My friend Andy Wallace gave me some test equipment back in the early 1990’s, ...
… and this HP 3456A was among them.
At power on, it would fail self-test #4 with an overload indication (“OL"), so I set it aside, fearing that it would be difficult to find and fix the fault. The HP 3456A is a 6 1/2 digit voltmeter, which was very precise for its time. It’s a microprocessor-based instrument (it actually has two, a Motorola MC68A00 and an Intel 8048) and is just completely full of analog and digital circuitry in a 20 3/4 inch deep 3U chassis. I expected to have to do some serious troubleshooting.
I’m getting serious about upgrading my calibration standards now, so I put the 3456A back on the bench. It’s not up to real Volt Nut levels of accuracy, but at least it's a step up from my 5 1/2 digit HP 3468A. I’ll be satisfied to claim Junior Volt Nut status.
I found the section in the manual on troubleshooting from self-test failure codes (section 8-334 on page 8-59 in my February 1982 manual). It pointed me to the procedure in Service Group B, Paragraph 8-B-8 or Service Group C, Paragraph 8-C-10. I tried Service Group B first.
Paragraph 8-B-8 starts by having you move a jumper and seeing if the overload indication goes away and the meter starts displaying measurements. Mine did, so the next step told me to check a capacitor to see if it was shorted (it was not), and if it wasn’t, to replace an LM339 quad comparator in the overload circuitry. I did so, and that’s all this 3456A needed to come back to life. Easy-peasy.
I dug out my Geller voltage reference to check out the calibration of the HP 3456A.
It’s been ten years since Joe Geller adjusted it to 10.00000 Volts on his calibrated HP 3458A, so who knows how much the reference has drifted since then.
I should find a cal lab that can measure it for me on the cheap.
The HP 3456A was well warmed-up, ...
… but the 10 Volt reference drifted a bit after turning it on.
My HP 3468A says the reference is a bit lower. I have no idea whether the standard, the 3456A, or the 3468A is closer to the true value of 10.00000 Volts. I’ll have to get serious about calibrating these now. I’ll need something better than my HP 6920B meter calibrator.
My 1 MegOhm resistance standard came from the MIT lab clean-out. I don’t trust it, as it doesn’t have any calibration data. The first photo in this post shows that the HP 3456A thinks it is 1,007,596 Ω.
I built a 100K Ω resistance standard using two 200K Ω ±0.01% Vishay foil resistors in parallel. This should give me 100K Ω at ±0.007% tolerance. That translates to an uncertainty of ±7 ohms.
I used a four-wire resistance measurement. A resistance of 100K Ω is low enough for lead resistance to matter at this level of precision.
HP says the 3456A should have a 90-day accuracy of ±0.003% plus 2 counts, but it’s been at least 20 years since this unit has seen a cal lab. HP says I should add ±0.0004% per month past 90 days, so that’s an additional ±0.096%, for a total of about ±0.1% accuracy. Still, the 3456A reads within the 100K Ω ±7 ohms tolerance for the Vishay resistors.