"Or Kirchoff's Law and the First Law of Thermodynamics are wrong"
Burkhard Heim has predicted the existance of the e^0, a chargless electron. Gregg's circuit must have a bad triode on one side that is converting some e- to e^0. Nobel prize?
"same resusts if you replace one triode with a Pentode...."
Uhh, except for screen current causing loss of current on one side, unless using a floating screen supply.
Burkhard Heim has predicted the existance of the e^0, a chargless electron. Gregg's circuit must have a bad triode on one side that is converting some e- to e^0. Nobel prize?
"same resusts if you replace one triode with a Pentode...."
Uhh, except for screen current causing loss of current on one side, unless using a floating screen supply.
So, rather than dwelling on such sillyness, how about playing with said circuit and see what you can come up with?
The Law of Science is an experiment must be able to be duplicated. Fred made the first one I came across:
RA-100: LTP/CCS PA Driver Improvement
I adapted it:
A 6V6 PP Project
It's been expanded upon in balanced SS amps by Tim Williams:
Re: Simple bridged amplifier design question
No need for physics bending
Cheers!
Last edited:
The cascoding device (tubes, FETs, BJTs doesn't matter) can not be modulated, the feedback must be connected to the active part in the CCS which decides the bias current.
Cheers Michael
That one won't work, the gate is shorted to the negative supply through Rgs so one can't modulate the CCS current with the CMFB signal.
Add a LED or resistor from the point where Rg and R||C (from CMFB) meets to the negative voltage, now we will have a higher voltage on the lower FETs gate so one must increase Rset to preserve the current setting of the CCS otherwise the current will increase quite much. A rough guide to start with.
EDIT: Seems SY and I wrote/replied simultaneously, well 2 versions..
Cheers Michael
Add a LED or resistor from the point where Rg and R||C (from CMFB) meets to the negative voltage, now we will have a higher voltage on the lower FETs gate so one must increase Rset to preserve the current setting of the CCS otherwise the current will increase quite much. A rough guide to start with.
EDIT: Seems SY and I wrote/replied simultaneously, well 2 versions..
Cheers Michael
Last edited:
Hello everyone,
Back on post #45, I suggested that the hypothesis of the imbalance being differences in the load resistors be tested by swapping them and looking for full reversal. This type of test is widely recognized in the world of structured problem solving as a method of validating a hypothesis as to what is happening. If it was tried, I did not see any data posted about the outcome. If there is no data to share, I will likely breadboard up Greg's circuit late this week or possibly this weekend to play with it myself as this type of stuff is just to much fun to sit here and hear others talking about it while I have nothing to go instrument up myself and play with.
Doug S. (aka Mickeystan)
Back on post #45, I suggested that the hypothesis of the imbalance being differences in the load resistors be tested by swapping them and looking for full reversal. This type of test is widely recognized in the world of structured problem solving as a method of validating a hypothesis as to what is happening. If it was tried, I did not see any data posted about the outcome. If there is no data to share, I will likely breadboard up Greg's circuit late this week or possibly this weekend to play with it myself as this type of stuff is just to much fun to sit here and hear others talking about it while I have nothing to go instrument up myself and play with.
Doug S. (aka Mickeystan)
Gentlemen,
I was doing more research on the bench tonight and discovered something interesting.
Up until now, I have only experimented with the CMFB method on high-bias LTP's... 6SN7 and 12BH7 running between 5 and 10mA, anode resistors 27K down to about 12K.
Tonight I was playing with a little 6J6 and found she likes ~3mA of bias with 47K on the anodes.
I began at 22K and 5mA and worked down the current scale until I was at 47K each anode. As the current went down, the effect of CMFB went down too until it was practically indescernable with the 47K load. DC balance was horrible, but AC balance was maintained.
I would surmise that no, CMFB is not required with low current LTP's. Higher current tubes or ones that need to drive harder loads, definately benefit.
Speaking of drive, load makes a big difference. If Rl ~= the anode load R (small grid resistor on the output tube vs. PI anode load), CMFB can again help. However if the load is light (output tube grid resistor is the RCA recommended 5 to 10 times the PI anode), then CMFB is of little use.
Cheers!
I was doing more research on the bench tonight and discovered something interesting.
