HP 5345A counter blank display except on reset

[DIPLover]

Hi,

I own a venerable 1979 HP 5345A frequency counter, that was still working quite well until this morning.

The only problem so far was that the position of the SAMPLE RATE knob on the front had no effect and the counter was basically always in HOLD mode (taking a single measurement until a change of the GATE TIME or a press of the RESET button).

This was easily sidestepped by putting the GATE CONTROL rear switch to EXT ARM and feeding in a 100Hz 1V sine wave to the gate control input jack. I planned to fix this in time, but current setup was ok.

Then disaster struck 

This morning while I was turning the DISPLAY POSITION switch toward AUTO, the display suddenly went blank.
Pressing and holding RESET does generate a correct lamp test, i.e. all segments light up. But as soon as I let go of RESET, display goes blank. Changing the GATE TIME position also does the same thing. All segments briefly light up then nothing.

The Hz, SEC and gate annunciators still work. All digits and dots are blank.

Counter still outputs a very good 10MHz signal from the FREQ STD OUTPUT jack, so it's not dead dead.

Any pointers as to where to look first would be greatly appreciated. The manual is fantastic, but also HUGE. This thing is crammed.

[fenugrec]

From that era and your description, I'm guessing it's an all-logic counter ? Or is there a processor ?

Unless the manual has better ideas (probably does), I'd try working from both ends :  follow the signal input to see where it gets gated / blocked, or if there's some counting going on; at the other end, schematics helping, finding what's missing to show digits. Segment drivers probably ok, but what about shift registers / bin-to-7seg decoders ? etc.

[DIPLover]

Although it appeared in the 1975 HP Catalog, it does indeed have a processor, though no suitable micros were available at the time. The processor is an all TTL job, with 4x 1K ROMs as instruction decoders and 3 16x4 SRAMs as working store. The processor operates on 16 BCD digit numbers, one decimal digit at a time, rather like a calculator.

The 5345A was a quite revolutionary device at the time, and HP Journal issue 1974-06 is entirely dedicated to it.
5MB scan from HP
MUCH better 25MB scan from HPArchive.com


- It has one of the first 500MHz ICs (A single flip flop, in a custom variant of ECL called EEL by HP)
- It is HPs first reciprocal counter
- It is part of the original HP-IB lineup, and certainly the centerpiece, with the 9820A calculator
- It uses a clever phase modulation randomizing scheme on the time base to improve accuracy when averaging
- It can take it's own new plugins, or any 5245L plugin through an adapter chassis
- It can measure frequencies to 18GHz through plugins.
- It has a very stable Ovenized time base as standard

Mine is calibrated to a Standford Research FS725 Rubidium Standard.

It's ridiculously large by 2018 standards, it weighs 50lbs, uses 15 Watts 24/7 for the oven and 120W in operation, it's stupidly loud at 53 dB and it's by far my favourite time piece in the lab. 

This is what it looks like with a 5245L plugin and adapter (not mine) :


I have ordered a paper manual, the scanned (faxed?) manual does ok with text, but the figures and tables are unusable. Should arrive in the next couple weeks.

I will document the repair process in this thread.