PXI-2533 SSR Matrix and damaged crosspoints

[sphat]

Hi

I have a PXI-2533 SSR matrix card and PXI-4070 in my chassis. I am using NI Teststand to do resistance and voltage measurements on my UUT.  I am measuring a high impedance on a couple crosspoints on my UUT. By placing a 1kohm resistor on two crosspoints and using NI-SWITCH and DMM softpanel, I verified that the SSR will not close and it measures in the megaohms.

I would like to know what's the best practice to determine a damaged crosspoint? Does NI have a tool that will test the SSR automatically?  Also, is there a table list the SSR relay crosspoint to the board reference designator? Something similar to the 2815 relay replacement guide.

http://zone.ni.com/reference/en-XX/help/375472A-01/switch/2815_relay_replacement/

Thank you,

Sopheak

[Now_With_Underscores · Accepted solution]

NI has an internal final verification test tool to verify all modules are operational before leaving the factory.  We don't have a public tool, but it is easy to replicate a similar system:

1. Short all columns together and connect to DMM-

2. Short all rows together and connect to DMM+

3. Verify that with no relays closed, the resistance is infinite.  If the resistance isn't infinite, then you've got a shorted relay somewhere.  To determine which relay is shorted, you'll need to diconnect half the rows or columns recursively until the short is gone.  Once you've determined which relay is shorted, replace it and repeat step 3 until there are no shorts.

4. Close all relays one at a time and verify the resistance is in specification (1.4Ω for 2533).  If a relay exceeds specification, make note and then once you've cycled through all relays, replace all out-of-spec relays.

We don't publish solid state relay replacement guides, as these modules shouldn't fail if used within specification.  However, if you let me know which crosspoints are broken, I'll let you know their location.  We do offer an RMA service, if you'd like NI to service the module, but it sounds like you've got everything under control and know how to use an iron ;).  FYI: The 2533 uses Avago's ASSR-1510-503E

Of course, you'll need to understand why the 2533 has blown out solid state channels... something in your test system likely exceeded voltage/current specifications.

[sphat]

Thanks for your help John.

It sounds like a lot of work to check the crosspoints manually. Have you come across any test sequence that can do this automatically through Teststand?

Could you please send me a table of all the crosspoint and their locations. This would help for future problems that might come up. 

[Now_With_Underscores]

The difficulty arrises whenever you have a shorted relay, as anything in parallel with a short still looks like a short.  There are algorithms to programmatically hone in on potentially shorted relays.  Without an additional matrix, your algorithm can't guarantee 100% success.  If you have an additional 2xN matrix, where N >= DUT_Rows+DUT_Columns, then you can connect one column to each DUT row and column, and then your algorithm can programatically divide and conquer whenever it finds a short, allowing you do programmatically determine which specific relay(s) are shorted, albiet not guaranteed in one run... you'd need to recursively run the test until there are no shorts.  NI uses this general method to test our boards... however, we actually use tristating I/O boards to find shorted relays, as they are faster and can test in parallel.  Alas, we don't provide this code externally, as it requires unique hardware cabling for each specific NI module.  With the above discription, it wouldn't be difficult to reproduce our methodology. 

The crosspoint<->layout table doesn't exist... Here's a picture of the layout:

There is no public table correlating Kxxx reference designators, but I'm looking at creating one and will post back soon.

[Now_With_Underscores]

And here's the back (mirrored, as you'd see it when turning the board around):

[Now_With_Underscores]

Here's a table I just derived from the layout... provided as-is, as it has not been reviewed, but it should be 100% accurate.  Note that servicing hardware that is not officially serviceable is blah blah blah, etc... you sound like you know what you're doing, so be careful... if you use leaded solder, you'll need to mark the board as leaded so as to not cause confusion if somebody else reworks the board in the future.

Note that the 2533 (4x64 matrix) and 2534 (8x32 matrix) share the same PCB, so you'll need to convert the software row and column names in the attached file from the native 8x32 to 4x64.  To do this, conceptually chop the 8x32 matrix in half (two 4x32 matrices), then move the bottom half (rows 4-7) over to the right and connect to the existing rows (row 0 and row 4 are the same 'row 0', row 1 and row 5 are the same 'row 1', etc).  Then add 32 to the right 4x32 matrix's columns (column 0 is column 32, column 1 is column 33, etc).  Let me know if this isn't clear and I'll draw a picture.

[S.Braun]

Hi John,

I have two PXI-2533 cards with three destroyed relays.
Do you have some updated Excel sheet as it is not totally clear to me how to transform the given data for the 8x32 matrix to 4x64.

Thanks & regards

Sebastian

[Now_With_Underscores]

Sorry on the delay... I'm not actually an NI employee anymore, but still check in every once in a while to do some work for free ;).

The relay reference designator is listed in the NI Switches Help file, under the specific device.  Each particular topology is listed in that same Help file.  That should be all you need to correlate a particular topology's matrix to its actual mechanical relay position on the PCB.