Measurements on emf-error of switches

[Echo88]

Since a lot of people here in this forum have nanovoltmeters, id like to ask if anyone of you has ever done a systematic comparison of different switches/relays/solid-state-switches regarding emf-error, with numeric results. If it has been done, can you publish it here?

Regarding relays: As far as i know suitable low-emf-relays are for example the G6AK/G5AK-series from Omron, Panasonics TXS-series or COTOs 3500/3600-series.

Ive seen that a few times solid-state-switches were used for low emf in high-end-scanner-cards, but never a mention of the used solid-state-switch-component (manufacturer, device name). Maybe someone knows more about it? 

[ManateeMafia]

On a Data Proof 160A with inputs shorted using copper tubing , I was able to measure between 50nV -100 nV on different combinations of switch connections. The data I have seen in their manual shows typical values under 50nV. My test conditions were far from perfect but were consistent. It was all automated with software and measured with a 34420A.

I have plans on doing the same with some TX-S relays next.

[Vgkid]

In one of the voltage intercomparison from bipm/nist??/a bunch of others. Used a automated Electroswitch C4 switching system.

[dacman]

Latching relays, which are in scanners such as the Data Proof 160A and Measurements International 4220A, will have lower EMF values than non-latching relays.

[Vgkid]

In one of the voltage intercomparison from bipm/nist??/a bunch of others. Used a automated Electroswitch C4 switching system.

https://www.google.com/url?sa=t&source=web&cd=5&ved=0ahUKEwjIyK3n0pDQAhWESiYKHcvYCNkQFgglMAQ&url=http%3A%2F%2Fwww.bipm.org%2Futils%2Fcommon%2Fpdf%2Ffinal_reports%2FEM%2FS28%2FEUROMET.EM-S28.pdf&usg=AFQjCNGPtV70IXTjBpYG1zqa7BQNNLk-Pg&sig2=yfK3JPMqw-opK81A6LTzIQ
https://www.cenam.mx/memsimp06/Trabajos%20Aceptados%20para%20CD/Octubre%2025/Bloque%20A/A3-ELECTRICA%20I-Patrones%20Cuanticos/A3-3.pdf

[ManateeMafia]

I should have been more specific that I plan on using the TXS2-L2 dual coil latching relays. They appear to have good low thermal specs.

[Andreas]

According to datasheet it would be 0.3uV
But for which contact?
The low level contact (option -1) or the standard contact?

With best regards

Andreas

[Echo88]

Ive found the AQV210 PhotoMOS-switch, which is stated to have 1µV emf-error....but it is only mentioned on the presentation page of 2 distributors, not in the data sheet: http://au.rs-online.com/web/p/solid-state-relays/1137210/

@Andreas: Maybe Panasonic does emf-tests like COTO: across 2 antiserial connected contacts, so they cancel the emf-error slightly -> http://disti-assets.s3.amazonaws.com/testco-inc/files/datasheets/17016.pdf

[Andreas]

Hello,

ok bad trick.
Unfortunately this is not the way I plan to use with my new relay scanner for my LTZ-references.
I want to be able to measure the difference between any of the 15 inputs.

In the old 7 channel scanner I have used TQ2-L2 relays. (and I still have some of them specced with 2uV).
But the question is if the thermal EMF plays a large role when latching relays are used.
The whole scanner (including processor) is powered one week with a 9V NiMh battery.

In the TXS2 datasheet if I interpret the EMF distribution diagram correctly they have used standard contacts for the test.
But on the other side: how can they get a sum quantity of 20 results out of 6 relay samples?
So also the relay type might be wrong.

I guess that the EMF is lower with standard contacts (same material for moving and fixed contacts).

With best regards

Andreas

[VintageNut]

The Keithley 7168 Mux card is specified to have contact potential of <30nV. The problem is that 10V is the maximum voltage.

The 7035 Mux has <1uV contact potential.
The 7011-S Mux card has <500nV contact potential.

All of these cards use the 7001,7002 switch frame.

For the 2700 family DMMs, here are some contact potentials

7708 Mux card, 1uV max contact potential, <500nV typical
7700 Mux card, 1uV max contact potential, <500nV typical

The relays in my 7708 card are a NEC EE2-3SNUH

[HighVoltage]

I am using the Keysight 34970A data aqusition unit with the 34901A 20 Channel Multiplexer Cards
These cards have relays and those contacts are rated as  <3uV Thermal Offset.

