I mentioned, some time ago, that Bob Widlar used to use his "Mexican computer,"
namely, Teledeltos paper. This is a kind of resistive paper that you could use
to make analogs or analogues of various physical items. Well, several people
wrote in to ask, where can one buy some Teledeltos paper, these days? I checked
around and found quickly that the last 3 places we had bought this paper were
not in business any more. That's one of the problems you get when you buy a
5-year supply of something—after 4 years, you find that the guy has gone
out of business—and, since you haven't bought anything for 4 years, no
wonder. Our Librarian Mary Holland was agreeable to trying to help me find,
where can you buy it? We still had several yards left, but we did not
have enough to give or sell to all the people who wanted some. Mary did a search
of businesses who might be selling anything by that trade name. No luck.
Then she started doing searches of the technical literature and journals.
She found some Englishmen who had used it as recently as 6 or 8 years ago, in
various technical studies.
Unfortunately, I was not able to contact any of those people to talk to. But
Mary did not give up easily. She searched for companies in the United Kingdom
who might be making or selling this Teledeltos paper, and she found a company—Sensitised
Coatings, Bergen Way, North Lynn Industrial Estate, Kings Lynn, Norfolk, England
PE30 2JL.
After a little planning and negotiating, we were able to arrange a business
deal, to keep it in a simple form, so anybody who wants to buy some Teledeltos
paper can do it easily:
Simply buy an International Money Order for 44.00 pounds sterling (about $70).
This pays for everything, including the paper, tax, packing, plus shipping,
air freight to anywhere in the U.S.A.. (Unfortunately, the fee for the Money
Order will be about $30, but that is an acceptable expense, if you are warned.)
Send this money order to Mr. David Eatwell at the address above. This will soon
get you a roll 29-in. wide by 45-ft. long, about 6 kilohms per square, Grade
SC20.
Or if you send a money order for 36.50 pounds sterling, you can get a roll
18 in. x 59 ft., tax and air shipping included. Either way, the price per square
foot is about the same, about 85 cents per square foot, fairly reasonable, as
most experiments take only 1/2 or 1 or 2 square feet.
Now, while you are waiting for the paper to arrive, you have to get some conductive
paint. You can buy 1/2 ounce of silver-loaded paint for about $36 from various
hobby stores. The only problem with this stuff is, after you open it once, you
come back in a month and it's all dried up. There are various kinds of conductive
sprays or paints, and I do not have much luck finding them in stock, or getting
them to work. At present, Planned Products, 303 Potrero St. (Suite 53), Santa
Cruz, CA 95060 sells a new product, 1/2 ounce of silver-loaded epoxy (mix equal
amounts of 2 components and it will cure in 10 minutes). I bought some, and
it works well (about $14.95).You can phone them at (408) 459-8088. Also buy
several inexpensive brushes and get some soft stranded copper wire.
OKAY—now that you have all this STUFF, what can you do with it? You
can make a resistive analogue (or analog), a model of the physical thing you
want to analyze. I have used it perhaps a dozen times since I was a sophomore
at MIT, and on each of these occasions, it was exactly the right thing to do
the job.
One of the early uses of this paper was to model the potentials in a vacuum
tube. You paint onto the paper, a conductive pattern that represents the cathode,
grids, and plate, and use conductive cement (or paint) to connect wires to those
regions. Then you apply some voltages to the elements, and look for the patterns
of the fields and the equivalent electron flow.
Now, you don't have to apply 300 V just because you are modeling a tube. This
is an analog, remember, and if 5 or 10 V is convenient, then all the signals
are proportional. You can probe carefully to see where the equipotential lines
are, using a sharp probe, very delicately. You can also conclude from this that
the current flow is orthogonal to the lines of equal potential.
Figure 1 shows a typical example of how
you can make an oddly-shaped resistor, where the outlines of the Teledeltos
paper are cut, and contacts to the resistor are made with conductive paint.
I've used it to compute the resistance of a low-ohms thin-film resistor with
the smallest legal shape I could make. I painted along the ends of the unknown
resistance. I just used a minimum amount of conductive paint to make the outline
of the ends of the resistor. Then I made the outline of a simple rectangle so
we could establish the sheet resistivity of the paper right adjacent to the
experiment. After all, if the sheet resistivity varies 5% in a few inches, you
can't be surprised.... Then we did some ratiometric tests and found what the
resistance was. If the resistance was too low, we could slice off a little.
Another time, we made some current shunts with precision voltage taps, made
out of N+ material. We had one geometry that gave repeatable results, and we
wanted to change the shape a little to get a slightly larger voltage. We made
up various shapes and compared them. See Figure
2. Of course, in this case, you have to be very careful not to nick the
paper when you cut it, or you'll have a bad flaw in your analysis.
Figure 3 shows how Bob Widlar used this
Teledeltos paper to define the sizes of several strangely-shaped emitter ballast
resistors in his high-performance circuits. If there are any old-timers who
have had other uses for Teledeltos paper, we will be interested to hear what
they say.
I talked to Mr. Ken Eatwell, who worked with Teledeltos paper in its early
days. He explained this paper was first developed in the United States by Western
Union about 1936, as a paper sensitive to electric pulses, for the new Facsimile
(Fax) machines. After the war, the Creed Co. (a subsidiary of ITT) started to
make the "L39" paper, under license in England. Originally, the carbon-loaded
paper was coated with dark gray lead thiosulfide, but for a better image, this
was changed to a lighter gray copper thiocyanate. In the 1950's, Creed decided
that the Fax business would never go anywhere, and got out of the business.
Mr. Eatwell thought they were wrong, and started his company, Sensitised Coatings.
These days, Fax machines use thermally-sensitive paper, or plain paper with
a laser engine or ink jets.
Now is this paper adaptable to a 3-dimensional study? Generally not. I would
love to have some 3-dimensional resistive stuff, to help me estimate thermal
impedances and thermal cross-talk on an IC die. But I don't think the Teledeltos
paper gives me any help there.
Three-dimensional flow is a whole 'nother ballgame. I have thought about making
a capacitive analog using a large tub of oil in the shape of a die, with little
electrodes in the shape of the heat sources. I am sure I could do this, but
I haven't gotten exactly desperate enough to make this one. Besides, one spill
and I'd be banned from our lab for life.
Could a powerful modern computer do any or all of this work? Quite possibly,
using Finite Element Analysis. But if there is an error in there, you are not
likely to find it. If you make an error with Teledeltos paper, you have to think
about the shape of the curves, and you may get enough insights that you would
catch a mistake. Also, for certain kinds of geometries, the Teledeltos paper
may be really easy to apply, when defining the shapes for a computer might be
excessively hard.
Is the Teledeltos paper really the hard way, the old-fashioned way? Maybe
it seems old-fashioned, but if it gives you some good insights, then it may
be a superior technique. And it may be easier than typing a whole bunch of data
into a computer.
All for now. / Comments invited! RAP / Robert A. Pease / Engineer
Address:
Mail Stop D2597A
National Semiconductor
P.O. Box 58090
Santa Clara, CA 95052-8090