hey-Hey!!!,
I am quite comfortable to lable the output Z of both nodes at 2/gm. So now I know to call this differential output Z.
Dave, that loading a PP OPT discussion is a useful one. Found some of it elsewhere whilst looking for something else.
It's interesting... - Henry Pasternack - Tube DIY Asylum
What I'd like is better understanding of what happens during cut off in an AB stage, or perhaps a just-barely-Class A stage. For example, I took the load line as a straight line with discontinuity at cut off for a 40% U-L 6L6GC stage and found a very satisfying looking load line right where the data sheet specified...ov course I suspect that it isn't that simple...heh-heh-heh.
cheers,
Douglas
I am quite comfortable to lable the output Z of both nodes at 2/gm. So now I know to call this differential output Z.
Dave, that loading a PP OPT discussion is a useful one. Found some of it elsewhere whilst looking for something else.
It's interesting... - Henry Pasternack - Tube DIY Asylum
What I'd like is better understanding of what happens during cut off in an AB stage, or perhaps a just-barely-Class A stage. For example, I took the load line as a straight line with discontinuity at cut off for a 40% U-L 6L6GC stage and found a very satisfying looking load line right where the data sheet specified...ov course I suspect that it isn't that simple...heh-heh-heh.
cheers,
Douglas
You shouldn't be comfortable with that because it isn't true.
Remarkably, the LTE from Burkhard Vogel to SY explicitly confirms precisely what I and Chris have been trying to convince SY and others here:
I say "remarkably" because this is the LTE alleged to confirm SY's analysis.
I've perused the letter and associated MATHCAD document and it is clear that Mr. Vogel confirms what I've said all along: when taking the output differentially, AKA balanced, use the differential output impedance.
Last edited:
IOW, used as phase splitters are used (two outputs, each referenced to ground, equal loads), my analysis is correct, as Burkhard explicitly says several times. He disagrees on two minor details (not under discussion here) to which I am writing a reply.
Do you continue to maintain that the anode and cathode output impedances are equal, when the circuit is used as you describe, despite Vogel's explicit statement and algebra to the contrary?
Without bothering to go and dig up a quote of yours, do please tell me how the output Z of each node, when taken together in the split-load PI, operating within its limits( as in no power final grid current, or PI stage saturation, etc. ) displays a different output Z at either node.
cheers,
Douglas
Do you continue to dispute Vogel's explicit statement that Ro.bal is not equal to the output resistance from anode to ground or the output resistance from cathode to ground?
Be careful here. If you dispute his statement re the output resistances of the anode and cathode, if you dispute his equations which undeniably show that the two resistances are not equal, then you undermine the analysis that follows.
You can't have it both ways. The equations are either correct or they aren't. If they are correct, the anode and cathode output resistances are unequal and very different from the differential output impedance. If the equations are incorrect, then the conclusions of the analysis using those equations are incorrect.
Be careful here. If you dispute his statement re the output resistances of the anode and cathode, if you dispute his equations which undeniably show that the two resistances are not equal, then you undermine the analysis that follows.
You can't have it both ways. The equations are either correct or they aren't. If they are correct, the anode and cathode output resistances are unequal and very different from the differential output impedance. If the equations are incorrect, then the conclusions of the analysis using those equations are incorrect.
No, his statement is correct for the general case. And under the conditions used for phase splitters (equal loads to ground), his rather complex derivation reduces to my simple one. You're trying to make up a disagreement where he in fact explicitly says there is none.
Then Vogel, Alfred, bcorso and I and others on this thread are correct. The plate to ground impedance far exceeds the cathode to ground impedance. This is true regardless of whether the circuit is balanced or not - the "general case". Note that this does nothing to invalidate the model of the 2/gm differential driver driving a balanced load.
But even if it were true only in the unbalanced case (which it is not), unbalance would occur when output tube grid current is drawn. Unless you are stating categorically that it is absolutely impossible under any circumstances for a Cathodyne to elicit output stage grid current?
But even if it were true only in the unbalanced case (which it is not), unbalance would occur when output tube grid current is drawn. Unless you are stating categorically that it is absolutely impossible under any circumstances for a Cathodyne to elicit output stage grid current?
Imbalance of load is fatal for Concertina, so in the real life it has to be maximally minimized.
How could that be? If SY's model is correct, imbalance of load should be no big deal.
Imbalance of load should only be fatal if the genuine anode and cathode output impedances are vastly different even when the loads are perfectly balanced.
Oh wait...
🙂
Plate error currents uncorrelated to the cathode see a high impedance.
Cathode error currents uncorrrelated to the plate see a low impedance.
Either location you might inject this error current, most of the voltage
of error appears always at the plate.
When error currents are correlated in phase. We see almost exactly the
same picture, but now these impedance values appear to be doubled!
The error currents see equal impedance only when they are correlated
equal and opposite. May be a likely situation near the quiescent state,
but far from garaunteed to remain that way with large signals as they
approach a forward bias or other non-linear challenge.
---------
Anyways, it makes no sense to me to speak of correlated impedance
in those circuits with errors that clearly do not correlate. Its the same
mistake as testing the output impedance of an amp against a signal
that is identical, and drawing a conclusion that the output impedance
must be infinte... You would never do such a silly thing, but with the
cathodyne we do???