[Echo88]

The Keithley 7168 also uses "CONTACT TYPE: Solid state JFET switch"  Now i really want someone to open such a card and expose which components are exactly used and how the layout is done. 

The mentioned NEC EE2-3SNUH-relays dont mention emf, but are quite cheap at 2,5€:

http://www.mouser.com/ds/2/212/KEM_R7002_EC2_EE2-540906.pdf

[doktor pyta]

Have You seen this: Tektronix 148-0140-00 Latching Relays ?
Great stuff for DIY.

http://www.ebay.com/itm/4-pcs-Tektronix-148-0140-00-Latching-Relays-2400-Serie-Oscilloscope-Attenuators-/381691204065?hash=item58de9191e1:g:U0UAAOSwkl5XeYdo

If only they were available in large quantity ...

[David Hess]

For a while Tektronix was making their own relays and OEMing them for others.  They pioneered the relay design where all of the contacts are at the bottom allowing good high frequency performance.  In the 1970s, they were selling the relays for like $50 each.  Now you can buy equivalents, but not with the superior Tektronix pinout, for $2.

The low EMF relays are not suitable for switching while under load so rugged designs use two relays in parallel with the tough relay switching first to protect the contacts of the low EMF relay.

[Echo88]

Infineon (formerly International Rectifier) has a bunch of Photo-FETs with stated Offset Voltage < 200nV:
http://www.infineon.com/cms/en/product/power/solid-state-relay/photovoltaic-relays/channel.html?channel=5546d4624eeb2bc7014ef42feeed30d4#goto_producttable
for example: http://www.infineon.com/dgdl/pva33n.pdf?fileId=5546d462533600a401535683a6832928 7 Eur/piece at digikey

The older PVA30-type even has a Offset Voltage-diagram: http://www.irf.com/product-info/datasheets/data/pva30.pdf on page 3.

Thoughts or can anyone see pitfalls?

[TiN]

There is also low leakage Omron G3VM-41GR6 released recently, however En not specified

[Kleinstein]

In all cases the actual thermal EMF depends on the temperature gradients. This is why latching coils are so much more attractive. For the Optomos and similar, there is always a sizeable power consumption driving the LED. So the thermal EMF depends on the layout - cooling.

If used right they could be a good choice, though the on resistance might be a problem in some applications. For example the HP34420 uses quite a few of the Photomos in the input stage.

For a low voltage scanner one could even consider standard CMOS switches - the low power consumption helps to keep temperature gradients down.

[Echo88]

Time to drill a hole a teeny weeny hole in those Photo-FETs, attaching a fiber and controlling it from far away to avoid thermal gradients. 

[dr.diesel]

So I included a few different relays on my last DK order:

Omron G6AK-234P-ST-US-DC5
Panasonic TQ2-L2-5V
Potter & Brumfield V23079E1201B301

My interest here are volt-nut scanner options, so at 10V with a >10G input impedance we're in the ~1nA range.  Starting with the Omron relay I unsuccessfully attempted to measure the closed contact resistance in the 1nA range, but couldn't get reliable results with any of my equipment, k2450, k155.

So then I setup a simple test, Fluke 731B > Omron relay > k7510, all fairly sloppily connected via Pomona Minigrabbers.

731B measured directly = 9.999888, in circuit above latched = 9.999888, very repeatable unlatching/latching (about 20 cycles).  (2uV noise/bobble)

I've not played with the IO on the back of the k7510, but might be possible to utilize that for switching the relays in a smaller scanner setup.

More testing required obviously, but a glimmer of hope maybe 

[3roomlab]

So I included a few different relays on my last DK order:

Omron G6AK-234P-ST-US-DC5
Panasonic TQ2-L2-5V
Potter & Brumfield V23079E1201B301

My interest here are volt-nut scanner options, so at 10V with a >10G input impedance we're in the ~1nA range.  Starting with the Omron relay I unsuccessfully attempted to measure the closed contact resistance in the 1nA range, but couldn't get reliable results with any of my equipment, k2450, k155.

