A common problem users face is how to decide what matrix size you should for an application. Cost is critically sensitive to matrix size - but so is performance. The larger the matrix the more loading the signal lines will experience unless additional switching is included to reduce the loading.
A common misunderstanding is that a matrix should connect test resources on one axis (Y) and UUT connections on the other (X). This usually results in an inefficient switching system design - one where the the matrix is over sized compared to what the matrix really needs to be. The matrix size and cost is usually directly dependent on the crosspoint count, the size of the X axis multiplied by the size of the Y axis.
A much more efficient approach is to place both the test resources and the UUT connection on the same axis and use the other axis (Y) to provide the connections.
This methodology requires the X axis to have a size equal to the number of UUT connection plus the number of connections to the test resources. The size of the Y axis is then determined not by the number of test resource connections but by the number of connections required at the same time. It is very rare for all the resources to be connected at the same time so typically this can be a relative modest number. For 4-wire measurements using a DMM for example a x4 Y axis is enough to do 4-wire measurements, but permits no other connections at the same time. For most test system applications x6, x8, x12 or x16 is an adequate choice, Pickering Interfaces offers x6 and by x12 matrices to allow users to have finer control of matrix size than would be the case if the Y axis size increases in binary steps (x4, x8, x16). Users can make a cost based decision on how many concurrent tests are going to be performed, the more it is the faster the system test could be but the higher the cost.
For applications requiring high Y axis counts LXI is generally a better approach than PXI because the matrix size can be increased without the cost of interconnection cables and the overhead of the PXI module infrastructure and chassis. Our large PXI matrix (BRIC) designs in PXI offer the ability to avoid interconnection costs but take more room in the chassis. Find more articles on our BRIC large PXI matrices here >>
The only disadvantage of using X connections for Y axis is that each path is connected by two relay contacts in series, so some paths could have a higher path resistance. Others though will have shorter paths because the connections are relatively close together, consequently reducing copper losses. In large matrices copper resistance is by far the dominant contributor to path resistance, not relay contact resistance.
On Pole Switched matrices, we strongly encourage users to use X connections for test resources - find more information on Pole Switched Matrices here >>.
To simplify cabling users may want to design their system so that the UUT connections are on one connector while the test resources are an a different connector. This may only be practical on physically larger switching systems where there is more panel space for the connectors. Many of our LXI solutions provide this simple connectivity option.