Up until now, I have only experimented with the CMFB method on high-bias LTP's... 6SN7 and 12BH7 running between 5 and 10mA, anode resistors 27K down to about 12K.
Tonight I was playing with a little 6J6 and found she likes ~3mA of bias with 47K on the anodes.
I began at 22K and 5mA and worked down the current scale until I was at 47K each anode. As the current went down, the effect of CMFB went down too until it was practically indescernable with the 47K load. DC balance was horrible, but AC balance was maintained.
I would surmise that no, CMFB is not required with low current LTP's. Higher current tubes or ones that need to drive harder loads, definately benefit.
Speaking of drive, load makes a big difference. If Rl ~= the anode load R (small grid resistor on the output tube vs. PI anode load), CMFB can again help. However if the load is light (output tube grid resistor is the RCA recommended 5 to 10 times the PI anode), then CMFB is of little use.
Cheers!
Doug, do it! One thing to be careful about is maintaining balance with the measurement- if you stick a scope probe on one output, you've unbalanced things. Use identical probes on each output. ALL bits related to the load, including the measurement apparatus, must be the same for balance. Two probes, dual channel scope.
Hello Sy,
I have a Tek 465 B scope with identical 10x probes to view the outputs with. I am an engineering manager with Xerox and previously worked at Tektronics for 26 years. My team at Xerox is very deeply involved with structured problem solving (statistically driven methods) for finding answers to difficult problems. I love to dig into things where measured results seem to not follow well established theory. Your point about equal balance when making measurements is spot on when doing this type of work. It will be fun to look at the behavior of this splitter at light loads as well as when it it pushed to higher plate currents with big swings in both cases.
Doug S. (aka Mickeystan)
I have a Tek 465 B scope with identical 10x probes to view the outputs with. I am an engineering manager with Xerox and previously worked at Tektronics for 26 years. My team at Xerox is very deeply involved with structured problem solving (statistically driven methods) for finding answers to difficult problems. I love to dig into things where measured results seem to not follow well established theory. Your point about equal balance when making measurements is spot on when doing this type of work. It will be fun to look at the behavior of this splitter at light loads as well as when it it pushed to higher plate currents with big swings in both cases.
Doug S. (aka Mickeystan)
Please read through until the end before making conclusions.
Others of us have also done tests .....
And I know mine, where I found they are equal!
As said, this is contrary to Kirchoff's Laws. End of matter. But also here I have done tests, set up experiments for students ... We used Tektronix equipment, blah blah blah ... need I carry on?
I found that the concertina, correctly designed according to basics, is one of the best/simplest phase inverters for audio. As long as those loads are equal impedances, they affect balance quite equally up to comfortably higher than audio frequencies. And as long as the tube is capable of not being overloaded by the two output loads in series, which is not always the case, particularly with the 12AX7.
You say 'big C-miller will do a number on it because it acts as a cathode bypass cap ...' - then why is there Miller-C on the cathode side only? - it will do an equal action on the anode side if it is there also, as it is likely to be in p.p.? What am I missing here?
Define 'perfect'. Better than the combined accuracy of your measuring devices?? Otherwise, my tests appear to indicate - er - otherwise.
OK.
I have particularly composed this post in a directly similar grammatical style
than yours, Geek. That NOT to be offensive, but simply an effort to indicate that one gets nowhere with the "My-father-is-better-than-your-father" style!
All my above statements are actually true - as I accept yours are; the outcomes at least. I hope you see the point. Both cases cannot co-exist, so there must have been something the matter somewhere. With respect, it is not good enough to say 'I believe my tests' - end of matter - in the process sweeping basic physics laws off the table simply because they don't fit!
Kindly accept my genuine respect for what you are and know, but I think there is a point to be made.
As said - been there, done that. And also at fairly low levels. Yes, balance is moot there (how moot?), but again quite acceptable for audio purposes.
Quite; that is not my difficulty. But then a question arises. Why would one need further improvement in ac balance (how far down the road should one continue), if the basic system is more than adequate? But you seem to disagree that it is, so perhaps we are back where we started! (My turn )
Only, let us stop re-inventing the wheel, and seek wisdom along productive avenues.
Regards
Others of us have also done tests .....
And I know mine, where I found they are equal!
