Picoamps at kilohertz anybody?

I spent a good chunk of my career working for a major automatic test equipment (ATE) manufacturer, mostly designing and testing analog instrumentation. One module found on most cards in an ATE system is a parametric measurement unit (PMU), which measures DC parameters – voltage or current – on a pin of a device under test (DUT) in response to the complementary stimulus – current or voltage.  Because time = cost during IC testing, as in so many industrial arenas, “DC” must be taken with a grain of salt.  These measurements are the slowest single measurements made on a pin, so there is always pressure from the market to make them as fast as possible. Still, a few milliseconds is not uncommon. On the other hand, the vast majority of the market doesn’t need to measure much below a nanoamp.

Recently a project came through the door which, despite the fact that it came from a completely different sphere of electronic product development, sounded strangely familiar once we translated the client’s desires into electrical requirements. Basically it was a PMU.  Only trouble was, it required measurements of not very many picoamps to be made within not very many microseconds. Fortunately, CMOS input op-amps have come a long way since my ATE days. In particular, the OPA320 was a good fit for the application, not least because it’s available in a five-lead SOT-23 package with the critical negative input pin well separated from the others.

Five lead SOT-23-3 package

Five lead SOT-23-3 package

These are not exactly jellybeans, but the cost is reasonable in quantity, which was important because there could be a large number of PMUs in the complete system. It turned out that specifying and finding the high-ohm resistor that is in many ways the heart of the PMU was more challenging. After many days of circuit simulations of the major operational modes and the transitions between them, I was confident of my design and ready for layout.

With requirements like these, the circuit on the CAD screen is only half the battle. Layout, especially if the number of PMUs does turn out to be large, is another major challenge. We had the research and the practical experience to be confident it would work with standard mass-production PCB materials, but only with care. Guarding and shielding techniques, learned well in my ATE days, would be critical. Not only that, I developed a way of partitioning the circuit so that, as long as the most critical parts were located near the DUT, the great majority of parts in each PMU could be located a couple of feet away—yes, feet: far enough to support layout on the largest PCB flats commonly available.