Our approach on deciding the utility of an N-Up tester is based on the following items:
- The amount of testing that can be done in parallel versus sequentially with the existing measurement hardware.
- For each step, how long does it take and what equipment is used.
- The costs of complete new test stations (including annual expenses such as maintenance, floor space, HVAC, and so on) as compared to additional fixturing which shares the same test equipment.
Look at this illustration showing an example of sequential versus parallel testing for two test sockets. If some equipment can be shared, then you can test two parts at once. In this illustration, the measurement I/O used to run the step labeled Test2 cannot be the same I/O as needed for Test1.
Also, from this illustration, you see that when the UUT1 is done with testing, you have to unload that unit and load in a new one into socket 1. If you can finish that unload/load before Test2 is done on UUT2, then the operator can be busy with load and unload tasks without waiting for part of the entire test to complete. The test time is then “free”.
How many test sockets are needed to be able to have “free” test time? This equation provides an estimate:
The calculation is actually more complicated because much depends on the amount of the test that can be parallelized. But in the example shown here for the parallel test illustration, if you can perform the unload/load in the time it takes to do the step Test2, then total test time on the numerator is twice as long as the load and unload time, so you need two sockets. The illustration shows this, since there are two UUTs in action at any time.
Another factor in deciding how to parallelize is the fact that some equipment may be very expensive and you will be motivated to share that particular hardware across all the sockets, regardless of the impact to the overall test time.
All these considerations can be rolled up into an analysis of overall profits. If you make a profit, P, per unit and M is number of additional parts per year produced before parallelizing, and E is cost of the test equipment. Then, if P*M is > E, then it pays to enhance with extra test capability.