Push-Pull Basics.
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#1 Push-Pull Basics.
I need to sort out the facts on this subject, so I will start with the simple differential input stage with both grids driven by a balanced signal, as that's where I am at the moment.
So we have a differential setup, which has been likened to two single ended amplifiers, placed back to back in opposing phase.
So in a normal common cathode, single ended stage I can appreciate the need for a cathode bypass resistor because it prevents loss of gain and increased Ra caused by feedback around the grid circuit at AC that without the cap would remain unchecked.
My problem comes when we have a differential amp, where two valves passing signals in opposing phase are connected together at their cathodes and share a common path to ground, via the tail resistor.
Surely in that situation there is no need for the bypass cap as the two AC signals generated across the cathode resistor will be in opposite phase. Given a perfectly balanced signal applied to both grids, the opposing phase signals will result in no audio on the cathodes, so no feedback, so no raising of Ra, so no need for the bypass cap.
The only situation where this could break down, will be in a Class A/B output stage where each valve is driven into cutoff on alternate cycles, unbalancing the stage and causing one side to have unfettered feedback via the grid circuit, then the other. To prevent this we either use fixed bias for class A/B or use a suitable cathode bypass capacitor, whose value is carefully chosen to be a compromise between distortion performance and too a slow recovery after overload.
Now this all sounds correct to me after having read numerous old books, and latterly Morgan Jones.
The question is, am I missing something?
If so, it would be very useful to know as this sort of thing is right in the area where I want to continue building.
So we have a differential setup, which has been likened to two single ended amplifiers, placed back to back in opposing phase.
So in a normal common cathode, single ended stage I can appreciate the need for a cathode bypass resistor because it prevents loss of gain and increased Ra caused by feedback around the grid circuit at AC that without the cap would remain unchecked.
My problem comes when we have a differential amp, where two valves passing signals in opposing phase are connected together at their cathodes and share a common path to ground, via the tail resistor.
Surely in that situation there is no need for the bypass cap as the two AC signals generated across the cathode resistor will be in opposite phase. Given a perfectly balanced signal applied to both grids, the opposing phase signals will result in no audio on the cathodes, so no feedback, so no raising of Ra, so no need for the bypass cap.
The only situation where this could break down, will be in a Class A/B output stage where each valve is driven into cutoff on alternate cycles, unbalancing the stage and causing one side to have unfettered feedback via the grid circuit, then the other. To prevent this we either use fixed bias for class A/B or use a suitable cathode bypass capacitor, whose value is carefully chosen to be a compromise between distortion performance and too a slow recovery after overload.
Now this all sounds correct to me after having read numerous old books, and latterly Morgan Jones.
The question is, am I missing something?
If so, it would be very useful to know as this sort of thing is right in the area where I want to continue building.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
-
- Old Hand
- Posts: 369
- Joined: Mon Oct 14, 2013 3:20 pm
#2
You are not missing anything - but the feedback resistors you are mentioning, between the anode of the driver and the anode of the output tube on each phase.
Without those resistors, you could choose to have a common resistor on the two drivers (as well as the two output tubes), and as a matter of fact having a CCS instead of the resistor would probably be beneficial towards a more perfect balancing.
Why "probably": because you need to have matched tubes as well, in order to split everything between them (same current, same gain, etc.). Really matched tubes are a rarity, and usually matching is "good enough" - thus I like to avoid situations where perfect matching is required, opting for solutions where there is some automatic mechanism to equalize the results as much as possible.
Thus the CCS under the common cathode becomes separate CCS for each cathode (both driver and output stage). While some of the advantages might seem lost (in the driver stage), having two tubes behave in almost identical manner is actually a good substitute. But the output stage with a CCS under each cathode will have to operate in class A, thus the increase of power possible with class AB (or even B) operation is lost. If power was a premium requirement, it looks best to forget about this solution for the output tubes, but if sound is the premium requirement, I guess that CCS is advisable for this stage.
