Larry received the second, non-functional eBay 9100B. Since the first unit worked (aside from the display anomaly), a few board swaps isolated the problem to the A1 ROM board. A little poking around isolated the problem to a single bad transistor:

HP uses weird part numbers and cross-references are rare, so Larry replaced it with a modern transistor that seemed electrically similar. It's working fine so far.

The "comet tail" display anomaly on the first 9100B was caused by the intensity being turned up too high, probably because the display is slightly dim. Here's the "fixed" display:

And here's the brighter display from the second unit:

Both units measure about 3200VDC on the display. Note that the second unit breaks numbers up into convenient groups of three...was this a feature that was rolled into production? It's a later serial number than the first unit.

9100s together!

My HP9100A also breaks numbers in groups of three. Is amazing that this problem was caused by a single bad transistor. Wonderful machines they are indeed. ;-)

Wait, your 9100 _A_ breaks digits in groups of three? Why would they do this originally, take it away in early 9100Bs, and then put it back?

I just don't know, David, but it seems your machine is different, as the first brochure I have about the operation of the HP9100 "A" show it with a display with group of digits of three.




The Hewlett-Packard Journal of September of 1968 dedicated to the 9100 also showed the original calculator with that feature.

I suspect that there is a specific circuit that causes the grouping. Perhaps a component in that circuit has gone bad?

3 digit spacing is normal behavior for floating point displays. The service manual for the 9100A/9100B discusses this and shows a waveform that drives the X deflection circuit to the CRT:

Quote:

---

Also, at times during A15C2's charging, (assuming a floating display presentation) ZDSL will pulse negative. When it does, both A15Q4 and Q3 will saturate, charging A15C2 rapidly. This creates a spacing of the alphanumerics seen on the CRT when FLOATING is selected. When the register has been displayed, ZDRL will discharge A15C2 and the cycle starts again.

---

In the images you can see that both calculator are set to floating point via the appropriate paddle switch. Perhaps it's just a failure in this switch, but I'll send the information to Larry.

The keyboard on the earlier unit (the one with the dim screen) failed yesterday due to a defective 2.2K resistor (probably) on the keyboard PCB. Replacements ordered from Mouser.

Edited: 4 Nov 2011, 11:38 a.m.