[SOLVED] Probing Test Pads in 0.35um

[ja2bo2t]

We are working in a 0.35um TSMC technology and I am in the process of designing a test chip that is to be used to characterize some analog amplifiers used in the actual design. In the actual design, these amplifiers require about 40 I/O pins due to digital configuration bits, digital power, analog power, and reference voltages. For testing, I was considering using a probe card with the ~40 pins put in a peripheral around the amplifier and moving from one amplifier to the next for characterization. Is this the normal application of a probe card? Do they have pre-made probe cards for standard footprints? How expensive would a custom probe card for this application be?

Also, is there an easier way to accomplish the task of getting access to a large number of I/O?

 

[erikl]

There are no such things as standard footprints. The standard procedure for testing prototypes and low volumes is to use a probe card with independently adjustable probe needles. For testing high volumes, a custom probe card pays off.

 

[ja2bo2t]

Are you aware of off the shelf probe cards that have adjustable probe needles? I haven't been able to find such a product as of yet.

 

[erikl]

I've heard about. Here are a few links from the **broken link removed** you could try.

 

[dick_freebird]

For characterization, chip-on-board could be more
economical than a probe card especially if you have
no further interest in the design. A production part
of course will need probe hardware but this does
not sound like that. COB offers much less inductance
and more opportunity to put close-in support stuff
like op amp measurement loops, etc. Of course it
helps if there's a nice lady with a wire bonder in your
lab....

 

[ja2bo2t]

I like the idea of chip-on-board. I think I'll invest a little more time in combining all the common amplifier signals to common pins, and wire bond these to a board. I can then probe each amplifier's input/output (differential, so 4 pins) with standard probes for characterization. As long as noise is not an issue with the common lines, I think this will work. Thanks for the help!