As said, this is contrary to Kirchoff's Laws. End of matter. But also here I have done tests, set up experiments for students ... We used Tektronix equipment, blah blah blah ... need I carry on?
I found that the concertina, correctly designed according to basics, is one of the best/simplest phase inverters for audio. As long as those loads are equal impedances, they affect balance quite equally up to comfortably higher than audio frequencies. And as long as the tube is capable of not being overloaded by the two output loads in series, which is not always the case, particularly with the 12AX7.
You say 'big C-miller will do a number on it because it acts as a cathode bypass cap ...' - then why is there Miller-C on the cathode side only? - it will do an equal action on the anode side if it is there also, as it is likely to be in p.p.? What am I missing here?
Define 'perfect'. Better than the combined accuracy of your measuring devices?? Otherwise, my tests appear to indicate - er - otherwise.
OK.
I have particularly composed this post in a directly similar grammatical style
than yours, Geek. That NOT to be offensive, but simply an effort to indicate that one gets nowhere with the "My-father-is-better-than-your-father" style!
All my above statements are actually true - as I accept yours are; the outcomes at least. I hope you see the point. Both cases cannot co-exist, so there must have been something the matter somewhere. With respect, it is not good enough to say 'I believe my tests' - end of matter - in the process sweeping basic physics laws off the table simply because they don't fit!
Kindly accept my genuine respect for what you are and know, but I think there is a point to be made.
As said - been there, done that. And also at fairly low levels. Yes, balance is moot there (how moot?), but again quite acceptable for audio purposes.
What is so hard to believe about a proven method of improving AC balance?
Quite; that is not my difficulty. But then a question arises. Why would one need further improvement in ac balance (how far down the road should one continue), if the basic system is more than adequate? But you seem to disagree that it is, so perhaps we are back where we started! (My turn
) Only, let us stop re-inventing the wheel, and seek wisdom along productive avenues.
Regards
You read something that is not there.
I have measurements. I make them public. People try and prove or disprove it. That's how development works. Everything else is semantics or personality.
As for the concertina, I did some work on a MK-III driver + 6AN8 the other day and she's the same until the load gets below 220K.
Cheers!
Last edited:
'k, here's some pics as per the MK-III concertina... 100K load.
10V/div:
20V/div:
Scope balanced before each test. Drive to the triode section was a sine going in. Just visible distortion for the second. Output isn't exactly a sine...
Best of three tubes tested. 480V B+ and supply resistors/caps as per Dynaco spec.
If there's something I did haywire, I'd like to know
Cheers!
10V/div:
An externally hosted image should be here but it was not working when we last tested it.
20V/div:
An externally hosted image should be here but it was not working when we last tested it.
Scope balanced before each test. Drive to the triode section was a sine going in. Just visible distortion for the second. Output isn't exactly a sine...
Best of three tubes tested. 480V B+ and supply resistors/caps as per Dynaco spec.
If there's something I did haywire, I'd like to know
Cheers!
Last edited:
Three things...
One. You can't cheat Kirchoff. Kirchoff owns all. Period! If the resistors are equal and the loads are equal... ya, the signals damn well are going to be equal. This is completely regardless of the active device, you can pair a 12AX7 with a BJT for all I care.
Two. The cathodyne and LTP circuits are Exactly Equivalent in terms of their output circuits. In both cases, Mr. K forces equal currents = equal and opposite voltages (in the cathodyne case, or opposite currents in the LTP case).
Three. How much is actually needed? When in doubt, consult the Ten Percent Rule. If the signals are within 10%, and the tubes are matched within 10%, and the resistors and so on, you aren't going to give a damn if the signal voltage at one particular point in the circuit is within 1% of equality. At any stage before or after that point, it's probably going to be worse matched due to circuit variation anyway. Do you even WANT a clean matched signal? No, you don't! You want some nasty ugly unbalanced distorted monstrosity -- cleanliness is for transistors!
Tim
One. You can't cheat Kirchoff. Kirchoff owns all. Period! If the resistors are equal and the loads are equal... ya, the signals damn well are going to be equal. This is completely regardless of the active device, you can pair a 12AX7 with a BJT for all I care.
Two. The cathodyne and LTP circuits are Exactly Equivalent in terms of their output circuits. In both cases, Mr. K forces equal currents = equal and opposite voltages (in the cathodyne case, or opposite currents in the LTP case).
