There are experts in the audio field out there who have their reasons that
certain tubes SOUND better in audio (related to distortion and nonlinearity
as well as other reasons such as hum, microphonics). Eric Barbour of
Svetlana, who writes for Vacuum Tube Valley (His articles on 6L6, EL34,
300B, and 6550 are found in the archives of VTV at
http://18.104.22.168/articles.htm), has published a lot on this.
If one looks at pure linearity as a function of the tube curves, for class
A and AB service, notice the compression at the low end of plate current,
and the high ends where cathode is emitting all it can. The plate current
increments for a given grid voltage change are compressed. Also the knee or
folds in the curves (for tetrodes) where significant grid and screen
currents flow, at low plate voltage. Taking all of these into the operating
space, it is evident that secondary emission of the elements as well as
space charge of the electron cloud and the virtual cathode location from
that, all cause nonlinearity from perfect operation. If these effects were
gone, the ratio of plate to grid current alone should have been a function
of the ratio of the plate and grid voltages, alone. The tetrode introduced
secondary emission from the screen, which distorts the current-voltage
characteristics. And so on....
Without a long discussion, I point to some excellent old texts on the
subject. Karl Spangenberg of Stanford Univ. wrote Vacuum Tubes, in 1948. It
is one of the McGraw Hill electronics series, edited by Terman.
Frederick Terman's books also cover a bit of the vacuum tube concepts.
Spangenberg spent a chapter just on space charge and discussed secondary
emission effects with examples from good old glass (GOG) tubes.
And the secondary characteristics of the many tubes out there are highly
dependent on metals used, geometries, and electron beam optics. Its an art
to make perfect tubes, and manufacturers have worked to make special high
power versions for higher linearity, i.e. 4CX1500B over the 4CX1500A,
4CX5000J over the 4CX5000A. Eimac says in their data books, 'low grid
interception' for the higher linearity versions.
In Care and Feeding (the old edition), "It is possible, by clever
manipulation of the electron ballistics within a given tube structure, to
alter the transfer characteristic and minimize the distortion products."
My guess on factors that effect linearity:
Alignment of the elements
Beam forming electrodes such as in the 6L6 and 6146B
Filament gun - uniformity of E beam
Plate material (do electronics emit and hit screen?)
I degress for a minute:
An interesting historical point was the Audion. DeForest had trouble making
good hard vacuum, so the early tubes had a lot of gas, and were very
unpredictable. Linearity was not even something to measure then. They
amplified with a lot of ions floating around. I read this in some old
1910-15 IRE papers, in which DeForest and Edwin Armstrong argued about
these points. It was General Electric, under I. Langumuir's guidance, that
made hard vacuum tubes more like what we use now. They called them
pliotrons. DeForest was furious over their claims to be superior; people
like Armstrong figured out how to build reliable oscillators and
amplifiers, and the better tubes made it less empirical and more
quantitative. The Audions eventually got better. Remember that PBS TV show
"the Empire of the Air". It touched the Armstrong/DeForest battles, and
later the Armstrong/RCA-Sarnoff battles.
I would love to hear from tube experts/builders out there about tube
linearity and how their designs work to improve it.
FAQ on WWW: http://www.contesting.com/ampfaq.html
Administrative requests: amps-REQUEST@contesting.com