Yes, one could lift the ground point if there were no individual sensing of each phase. Unfortunately, the actual loads (output tubes) don't see that and they need to see the outputs of each phase individually, referenced to ground.
Are you familiar with the Thevenin Theorem? It was stated only for resistors, but it can be generalized for R's L's and C's. Wikipedia actually has a decent statement of it plus a description of the method of how to calculate the equivalent impedance. This is exactly what I am asking you to do.
Not really. Here's my specified constraint for the voltage amplifier: the voltage at the output exceeds that at the input.
I see no relevence to the number of nodes involved.
I have no idea what your second paragraph means. Perhaps you would care to explain it?
I see no relevence to the number of nodes involved.
I have no idea what your second paragraph means. Perhaps you would care to explain it?
No one is recommending that we add impedances in series with the cathode and not the plate of a balanced Cathodyne.
Might be possible. Suppose we accept for the sake of argument that you can't unbalance the Cathodyne in any way to measure its impedances.
OK, no grid voltage excitation. Equal and opposite ground-referenced AC current sources ip and ik drive the plate and cathode. The circuit remains balanced. Derive expresssions for Vplate and Vcathode in terms of ip and ik.
A definition of impedance between nodes A and B is the change in the voltage between them due to a current flowing from A to B, divided by that current.
Since only ip flows between the plate and ground, divide only the ip portion of Vp by ip. The same for the Cathode. The results are Zp >> Zk.
Of course, all this effort is silly. Superposition says we can turn off the ik and the contribution of ip to Vp will not change. So just insert ip into the plate with ik = 0, measure Vp/ip and be done with it! ikewise for the cathode.
OK, no grid voltage excitation. Equal and opposite ground-referenced AC current sources ip and ik drive the plate and cathode. The circuit remains balanced. Derive expresssions for Vplate and Vcathode in terms of ip and ik.
A definition of impedance between nodes A and B is the change in the voltage between them due to a current flowing from A to B, divided by that current.
Since only ip flows between the plate and ground, divide only the ip portion of Vp by ip. The same for the Cathode. The results are Zp >> Zk.
Of course, all this effort is silly. Superposition says we can turn off the ik and the contribution of ip to Vp will not change. So just insert ip into the plate with ik = 0, measure Vp/ip and be done with it! ikewise for the cathode.
It doesn't just boil down to semantics, Alfred. I've been fighting this fight for years with SY. You make cogent, well reasoned points. Keep it up!
Yes , I think so too , the output impedance of the anode affected by the Anode resistor and the Rp of the triode and the cathode resistor which are in series , but the cathode impedance is equal to 1/gm and it will affected by the anode resistor too , from this it is obvious to see that the anode has higher impedance , and no one recommended that how could the anode impedance be equal with the cathode impedance anyway even if they are in the same circuit .
I did, and at least twice.
Method #1) Reduce cathode feedback to 1/Mu, see post #55.
http://www.diyaudio.com/forums/tubes-valves/189153-phase-splitter-issue-6.html#post2800139
This trick will allow both impedances to resolve uniformly high.
And also give the cathodyne +Mu/2 and -Mu/2 voltage gains.
You need a piece of P-sand to pull this off, but its direct coupled,
and totally slave to the triode with no voltage gain of its own.
Method #2) Close a global NFB loop with *any* splitter inside.
Impedances that might begin different, will remain different.
But both can be reduced so much that difference will become
insignificant. Crushed to low numbers by massive feedback, the
absolute difference falls away, even if ratio remains the same...
And: both methods can be used together, no problem...
Last edited:
No, the result is Zp = Zk. The voltages developed at both anode and cathode are identical, as you will see if you do the maths.
Obvious, but wrong.Dimitris AR said:
CPaul, have you discovered the Edit button yet? It might save you making six consecutive posts.
FWIW, I've attached some hand written notes I quickly worked up this morning explaining why voltage measurements at the plate and cathode to determine Zp and Zk cannot be done with balanced output currents. Essentially, unless you know the transfer impedances from p to k and k to p, you cannot separate those contributions to the node voltage from the contributions due to Zp and Zk.
I've got lots on my plate today so I'll check back in this evening or tomorrow morning.
I've got lots on my plate today so I'll check back in this evening or tomorrow morning.
But the experimental results sit there, staring you in the face. I'm still amazed that the theory-spinning gets more and more complex and strident without dealing with the fundamental results of experiment.
Again, please suggest any matched loads (they can be independent, just like the grids and grid leaks of output tubes) that will cause my Thevenin model to give incorrect experimental results. I will do the experiments and post results.
Again, please suggest any matched loads (they can be independent, just like the grids and grid leaks of output tubes) that will cause my Thevenin model to give incorrect experimental results. I will do the experiments and post results.
Now I'm confused, what is the problem? I thought everyone but CPaul (aka Martin A) were satisfied that the effective output impedances for balanced conditions were equal, and that for unbalanced conditions they were not?
I think the issue is still whether there is any significance to independent Zp and Zk impedances under the balanced condition.
Though I must admit I wish I had Linear Volume 0 (I bought volumes 1 and 2 from Jan at this years burning amp, and have very much enjoyed them. I had vol 0 loaned to me and remember seeing SY'a article but don't recall the details).
Thanks
-Antonio
Though I must admit I wish I had Linear Volume 0 (I bought volumes 1 and 2 from Jan at this years burning amp, and have very much enjoyed them. I had vol 0 loaned to me and remember seeing SY'a article but don't recall the details).
Thanks
-Antonio
SY, it's your interpretation of the results that I disagree with. You're claiming to have measured the Zp and Zk but you haven't. You've measured the differential output impedance. It really is that simple; you've made some measurements and interpreted them incorrectly.
The fact is, you could replace the two equal AC loads with a single floating load of twice the value. Then by measuring the voltage across that single, differential load, as you vary it, you can calculate the differential output impedance. You'll come up with the same answer as you claim the plate and node impedances to be thus proving beyond any reasonable doubt that you have, in fact, measured the differential output impedance, the impedance between the plate and cathode.
The fact is, you could replace the two equal AC loads with a single floating load of twice the value. Then by measuring the voltage across that single, differential load, as you vary it, you can calculate the differential output impedance. You'll come up with the same answer as you claim the plate and node impedances to be thus proving beyond any reasonable doubt that you have, in fact, measured the differential output impedance, the impedance between the plate and cathode.
My measurements accord exactly with the Thevenin model in my article. So far, you still haven't suggested an experiment that will falsify that model under the equal-load constraint. When you do, I'm all ears.
SY, I have not seen your article, but from your posts it sounds like you took the equivalent circuit ('black box') of the cathodyne, with its two outputs, and applied a fictitious source between these two outputs to find the output impedance. If that is the case, then you must have found the impedance between anode and cathode, right?
Zak= ra (Ra+Rk) / [ra + Ra + Rk(mu+1)]
The effective output impedance of either output under balanced conditions (Ra = Rk = R) is simply half this figure, or:
Z = ra R / [ra + R(mu+2)]
Last edited: