stint on the bypass cap: you are not bypassing the resistor, you are bypassing the tube's cathode impedance (1/Gm) which may be 100 or even 50 ohms in fat tubes. And you really want to bypass small differences between the two tubes. 100µFd is usually a minimum, and 1,000µFd may be audibly better. With 6L6 et al, 1,000uFd 50V is not expensive.
> no need for adjustment...
With one CC source, a P-P stage is not sure to balance, as Tim says and I'm sure you realized. You could use two, but they must be bypassed.
> you can drive from one side
Sure. That's a long-known cheap trick. You need double the drive, but with pentodes that may not be a problem. The real issue is that it goes out of true push-pull at high level. In effect it morphs from push-pull at low level to a funny kind of Single Ended at high level. You retain the advantage of no-DC in the transformer which is nice. You may save a tube. Yet you just do not see this scheme in commercial designs from the Tube Era. Either you get a cheap SEP, or you get full push-pull.
Using only hollow-state technology, the current source is either a fatter tube than the outputs, or wastes a lot of power.
With a sillycon device, it is quite practical.
This does of course force the stage to ONLY work in Class-A. If one side goes near cut-off, the output just clips. In contrast, most tube "Class-A" stages really do get into Class AB at full power, one side near cut-off and the other side covering the difference.
Distortion is in fact higher unless you bypass. When one side goes near cut-off, its cathode impedance rises. That reduces the effective gain of the other (active) side, causing soft-peaks. If it stayed fully Class-A, the variation of the two cathode impedances nearly cancels. Or with resistor bias and going just a little into cutoff, the resistor tends to be of a value to limit the loss of gain on peaks. Bypassing the resistor is not needed in true A operation but usually helps reduce peak distortion because most "A" amps do swing pretty near cut-off.
My gut reaction is that for audio and loudspeakers, it would be better to let the resistor and capacitor support AB swings. A transient at a speaker low-Z dip may swing into Class AB even if the amp has plenty of 8Ω power in Class A. Distortion rises as it goes into AB but not as bad as the clipping imposed by a constant current source.
However I have seen a guitar amp that in effect used constant current drive. It is certainly practical and may have unexpected advantages.
In a VERY different field, vari-gain circuits for recording limiters, I finally proved to myself that constant current biasing actually increases distortion and/or reduces usable input level, because of loss of Gm near peaks. If the CC source is bypassed with a resistor, it gets better. A good value for the resistor is very nearly the value of bias resistor you would use for a non-vari-gain stage. Which confirms why many P-P amps with common resistor and no bypass sound fine.
Try it. With and without bypass(es). At low level it will work fine unbypassed. At high levels you may (or may not) prefer the sound with a bypass. Don't stint on the bypass cap: you are not bypassing the resistor, you are bypassing the tube's cathode impedance (1/Gm) which may be 100 or even 50 ohms in fat tubes. And you really want to bypass small differences between the two tubes. 100µFd is usually a minimum, and 1,000µFd may be audibly better. With 6L6 et al, 1,000uFd 50V is not expensive.
> no need for adjustment...
With one CC source, a P-P stage is not sure to balance, as Tim says and I'm sure you realized. You could use two, but they must be bypassed.
> you can drive from one side
Sure. That's a long-known cheap trick. You need double the drive, but with pentodes that may not be a problem. The real issue is that it goes out of true push-pull at high level. In effect it morphs from push-pull at low level to a funny kind of Single Ended at high level. You retain the advantage of no-DC in the transformer which is nice. You may save a tube. Yet you just do not see this scheme in commercial designs from the Tube Era. Either you get a cheap SEP, or you get full push-pull.
Postscript
Hi,
In essence it is no different from any other auto-bias system. It does however have some disadvantages:
The signal must pass through it.
It assumes that the valve needs constant current, limiting it's application to A1.
A valve failure may take-out the bias circuit.
A very small sense resistor in the cathode, who's voltage is used to determine grid voltage, does not suffer these drawbacks, but lacks the simplicity.
Also, there is a view held by some, that PP sounds better when slightly off-balance.
Cheers,
Hi,
In essence it is no different from any other auto-bias system. It does however have some disadvantages:
The signal must pass through it.
It assumes that the valve needs constant current, limiting it's application to A1.
A valve failure may take-out the bias circuit.
A very small sense resistor in the cathode, who's voltage is used to determine grid voltage, does not suffer these drawbacks, but lacks the simplicity.
Also, there is a view held by some, that PP sounds better when slightly off-balance.
Cheers,
Hi,
John,
Those few, and you can count me in, are actually in the good company of RCA who actually described what was going on in their manuals.
I have the article somewhere in my archives...
Peter,
If that's the Beard I think it is, I can only say that it was almost impossible to bias it correctly for longer than a week.
That one was a techs' nightmare, no matter how well you paired the valves on the testers it just was no good.
To add insult to injury, B+ was higher than usual making it only fit for the more sturdy E84L/7320 which don't come cheap if you want a decent pedigree.
