|Re: WOW talk about your retro calculator HP 9100A|
Message #5 Posted by paolo on 10 Oct 2009, 6:32 a.m.,
in response to message #1 by Donald Williams
I too stumbled on a HP9100A early this year, it was not working at first but the problem was just a couple of blown rectifying diodes in the power supply so I could fix it quite easily.
I wonder how many such units are still in working condition all around the world, may be forum could held a "census" about it ?
Also in my case I had the problem of missing magnetic cards so I tried to build some, at the end I made some working magnetic card for the HP9100A which I describe here below in case you would like to perform some test.
To start with I used the card dimensions, kindly told to me by a forum member (Etienne), to scale a printout of an actual card I found on the HP museum site.
I then glued it to a cardboard to reach the given thickness.
On the rear side I glued two strips of magnetic tape from a wrecked VCR cassette. Take care that the active side of the tape is that which faces outward w.r.t. the reels hubs, if you use the other side the support plastic film will prevent the head to get enough signal.
The exact position of the strips I derived from measurements of the head position in the card reader. It took me four tentatives to get a working prototype, I realized that the main problem are:
a) to have the correct width for the card (I found it is 50.5 mm) to fit exactly between the rails of the reader, otherwise the card tilts on the right when driven out by the mechanism (which has a single rubber wheel on the left side) and alignement is lost with the data.
b) to be very careful in having the tape smoothly glued on the cardboard, any wrinkle causes poor performances. The trick seems to glue the tape with the card slightly concave (U-shaped) so that when allowed to relax it will "pull" the tape flat. A quality check is that the tape surface must be reflecting an incident lightbeam as it would be a mirror.
I tried the card many times, at first recording a sequence of numbers (which is useful to check if the process "skips" a data) and later saving and reloading simple programs.
After the first success with short (20-30 lines) programs I tried to load on the calculator the full "diagnostic program" which is 196 lines long (i.e. the whole RAM capacity) which was used by HP field engineers to test the overall functionality of these machines.
The prime result was that my HP9100A passes the test, which is the final prove that the HW is all working.
I did some other experiments to find a reliable enough procedure to reconstruct cards which works, the point seems just to be very precise with dimensions and tape gluing.
On two of them I was able to load the full 196 steps sequence of the 9100A diagnostic program and the data were still retained one month later.
I have in the back of my mind the idea to write a "paper" about this topics to put in the related forum section but up to now I didn't fond the quiet times to prepare it, may be the renewed interest in HP9100 can trigger it.