• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Stability question for the gurus

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Yes, pf.

The impedance Xc (sort of like resistance except phase changing) of a capacitor is proportional to freq. and capacitance. Xc = 2 pi f C

With the higher AC voltage at the plate, we need to scale up the impedance Xc so the same current flows back thru the N Fdbk path as before. I = Vp/Xc
So the input circuit sees the same Fdbk signal level.
 
Woops, that should be Xc = 1/ (2 pi f C ) above

-------------------------------------------------------------------

I've been reading thru the Wolcott White Papers linked earlier and I'm finding some claims that I don't think hold up.
1) the claim that the difference of two differential gain channels cancels odd harmonics.
---- compressive or expansive distortion to the signals is still comp. or expansive when differenced
2) the claim that positive and negative Fdbk loops both reduce in gain when clipping occurs
---- looks to me that only the outer neg. Fdbk loops saturate, leaving an active inner pos. Fdbk

More reading to do. Want to find out how the OTs were made.
 
Last edited:
Member
Joined 2006
Paid Member
I've been reading thru the Wolcott White Papers linked earlier and I'm finding some claims that I don't think hold up.


I did go a bit through this paper (because I use zero impedance also to drive the OPT, and also nfb with pfb inside of it, works like a charm in simulations), am not to confident in what he says because he said in an other paper:

"The amplification factor µ depends onby upon the physical structure of the vacuum tube; principally upon the distance from grid to cathode and grid to plate and the fineness of the grid mesh (i.e., the closeness of spacing of one grid wire to the next within the electron stream between cathode and plate."

Allthough myy is the most stable parameter of a tube it depends on other factors too, otherwise a "variable myy tube" could not have been possible.

Because of this, one must be carefull not to swallow unchewed...
 
Last edited:

PRR

Member
Joined 2003
Paid Member
... one must be carefull not to swallow unchewed...

True.

But Mu is the most constant of the three tube "constants".

In an ideal tube, Mu is constant at about any current. You can calculate it from ruler measurements.

In a Real tube there are end-effects. When the grid "fence" shuts-down most of the electrons, some electrons go the long way "around the fence". The fence (grid) has poor control of these. The effect is a low Mu at low current, typically far below where the tube was intended to work )and typically far below where we would work a Hi-Fi stage which must drive stray capacitance past the audio band).

On the other hand, Gm and rp vary as low powers of plate current. People think the data-sheet Gm and rp are true; they are true only for the current they are rated at. Which is often a show-off current higher than we audio designers would normally run. I was just looking at 6SL7. When you go from the sheet's 2.3mA to a (guitar-amp) 0.5mA, Mu falls 7% but rp rises from 44k to 77k and Gm falls about as much. More than 2X change. (If you do not have detailed data, assume square-root of current.)

Large changes of Ip will make small changes of Mu and large changes of Gm and rp. However once we bias a low-THD audio tube we do not make large changes of current. A "high output" stage may be swung 3:1 (from 50% to 150% of idle current). For lowest THD we swing much less. So the change of Mu may be very small. Yes, Gm and rp do vary significantly. Plate resistor linearizes triode significantly, cathode resistor more.

Mu also seems to fall at very high current. I do not have an excuse. This is usually close to the tube's limits.

Variable-Mu tubes are *different*, and always identified as such. Ideal tubes have uniform grid-wire pitch. Most real tubes have pitch as uniform as possible; for linearity and because that is the simplest mechanism for winding. Variable-Mu tubes have deliberately *non-uniform* grid windings. Close-spaced at one end, tight-spaced at the other end. (Some forms have skipped-tunes in the middle.) The same geometric laws of Mu apply, except you have to calculate for each turn down the whole length and superimpose the results.
 
Member
Joined 2006
Paid Member
Waveborn, you did not get what I said, I said: ...otherwise a variable myy would not be possible. I also said: ...it depends on other factors too.