The real problem arises when you want to combine a CCS instead of an un-bypassed cathode resistor with the anode to anode feedback. The feedback ratio will be greatly reduced, and the only way to get it back would be increasing the feedback resistor.
Increasing the feedback resistor is not something you can do for free, because feedback is not only in voltage but in current as well. Some current is drawn through this resistor and the lower the value (necessary in order to increase feedback) the more it will skew the current ratio between anode resistor and feedback resistor...
Thus whether it can or cannot be done should first be calculated. The calculation requires taking several factors into account. Thus it is much easier to perform simulations than to get a pen and a calculator and start the hard work. For accurate simulations you require accurate models, which is (still) easier to do for triodes than pentodes...
In the end, what you get might not be an improvement, and you might have devoted a lot of effort in trying to manage the solution. It is usually best to avoid this path, because of the diminishing returns.
Last but not least, you are using pentodes. Pentodes are rather "constant current devices" when compared to triodes, because of the influence of the screen grid (mostly, although the third grid/beam former has some effect as well). Thus while CCS may deliver a lot with triodes, the same device will deliver much less with pentodes.
That said, the only reason I use a CCS under the cathode of output tubes (in most cases pentodes/beam tetrodes) in my amps is drawing a constant current regardless of tube type/condition/age/manufacturer - which gives more precision and reliability to the operation of the amp. The other potential improvements are almost unimportant in this case: but in a PP solution drawing a constant current for each tube means perfect DC balancing which is beneficial to the operation of the output transformer (without the necessity to match tubes).
I hope this goes a long way towards explaining. Now, it is possible that someone might step up and state that everything I wrote is rubbish. That would only go a long way towards proving that I am rather stupid to loose time writing such posts...
Without those resistors, you could choose to have a common resistor on the two drivers (as well as the two output tubes), and as a matter of fact having a CCS instead of the resistor would probably be beneficial towards a more perfect balancing.
Why "probably": because you need to have matched tubes as well, in order to split everything between them (same current, same gain, etc.). Really matched tubes are a rarity, and usually matching is "good enough" - thus I like to avoid situations where perfect matching is required, opting for solutions where there is some automatic mechanism to equalize the results as much as possible.
Thus the CCS under the common cathode becomes separate CCS for each cathode (both driver and output stage). While some of the advantages might seem lost (in the driver stage), having two tubes behave in almost identical manner is actually a good substitute. But the output stage with a CCS under each cathode will have to operate in class A, thus the increase of power possible with class AB (or even B) operation is lost. If power was a premium requirement, it looks best to forget about this solution for the output tubes, but if sound is the premium requirement, I guess that CCS is advisable for this stage.
The real problem arises when you want to combine a CCS instead of an un-bypassed cathode resistor with the anode to anode feedback. The feedback ratio will be greatly reduced, and the only way to get it back would be increasing the feedback resistor.
Increasing the feedback resistor is not something you can do for free, because feedback is not only in voltage but in current as well. Some current is drawn through this resistor and the lower the value (necessary in order to increase feedback) the more it will skew the current ratio between anode resistor and feedback resistor...
Thus whether it can or cannot be done should first be calculated. The calculation requires taking several factors into account. Thus it is much easier to perform simulations than to get a pen and a calculator and start the hard work. For accurate simulations you require accurate models, which is (still) easier to do for triodes than pentodes...
In the end, what you get might not be an improvement, and you might have devoted a lot of effort in trying to manage the solution. It is usually best to avoid this path, because of the diminishing returns.
Last but not least, you are using pentodes. Pentodes are rather "constant current devices" when compared to triodes, because of the influence of the screen grid (mostly, although the third grid/beam former has some effect as well). Thus while CCS may deliver a lot with triodes, the same device will deliver much less with pentodes.
