USB picoammeter resolves 20pA-2mA

Got a nice note from Myron Lee over at Ix Innovations. They have a new picoammeter Myron described as “the first major picoammeter innovation since the Keithley 6485.” The PocketPico costs $1,199, features include:

• Range 20pA-2mA Accuracy ± 0.5%
• Min resolution 100fA
• Max read speed 15,000/s
• Ranging Continuous, SafeRanging
• Software interface SCPI subset, PocketPico Reader
• Input voltage range -1V to +4V DC
• Measurement method Unidirectional current sink
• Analog input BNC
• Data & power USB 2.0
• Size 2.7in × 1.4in × 3.7in (69mm × 36mm × 94mm)
• Weight 8oz (227g)
• Included accessories Low-noise BNC cable, USB cable, custom travel case, installation instructions, user manual
• Operating system Windows XP or greater

The pocket pico website has a link to the theory of operation that explains how the unit works. I sent the document to several analog guru friends and here are their comments:

Paul Grohe-National Semiconductor– It’s a “Log Compression” ammeter, or an ammeter with a logarithmic response. I have pondered them… Simple to implement if you have a good look-up & corrections table and a processor. But since the output is log and ADC has linear steps, the ADC resolution creates variable accuracy and resolution over the measurement range, creating “steps” in the accuracy. “Ranging” would still be beneficial, unless someone comes up with a 64 bit A/D. Cute device, though. But I’ll stick with my “bulky” GPIB Keithleys (who is being bought-up by Tek, if ya’ did not know). The Keithleys use a conventional “op-amp” resistive feedback I-V converter (trans-amp) circuit, but with an output voltage swing of up to +/- 100V for dynamic range. The “transients” he speaks of are when the Keithley switches between feedback resistors for the sub-ranges (charges in the stray C). The biggest jump is the switch into the mA range (as they point out) since they are kicking out the gig-ohm resistors. If you have a delicate circuit, you lock the range so it does not switch. I learned that a long time ago….

Barry Harvey-Intersil– Nice way to eliminate range switching, by using a log circuit.  Nothing wrong with it as long as calibration uses temperature sensing near the fet.  The gate voltage will be very variable with temperature in sub-threshold mode. The circuit will of course be sensitive to source capacitance with respect to op amp stability.  I hope they have an inner high-frequency feedback path to bypass external capacitance. The output will be much noisier than a shunt resistor due to the high-gain operating point of the fet.  The 1/F noise of the fet will be worse than a shunt approach. One nice feature is that the voltage burden on the source is constant and can be made small, like 10mV.  On the other hand this increases noise even further. I hope they use a chopper amplifier (hey Intersil makes them now). I’m not sure why there was a problem in the linear shunt approach.

Glen Brisebois-Linear Technology– This should work in principle, but I would have expected the compression to be square root, not logarithmic.  One error: his feedback is to the -input.  It should be to the +input because the FET is inverting. Leakage and temperature should both be problems, which he may be compensating for but not showing how. Oh, and another thing, he’ll definitely need compensation because the FET is in gain in the loop.

So there you have it, a neat, compact, picoammeter that works in a clever way, as pointed out by my analog pals.