LMC6001: Recommended PCB Material for High Precision Applications

Kevin,

This external article written by Paul Grohe, Texas Instruments discusses PCB material and other important considerations for femtoampere amplifier layouts.

www.edn.com/.../Design-femtoampere-circuits-with-low-leakage-part-one

Hi Keven,

Ron beat me in tooting my own horn, but some basic info is there in the article, as well as the Keithley Low Level Measurements handbook, should be enough to get you started.

Basically, the things that are good for microwaves are good for femtoamps. Teflon (PTFE), controlled dielectrics, shielding/guarding, Teflon dielectric cables, sealed shield boxes, etc. The RF shields with the tight-fitting pop-lids commonly used for oscillators and RF decks work well.

The PCB material used for microwaves, like Rogers, can also be used. Mainly because the dielectric constant is tightly controlled at the panel level, unlike FR-4. So this helps control the stray capacitance contribution, as well as the dielectric absorption. Controlling these will control your settling times. One problem is some of these are ceramic based, and can be triboelectric.

We tried Rogers 3000 and 5000 and compared them to FR-4, and while they did reduce stray capacitance and leakage - up to 30%, the cost also went *way* up  - and only certain board houses can properly process these materials. Expect to pay 3-4x the cost of FR4.

Plain ole' FR-4 can work fairly well if properly guarded. Think in 3-D; as the traces need to be guarded in all directions. It can help to run guard via "fences" down next to traces to reduce "sideways" leakage from internal traces along with guard traces/shields underneath and above. The input trace should only see guards in all directions.

Be careful with routing out the boards. While it will eliminate the surface and cross-substrate leakage, it also exposes the now bare edges of the boards to moisture, which are closest to the critical nodes. The edges of the boards would need to be sealed.

The "big boys" like Keithley and Agilent still use up-in-the-air wiring on critical nodes. Remember you only need the critical nodes in the air - not the entire circuit. The lower impedance side of the components can be mounted conventionally in the PCB. You just need to analyze the cost between using exotic materials and manufacturing vs. some manual soldering with cheaper components/PCB's.

The key is to keep out the environment and movement. Moisture is the long and short term enemy. Seal or pot the boards if possible. Seal the sensitive circuitry inside a shell if the entire circuit cannot be put into a sealed enclosure. It may work fine in Arizona, but not so well in Malaysia during monsoons.

Obviously - minimize the number of other circuit in the area - and avoid running power traces or digital lines anywhere near the critical areas. Make sure there is a shield (GND) and a guard between them.