That said, the only reason I use a CCS under the cathode of output tubes (in most cases pentodes/beam tetrodes) in my amps is drawing a constant current regardless of tube type/condition/age/manufacturer - which gives more precision and reliability to the operation of the amp. The other potential improvements are almost unimportant in this case: but in a PP solution drawing a constant current for each tube means perfect DC balancing which is beneficial to the operation of the output transformer (without the necessity to match tubes).
I hope this goes a long way towards explaining. Now, it is possible that someone might step up and state that everything I wrote is rubbish. That would only go a long way towards proving that I am rather stupid to loose time writing such posts...
Last edited by Alex Kitic on Sun Jul 20, 2014 11:04 am, edited 1 time in total.
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#3
Thanks
I don't understand all of it but it does throw some light into quite a few dark corners.
Cheers Alex.
I don't understand all of it but it does throw some light into quite a few dark corners.
Cheers Alex.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
#4 Requirements list
Maybe it would be helpful if we declared the design goals and worked backwards. There needs to be no reasons given.
1) efficiency of the speakers to be used
2) how loud in what sized room
3) what type of music is listened to (this determines how much headroom is "sensible"
These first 3 parameters will get us into the ball park as to how much output we need. Great sound quality is a given.
Now we can itemize constraints:
4) time and money available to develop
5) chassis size
6) desired architecture (PP class A, A/AB, AB/B), feedback
7) input is balanced
8) cap or transformer coupled?
I think generally if we design and build to a requirements list, it is much easier to know where and when to start and when we have reached the goal.
Now we can design something.
Jumping the gun a bit, here is the link to Gary Pimms 47 PP with CCS:
http://www.pimmlabs.com/web/47.htm
Interesting to note, he built it textbook first, got it to work and then describes the process of adding technology - one step at a time, with the sonic results - often shared with Lynn Olson. The final result is a very complex amplifier that for sure sounds wonderful. I chose this link rather than the Tabor as it shows the complete process that even Gary Pimm had to go through already having had extensive CCS experience.
1) efficiency of the speakers to be used
2) how loud in what sized room
3) what type of music is listened to (this determines how much headroom is "sensible"
These first 3 parameters will get us into the ball park as to how much output we need. Great sound quality is a given.
Now we can itemize constraints:
4) time and money available to develop
5) chassis size
6) desired architecture (PP class A, A/AB, AB/B), feedback
7) input is balanced
8) cap or transformer coupled?
I think generally if we design and build to a requirements list, it is much easier to know where and when to start and when we have reached the goal.
Now we can design something.
Jumping the gun a bit, here is the link to Gary Pimms 47 PP with CCS:
http://www.pimmlabs.com/web/47.htm
Interesting to note, he built it textbook first, got it to work and then describes the process of adding technology - one step at a time, with the sonic results - often shared with Lynn Olson. The final result is a very complex amplifier that for sure sounds wonderful. I chose this link rather than the Tabor as it shows the complete process that even Gary Pimm had to go through already having had extensive CCS experience.
Whenever I feel blue, I start breathing again.
-
- Old Hand
- Posts: 369
- Joined: Mon Oct 14, 2013 3:20 pm
#5
I'm replying mostly to stay on topic. I am not satisfied with the example proposed, because from the very beginning technical inconsistencies are shown, as well as incorrect or imprecise explanation of the work being done. Also, lots of current source symbols without explanation of the actual circuitry of such current sources.
This is a poor example both because of inconsistencies and vagueness, and incomplete schematics. The beginner will learn nothing, except implied greatness of the author whose CCS devices should by all means be acquired, as the man is a genius (?).
As for myself, I could not force myself to read further from the explanation between the second and third schematics... somewhere between disbelief and astonishment. When I call such stuff "mumbo-jumbo", aficionados find themselves insulted. It reminds me of the Hazen mod... of course it sounds different, the kink gets into play...
I will not give further comments on this, so everyone can go on with their business about it.
This is a poor example both because of inconsistencies and vagueness, and incomplete schematics. The beginner will learn nothing, except implied greatness of the author whose CCS devices should by all means be acquired, as the man is a genius (?).
