Thanks Katie!

[Harald]

For this article:

http://www.hpmuseum.org/cgi-sys/cgiwrap/...i?read=267

I seem to have just "repaired" my newly acquried HP 27 by inserting a 20k resistor into the 1.6kHz clock line.
When I received the calculator, at first I thought it was dead. I hooked it up to my power supply and started measuring all the supply voltages. To my surpris after turning it of and back on it worked. I put it back togather thinking that I had probably fixed it by unknowingly resoldering a bad solder joint. Well, next time I turned it on it did not work anymore.
Then I found your article and could reproduce the problem. Cooling the calculator doubled the clock signal on pin 20. I inserted the resistor (by the way, is it 20k because it is on pin 20 ;-P or how did you get to that value?) and now it continues to work even when I cool it down.
I still can't really belive this solved the issue, but it does look like it so far. I recon it is noise getting into the ACT that causes the problem. Red is the signal at the ACT, blue at the other end of the resistor:

I wonder where the noise comes from, maybe in reality this is a bad capacitor somewhere that causes it.
Did you try and replace capacitors to solve the problem?
Anyhow, I seem to have a working HP27 now and am happy

Thanks, Katie!

Harald

Edit: I just had to open the calculator again and check. There was quite a lot of noise on the battery voltage. I replaced the cap and it did get a lot quieter. No change on the clock signal (I think this is actually the sync signal) though. Some of the pins on the RAM/ROM chip are unconnected and have noise on them. Grounding these didn't change anything either. So I am non the wiser where the noise is coming from....

[PANAMATIK]

(10-15-2015 08:40 AM)Harald Wrote:  I seem to have just "repaired" my newly acquried HP 27 by inserting a 20k resistor into the 1.6kHz clock line.....

Congratulations for having a working HP-27 now!

The clock line is actually called SYNC. Katie discovered, that the SYNC line shows 1.6 kHz when the calculator works correctly and 3.2 kHz if it doesn't. This is true, but depends on the instructions executed.

The SYNC line is high for 55 us in each 308 us instruction cycle, except for "go to" instructions. What you see on your scope is the idle loop, when the calculator waits for key pressed, which is a repeatedly loop of a test instruction followed by a go to instruction.

Code:


loop:  if 0 = s key  15     ; is no key pressed ? status bit 15 = 0        
         then go to loop    ; then test again

This generates a SYNC pulse every second instruction cycle, which results in a 1.6 kHz signal. If the calculator fails to stay inside the idle loop, perhaps because of executing permanently nop instructions, the SYNC pulse is given every instruction cycle and results in a 3,2 kHz signal.

The reason why the SYNC signal is not given in "go to" instructions is to prevent peripheral chips like RAMs to interpret the "go to" address as an opcode.

Bernhard

[Katie Wasserman]

This was 13 years ago, so my memory of the repair is a little fuzzy. However I think that the idea of limiting the current into the sync pin occurred to me becasue the calculator failed at low temperatures, meaning that probably more current was flowing. I experimented with a few different current limited resistor values and found that 20K worked! As I recall many other values worked too, but this was in the middle of the range.

Over the years several people reported that this fix worked for them -- although it doens't always work -- I'm glad that you found it useful too.