Triode plate resistance == internal negative feedback?

[jcx]

On another thread I posted an aside as a addendum to a otherwise on topic technical contribution:

Originally Posted by jcx

...it is hard to not comment on ... several swipes at negative feedback:

use enough closed loop gain in a negative feedback amp with only smooth, low order nonlinearities and you get decreasing levels of harmonic distortion with harmonic order - not the 10-20 dB some "no-feedback" types may tepidly apply - it is easy to have >60 dB loop gain at all audio frequencies

there is no other practical way to obtain useful damping ratios - cathode, source, emitter followers are all 100% (local) negative feedback circuits - there are no "inherently low output impedance" amplifying devices

the "no-feedback" crowd's beloved Triode vacuum tube has loads of internal negative feedback - directly causing the observed plate resistance - a tube without internal feedback is a correctly biased Pentode{emphasis added}

negative feedback is a foundational principle and should be properly understood and used correctly - ignore audiophool mythology if you want to learn how to design sucessful circuits

another poster has gone off topic and is pursuing the throwaway comment about Triodes


so the title is the topic of This thread – is there “negative feedback” built in to triodes – is limited voltage gain, plate resistance a consequence?

I don’t remember where I 1st encountered the idea – is it just a feedback engineer’s “urban legend”?


Many circuits predating the wide-spread adoption of feedback analysis are still usefully analyzed with the more recent intellectual tools - followers, degeneration are taught today as feedback circuit techniques

Anyone interested enough to follow up with references? – if I cared enough I would look at 50’s electronic textbooks as Bode’s work, Blackman’s theorem only became widely distributed after WWII

maybe a start:

Originally Posted by jcx

..for those still interested in tube theory just try some search: Langmuir-Childs law

Improved vacuum tube models for SPICE, Part 1

basically the Vgc and Vgp voltages control the current via the same nonlinear law

[sawreyrw]

A triode can be modeled as a voltage controlled current source (gm*Vgc) in parallel with the plate resistance (Rp). If the cathode resistance is zero or it is bypasses, then there is no feebdack through Rp. Rp will be in parallel with the anode resistor. If there is a cathode resistor, then Rp could be thought of as providing negative feedback. However, Rp varies with operating point so the feedback in nonlinear for large variation in plate voltage.

[Wavebourn]

Originally Posted by jcx

another poster has gone off topic and is pursuing the throwaway comment about Triodes

It was me, who "has gone off topic pursuing the throwaway comment". Actually, I reacted on your off-topic comment about triodes as pentodes with feedback, asking the question that you could not answer.

Please try once more... How come that feedback inside of pentode that makes triode eliminates higher order distortions? And, how it may justify application of feedback to other devices where it increases order of distortions?

[smoking-amp]

Uh, the plate produces feedback even if there is no cathode resistor. The figure called Mu gives the ratio of the grid's voltage affect on cathode emission current versus the effect of plate voltage on the cathode current. Individually the plate effect is called Rp (1/Rp = Gm(p) for the plate) and the grid effect is called Gm (1/Gm equiv. to Rg). They both vary non-linearly with current due to the Childs 3/2 power emission law approx.

In practical devices, the grid does not follow the 3/2 power law so well due to grid wire proximity effects at the cathode, square law emission (grid1 effected) might be a better approx. for some high gm tubes and power tubes.

[kenpeter]

Schade Fig 35, go figure...

Conversely, a triode with internal feedback sabotaged: a Pentode.
Most obvious example of the reversed principal at work is Cascode...

[jcx]

hoping debate can continue in this thread I'm copying this to here

Originally Posted by Nelson Pass

You can't have it both ways.

If triodes have internal feedback, then so do pentodes,
since they exhibit a finite plate impedance due to their
"feedback".

Originally Posted by jcx

As engineers we can perhaps argue over what defines "good approximation" in terms that make sense to both of us

I am happy to engage in the original meaning "dialectic" - not that I think we really have uniformly "opposing" positions or that every engineering issue can be reduced to simplistic "two sided" sets of propositions with simple binary true/false resolutions

(which doesn't mean that there's no definitions, standards that have to be agreed - and are agreed by the EE community - some statements can be "nonsense" or "true/false" by the rules that a community like professional EEs use to "construct"/express/represent knowledge )

for tube internal feedback we should continue in my thread in the tube forum?

the better measure of internal feedback in vacuum tubes is probably (inverse) "mu" - which is easily 100x higher in pentodes vs triodes

so pentodes would seem to at least qualify as ~ "100x better approximations" to "no (internal) feedback" amplifying devices by having ~100x less internal feedback?

my tube expertise is pretty limited to having been the only one with any idea of how to replace/rebias/maintain a Dynaco tube amp in a community music room in my college days I could easily be wrong – care to put up some numbers to compare or point to some good tube modeling sites?
In one of those life ironies George Valley hung around the undergrad program I was in but before I realized I wanted to do electronics he fully retired

Since my education in control theory started with Mech E courses and my career has involved motion control with a variety of motors, hydraulic valves as well as electronics I do think I can figure out diverse "gain device" parameters if anyone wants to “put some numbers” on the floor to debate

this source kind of fades away before reaching pentode properties: Tubes 201 - How Vacuum Tubes Really Work

[Miles Prower]

Originally Posted by jcx

so the title is the topic of This thread – is there “negative feedback” built in to triodes – is limited voltage gain, plate resistance a consequence?

