I linked you to the relevant page at Duncan Amps for models.
The 2nd 6l6 model was and is dm6l6.inc. You couldn't find it
the first time, and after I explained where to find, you went
back and did the exact same thing and it still doesn't work.
I never told you to do "exact same thing". You just do what
you do, and its my fault you can't get a result. Coincidence
your arguments against Sy ring much the same? Sy is still
dead wrong (even though he's absolutely right because I
just said so, kinda like what happens at the plate), but you
will never be the one to prove it by whining that we don't
give you stuff.
There is a problem with assumption of correlated independent
causes always giving the same effects. This assumption, and
not absurd models or calculations that may follow, is at fault.
Yet this assumption agrees with real amps pretty darn well,
even in worst realistically conceived case that should poke
big holes, not faulty enough to make a thousand+ posts over.
Kenpeter, bottom line, I couldn't get the files to work on my PC. I never blamed you for the problem.
If you want to characterize a request for help as "whining" and respond with libelous comments about SY and me in your PM, anyone who might want to ask you for help in the futture should take note.
It's fine that you don't want to help. The vitriol is unnecessary.
If you want to characterize a request for help as "whining" and respond with libelous comments about SY and me in your PM, anyone who might want to ask you for help in the futture should take note.
It's fine that you don't want to help. The vitriol is unnecessary.
In case of a triode with Mu=100, and the assumption that load behaviors correlate.
What happens at the cathode end is 99% impedance, and 1% arbitrary assumption.
What happens at the plate end is 1% impedance and 99% arbitrary assumption.
I can make you a hybrid where both ends are 50% impedance 50% assumption.
But strangely, it doesn't work any different. My conclusion is that the assumption
was a good one, despite my feelings that it has no merit and ought to be invalid.
What happens at the cathode end is 99% impedance, and 1% arbitrary assumption.
What happens at the plate end is 1% impedance and 99% arbitrary assumption.
I can make you a hybrid where both ends are 50% impedance 50% assumption.
But strangely, it doesn't work any different. My conclusion is that the assumption
was a good one, despite my feelings that it has no merit and ought to be invalid.
I believe the phase in question was, "You and Sy need to stop smoking crack."
SY, you seem to be having some trouble responding to p. 1016, and 1017 in light of 1016. Perhaps if I summarized?
First, please review p. 1012, which establishes that model 1 in p. 1016 is the same as the model in Figure 3 of your article. (If you disagree, be explicit as to why. Simply saying that models are wrong is meaningless. Wrong? How? Wrong color? Why, are black and blue no good?)
You can create model 2 from model 1 by adding a resistor of arbitrary value between ground and the sources. Now get rid of that ground and resistor to create model 3. Subject all models to the loads and signals in the square wave test you published in your article. The simulation results (p. 1016) are identical and conform to your test results. So the models aren't “wrong” and “incorrect.” We also see that the ground is non-functional. It plays no part in explaining how a balanced Cdyne works. No great surprise here – there is no ground lurking within a triode.
It makes no sense to defer to models with non-functional components (models 1 and 2) if we have one without them (model 3.) In the absence of a ground, model 3 can even be simplified by combining the sources and resistors to get Burkhart’s model. That model is favored not only because it alone contains only functional components, but because with half the components and nodes of model 1, it would be the choice of Occam (parsimony) and Einstein (as simple as possible, but no simpler.)
But there’s a problem – model 3 states that Zpg and Zkg are a little more than Rk/2 = Rp/2. Actually, all three models contradict one another on that point. But on second thought, it’s really not a problem. There is no experimental evidence available from a balanced Cdyne to favor any of these values. Crafted to reflect only the empirical measurements of a balanced Cdyne, there is no reason to believe that any of the models offers any reliable information about parameters which can’t be measured – Zpg and Zkg.
Neither these nor your model can establish the values of Zkg and Zpg.
First, please review p. 1012, which establishes that model 1 in p. 1016 is the same as the model in Figure 3 of your article. (If you disagree, be explicit as to why. Simply saying that models are wrong is meaningless. Wrong? How? Wrong color? Why, are black and blue no good?)