Three. How much is actually needed? When in doubt, consult the Ten Percent Rule. If the signals are within 10%, and the tubes are matched within 10%, and the resistors and so on, you aren't going to give a damn if the signal voltage at one particular point in the circuit is within 1% of equality. At any stage before or after that point, it's probably going to be worse matched due to circuit variation anyway. Do you even WANT a clean matched signal? No, you don't! You want some nasty ugly unbalanced distorted monstrosity -- cleanliness is for transistors!
Tim
Well, I'm going to print this whole post 76, put it in a cornice and nail it to the the wall in front my soldering desk. It will serve me as spiritual guidance every
time I feel I'm falling in the irrational- lust- of- perfection !
All I can really add to this is: It's too easy to second-guess yourself in the world of electronics design and especially troubleshooting. Keep trying different perspectives, keep a fresh mind (sleep well and go at it the next day if you need to), and best of all, be completely open to the fact that anything you hold as gospel might fall apart before you reach a useful conclusion!
I try and ask myself: What, in this system, affects the results I'm getting? (This includes my observations because the observer is inextricably bound to the system under observation and most failure-prone of all.) Especially check inputs to circuit subsections when the outputs aren't as expected. Many a time I've tried to fix a working circuit because I believed falsely that its input stimulus was as expected!
I try and ask myself: What, in this system, affects the results I'm getting? (This includes my observations because the observer is inextricably bound to the system under observation and most failure-prone of all.) Especially check inputs to circuit subsections when the outputs aren't as expected. Many a time I've tried to fix a working circuit because I believed falsely that its input stimulus was as expected!
Hi Geek,
I am at a disadvantage. I received, in the usual manner, a notification from diyAudio that you have posted a reply to my previous post, including the contents thereof. But for some reason that was taken off the thread (it was not offensive to me). I can therefore in decency not reply to it as you requested.
I hope to be excused, however, for quoting the non-personal part of it, as it is technically relevant:
"Enough theory people, just do it. SY has, I have. I am awaiting more input."
As I have, as indicated in my previous post. As for my results, view SY's post #3. Again I am at a disadvantage to show them (not that that seem to convince you) - the tests (tutorial) have been done some >30 years ago; the results are probably languishing somewhere in my previous employer's files, if not turned to dust by now. We have progressed past tube basics.
So what are you asking us to do? Believe you, or accept Kirchoff - since, by your evidence, the two are in essence mutually exclusive (barring trivialities)?
The (accepted) principle has already proved itself. It is not development to re-prove it, it is only about finding where disagreeing results have gone astray. As you asked before: Why is that so difficult to get?
I rest my case .....and apologise to members for perhaps having bored them by trying a different way, (even if unsuccessful) to get the idea across.
End of my contribution along this line.
Regards
_________________________________________-
PS: I am not refraining from also showing test results, I only deem it unnecessary to re-set up the experiment since others are doing as well and better than I am now able to.
I am at a disadvantage. I received, in the usual manner, a notification from diyAudio that you have posted a reply to my previous post, including the contents thereof. But for some reason that was taken off the thread (it was not offensive to me). I can therefore in decency not reply to it as you requested.
I hope to be excused, however, for quoting the non-personal part of it, as it is technically relevant:
"Enough theory people, just do it. SY has, I have. I am awaiting more input."
As I have, as indicated in my previous post. As for my results, view SY's post #3. Again I am at a disadvantage to show them (not that that seem to convince you) - the tests (tutorial) have been done some >30 years ago; the results are probably languishing somewhere in my previous employer's files, if not turned to dust by now. We have progressed past tube basics.
So what are you asking us to do? Believe you, or accept Kirchoff - since, by your evidence, the two are in essence mutually exclusive (barring trivialities)?
The (accepted) principle has already proved itself. It is not development to re-prove it, it is only about finding where disagreeing results have gone astray. As you asked before: Why is that so difficult to get?
I rest my case .....and apologise to members for perhaps having bored them by trying a different way, (even if unsuccessful) to get the idea across.
End of my contribution along this line.
Regards
_________________________________________-
PS: I am not refraining from also showing test results, I only deem it unnecessary to re-set up the experiment since others are doing as well and better than I am now able to.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.