So then I setup a simple test, Fluke 731B > Omron relay > k7510, all fairly sloppily connected via Pomona Minigrabbers.

731B measured directly = 9.999888, in circuit above latched = 9.999888, very repeatable unlatching/latching (about 20 cycles).  (2uV noise/bobble)

I've not played with the IO on the back of the k7510, but might be possible to utilize that for switching the relays in a smaller scanner setup.

More testing required obviously, but a glimmer of hope maybe 

i was reading the omron 6A pdfs, and then looking at 125VAC resistive specs, for 274P @ about 0.1A, its over 50million ops, did i see correctly? that much? (the 234P seems lower at 15million ops?)
i am thinking, if to induce a fake sense of heating, what if a small 1w resistor is stuck to 1 side of the relay and some temperature diff is introduced? maybe then with this, a visible emf can be measured? then an actual uV per celcius can be found?
another idea, what if the relay can be wired paralleled with 1 in reverse to self cancel its EMF? (assuming only SPST operation). parallel also halves the contact resistance.
downside is 2 sets of coils to fire instead of 1. double the price for no EMF?

[ap]

i was reading the omron 6A pdfs, and then looking at 125VAC resistive specs, for 274P @ about 0.1A, its over 50million ops, did i see correctly? that much? (the 234P seems lower at 15million ops?)
i am thinking, if to induce a fake sense of heating, what if a small 1w resistor is stuck to 1 side of the relay and some temperature diff is introduced? maybe then with this, a visible emf can be measured? then an actual uV per celcius can be found?
another idea, what if the relay can be wired paralleled with 1 in reverse to self cancel its EMF? (assuming only SPST operation). parallel also halves the contact resistance.
downside is 2 sets of coils to fire instead of 1. double the price for no EMF?

In order to minimize th. EMF (if it is really critical) good practice always is to use one relay with two contact pairs in series in such a way that the EMF of both contact pairs cancels each other out as good as possible. Also, one would try to make an isothermal design (e.g. avoid heat sources arround...). 

[Kleinstein]

Using two contacts in series is a two sides thing. It can reduce thermal EMF if the two contacts compensate, but if thermal gradients are not well defined the errors can also get larger. So it depends on the type of relay. With reed relays using two in series is often a good idea - with a latching relay, a single contact could be better.

For measuring there should be no need to have a voltage applied. Just measure a nominal short with maybe a very small bias current (e.g. nA range) should be enough. The trouble could be that one might want to reverse the DMM / amplifier as well, but this adds some extra switches. So one could use a circuit with several switches in parallel an compare the voltage of using one switch at a time. So the result would be the difference in thermal EMF between different contacts. With several of the same type one should get a reasonable value - there will be a tendency to have different values anyway. I would expect that one might need a kind of nV meter - the normal 100 mV of a DMM might not be good enough.

[janaf]

Revitalizing bump for this thread....

Hardly any of the OP's questions where answered. No more input?

Any measurements?

Anything on IC MUX and emf?

I'm about to build a MUX for my voltage references.....

How measure?!!?

[Andreas]

Hello Janaf,

its a long time since I last heard something from you. (almost missing your comments).
I also want to built some MUX (since around 1.5 years now).

But this is still "work in progress"
In the end I will use that what I have already in the drawer:
TQ2-L2 latching signal relays which do not heat up since they are powered only some ms during switching.

With best regards

Andreas

[janaf]

Hi,

Today Iv'e posted here for the first time in about two years. Time flies.

I've got a bunch of TXS2SA-L2 that I was going to use. Some datasheet say 0.3uV emf while others just recommend the AgPd option, without specification. Mine are not the "-1" AgPd option.

I was probably going to switch them directly with some PCA9539DWR. Simple.

The attached pic is from the page:
https://na.industrial.panasonic.com/blog/thermal-electromotive-force-emf
"How To Offset EMF: Change in contact materials; Silver Paladium (AgPd) is an excellent material for contacts" Does not give any help really....

This overview:
http://www3.panasonic.biz/ac/e_download/control/relay/signal/catalog/mech_eng_chart_signal.pdf?f_cd=402060
mention GN and GQ series have AgPd contacts but no specifics.

High sensitivity seems to refer to low coil power only.

Anyway, if you want some TXS2SA-L2 I can send some!

But what do you mean by "almost missing"