What happens at the plate is mostly because of the cathode, and it
is ridiculous to conclude it has an equal impedance merely on account
of a fair weather correlation that may not always exist.
-----------
Note the current sources are infinite impedance, and cannot change
the output impedance, nor the load. The illusion that they do is an
effect of correlation only...
Cathode error currents uncorrrelated to the plate see a low impedance.
Either location you might inject this error current, most of the voltage
of error appears always at the plate.
When error currents are correlated in phase. We see almost exactly the
same picture, but now these impedance values appear to be doubled!
The error currents see equal impedance only when they are correlated
equal and opposite. May be a likely situation near the quiescent state,
but far from garaunteed to remain that way with large signals as they
approach a forward bias or other non-linear challenge.
---------
Anyways, it makes no sense to me to speak of correlated impedance
in those circuits with errors that clearly do not correlate. Its the same
mistake as testing the output impedance of an amp against a signal
that is identical, and drawing a conclusion that the output impedance
must be infinte... You would never do such a silly thing, but with the
cathodyne we do???
What happens at the plate is mostly because of the cathode, and it
is ridiculous to conclude it has an equal impedance merely on account
of a fair weather correlation that may not always exist.
-----------
Note the current sources are infinite impedance, and cannot change
the output impedance, nor the load. The illusion that they do is an
effect of correlation only...
Attachments
Last edited:
Alfred;
For his birthday sake please stop teasing the guy! He never said that it can't be imbalanced. That means, if imbalances the theory does not apply. The theory is about balanced Concertina.
Have you ever started building models from ball-shaped horse in vacuum, when asked how to harness the beast? It is common for all good scientists!
For his birthday sake please stop teasing the guy! He never said that it can't be imbalanced. That means, if imbalances the theory does not apply. The theory is about balanced Concertina.
Have you ever started building models from ball-shaped horse in vacuum, when asked how to harness the beast? It is common for all good scientists!
Well, Happy Birthday to SY!
Wavebourn, he may not have ever said that it can't be imbalanced, but he strives mightily to avoid considering the situation. A good example is in post 350, p. 35.
I think Merlinb sums up the situation well, especially in the last paragraph of post 326, p 33.
Wavebourn, he may not have ever said that it can't be imbalanced, but he strives mightily to avoid considering the situation. A good example is in post 350, p. 35.
I think Merlinb sums up the situation well, especially in the last paragraph of post 326, p 33.
Unfortunately, my model doesn't say that, nor have I- I did not do a sensitivity analysis on the effect of deviating from the explicit constraint (equal loads, each connected to ground) that you continue to sidestep. I really would prefer it if you dealt with what has been written or stated rather than attacking straw men.
No matter what the two loads are, if they're equal, the model gives correct results. I still await any suggested loads for which this will not be true. (SFX: crickets chirping...)
hey-Hey!!!,
The sensitivity issue is ov course a useful one. I can't help thinking that triodes with both lower gain and plate resistance will mind load imbalance less( not to mention the ability to have them idle at higher current ).
In the mean time, when building Class A amps, I am always pressed for more voltage output from the PI, and so I stick with LTP as both input stage and PI. Split load is just too wasteful of B+ wrt its output votlage capability.
cheers,
Douglas
The sensitivity issue is ov course a useful one. I can't help thinking that triodes with both lower gain and plate resistance will mind load imbalance less( not to mention the ability to have them idle at higher current ).
In the mean time, when building Class A amps, I am always pressed for more voltage output from the PI, and so I stick with LTP as both input stage and PI. Split load is just too wasteful of B+ wrt its output votlage capability.
cheers,
Douglas
Congrats SY..my birthday is also in Dec on the worst day that one could have it.
ALot has drifted past on this thread and I've been pretty busy on finalising the power supply metalwork and performance for the 2x200W amp....
Bandersnatch is there with triode thinking..but I haven't seen a tread suggesting the ideal tubes for a concertina....choose low mu tubes for concertina work, ECC82,6SN7,ECF80 triode sections etc. I often parallel up both sections of an ECC88. In general, allocate 33% of the voltage across the tube and two resistors.
richy
ALot has drifted past on this thread and I've been pretty busy on finalising the power supply metalwork and performance for the 2x200W amp....
Bandersnatch is there with triode thinking..but I haven't seen a tread suggesting the ideal tubes for a concertina....choose low mu tubes for concertina work, ECC82,6SN7,ECF80 triode sections etc. I often parallel up both sections of an ECC88. In general, allocate 33% of the voltage across the tube and two resistors.
richy
I think kenpeter hits the main point which is whether the signal is correlated or uncorrelated.
a 1V signal on the grid will give about .95V of correlated signal on the output. Isn't it this number we need to work against to get the correlated to uncorrelated ratio?
can anybody put some real noise numbers into the mix for the currents of uncorrelated signals we are talking about?
as for resistor noise, given that it is a series circuit, wouldn't that force the noises to show up correlated?
dave
a 1V signal on the grid will give about .95V of correlated signal on the output. Isn't it this number we need to work against to get the correlated to uncorrelated ratio?
can anybody put some real noise numbers into the mix for the currents of uncorrelated signals we are talking about?
as for resistor noise, given that it is a series circuit, wouldn't that force the noises to show up correlated?
dave
Absolutely right , because if the concertina drives output tubes , that tubes will never be identical loads for it .