I have been asked to try to repair an HP 34740A. It is a part of the HP 3470 measurement system. Specifically it is one display module (the other available display model is the HP 34750A).
The complete instrument is composed by the mainframe display module and a plug-on function module, an HP 34702A in our case.
The symptom was easy to spot: the display was blank except for the decimal point and the minus sign when the instruments was turned to DC measurement.
Luckily, thanks to www.hparchive.com, the manual was easy to get. Moreover HP manuals are usually very through.
The manual really helps you: not only there are the schematics, but the principle of operation is also clearly explained. Moreover you also get the timing and data waveforms. A table that shows you the 27 major signals, how they are generated and which component may be faulty if they are not present is really usefull too.
I started my troubleshooting from Figure 7-11 and then went backwards.
Was it possible that the problem was not the 34740A but the HP 34702A? I didn’t think so and a quick check with a multimeter told me that I was right: the module was working properly but the 34740A wasn’t.
Looking at the schematic I felt a bit troubled by three IC that were deeply involved with how the digits were displayed: A2U1 (1820-0635, IC:DIGITAL, scanner), A2U2(1820-0571, IC:TTL DIGITAL, character generator) and A1U9 (1820-0798, IC:TTL SIX DECADE, data accumulator… which is really involved in everything the module does). They are HP proprietary parts. If one of them wasn’t working, my chances to find a substitute were zero.
I could probably replace one or more of them with a SPLD and/or a microcontroller but that would be a somewhat bigger project.
I started to check the most important one: 1820:0798.
The first thing I did was to check if it power supply was good. It was as the voltages were within tolerance. So, 1820:0798 supply was ok. I also checked the supply of the other two IC. They were both as good as expected. I checked all the voltages that were involved with the display and they too were fine.
I had to go back to A1U9. I decided to start with the data clock and see what was going on with an oscilloscope on pin 9.
There was a nice square wave but the frequency was wrong: 5MHz while the manual told me that it should have been 3MHz.
Once again the manual provided me with an answer: the instrument came in two options: Option 060 and Option 050 for optimum operation ad 50Hz or 60Hz line voltages. Each option had its own quartz.
So now I had to remember that the frequency of every signal derived from the main clock would not be the one written on the manual. I knew that I could calculate the correct frequencies anyway if I needed to.
Knowing that the clock was ok, I checked pin 10 of A1U9. There was also a square wave as I expected which meant that A2U1 was working at least partially.
After pin 10, I checked pin 1,2,15 and 16. They weren’t stuck neither at gnd nor at Vcc but showed clear transitions. Those 4 lines were BDC data from A1U9 to A2U2 which is the character generator.
As A1U9 seemed to work I decided to check A2U1, pins 1 through 9. The signals were as expected.
It was time to check A2U2. I didn’t see any activity on pins 5,6,7,10,11,9,4,20,2,1 which wasn’t what I expected and I started to fear for the worst.
I checked pin 18 and the signal was as expected while pin 14 through 17 were stuck high. That were good news as pin 14 through 17 were BCD data and 1111 was Display Blank Code. I couldn’t say for sure that A2U2 was working but at least its behaviour was right.
Multimeter and oscilloscope brought me far but it was time to start with a logic analyzer.
I went back to A1U9 and put micrograbbers on almost every pin. A quick acquisition and I was sure that the IC was really working: the BCD data lines had the right data at the right moment.
A2U1 was also working.
At this point I was rather sure that either U2A2 had failed or that there was a failed component somewhere between A1U9 and A2U2.
I decided to start from U2A2 and move back to A1U9, so more micrograbbers went on the IC A2U3 (SN7432N IC:TTL QUAD 2-INPT OR GATE).
And there was my failure.
Even when two related inputs were low the related output was still high which isn’t what you would expect from an OR gate.
I desoldered A2U3. Even without it, pins 14 through 17 of A2U2 were still high but this was somewhat expected.
I soldered some resistors on and I tried to pull low manually the BCD lines. The display started to do again its work and by pulling low the correct lines I could display every number from 0 to 9. Obviously it was the same number for every digit except the first digit, which during normal operations could be either off or 1.
What was wrong with A2U3? It seemed that its outputs could not sink enough current, which is strange.
Anyway I replaced A2U3 and the multimeter was back to its working condition.
I let it run for 24hrs without any problem so I suppose everything is ok. However I regret that I didn’t measure how much current A2U3 outputs were sinking as I am still puzzled: I tested the SN7432N I replaced on a breadboard and it seems to work. I guess I will have to test how much current it can sink sooner or later.
And… this is all.
Writing this post took me way more time than troubleshooting and repairing the multimeter which is a real shame.