HP 3478A behavior on low ranges

[radar_macgyver]

I have an HP 3478A that shows a slowly increasing count on lower DC volts ranges (anything less than 30V). Eventually, it goes overrange, and starts faintly beeping and zeroes itself, then repeats the cycle. Any resistance applied between the probes (I tried several 1M resistors in series) will bring the voltage back to zero, as will putting it into the 30 or 300V range. I suspect the latter is due to a voltage divider. Is this normal behavior?

The instrument has not been used in many years (possibly decades), I cleaned it out with compressed air but am afraid to use any solvents after seeing the weird hybrid inside. I also need to replace its cal battery... it measures about 1.2V, and it somehow is still calibrated!

Fun fact: it was actually a prototype unit (the son of a former colleague used to work at HP and was apparently on the design team). Some folks from Keysight visited our lab and I showed it to them, they remarked that the missing front badge and no serial number was a good sign it was indeed a prototype.

[bdunham7]

Completely normal due to the small pias current (tens of picoamperes) charging up the input capacitance.  The resistance across the input is effectively nearly infinite.  Once you add 10M in parallel, the tens of picoamperes of bias current only result in hundreds of microvolts.

[alm]

Yes, this is normal. Happens to pretty much every meter that is high impedance (GOhms) on the lower DCV ranges. This won't happen if it's actually measuring a signal with a finite impedance. Fix is to set it to a higher range where the input impedance is only 10 MOhm.

[adam4521]

Before changing the battery, you can read out your calibration constants if you have got a GPIB/USB cable. The feature is documented on the forum here:

https://www.eevblog.com/forum/repair/hp-3478a-how-to-readwrite-cal-sram

I've attached a python script I made a few years ago to extract the calibration. I haven't made the reverse one, but it would be easy enough to create that. Unfortunately the pyvisa module needs one of those gigantic VISA runtimes to be installed on your computer, in order to work. I think I used the Keysight one.

[adam4521]

I just saw in the pyvisa docs that there is a pure python implementation as well (pyvisa-py), maybe that would be enough. It might depend on what type of interface cable you have got.

[radar_macgyver]

Thanks all, that's a relief. I verified the unit against a DMMcheck and an in-cal Keithly DMM6500 and it's within 1 digit. Interesting that the DMM6500 has a similar behavior in the 1V and 100mV ranges, but not in the 10V range that it defaults to in autorange mode. I guess it's to prevent newbies to > 4 digit DMMs like me from worrying too much.

@adam4521 I've been lurking on that thread for a bit, thanks for the script. I see that someone's written a wrapper for the Prologix GPIB to Ethernet adapter that I have, so I can adapt the script to use that since it seems like PyVISA does not support it natively. Being a believer in belt-and-suspenders, I'll also have a battery hooked up to keep the RAM alive during the swap.

[J-R]

Input impedance on HP3478A:
30mV, 300mV, 3V range: greater than 10G Ohms
30V, 300V: 10M Ohms ±1%

DM6500:
Input Impedance 100 mV to 10 V Ranges: Selectable: (>10 GΩ or 10 MΩ ±1%) in parallel with <400 pF.
100 V to 1000 V Ranges: 10 MΩ ±1% in parallel with <400 pF

[alm]

DM6500:
Input Impedance 100 mV to 10 V Ranges: Selectable: (>10 GΩ or 10 MΩ ±1%) in parallel with <400 pF.

What is the default input impedance setting of the DMM6500 on the 100mV, 1V and 10V ranges?

[radar_macgyver]

Looks like it defaults to 10 MΩ for the 10V range. If I select 'Auto', it picks 10 GΩ and the readings start to drift due to leakage. On the 100 mV range, it shows overrange regardless of the termination.

I was able to read the calibration data from my 3478A and it looks... weird. Posted on the other thread.