How to test PXI switch matrix

[johngardner58]

Hello,

I have a PXI-2530B and was wondering is there an accepted practice for verifying relay viability apart from just looking at relay counts?  I want to know if I've damaged a relay.  Based on what I've read so far,  I would have to create my own test harness connecting the switch matrix to my PXI-4071 DMM to check the reed condition.   Is there another way.  Has NI considered adding a self-test for relay condition?

[Now_With_Underscores]

Hey John,

The 2530B uses reed relays, which typically don't increase in resistance very much until they reach end of life, at which point the resistance rapidly rises (aside: this is different from electromechanical armature relays, which typically have erratic resistance as they reach end of life).

The 2530B does not have an onboard resistance checking mechanism, so an external resistance measurement device is needed.  Any DMM that can perform 4-wire measurements will suffice (4065, 407x, etc).  You will need to create your own test harness.  I recommend two harnesses for maximum coverage in a minimal amount of time with minimal relay cycles.

The first harness ties all rows together to the DMM+/Sense+ leads on the DMM.  All columns tie together to the DMM-/Sense- leads.  All relays start in their open position.  Close relays one at a time and measure the resistance.  Replace any relay with a measured resistance that exceeds the end of life value (3Ω for 2530B). 

Thanks for your feedback regarding a built in self test feature to ensure all relays are functional.  The NI-SwitchBlock contains a built in self test that can identify failing relays, but the 2530B does not.

[Now_With_Underscores]

Disregard post above... I ran out of edit time ;).  Here's the entire post:

Hey John,

The 2530B uses reed relays, which typically don't increase in resistance very much until they reach end of life, at which point the resistance rapidly rises (aside: this is different from electromechanical armature relays, which typically have erratic resistance as they reach end of life).

The 2530B does not have an onboard resistance checking mechanism, so an external resistance measurement device is needed.  Any DMM that can perform 4-wire measurements will suffice (4065, 407x, etc).  You will need to create your own test harness.  I recommend two harnesses for maximum coverage in a minimal amount of time with minimal relay cycles.

The first harness ties all rows together to the DMM+/Sense+ leads on the DMM.  All columns tie together to the DMM-/Sense- leads.  All relays start in their open position and one initial resistance is taken to ensure that none of the relays are stuck closed.  Close relays one at a time and measure the resistance.  Replace any relay with a measured resistance that exceeds the end of life value (3Ω for 2530B). 

The first harness will not work well if any of the relays are stuck closed because all subsequent measurements will indicate the closed relay resistance in parallel with whatever the resistance of the relay under test is... this is always bad, but particularly bad if the test relay resistance is, for example, 199MΩ (i.e. the relay is stuck open).  Hence the need for a second harness, which connects an Nx1 multiplexer to the N rows and another Mx1 multiplexer to the M columns.  The DMM+ lead connects to the common on the Nx1 mux, while the DMM- lead connects to the common on the Mx1 mux...

...instead of the second harness topology above, you can use a digital I/O module and tie each pin to one of the NxM matrix leads.  You can then use the DIO board to sweep through and find out which relay is broken (I like to call this the "you sank my battleship" approach).  It's much faster and doesn't require any additional multiplexers.

...or instead of the second multiplexer, you can manually use your DMM to see where the stuck closed relay is.  Just open all of the relays and then manually probe until you find a short.

Thanks for your feedback regarding a built in self test feature to ensure all relays are functional.  The NI-SwitchBlock contains a built in self test that can identify failing relays, but the 2530B does not.

[johngardner58 · Accepted solution]

Thanks for your quick response.  

I thought as much but I wanted to be sure.  While I know the reeds have a long life-time, it is possible to blow them or damage them.   And since there is no calibration procedure for the switch matrix,  I was wondering how to verify that this component of the system is still usable.   We are using the PXI system as a means of testing and calibrating an in-house pacemaker tester with 112 connections.  We are partitioning the 2530B into 5xBlocks,  1xBlock and 2xBlocks in independent mode and then using a custom in-house test adapter to provide all the interconnects.  The DMM and Multi-IO are spread across these partitions.

I would recommend for this reason that future matrix systems have a rudimentary way to self-test that relays are below 3 ohms (or whatever applies concerning a particular switch matrix).   Pickering is advertising this for cards supporting Birst. 

[johngardner58]

BTW,  we love the 2530B.   Especially the independent mode.  By isolating it into various partitions it becomes super nice.   I haven't seen anything comparable.

[Now_With_Underscores]

Hey John,

We definitely value your feedback.  Thanks for taking the time to let us know which features are most important in your application. 

It sounds like you're designing a customer PCB for the 2530B.  If this is the case, you can maximize reed relay life by adding 100Ω series resistors to any voltage or 4-wire resistance measurement.  This will reduce parasitic/discrete inrush current and flyback voltage and thus increase reed relay life.  For current measurements, I recommend turning off the power supply before switching for maximum life, or if this is not possible, I recommend adding flyback protection diodes reverse-biased and in parallel with each relay.  While the above circuit elements will greatly increase relay life, you are correct that all mechanical relays will eventually fail.  If you have any leftover sub-multiplexers on the 2530B, you could easily create an in-production relay self-test multiplexer using the methodology I previously described.

Let us know if you have any questions.  Have a great day!

[johngardner58]

Thanks!  All switching will be cold.  DMM impedance measurements parameters will be chosen so that current used is limited.   The tool under test generally has very low currents as the voltage is low (20 mV across a 500 ohm load..  Our constant current sources are less than half a milliamp.   Also the number of times we actually switch is relatively low compared to most of your customers.  Since the tool under test is an in-house device we have a limited number and calibration occurs once a year.

But we expect the PXI system to be used in other scenarios as well and so when it comes to test time we need to do a check.

I am not so worried about the normal switching but abnormal behavior like the test engineer swapping our 2 50 pin cables and driving +15V @ 1 amp through the matrix.  This supply is not connected throught the matrix normally but is direct wired to an Analog Input of an PXIe-6363 via a voltage divider.   However if someone swapped our cable then this would be connected to the matrix.

This should never happen, but it would be really nice to be able to do a health check on the system prior to running tests.