As for myself, I could not force myself to read further from the explanation between the second and third schematics... somewhere between disbelief and astonishment. When I call such stuff "mumbo-jumbo", aficionados find themselves insulted. It reminds me of the Hazen mod... of course it sounds different, the kink gets into play...
I will not give further comments on this, so everyone can go on with their business about it.
- IslandPink
- Amstrad Tower of Power
- Posts: 10041
- Joined: Tue May 29, 2007 7:01 pm
- Location: Denbigh, N.Wales
#6
Quite unbeliveable disrespect to one of the most intelligent and nicest guys in the business - who also happens to be a professional electronics designer in the US . All of the CCS options, circuit diagrams and application notes are available on his site, for anyone intelligent enough to remove the characters "web/47.htm" from the URL .
http://www.pimmlabs.com/DIY.htm
Both Paul and I have met Gary and know that he's a genuine hero .
http://www.pimmlabs.com/DIY.htm
Both Paul and I have met Gary and know that he's a genuine hero .
"Once you find out ... the Circumstances ; then you can go out"
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#7
Why do I start these things?
I think I'll STFU altogether about the whole PP thing.
Probably best thing I can do is not post unless I've something sensible to talk about. I was reasonably sure I was understanding things, was making some good SE amplifiers.
What the fook possessed me to think about PP with local feedback?
It's quite clearly an offensive circuit topology
My nerves can't take it anymore.
Maybe it's time to ask Steve if I can buy back my Sowter 3K5 SE OPTs.
Long live the triode.
I think I'll STFU altogether about the whole PP thing.
Probably best thing I can do is not post unless I've something sensible to talk about. I was reasonably sure I was understanding things, was making some good SE amplifiers.
What the fook possessed me to think about PP with local feedback?
It's quite clearly an offensive circuit topology
My nerves can't take it anymore.
Maybe it's time to ask Steve if I can buy back my Sowter 3K5 SE OPTs.
Long live the triode.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
- slowmotion
- Old Hand
- Posts: 282
- Joined: Fri Aug 21, 2009 4:07 pm
- Location: Norway
#8
< opens a beer and keeps reading >
edit:
This could be a very useful thread, and it would be great if we could keep in mind that there is very many perfect ways to do things ( more than 42, and maybe as many as 144 ).
I'll continue to enjoy reading both Alex' and Gary's writings.
edit:
This could be a very useful thread, and it would be great if we could keep in mind that there is very many perfect ways to do things ( more than 42, and maybe as many as 144 ).
I'll continue to enjoy reading both Alex' and Gary's writings.
- Jan -
#9
I know that was tonge in cheek, but they are in a box somewhereSteveTheShadow wrote:Why do I start these things?
I think I'll STFU altogether about the whole PP thing.
Probably best thing I can do is not post unless I've something sensible to talk about. I was reasonably sure I was understanding things, was making some good SE amplifiers.
What the fook possessed me to think about PP with local feedback?
It's quite clearly an offensive circuit topology
My nerves can't take it anymore.
Maybe it's time to ask Steve if I can buy back my Sowter 3K5 SE OPTs.
Long live the triode.
I would like to try a push pull build as i think i now know what takes away the dynamics as compared to single ended
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
- slowmotion
- Old Hand
- Posts: 282
- Joined: Fri Aug 21, 2009 4:07 pm
- Location: Norway
-
- Amstrad Tower of Power
- Posts: 10552
- Joined: Wed May 30, 2007 12:25 am
- Location: South Yorks.
#11
Bah! you rumbled mesteve s wrote:
I know that was tongue in cheek, but they are in a box somewhere
I would like to try a push pull build as i think i now know what takes away the dynamics as compared to single ended
Sgt. Baker started talkin’ with a Bullhorn in his hand.
#12
Well having heard quite alot that is my impressionslowmotion wrote:Why
would a PP amp be less dynamic that a SE amp?