You can create model 2 from model 1 by adding a resistor of arbitrary value between ground and the sources. Now get rid of that ground and resistor to create model 3. Subject all models to the loads and signals in the square wave test you published in your article. The simulation results (p. 1016) are identical and conform to your test results. So the models aren't “wrong” and “incorrect.” We also see that the ground is non-functional. It plays no part in explaining how a balanced Cdyne works. No great surprise here – there is no ground lurking within a triode.
It makes no sense to defer to models with non-functional components (models 1 and 2) if we have one without them (model 3.) In the absence of a ground, model 3 can even be simplified by combining the sources and resistors to get Burkhart’s model. That model is favored not only because it alone contains only functional components, but because with half the components and nodes of model 1, it would be the choice of Occam (parsimony) and Einstein (as simple as possible, but no simpler.)
But there’s a problem – model 3 states that Zpg and Zkg are a little more than Rk/2 = Rp/2. Actually, all three models contradict one another on that point. But on second thought, it’s really not a problem. There is no experimental evidence available from a balanced Cdyne to favor any of these values. Crafted to reflect only the empirical measurements of a balanced Cdyne, there is no reason to believe that any of the models offers any reliable information about parameters which can’t be measured – Zpg and Zkg.
Neither these nor your model can establish the values of Zkg and Zpg.
If you look at that extremely complex non-Thevenin model, then look at my very simple (and experimentally accurate) model and think they're the same, you need glasses. Try counting components- mine, two voltage sources, two resistances. Yours, well, you can count.
The other, simpler models you presented, which I already cited to you, are not the same as mine because they give totally wrong answers for the output voltages.
I think I'll tip my hat at this point and say, "Adieu" until you experimentally show that my model, under the given constraint, does not agree with experiment. Fume away.
Driving a push-pull pair of 6L6 into A2 open loop with equal and unequal
models was sufficient challenge to satisfy that both models produce very
nearly the same end result. Any model that says otherwise probably isn't
relevant to the need for this sort of phase splitter anyway.
Cap pumping was the killer, not the subtle effect of huge differences in
AC impedance. Chasing our tails around ridiculous proofs for nothing.
models was sufficient challenge to satisfy that both models produce very
nearly the same end result. Any model that says otherwise probably isn't
relevant to the need for this sort of phase splitter anyway.
Cap pumping was the killer, not the subtle effect of huge differences in
AC impedance. Chasing our tails around ridiculous proofs for nothing.
Last edited:
I find SY's reply inscrutable.
What complex model? model 1 is SY's model, model 2 adds one resistor to it, and model 3 subtracts a ground from it!
Another SY special: unfounded assertions again. What's non-Thevenin about it?
Unfounded assertions again! This time, in the face of simulations which clearly disprove the claim. Where is the support for "totally wrong answers"? What are the salient differences between the sims I presented and the test results in SY's article?
If after all this time you think I've been challenging your experimental results, you need a semester or two of reading comprehension drills. I challenge the conclusions you draw from a model that are impossible to test given the constraints you have imposed, never your test results. You don't get to exit as cleanly as you hoped.
We have reached the point where we now are living in different realities. Not only do we get unsupported claims, but we are getting ones which seem to defy what is right before our eyes. If no one who has been following this exchange can explain to SY and me each others' fundamental perceptions, then further communication on this topic is indeed useless.
I'd welcome anyone who has been following this exchange to step forward at this point with constructive explanations to one, the other, or both of us.
What complex model? model 1 is SY's model, model 2 adds one resistor to it, and model 3 subtracts a ground from it!
Another SY special: unfounded assertions again. What's non-Thevenin about it?
Unfounded assertions again! This time, in the face of simulations which clearly disprove the claim. Where is the support for "totally wrong answers"? What are the salient differences between the sims I presented and the test results in SY's article?
If after all this time you think I've been challenging your experimental results, you need a semester or two of reading comprehension drills. I challenge the conclusions you draw from a model that are impossible to test given the constraints you have imposed, never your test results. You don't get to exit as cleanly as you hoped.
We have reached the point where we now are living in different realities. Not only do we get unsupported claims, but we are getting ones which seem to defy what is right before our eyes. If no one who has been following this exchange can explain to SY and me each others' fundamental perceptions, then further communication on this topic is indeed useless.
I'd welcome anyone who has been following this exchange to step forward at this point with constructive explanations to one, the other, or both of us.
Just because his definition of impedance includes anywhere from 1 to 99%
cross coupled arbitrary noise. By pure coincidence, usually but not always,
happens to quack so muck like impedance, its own mother would not know
where real impedance ends and noise begins...