PRR got it correct, in a ideal tube you would get a AVERAGED CONSTANT myy no matter what the pitch. Shared current distribution changes makes myy variabel even in tubes with evenly wound grids. And this effect is put to use in a "variable myy" tube with non-uniform grid windings.
 
Last edited:
The principal other factor is grid wire spacing, which he specifically mentioned. I don't understand how gorgon53 can disagree with someone who says exactly the same thing. I think we are arguing over nothing.

One must never swallow anything unchewed, but it would be unwise to accuse a writer of mistakes and 'prove' it by quoting a sentence which is completely true.
 
And this effect is put to use in a "variable myy" tube with non-uniform grid windings.
I'm going to pick a bone here and point out that the principle reason for variable grid pitch was to greate variable gm, not variable mu. This does indeed affect mu somewhat too, but that is more of a by-product than the design expectation. mu is mainly due to spacing, gm is mainly due to pitch. So-called variable-mu tubes don't actually have hugely variable mu, and the name 'variable mu tube' was considered a misnomer even in its day.
 
Last edited:

PRR

Member
Joined 2003
Paid Member
I'm going to pick a bone...

I claim a tooth on your bone. I was *going* to say that; Mu rarely varies by factor of 4 at any useful current even for "variable-Mu" tubes. OTOH the Gm is commonly worked down by factor of 100(*), although at currents which only make sense for tuned tank loads. Radio receivers often need more than 30dB (30:1) gain reduction; good ones over 50dB (300:1), so we find 1, 1.5, and 2 vari-Gm stages in the AVC loop.

(*) Granting always that Gm "can" and does go to Zero for zero current.

You may have a point about spacing versus pitch. Without hard thought, I observe that if Gm goes way-way down, at constant Mu the rp must go way-way up, which it won't due to end- and stray-effects. So the fall of Mu in a "variable-Mu" tube is perhaps more a side-effect than The Point. (A true wide-Mu triode would be handy in audio, but does not exist.)

However Manley could not fit "Variable Transconductance" on the front of their limiter, "VARIABLE MU" fit, and now we are stuck with it.
 
Member
Joined 2006
Paid Member
I (wrongly?) presumed that it is NOT necessary to specifically point out how the tubes in question are built.

I presumed that would be clear to everybody here.

I wanted to draw your attention to the fact that the mu of any real tube is at least to some extend depending on the operation condition.

I also wanted to draw your attention to the fact that this dependency of operation conditions is exaggregated put to good use in tubes called "variable mu" wich offcourse are built the way many of you pointed out.

My apology in case I presumed to much and distracted you from the real point I tried to make.

To my knowledge, tubes intended to be as linear as possible, where built with great attention to even gridspacing, grid wire thickness, centering, grid-cathode spacing, outlay of boundries a.s.o. to get the most uniform field and current distribution practical possible.

All the datasheets I have ever seen of real tubes show the interdepence of gm, mu, and Rp and there depence on operation condition.

In my view, alltough variation of mu is smaller than the rest, it cannot be said that a real tube has constant mu under other than statical conditions.

I hope you all can understand now better why I said to Waveborn "you dont get it"
I should have said "you did not get me" instead.
I was a bit pissed, because I am pretty sure Waveborn understood just as good as
PRR did what I was talking about, namely, that mu in real tube is not constant and depend on operation condions.

Since mister Wolcott based his claim on the assumption of constant mu I said:
"better chew before swallow".

I went through some of mister Wolcotts papers and "swallowed" just about everything he said, in other words, I had great trust in what he said until I came across the paper where he based his design on the assumption that mu is constant.

To this day, and I have 50+ years of experience in electronics behind me, I have not yet come across a datasheet where mu is constant, so, until someone can show me otherwise, I will stand behind of what I said, and be suspicous if someone tries to tell me otherwise.

I would have hoped for a more fruitful discussion, to sad...
 
Last edited:
Member
Joined 2006
Paid Member
To PRR,

Thanck you for giving that "mu thingy" some deeper thougths.