I don’t remember where I 1st encountered the idea – is it just a feedback engineer’s “urban legend”?

It's not an "urban legend". It's quite real. Triodes are unique in that the Vpk has a decided effect on plate current. The plate current of a pentode, and the collector/drain current of a transistor are largely independent of the Vpk (Vce, Vds). This is why the parameter (u-factor -- amplification factor) is so useful for triode analysis, but not so useful otherwise.

If the signal is moving Vgk positive, the plate current increases, and the Vpk falls, as a consequence of increased voltage drop across the plate load. The decreased Vpk will therefore try to reduce the plate current.

It works the opposite if the signal voltage is taking Vgk more negative: less plate current, and a rising Vpk that will try to increase plate current.

Vgk and Vpk are pulling in opposite directions, trying to prevent Ip from changing at all, and that is the definition of negative feedback. The effect is largely absent from every other active device, solid state or hollow state.

As for the effectiveness of Vpk in controlling the plate current, that's given by dynamic plate resistance (r(p)). The greatly reduced effectiveness of this sort of plate feedback in the high r(p), low current, low Gm triodes (e.g. 12AX7, 6SF5, to a lesser extent, 6SL7) with their high r(p)'s, tends to make these sound like pentodes, which is why some consider them sonically inferior to the lower-u small signal triodes like the 6SN7 or 6FQ7 (nine-pin mini version of the 6SN7). It's not that you can't make the high gain triodes (or pentodes or transistors) sound good, it's just that it doesn't come quite so easily. You'll have to pay more attention to loadline selection, and, with pents, probably have to do some empirical breadboard tweaking of screen voltages and bias points to find out where the low distortion point really is. Also, pents will probably require added attention to stiffening screen voltages and screen bypassing. You may also have to resort to fully or partially active plate loading.

[mfc]

Originally Posted by jcx

Anyone interested enough to follow up with references?

Hi Jcx,
Sure a triode has internal feedback. The input is not protected from the output.

Here's a reprint from 1953:
stockman

Shows it in terms of classic feedback theory.

As far as no feedback goes, internally the input needs to get protected from the output. Hence the term no feedback. A pentode tries to do this, as well as two cascode connected triodes.

Mike

p.s.
I don't understand why some say cathode/emitter degeneration is not feedback. Just ask the question: Can the output influence the input? Then its feedback!! I know I'm preaching to the choir here.

[DF96]

Yes, of course a triode has internal feedback. However, as it says in post #4, this is not linear feedback but shaped in a similar way to the grid response. Hence an ideal triode in an ideal circuit can have zero distortion with non-zero voltage gain. There are no ideal triodes or ideal circuits, but it is nice to know that at least in theory it works.

Pentodes (and tetrodes) have very little internal feedback. This is because the whole point of introducing g2 was to prevent this feedback from happening. An ideal pentode would have a 3/2 grid law, and infinite output impedance. External negative feedback, such as the anode follower circuit, can reduce distortion (and add higher order components) but it will only get near zero when the gain gets near zero too. This is because the external feedback is linear, so it can only suppress distortion not cancel it as the triode does.

If you define negative feedback as being linear and sampling voltage only then the triode does not use feedback and cathode degeneration is not feedback. This, however, would be a perverse definition of feedback intended to artificially bolster a weak argument. On any normal definition of feedback (sample the output, to affect the input) then triodes and cathode degeneration use feedback. This means that many "zero feedback" circuits work only by adopting a perverse definition of feedback, because some people have an irrational fear of feedback - I suppose we are all afraid of things we don't understand!

[ruffrecords]

I remember this being debated ad nausem on rec.audio.tubes some years ago.

I suggested the following experiment. Wire up a pentode as a triode and measure its stage gain and distortion harmonics at a given frequency and output level.

Next wire up the same pentode as a pentode with NFB so that it has the same stage gain. Again measure the distortion harmonics at the same frequency and output level.

If a triode truly is a pentode with NFB then the results should be at least very similar i.e. the relative levels of the harmonics should be very similar.

In practice they are not.

Cheers

Ian