Vacuum Tube Modeling
One of my long-term projects is trying to produce data for reasonably
accurate models for various common audio tubes.
An intermediate step will be tracing plate and screen currents as a function
of plate, screen, and grid voltages. Yeah, a lot of this data exists, but
a lot of it doesn't apply to what's available now. There's also precious
little information on both the screen current function and data for plate
currents at more than a few screen voltages for most tubes. I needed both
of these for a good model, so here we are.
If you're interested in this, a couple of resources available to you are:
- Radiotron Designer's Handbook. Chapters 1-3.
(RDH was recently reissued on CD-ROM (Old Colony Sound Labs) and soon to be
reprinted
in 8.5" x 11" (Antique Electronic Supply)
- Vacuum Tube Models for PSPICE Simulations, Reynolds, Scott. Glass Audio, 1993 Vol #5, Issue #4)
- Audio Designer's Tube Register, Vol 1. Common Low-power Triodes
Tom Mitchell, ISBN 0-9628170-1-5
Modeling Parameters
Watch this space for parameters to model various tubes. I'm primarily
interested in audio (and musical instrument audio at that), but I'll
be happy to put other modeling parameters for other tubes if you wish
to send them to me.
I have some tweaked parameters for an EL-84/6BQ5 that provide a pretty good
fit. Hope to put it here soon.
Curve Tracing
I have a multi-part project which is on the back burner at the moment,
that aims to provide models suitable for simulation in SPICE or other
simulation tools. If you consider this a noble cause, and would like
to help, please drop me a line for how you can help.
I'm primarily interested in modeling power pentodes, as the raw data
for most of the low power triodes in use in audio is already available
for modern production tubes in the Mitchell book.
The basic plan is to build three channels of voltmeter, and two channels
of ammeter, and use programmable power supplies for the plate, screen,
and grid.
I really don't relish the idea of entering oodles of data points by hand,
so automating the data collection is the only way this seems feasible.
At least when I'm done, I should have a very nice bench setup.
I have a temporary setup of a linear PS + variac good for 0 - 400vdc at about
150 mA or so. This will hopefully be replaced this summer with a pair of
slightly more clever regulated power supplies, and a bias supply.
I also have one curve (see below).
I wouldn't bet the farm on this one, as I used a population
sample of one. (And it was orphaned after considerable use. Its mate blew,
taking out the screen decoupling resistor in my Carvin X-60.)
I won't be gathering any
more data (except on an as needed basis) until I can automate both data
collection and testing. (I'm working on making all power supplies programmable.)
What data I do have, you can find below:
Sovtek 5881
-
Graph
Plate current vs Plate voltage curve for
a triode connected Sovtek 5881. (Vg = 0 V)
-
Raw Data
for graph above.
Instrumentation
-
Data Acquisition
-
HV Programmable Power Supply -
I've tossed out the optoisolated design
that appeared here previously for an identical pair of 800V HexFETs - the
first working as a low power inverting amplifier, and the second a source
follower. If the prototype works, I'll put it up here.
Back to Nate's
Project Page
Nathan Stewart,
npstewart@pobox.com
$Date: 1997/02/08 08:23:58 $
$Revision: 1.3 $
