HP 3455A strange problem around 14 - 80 V DC Input

[kblue]

Hello there!

I rarely use Voltages above 10 Volts, so the discovery about this specific problem was kinda random (testing a DC power supply). With 50 VDC input, the meter jumps rather random between values like 22, 26, 34, 28 or sometimes nothing at all (blank). If i hit the 70+/- VDC mark, everything goes back to normal operation. Same with 14 Volts or below. So far nothing was changed within this specific meter (no caps, no nothing). Its rather untouched including the superb fan of unholy loudness

Problem starts around 15 V input and stops at around 70 or 80 Volts DC ... between 15 and 80 it seems kinda "unstable". Sometimes 60 VDC input shows at 60 VDC, sometimes not. It seems to be within spec with 10 Volts input (compared to other meters 5-6 digit meters) or 100 VDC input. Mayhaps this small clip can help to "illustrate" the behaviour.



I'm thinking about some dodgy relais or switches, dunno.

Best wishes from germany!

[Kleinstein]

A dodgy (e.g. oxidized contacts) relay is a possibilty. With 10 M input resistance the current is rather low and may not be enough to reach the wetting current of K5.
Switching between the 10 V and 100 V range multiple times could be a 2st easy try to clean the contacts from usage.

Is there a similar problem at around 200-600 V (1000 V range using a different divider setting) ?

The test is running rather fast in the critical range. So it is hard to see if the test voltage is really stable, though the modern fluke meter looks a little more stable.It is still possible that the supply has a problem in that range and the 3455 is just faster to see the problem, maybe still assuming 60 Hz ?.

[kblue]

Hello and thanks for the reply!

I dont have a DC power supply that goes above 150 Volts. I tried another DC bench supply with the HP meter and the problem still occurs. [If i set the other (modern) DC power supply to 15.00 VDC the HP 3455A reads 14,9999 and is stable ... once i set the modern linear DC supply to 15.10 VDC the HP 3455A goes haywire. Going back to 15.00 VDC its normal again and reads accurate. So it seems to start with just above 15.00 volts DC.

Everything here is running at 240 V 50 Hz. The "sample length switch" at the back of the 3455A is set to 50 Hz.

Here is another video with slower increase in volts DC with some stops. Thanks in advance!

[Kleinstein]

For a short time at the very top of the range the meter switched to the 1000 V range and was stable there.
The time when the meter showed OL was a problem with autoranging. The wrong reading in the 100 V range made is switch between the 10 V and 100 V range rather often. This should already exercise the relay. So likely no luck there.

One could try using manual 1000 V range and check the reading at some 20-50 V. If it is just the relay (K5) the reading should be unstable in the 1000 V range too, not just the 100 V range.

To see if the problem is with the auto zero cycle one could try an AZ off mode (if available) and maybe a slower integration (should give less switching effect).
Just for completeness a negative voltage could also give some insight.

[kblue]

Hello again!

Did some more tests ...

Input 15.00 VDC from my modern DC power supply.

Range Switch >10< shows 14.9999 VDC
Range Switch >100< shows gibberish
Range Switch >1000< shows gibberish

100 and 1000 are sometimes stable for a random amount of time. 10 and below is rock solid.

Reversing the polarity is the same as above just ... well ... reversed.

Decreasing the Sample Rate does not change the gibberish values with 100 or 1000 nor does AutoCal.

So we could assume K5 is the culprit? As far as i know, these are no "ordinary relays". :/

Thanks for all your input so far!

/edit: Did a bit more research and found this thread about "heating up" a reed-relay here on eevblog: https://www.eevblog.com/forum/repair/recovery-of-the-relay-k101(0490-1914)-in-hpagilent-34401a/

So i took my soldering iron and placed it at every wire/contact of K5 (3 on either side so 6 in total) and voila! Reading is fine but goes back to gibberish after some minutes of runtime. Now i need to source some of these relays i guess. They look kinda rare though

[Kleinstein]

The relays are kind of special types for high voltage and low thermal EMF. Chances are they are (could be internal) 2 contacts in series so that the voltage rating adds up and thermal EMF largely compensates.
If simple heating does not work, there may still be a chance to fix the old relay. As far as I understood it, the problem could be with the side where 2 reed contracts are joined, in some cases with some spot welding that did not last for ever, but one may be able to get to that joint.

[kblue]

I sourced the "last" 0490-0663 "NOS" from ebay for around 30 € including shipping. So i hope it is indeed "NOS" and will work for the next decade or so. After that, i can try to fix the old one.

So ye, simple heating only works for a couple of minutes, before it goes haywire again. I was a bit confused about all these contacts but it seems 2+2 are connected with each other measuring around 220 Ohms (guess its the coil) while the more "gold plated looking" contacts in the middle are the actual reed-relay i guess.

/edit: Somehow attachments wont work, but img-tag does.

[Kleinstein]

From the resistance 220 ohms, especially if stable make sense for the coils. In this case the design looks more like a classic high voltage reed relay, not a special low EMF solution.

The question is than of the problem is more with the contact or the coil. If really needed rewinding of the coil would still be an option.

[kblue]

Need to correct myself ... 220 Ohm are the smaller ones. K5 is around 340 Ohms which is the same as other relays with the same part-number.

I already removed K5. Applying 5 VDC the "normally open contacts" are closing (i can hear it too) but the lowest resistance i got was around 7 Ohms and the highest value around 200 Ohms (or rarely even megaohms). Most times it settles around 60 Ohms per actuation which sounds way too high.

So ye ... i guess K5 is the culprit all along. Big thanks to you, Kleinstein!

[kblue]

The reed-relay arrived (was the last one sold / NOS) ... its the same part-number and appearance (it arrived in a cylindrical HP paper box) but while the old one had big copper-like pins for the relay-switch itself, the "NOS" part all legs are silver and react to magnetism. Kinda strange (like cheepish resistors from ali). It does look legit and not reprinted but mayhaps someone re-stuffed whatever-reed-relay into the actual part. eBay-seller has overall very good rating.

Aaaanyway it works!