In the last post, we saw how a custom dedicated PCB allows powered curve tracing by grouping pins until the total pins to test were less than the pin count of the switch matrix. Now we consider an even more extreme situation where the device is more than 2x the pin count of the device, and, even using the reduction method, you still end up with more testable pins than the tester.
Multiple DUT board designs
In some situations, it actually makes sense to split the test requirement into 2 or more PCB designs, and either get more sockets or move the sockets from board to board. Fortunately, RTI pogo pin sockets are easy to remove and attach to a different PCB with simple tools. The socket could be universal or dedicated, but the DUT board layout makes the whole solution dedicated to a specific part code or family with similar power and IO pins.
Take the example of a user with a 108-pin test system. They have a BGA device that is 225 pins with a 15×15 array device but only 64 of the pins are used for power leaving 161 IO pins that need to have testable access. In this case, a design is started by bussing the VDD1, VDD2, and GND pins into 3 groups and dividing the IO group into 2 parts each with ~80 pins. But some pins could be common between the two groups, especially pins used to set the state of the device like control pin sets. For these, a decision is made on how to split the IO group based on some knowledge of the device function and what needs to be biased for powered curve trace to work optimally.
A variation of this for the MT century is a single fixture but with multiple connectors. A strategy can be worked out where one connector at a time can be tested while either leaving the others open or using 3-pin jumper systems to manually configure the remaining power pins. Shorting jumpers can also be used to ground all pins on unused connector positions
Pros
- More test conditions are possible including powered curve trace
- Good solution for ultra-high pin count >2,000 pins when testing on 1,080 pin MegaTrace
Cons
- Higher cost to design 2 boards, more NRE to figure it out
- More effort to test the device. Laborious to swap socket between boards which means user will likely focus only on the problem area and ignore other pins not reported as failed leaving a possible blind spot.
Contact RTI for technical details and pricing on the universal shorting boards discussed above.