HP-3400A RMS Voltmeter - Gerät 2

HP3400A is a very unusual AC voltmeter, unique in capabilities and design.

Even nowadays, it is hard to find something that can compete. HP manufactured this thing through several decades, with technical changes inside, but keeping the original spec.

This model is an early version. Sold as broken, without power cord, dirty.

Serial number G541-01020, means design date 41th week 1965, manufactured in Germany.

A second one, from 1970 or later, is described here

EBay Images

Inside HP3400A

Power supply board:

Chopper Amplifier:

Newer chopper amplifier, from other device. Cool thing is, my device does work with that board. So the error IS on the chopper amplifier board:

Repair

Replacement of ancient power socket:

One of the large capacitors from power supply section has lost its capacity. I replaced it with a nowadays capacitor (which is much smaller).

Neon bulb driver voltage, signal measured:

Neon bulb driver voltage, , how it should look like - very close! :

Input to demodulator, signal measured:

Input to demodulator signal, how it should look like:

Fact is, there is no signal response on demodulator output side (LDRs) at all. No signal change in output, whatever happens on demodulator input. I suspect some issue inside the NEON/LDR coupler element.

Thermo coupler opened:

Inside Neon/LDR coupler element:

I pulled out all LDRs:

Two LDRs seem to work (modulator side), two behave strange (demodulator side).

The good ones. They have some hundred KOhms in dark, some hundred ohms when in light:

The bad ones. They have many Megaohms when in dark, and still >2Megaohms when in light:

Bad ones in detail:

Good ones in detail. They have some colored dots at the bottom:

It looks like some brown fluid flow across the light sensitive area:

Even on the side of the LDR, traces of that fluid can be seen. There is nothing like that, or very little, on the two working LDRs.

In theory, I could replace these LDRs. Some research in Internet forums shows that the LDRs used by HP are very fast ones (have a very fast response time »100Hz), and they are used with a frequency in range 90..110Hz.

LDRs manufactured today are slower and do not react well on these “high” frequencies. So some time-consuming research, trial out of many LDRs, would be required going that way.

Another path is to replace the Neon bulb/LDR technique with LED+Photo-Diode/Transistor technology, which is much faster. 100MHz or so is possible there. But this would also require major rework and research for the chopper amplifier. Some people (see EEVblog) went that path with success.

I am lazy and looked for a solution without my own research and such. And there is one such solution!

Chopper Amplifer Replacement

The whole board with Neon-Bulbs/LDRs can be replaced with a drop-on solution!

In the latest versions of HP3400A, HP used a newly developed scheme for the chopper amplifier, based on ICL7650 OpAmp. The main features that qualify this chip are:

• Low Offset Voltage: 1µV
• No Offset Voltage Trimming Needed
• High-Gain CMRR and PSRR: 120dB min
• Low Offset Drift with Time and Temperature
• Low DC Input Bias Current: 10pA

OpAmps chips were not available in 1965 design, and the ICL7650 is a development especially to work in chopper amp replacements. So, when chips were available, HP redesigned the chopper amp.

HP original scheme with ICL7650:

There is a development by Scott “Defpom” which is a plugin replacement for the A6 Chopper Amplifier board. Described e.g., in this video: https://www.youtube.com/watch?v=hR4cEz629A0&t=1764s , PCB described here: https://www.pcbway.com/project/shareproject/HP_3400A_A6_Chopper_Board_Replacement_06465eb7.html

His updated scheme, using nowadays components and partially using SMD form factor elements:

Defpom A6 PCB, upper side, with through hole and SMD parts:

Defpom A6 PCB, bottom side, with SMD parts only:

Finished board:

I ordered 1N5817 as a through-hole part by accident. But the THT part could also be soldered onto the SMD pads.

All resistors from old potentiometers (these are replaced with three new ones on the new PCB) need to be desoldered from the original socket. Also, one wire (white wire to CAL trimmer) needs to be desoldered at the trimmer and directly soldered at pin 10. This is all described on defpoms web page in detail.

I left all old resistors in place, just in case… The new PCB also has a screw hole to fix it to the housing.

New boards fits nicely.

The device needs new calibration after replacing the board. I did this and used only the three trimmers on the new PCB, the “Full scale” old trimmer (connected to the meter, still in function after replacement) and the 1V/400Hz trimmer elsewhere. The calibration needed some time. But finally I have nice readings, check 0dB below.

Nobody knows for how long the HP3400A was not working due to the buggy old PCB. I am glad to have two HP3400A meters now.

Costs of replacement

Because I had none of the SMD parts, I had to order everything in 100-amounts. For one of the values, I needed to order even 5000 pieces (minimum amount they sell). So this was costly, about 55€. But I needed only parts worth about 10€. Then the board replacement itself, which also was 55€ (included two additional PCBs that I did not need, but I bought it as a set). So, as you would calculate, in range 65-110€, a well-working replacement.