Relay with ultra low contact resistance

[Amr Wael]

I am designing a current measurement circuit. The circuit involves more than one shunt resistance in order to cover different ranges of current. I would like to access any of them using 1 channel and switch between the shunts using relays but the minimum shunt resistor I am using is 50 mOhm and the relays have contact resistance within 30 to 100 mOhms so the values of the contact resistance is going to completely change the milli volt values that I am sensing over the shunt. My question is what are the alternatives , are there any relays with ultra low contact resistance 1 mOhms forexample? I can't use transistors because I need guaranteed isolation. Thank you very much in advance!

 

[FvM]

The question makes no sense. A 50 mohm shunt must use 4-wire (Kelvin) connection to maintain the specified accuracy of the part. If you are multiplexing shunts, you need to switch the sense lines, too.

 

[KlausST]

Hi,

I agree with FvM.

Mind if you use relays for sense line switching ...they may introduce thermocouple voltages. There are dedicated "low thermocouple effect" relays.
I'd try to avoid shunt switching. Do error calculations about the benefit and drawback of shunt switching.


Klaus

 

[Amr Wael]

The question makes no sense. A 50 mohm shunt must use 4-wire (Kelvin) connection to maintain the specified accuracy of the part. If you are multiplexing shunts, you need to switch the sense lines, too.

Yes I am using a kelvin connection and I am switching the sense lines too but I have no problem with the sense lines because I can use Analog switch or anything But for the current path If I used a relay with high contact resistance it will affect the millivolt value developed on the shunt resistor.

 

[KlausST]

Hi,

But for the current path If I used a relay with high contact resistance it will affect the millivolt value developed on the shunt resistor.

Why?
Show your circuit.

Klaus

 

[Amr Wael]

OK Here is a representative schematic for what I am trying to do. The position of the shunts (R6,R10,R7 and R13) won't be always between the VCC and the Load. I want the circuit to work in low side configuration as well as the high side.

--- Updated Mar 23, 2021 ---

OK Here is a representative schematic for what I am trying to do. The position of the shunts (R6,R10,R7 and R13) won't be always between the VCC and the Load. I want the circuit to work in low side configuration as well as the high side.

i don't want to provide different measurement point for each Current range , just 2 Probes , that's my problem.

--- Updated Mar 23, 2021 ---

That's why I do care about the resistance of the relay's contact resistance because 100 mOhm will be in series with my 50 mOhm shunt

 

[danadakk]

Your range of shunts is 50 / .05 ~= 1000 or 10 bits.

So if you used a 20 bit A/D would that be enough resolution and range just using one
.05 shunt ? Kelvin sensed of course.

This will give you ~41A range, and a resolution of ~39 uA. The onboard Vref is good to +/- .1%

Regards, Dana.

 

[Amr Wael]

@danadakk I don't want to use an ADC at this point , the circuit is a part of a system that is going to be centrally controlled through an MCU which is going to transform it digital later. I want to keep the signal analog at this stage.

 

[FvM]

The schematic clarifies that your statement in post #1 is incorrect.

the values of the contact resistance is going to completely change the milli volt values that I am sensing over the shunt.

Actually the contact voltage drop does not appear in the measurement. Of course, the contact resistance (as well as any trace and external connection resistance) causes some losses and load voltage reduction. It can be handled by using a relays with suffcient current rating.

 

[dick_freebird]

Check out relay mfrs like Pickering, high end test
relay matrices are their thing. They sell the piece
parts too.

The contact resistance rating is likely "sandbagged"
to encompass wear and aging. You might find within
a bunch of relays, better and worse to hand-down-
select. But knowing how they age might be an issue
if this thing you're building has to stand up repeatably
over time and abuse.

Gold contacts are nice for initial resistance but not
so good for abuse and contact-cycles. There's a
variety of contact materials that make various
trades.

Depending on the application and how you want
to architect the resistors and their switches, you
might find that logic level MOSFETs or eGaN FETs
provide a superior on resistance and reliability.
These can be had down to single digit milliohms.
You'd have to assess any leakage concerns but
at the low voltages mentioned (did not see anything
about the other end of the range) that might be no
concern (to a point).

 

[KlausST]

Hi,

I just see some kind of "3 wire" measurement. This way the relay contact resistance causes no problem, but still trace resistance may cause errors.

Klaus

 

[Amr Wael]

@FvM When switching on the relay , the resistance between the contacts will be in series to the Shunt resistor and the load. I don't understand why won't it's effect be similar to the effect of the PCB Traces that are also in series to the shunts. @KlausST yes there is a third wire but the current through this third wire is very low so they are still considered series same as the traces. I have tested a similar circuit on a PCB and literally any series resistance had a horrible effect even 2 mm Difference in trace width cause 300 mA error when measuring 1 Ampere on a 5mOhms shunt. Here luckily the shunt values are higher but still the resistance should be having an effect.
Anyway I have decided using multiple channels for each Current range/Shunt.

 

[KlausST]

Hi,

From your post I guess you did not understand 4 wire measurement correctly.
Show the PCB layout with the 300mA error.

Also read application notes about precision current sensing.

Klaus

 

[Easy peasy]

Put the shunts and relays in the gnd side - with the shunts going to gnd - then you can easily do gnd referenced 4 wire measurements that will be accurate ...