I have been attempting to repair an HP-27 that has the common thermal power on problem. The symptoms are as described in other threads. At power on the display is blank until the ACT eventually "warms up". The HP-27 can then be power cycled and it functions correctly.
I traced the problem to the ACT 1.6KHz clock output that is incorrectly doubled in frequency to 3.2KHz. I made an initial assumption that the ACT might have an internal hold time problem or a race condition. In order to slow the circuitry down, I decreased the VGG supply voltage. No effect on the thermal problem. I then swapped the DC-DC converter zener diode with sequentially lower value zener diodes to decrease all supply voltages. No effect. I individually varied each supply using an external lab power supply. No effect. I tried the suggested grounding solution, in the unlikely event that there was a ground bounce problem. No effect. The VGG supply is somewhat noisy, so I filtered that. No effect. At this point I am stumped by the problem. Other than an internal ACT pre-heater this issue seems unsolvable. The ACT temperature doesn't need to be raised much above ambient in order for the calculator to function correctly so it could be easily warmed up.
Just curious if anyone has identified other solutions to this problem? It would be interesting to have a schematic diagram of the internal ACT circuitry, to understand the source of the problem. I also searched the various retro IC databases in case any old stock of the ACT was still in existence. Couldn't find anything.
The performance of the HP-27 was impressive in that it functioned over an amazingly large voltage range. The lower limit was the point at which the LED display was no longer visible. Other than that, the HP-27 continued to function correctly.