# How to reliably switch a DC analog signal that can vary from 10mV to 12V, and 1mA to 100mA

Preface: I'm a Controls Engineer, not an EE.

I currently have a system that tests resistance through a Device Under Test using voltage drop. I am using 12 Phoneix Contact PLC-RSC relays to alter the test circuit to perform various tests. I have had some periodic false failures and bizarre test values. I believe I have isolated this problem to inconsistent contact resistance in the relays. I also previously replaced one relay that was reading consistently high across the contacts (around 1 Ohm).

So to my question, is there a type of transistor/SSR/MOSFET etc that can switch a voltage this low and a voltage range this wide (10mV - 12V), and a current this low 1mA to 100mA?

I have been googling for a while but I am really not sure what terms to be searching for.

• I'd like to ask, have you considered the use of mercury wetted reed relays? Or tried them, at all? Semiconductors are a bit more complicated (as you found out.) So I just want to make sure you've explored something that comes to my mind, first.
– jonk
Commented Jan 10, 2018 at 21:19
• Digikey has good search options, an example search for SSR with 'AC, DC' out, 10mOhm to 1 Ohm Ron in DIP packages: digikey.com/short/q8z8pz Commented Jan 11, 2018 at 0:55

You can try to find the best relays, etc. But the proper engineering way to measure resistance accurately and also detect contactor faults is to use a 4-wire (aka Kelvin) resistance measurement.

Here's a diagram from this website:

You run your test current through one set of contacts and then measure the voltage through another set of contacts. By contacts I mean separate relays, switches, and separate wire connections.

Let's say your test current is set to 100 mA and can hit a high of say up to 10 Volts, and your DUT is 1 Ohm. So you would expect 100 mV across your DUT.

If your relay is bad and adds an extra 1 Ohm, your current source voltage will go up 0.1V to maintain the exact 100 mA you set it to, so that extra resistance will not affect how much current is running through your DUT.

Let's say a bad relay connecting the voltage sense also has 1 Ohm extra resistance, well the current through the voltage sense is 0 so that extra resistance is ignored also.

Let's say your relay is super bad and cannot flow current. Then your current source will try to push 100 mA through the DUT and it will hit its upper limit of 10 Volts. We can set our 4-wire meter to give a fault indication when it cannot supply the test current we have set it to.

So basically, with a 4-wire measurement circuit, you can:

1) Ignore resistance in test probes, cabling, and switch contacts.

2) Determine when there is a gross fault in the measurement system.

Keep in mind you don't need a separate current source and voltage meter; there are plenty of bench meters that do this 4-wire sense.

This webpage has a very thorough explanation of 4-wire measurements. https://www.cirris.com/learning-center/general-testing/special-topics/40-4-wire-kelvin-testing

• I would love to do this, however the system is already in place and I have a PLC with single ended voltage inputs. To my knowledge this technique won't work if the 0V sense can't float. Correct? Commented Jan 15, 2018 at 22:53
• That is true. You could, however, use two single ended voltage inputs and subtract the readings to get a pseudo-differential voltage measurement. You'll have to check your measurement resolution and offset specs to see if this will work (and if it needs an additional offset calibration step), but it can work fine. Commented Jan 20, 2018 at 0:46

My impression > Problem: bad relay choice for application

Low current loads can cause oxidation failure without sufficient wetting current and give erratic results. For example all contacts that are rated to switch less than 2A MUST BE gold plated over silver alloy. but there is no spec on how thick this plating must be.

It is clear to me that the datasheet is rated for 3A and thus is NOT gold plated ( industry rules) and thus prone to oxidation at low currents.

( verified: Spec indicates contacts are AgSnO and min current is stated 10 mA which I suggest is not enough from my experience)

I had this problem on P&B power relays which had sense contacts which i used for remote feedback. Due to high failure rate with TTL logic level currents (Circa '77) I figured out the problem and resolved it with a solid tantalum capacitor across the contacts to ensure adequate wetting current but not too much energy to burn the gold plating rapidly.

Relays with a 3A rating can handle more surge current but at the expense of accelerated aging.

OMRON are much explicit about derating contact currents for different reactive loads and 1 Ohm is certainly a failure.

As Jonk suggested glass sealed mercury relays were very reliable and had no contact bounce and once very popular.

## recommendations

Choose relays with Gold plated contacts rated for < 2A

• or add 10uF 16V polarized tantalum caps across contacts with pullup R if necessary and very if contact resistance improves down to <20 mOhm.

• in future consider Automotive smart switches on circuit boards with opto isolation ( avail online with 1 to 8 channels) e.g. chips