You seem to forget: Sy's absurd claims are supported, "Because I said so."
Also because the A2 load experiment proved so. No other reality matters...
Time for me to sum up and put a coda on my posts to this thread. (And there was much rejoicing!)
I built the following familiar circuit and drove its grid with a square wave having a 30mV step:
I also measured the parameters gm ( 8.83mS ) and µ ( 27.6 ) of the triode under the circuit’s operating conditions. I used them in the following analytic expression and models, all of which were developed to predict the plate and cathode voltages of the physical circuit:
The outputs of the circuit, expression, and models appear in the following graph:
Clearly, the all-around agreement is excellent, validating the expression’s and the models’ voltage predictions. Not so much so with the models’ predictions of the low frequency P and K to ground impedances Zpg and Zkg though – in fact, they contradict one another, with values of approximately 1/gm for Model 1 and R/2 for Model 2. Obviously, at least one of the models gets it wrong. And since that one got the voltages right, we have proof that getting the voltages right does not ensure doing so with the impedances. Therefore, there is no reason to believe that either one of them gets it right. So what to do?
Check out the next post!
I built the following familiar circuit and drove its grid with a square wave having a 30mV step:
I also measured the parameters gm ( 8.83mS ) and µ ( 27.6 ) of the triode under the circuit’s operating conditions. I used them in the following analytic expression and models, all of which were developed to predict the plate and cathode voltages of the physical circuit:
The outputs of the circuit, expression, and models appear in the following graph:
Clearly, the all-around agreement is excellent, validating the expression’s and the models’ voltage predictions. Not so much so with the models’ predictions of the low frequency P and K to ground impedances Zpg and Zkg though – in fact, they contradict one another, with values of approximately 1/gm for Model 1 and R/2 for Model 2. Obviously, at least one of the models gets it wrong. And since that one got the voltages right, we have proof that getting the voltages right does not ensure doing so with the impedances. Therefore, there is no reason to believe that either one of them gets it right. So what to do?
Check out the next post!
Because there has been some controversy regarding the Thevenin theorem in this thread, I’ve turned to the folks who arguably are the ultimate authorities on this matter: The Institute of Electrical and Electronics Engineers. They publish The new IEEE standard dictionary of electrical and electronics terms. In it can be found the following:
Certainly the Cathodyne can be considered to be a linear network, and among its several “any two terminals” are the plate and ground. When we apply Kirchoff’s laws and Thevenin, it’s trivial to derive
Thevenin’s Theorem. States that the current that will flow through an impedance Z’, when connected to any two terminals of a linear network between which there previously existed a voltage E and an impedance Z, is equal to the voltage E divided by the sum of Z and Z’.
Here we have a simple method by which we can determine the impedance between “any two terminals of a linear network between which there previously existed a voltage E and an impedance Z”.
Certainly the Cathodyne can be considered to be a linear network, and among its several “any two terminals” are the plate and ground. When we apply Kirchoff’s laws and Thevenin, it’s trivial to derive
Zpg = [ Rk· (1+µ) + µ/gm ] || Rp
If instead we choose the cathode and ground as the terminal pair, we get
Zkg = Rk || [ (µ/gm +Rp)/(1+µ) ]
Finally, for the plate and cathode, the result is
Zpk = 2/gm / [1 + 2/µ + 1/(gm·R’)] ≈ 2/gm, where R’ = Rp = Rk
There are several take-aways from this:
1. Mr. T. deals with two nodes, not three (P and ground for instance, not P, K and ground);
2. There is no need to keep a balanced circuit balanced while measuring its impedances, and
3. Zpg and Zkg are quite different indeed. In particular, Zkg is just north of 1/gm, while Zpg ≈ Rp >> 1/gm.
And that’s all that needs to be said on the matter.2. There is no need to keep a balanced circuit balanced while measuring its impedances, and
3. Zpg and Zkg are quite different indeed. In particular, Zkg is just north of 1/gm, while Zpg ≈ Rp >> 1/gm.
I was thinking of simple ways to interface my single-ended CD player with my totally differential main system; is there any reason for not using a concertina phase-splitter with(necessarily) low values, 1k5 say, of R anode and R cathode assuming that the high tension is decoupled with a suitable RC network?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.