Measurements on emf-error of switches

[Muxr]

Really cool Andreas! I was looking at TI's MUX506 or MUX507 as well which can handle 36V, but they don't mention much about protection, they also advertise low leakage. I don't really need anything above 16V.

[Andreas]

What do you mean with protection?

Most of the muxes have no protection against input voltages above supply voltage.

I am now trying some FETs (BSS138) on my 2:1 dividers to protect the unbuffered LTZs if I have a flat battery/bad battery contact on my ADCs.

On my relay mux I have high ohmic output resistors which can be shorted if the output voltage is within 1 mV of the expected value.

With best regards

Andreas

[Muxr]

Indeed, I am finding that most analog muxes don't have protection. The MCP508 zhtoor suggested does however:

[Andreas]

Hello,

for a unbuffered LTZ a 1K resistor is relative low (compared to the 7V/4mA zener current).
I want a cirquit that limits the current below 2mA.
On the other side: a 1 K series resistor (additionally to the switch of 300-500 Ohms) is relative high when regarding leakage currents of up to 10 nA.

with best regards

Andreas

[Muxr]

Ahh, I see. All the references I will be measuring are buffered. Just the hodge podge of all the references I own and perhaps some more I make in the future.

[doktor pyta]

Just to have this in metrology section /if someone reads only metrology/:

https://www.eevblog.com/forum/projects/diy-low-thermal-emf-switchscanner-for-comparisons-of-voltage-and-resistor-stand/msg1366769/#msg1366769

[janaf]

How do you measure EMF of relays?

Just zero the voltmeter, short the inputs, and then connect a (closed) relay and take a reading? Or what?

I saw this method somewhere (picture): connect a relay/switch, open and close the relay, take readings, the difference would be due to switch EMF. What's do you think is a reasonable resistor value, low to keep noise down but high enough not to load down the EMF? 1K?

(Otherwise, I kind of like that Japanese paper; it seems they measured a number of relays, selected pairs, two per channel, that more or less cancel each others EMF).

[janaf]

...and the picture.

[Kleinstein]

1 K might be more on the high side - the switch resistance should be much lower than even 10 Ohms. I would not be worried so much about noise, more about bias current. 1 K and 100 pA would already be 100 nV.  I would be less worried about the loading effect that might change a few percent of the voltage at most.

The other point would be possible thermal EMF from the resistor, if not at a constant temperature.

I would consider testing a few contacts at a time, by switching between the contacts.

If temperature is really constant all over the voltage reading should be zero, even for poor contacts. So one would also get a kind of typical, not so good thermal condition. How this is defined is not clear to me. Anyway a few temperature readings would be needed. With a mono-stable relay the self heating of the could could be the relevant heat source - so maybe check with a voltage close to the upper limit to get more effect.

[janaf]

I'd use a 7.5 digit DMM, don't think there'd be anything like 100pA there but would have to check what it looks like...

Once I tested thermocouples and was surprised what load they could drive.