PRR,
Once again your posts are a joy to read.
Too bad it needed proof to realise what was happening when the cathode R isn't bypassed but you just confirmed what my ears often told me too all those years.
With cathode bypassed removed on gear that wasn't specifically and competently desisgned for it I often noticed abrupt clipping.
Not exactly a pleasant sounding experience I hasten to add.
Since I never came across a bypass cap that was sonically transparent I try to avoid them in stages where voltage amplification is high.
Thanks for the great posts.
Cheers,
John,
Those few, and you can count me in, are actually in the good company of RCA who actually described what was going on in their manuals.
I have the article somewhere in my archives...
Peter,
If that's the Beard I think it is, I can only say that it was almost impossible to bias it correctly for longer than a week.
That one was a techs' nightmare, no matter how well you paired the valves on the testers it just was no good.
To add insult to injury, B+ was higher than usual making it only fit for the more sturdy E84L/7320 which don't come cheap if you want a decent pedigree.
PRR,
Once again your posts are a joy to read.
Too bad it needed proof to realise what was happening when the cathode R isn't bypassed but you just confirmed what my ears often told me too all those years.
With cathode bypassed removed on gear that wasn't specifically and competently desisgned for it I often noticed abrupt clipping.
Not exactly a pleasant sounding experience I hasten to add.
Since I never came across a bypass cap that was sonically transparent I try to avoid them in stages where voltage amplification is high.
Thanks for the great posts.
Cheers,
Brett: Most driver/gain stages have some degree of imbalance to begin with. You can polish a turd by introducing a complementary imbalance in the output stage.
I suspect that we'd agree that it's best to make the whole shmeer well-balanced to begin with so that we don't have to do unstable tricks like that in the first place.
I suspect that we'd agree that it's best to make the whole shmeer well-balanced to begin with so that we don't have to do unstable tricks like that in the first place.
Hi,
Ever specced a xformer and measured what you actually paid for?
Ever wondered why those tubes that measured so well on the the tester didn't match so well in the amp?
Shall I go on? Or shall we open up a book and dig into AC current flow?
Maybe I should publish one of those rags too?
Come on you guys, use those little greys cells for a change.
Cheers,
Ever specced a xformer and measured what you actually paid for?
Ever wondered why those tubes that measured so well on the the tester didn't match so well in the amp?
Shall I go on? Or shall we open up a book and dig into AC current flow?
Maybe I should publish one of those rags too?
Come on you guys, use those little greys cells for a change.
Cheers,
SY said:
My preference is to have it balanced well all the way, usually with CCS's in the tails. However, fiddling with a paraphase splitter has been interesting. At about 5% imbalance it seems to fatten up a little bit (more robust sounding) with no real negative sonic consequences, so far. I'm still in the courtship stage with this topology so after the flush or harmonic hormones has worn off, I may find myself having an epiphany, finally understanding why the Scots call them lassies, and saying a quick farewell.
Congrats, Brett, you've discovered the secret of SET magic! And with open eyes yet, which puts you one up on most people.
Frank, yes, I've done one or three of those measurements. The real issue I've seen in balance at the low and mid frequencies versus the balance at high frequencies. The latter can be brought into line, if need be, by some compensation capacitance. Or, if you've got a triode class A stage, by neutralizing caps.
Frank, yes, I've done one or three of those measurements. The real issue I've seen in balance at the low and mid frequencies versus the balance at high frequencies. The latter can be brought into line, if need be, by some compensation capacitance. Or, if you've got a triode class A stage, by neutralizing caps.
SY said:
LOL. I don't think that's all the SE 'magic'; there's an o/p Z that modulates with the signal waveform, low inductance OPT's and often really poor PS design too.
The EL84 amp I mentioned above is nice, but isn't a patch on the PP amp I use as my main unit. I was just surprised (again) at the effect it had. The amps going into the bedroom system so it will be fine driving a crappy Tannoy Red.
back to bias
Firstly: thank you analog_sa for the Beard circuit.
OK, so CCS's are a possible bias solution. Now has anybody considered differentially driving an output tube as follows (sorry, can't supply drawing right now...)
P-channel FET in cathode circuit, using a small separate CCS to fix Vgs and therefore Is. Drive gate and tube grid equally, in anti-phase from a LTP...
I'd like to hear the tube guru's input on this (no pun intended)...
Firstly: thank you analog_sa for the Beard circuit.
OK, so CCS's are a possible bias solution. Now has anybody considered differentially driving an output tube as follows (sorry, can't supply drawing right now...)
P-channel FET in cathode circuit, using a small separate CCS to fix Vgs and therefore Is. Drive gate and tube grid equally, in anti-phase from a LTP...
I'd like to hear the tube guru's input on this (no pun intended)...
fdegrove said:
But did you ask for perfect balance? It's not actually a problem to make transformers that have (near as dammit) perfect balance, and that's why broadcasters use them for microphone and line-level signals in noisy television studios. Exactly the same techniques can be used in an output transformer, but if you don't ask, you don't get.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.