I know thats a generalisation...
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
- IslandPink
- Amstrad Tower of Power
- Posts: 10041
- Joined: Tue May 29, 2007 7:01 pm
- Location: Denbigh, N.Wales
#13
Are we talking differences in micro-dynamics, or bass dynamics ? Most PP amps seem superior in bass dynamics to SE amps unless the SE amp has a shunt-regulated supply . In microdynamics it can be the other way round . On the other hand the Aurora that I built , and Gary P's 47 PP amp were as dynamic as anything I've heard.
"Once you find out ... the Circumstances ; then you can go out"
#14
My personal experience with my 300B PP against my 2A3 PSE is that the PSE probably reaches deeper in the bass but doesn't have the same speed or control. That might change when my new PSU for the PSE is built. I agree that the PSE reveals better micro dynamics over the PP, but, to put a spanner in the works, running my PP fed by the LDR passive (passive is probably an incorrect term in this instance), I gained very noticeable improvement in the micro dynamic presentation particularly in the mid and treble ranges. I noticed nuances on both male and female voices not heard before. High percussion was far more realistic and revealingly detailed. The PSE is stripped down at present so the LDR Pre has as yet not been tried with it.IslandPink wrote:Are we talking differences in micro-dynamics, or bass dynamics ? Most PP amps seem superior in bass dynamics to SE amps unless the SE amp has a shunt-regulated supply . In microdynamics it can be the other way round . On the other hand the Aurora that I built , and Gary P's 47 PP amp were as dynamic as anything I've heard.
As it is all about the music, I remain steadfast in my enjoyment of both PP and PSE presentation. They are different but I would not say one is better than the other.....just enjoyably different.
Isn't it enough to see that a garden is beautiful without having to believe that there are fairies at the bottom of it too?
Douglas Adams (HHGTTG)
Douglas Adams (HHGTTG)
-
- Old Hand
- Posts: 369
- Joined: Mon Oct 14, 2013 3:20 pm
#15
This is exactly the kind of reply Iam soooooo sick and tired of!IslandPink wrote:Quite unbeliveable disrespect to one of the most intelligent and nicest guys in the business - who also happens to be a professional electronics designer in the US.
Unfortunately, you never met me, so you cannot say anything so nice about my personality, or be impressed by my genius or something like that.
Plus, I am not selling anything, so I don't need the "nice guy" aura.
Last but not least, I dare imagine you really have no idea what was I criticising? After all, why should you? You are just a nice well behaved hobbyist who happens to be born on the better side of the pond.
I would appreciate a direct link to those CCS used in the 47 amp. It seems that I am either plain stupid or internet illiterate and cannot find them, as well as misguided by the fact that people sell such simple stuff as solid state CCS to paying DIYers, while I seem to be publishing complete amplifiers for free.... because I'm stupid, or maybe because I live in Serbia, country with laws so antiquated that years after the demise of socialism citizens are still denied the right to sell abroad?!IslandPink wrote:All of the CCS options, circuit diagrams and application notes are available on his site, for anyone intelligent enough to remove the characters "web/47.htm" from the URL .
http://www.pimmlabs.com/DIY.htm
Great, so now he is the hero, and I am the villain?IslandPink wrote:Both Paul and I have met Gary and know that he's a genuine hero .
Thank you very much!
Let me tell you one more thing: we might be living in darkness if Nikola Tesla did not emigrate to the USA. In his native village he was considered almost retarded or deluded.
If your heroes were in Serbia instead of where they are, not only they would not be your heroes, but you would have never heard of them!
And if you were in Serbia, you would probably not be devoting any time to this hobby.
Your comment is shameful. If you wanted to make a "nice show" of yourself, you could have explained to us all what is it about the amplifier in question and the explanation thereof that I missed... the consistency of design, the openness of solutions presented, the clear and correct explanations... in one word, the makings of a true DIY hero. But you did not. You know why you didn't.