May I ask you to consider that, since "variable mu" tubes have been built for use with automatic gain control in receivers by regulation the " small-signal stage-gain" with a comparable large dc that mu may very well be variable in those tubes.

And to Waveborn, yes I know that stagegain and mu is not the same so please do not bring that up, I beg you to show some "goodwill" and try to understand what I am after, I want that mu-thingy getting sorted out:):)

To PRR

I also would be very interested in your thougths about that datasheet thing I brougth up.
I am sure you have seen graphs that show the interdependence of mu, gm and Rp depending upon operation conditions.
I would value your opinion on that, are those manufacturers getting it all wrong?
 
Last edited:
Member
Joined 2006
Paid Member
I'm going to pick a bone here and point out that the principle reason for variable grid pitch was to greate variable gm, not variable mu. This does indeed affect mu somewhat too, but that is more of a by-product than the design expectation. mu is mainly due to spacing, gm is mainly due to pitch. So-called variable-mu tubes don't actually have hugely variable mu, and the name 'variable mu tube' was considered a misnomer even in its day.

I fully agree with what you say
 
Clarification please. I am specifically referring to the capacitor in the global negative feedback loop, not the one in the input stage Zobel network. The former (as I understand it) has to do with phase compensation in the feedback loop, whereas the latter has to do with limiting high frequency bandwidth, both for stability, but operate differently.

The input capacitance can cause a lot of phase shift so the feedback capacitor cancels this out. Removing it could (under certain conditions) cause positive feedback and squealing.

I have just had problems with two amplifier circuits with the same problem. A little feedback capacitance fixed both problems.

Running things on the ragged edge doesn't always work. I bought an old 1980's Maplin amp that needed new output transistors. I put them in and it oscillated badly. I had to add capacitance to the VAS stage to tame it.

I had a chip amp oscillating and again a little capacitance across inv and non inv inputs fixed it.
 

PRR

Member
Joined 2003
Paid Member
..."swallowed" just about everything he said, in other words, I had great trust in what he said until I came across the paper where he based his design on the assumption that mu is constant....

I am sorry that you worked in the field so long and did not realize that EVERYBODY gets something wrong.

...May I ask you to consider that, since "variable mu" tubes have been built for use with automatic gain control in receivers by regulation the " small-signal stage-gain" with a comparable large dc that mu may very well be variable in those tubes...

Why should I "consider" it? (I might get it wrong.)

LINK to 6386 twin remote-cutoff triode datasheet. 6386 is one of the few "variable Mu" (remote cut-off) tubes which give Mu curves.

This is _not_ a typical broadcast band radio IF pentode for AVC operation, but an in-between. It is for VHF (~~100MHz) radios, where triode impedances do not damp practical tuned circuits. It gives lower noise than a pentode. And the particular virtue of the 6386 is a very high input overload level. You need that when tanks or aircraft may be far apart OR practically on top of each other, when input levels run from microvolts to whole Volts.

Bottom page 3, Mu may fall from say 19 to about 7 (@100Vp), or 17 to 4 (@300Vp).

Worth noting also, page 2, that Gm fall-off is quoted for a 40:1 range. (IF pentodes generally cite a 100:1 range.)

While meant for aircraft radio, the 6386 is also used in linear audio limiters/compressors. The first-order nonlinearity (and control-voltage bleed/thump) is cancelled by push-pull operation. 6386 seems to have the highest input overload of any small tube suitable for gain control. However the 4:1 (12dB) Mu fall-off does not cover the 15dB-25dB range of control; it works variable-Gm (and rp) not vari-Mu.
 
gorgon53 said:
In my view, alltough variation of mu is smaller than the rest, it cannot be said that a real tube has constant mu under other than statical conditions.
An ideal triode has constant mu. It will not have constant gm or rp, but follows a 3/2 law. Unfortunately we cannot make an ideal triode, but some triodes over some limited range of their characteristics can approximate an ideal triode. Hence it is quite reasonable to design on the basis of constant mu; it is a good enough approximation for both real engineering and teaching.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.