Transconductance.

[lndm]

Can anyone enlighten me as to what makes a 'high transconductance' device.

ie. is it the gm figure absolute, or must it be in combination with other parameters.

[gingertube]

Yep!
From your previous posts I may be "preaching to the converted" BUT

The gm IS the transconductance - that is, delta Ia vs delta Vg1-k.
where delta means "change of".
In words,
gm = change of Anode Current for a given change of grid 1 to cathode voltage. Its a dynamic measure.

It does change with Ia operating point - generally higher Ia means higher gm.

High gm tubes generally means that the grid is wound with very closely spaced wires and the grid structure itself is very close to the cathode. Usually also means that the cathode has plenty of emission and that heater current is higher than usual. Thats just the physics of achieving a high gm.

The usual algebra applies:
from:
mu = delta Va / delta Vg
gm= delta Ia / delta Vg
rp= delta Va / delta Ia


you arrive at all the other formulae:
mu= rp x gm
gm = mu / rp
rp = mu / gm

Cheers,
Ian

[lndm]

Yes, I am. I've been trying to digest and build what has been explained to me.

If those posts taught me anything, it is a new view of what makes a device good. I used to look at mu and plate curve linearity.

From what you wrote, it seems a triode needs a low rp to have a high gm as we don't want to use high mu triodes.

I also think you are saying the transconductance figure can be compared directly to that of other triodes without considering the other parameters (within reason).

Assuming current/Zo/Vswing are achieved, and notwithstanding linearity, is gm the most important figure of merit?

I have a 6DJ9/6ES8, it is similar to a 6DJ8 only remote cutoff. Do you have an opinion here?

Thanks for the formulas. I've seen them before but didn't see the point in remembering them at the time. Now I have the inclination, I will.

[ilimzn]

Your question has a rather trivial answer: there is no ONE single figure of merit, otherwise, that would be the only thing in the datasheet! It depends on the application, of course.

[anatech]

Hi lndm,
Remote cutoff means variable mu, which means non-linear response. Great for RF signals that vary widely. Not so good for linear audio amplification, unless you are building a circuit that gives a constant level out for variable level in.

-Chris

[Johan Potgieter]

Hi Indm,

...but now for the crunch. Added to Ian's fine post, when one is talking about audio applications, I want to warn about distortion. That bit where Ian stated that gm is dependant on Ia (as will be rp) can be quite important. These graphs are not linear.

I will summarise without too many figuers, e.g. an ECC88 is normally regarded as a high gm tube, and that is true at Ia=18 mA where gm=12,5 u.mho; rp=2.6K (Va=90V). But at an Ia of 2 mA, which can be the case in some voltage amplifiers, gm=2 u.mho only, with rp=13K. Compared to an ECC83 at roughly the same Ip and Va: gm=1.8 u.mho and rp=55K.

Cutting short, one finds that at these anode currents, a signal swing in Ia from 1 mA to 1,5 mA will result, for an ECC88, in an increase in gm of roughly 35% and a decrease in rp of roughly 24%. For the ECC83 the respective gm increase is 12% and rp decrease is 11%, roughly. (These according to my data book.)

The point: There will be a difference in distortion between the two types. Thus, one must watch at which Ia characteristics are given. Using a high gm tube can be fine, but operating it at well below its normal application can leave one working at such an incline (up or down) of the respective parameters that advantages are cancelled or even reversed.

I am not making a point for/against the specific types, I just used two quite different types to illustrate. (I fear the figures for the ECC82 are worse, depending.) It is also true that Va comes into the picture - the matter is not simple, and one must not oversimplify by e.g. the statement that I have seen (not here) that "high gm tubes are better" or such.

These points will be "old hat" to some; just thought I will include them for completeness.

Regards.

[lndm]

These posts are great, thanks to all.

After reading Johan's post I feel that linearity would be easier to comfirm by looking at a gm curve rather than plate curves.

[gingertube]

Indm,

Mu is the most stable of the tube parameters with changes of operating point. That is, as Ia goes up gm increases and rp decreses BUT mu stays much the same (except for variable mu tubes).

From a simple equivalent circuit point of view a common cathode Amp (with bypassed cathode resistor) is a Voltage generator of mu x Vg1-k with the tubes rp and the load impedance (ZLoad = anode load resistance in parallel with the next stage input resistance and capacitance etc).

That is rp and Zload form a voltage divider AND distortion is entirely due to variation of rp with signal current swing.

So for low distortion you want rp to be small compared to Zload. Also this explains why an active load will give lower distortion (constant current mode => little change in rp). It also explains the "WHY" of all those rules of thumb - like keep the next stage load equal to 3 to 5 times the Anode load resistor.

It also shows why you pick an operating point where the rp is most linear and try to keep the signal swing to a value which doesn't cause the tube operating point to transverse into a region of high rp slope or curve.

SO I would suggest to you that the rp vs Ia curve is the most important curve to look at when designing for low distortion. It will tell you the best operating point to pick and define the maximum singnal current swing you can tolerate.

Cheers,
Ian

[lndm]

I feel empowered to rebias my amps

With this info, I feel I could go through my device collection and rate them for usefulness. Thing is, I have 100's of small signal pentodes and t/p combos, I feel I would be wasting my time working them out, but some people use combos with success.

Why is it so difficult to find reviews/practical reports on devices other than the common ones everyone talks about?

[Johan Potgieter]

I would agree entirely with the further elucidation given by Ian's post #8. While I was trying to indicate that one must be careful not to work too far down a tube's "normal" operating conditions, he gave a good extension of design considerations. (Although, Ian, I did see some mu's go down at the "left" end of the Ia spectrum, but I think that was a little far out.)

I might just add that these considerations are for triodes; with pentodes rp is usually much higher than Zload - but that is a different story.

Thanks to Ian.