Micro Power (< 125 uA) OpAmp Questions

[ruku]

So I've been working on fixing a design for a remote sensor at work. Right now, a voltage based output sensor is being fed into an ADC on an AVR microcontroller in a rather brutish manner. A +/-5 volt supply powers the sensor, which outputs a +/- 2 volt signal (frequencies < 100 Hz). The +/- 5 V supply is divided with resistors to provide a virtual ground at -1.65 V. This is used at the microcontroller, so that voltages seen by the ADC are -0.35 to 3.65 V. I'm not sure this is a very good design, and I'd like to fix it! The problem is that the remote sensor is being powered by solar panels, so I've been given a VERY tight power budget! I can't take more than 2 mA in total.

Naively, I'd use a non-inverting ideal amplifier to scale the signal to +/- 1.65 V (gain of 1.212), and then a summing amplifier to add 1.65 V to put the signal between 0 and 3.3 V. I'm sure this could be accomplished with one amp though, I'm drawing on all my circuit theory here. One thing that I've learned from my (little) experience though, is that nothing is "ideal", and the further specialized something is, the more quirks there are to deal with.

I'm looking at two chips by National Semiconductor right now. (For such a weird chip, I didn't start on Mouser and DigiKey. National had a section devoted to micropower amplifiers, so I started there.)

http://www.national.com/ds/LM/LMC6462.pdf
http://www.national.com/ds/LP/LP2902.pdf

A few questions about them.

• What am I missing? This seems too easy!
• The input bias current is extremely low (20 nA, picoamps). Is this a good statistic to have low? How should it influence my design?
• Are there any OpAmp circuits which scale and bias the signal with one amplifier?
• Are there any analog multiplexers which would take my +/- 2V signal so I only need one set of amplifiers?
• Am I going about this all wrong? Are there any low power DACs that operate at +/- 5V and have built in scalers?

[scrat]

Beware of the output voltage swing of the LP2902, it seems too tight for you.

[mikeselectricstuff]

For low-powered sensing applications, you typically don't want a low-power opamp, as with low power comes slowness, so you have to have the opamp and sensor powered for a long time, and maybe also the MCU out of sleep mode for a long time
It is normally better to use a faster opamp and power up the sensor and amp for the shortest time needed to get a reading, then power down.

[ejeffrey]

What is the output impedance of your source?

Gain + level shifting is trivial to achieve with a single inverting op-amp and three resistors but the input impedance will be set by your input resistor and should be large relative to the sensor to avoid voltage divider errors.  A big advantage of this is that the op-amp inputs are always close to the virtual earth, so you can easily tie a +/- 2 volt sensor input to a 0-5 V rail-to-rail op-amp.

Input bias current is important for very high impedance circuits.  In a non-inverting op-amp circuit, the bias current from the positive input must pass through the sensor, which is a problem with capacitive sensors -- they require a resistor in parallel to the sensor, which causes an RC high-pass filter and a DC offset.  The lower bias current the bigger the resistor you can use.  For both styles of op-amp, the bias current from the inverting input flows through the gain network causing an offset voltage, but only very rarely is this less than the amplifiers voltage offset.

++ on power switching the op-amp.  If you get a 5V rail-to-rail op amp you can probably power it directly from a digital output pin on your AVR.  With a 100 Hz bandwidth you don't need to sample any faster than 1 kHz, and an ADC conversion can probably be completed in 10 uS -- if the op-amp is only on 1% of the time you hardly need a low power device.  The only thing to watch out for is SCR latchup if you leave the sensor powered up all the time -- when the op-amp is off, feedback will no longer keep the input voltage at ground, and some current will flow through the input protection diodes.  Check the op-amp specs and pick an input resistor that limits the current to a safe value.  This is a good idea anyway as it reduces the current consumption.

Is it possible to run the sensor on 0/+5 volt instead of +/- 5?  If so, you might be able to power switch the sensor as well.