ESD protection for pH measurement electrode input

Question

I want to puild a pH meter, which uses a normal glass electrode. Such an electrode behaves as a pH dependend voltage source (<+-500mV) with very high internal resistance (10MOhm - 1GOhm, depending on the electrode). As a consequence, the voltage measurement needs to be performed with very low current. Further, I read that current flow can have a significant impact on the lifespan of these electrodes, so keeping it as low as possible is crucial.

I looked around and found a few promising amplifiers with input currents in the femtoampere range - e.g. the INA116, LMP7721, and LMC6001.

However, I'm wondering now how to protect these amplifiers against ESD, since the device will have to pass standard EMC tests for household appliances. Some of the amplifiers have limited protection built-in, but as far as I understand, this is not enough to protect against the tests that are performed on external device connectors. Adding diodes to the power rails or a TVS diode seems to be out of the question due to the substantial added leakage of even low-leakage types.

However, since the voltage source I want to measure has a very high resistance anyway, would the amplifier be adequately protected by a very high resistance (e.g. 10MOhm) series resistor? Or do I have to use a connector / electrode design which mechanically prevents direct contact?

Answer 1

I've done some low leakage ESD protection recently. The scheme I used was fairly simple.

^{simulate this circuit – Schematic created using CircuitLab}

It's hard to show in circuit lab, but there are only 3 components here. 2 0603ESDA bidirectional clamps, and 1 SMAJ14CA bidirectional TVS diode. The TVS has too high a leakage current on its own. The birectional clamps of a leakage current of less than 100pA typical. They keep the TVS out of the circuit until they clamp, and then the TVS takes over. It satisfied IEC61000-4-2 Level 4, criteria B.

Answer 2

I don't know about PH sensors. I'd start by "cheating" and see how other people had done it. If you do end up needing a low leakage diode then you might want to look into using a small transistor. The c-b diode (open emitter) is low leakage (1-10 pA) but slow (100's of ns) The c+b - e (diode connected transistor, c/b shorted) is faster, but has a lower breakdown voltage. I've measured a few 2N3904/6's. and Bob Pease recommends the 2n930, 2N3707 and 2N4250. But you won't find these leakage number on the spec sheet. (For B. Pease reference search the web for B. Pease "Bounding and Clamping".)

Answer 3

You can use a simple RC network. A 10 meg resister and a 100 pF capacitor. The cap must be low leakage, a C0G/ 100v ceramic will do. With the right current limiting the op-amps own diodes will protect it.

Answer 4

For reference: I ended up using the INA116 because it was designed specifically for this application, and because of its advertised built-in protection. For the ESD protection I first put a large value resistor in series with the inputs, similar to user68868's answer, but without a capacitor. In theory, the resistor should already limit the current to a tolerable level for the Op Amp's own protection circuitry, but I was concerned a high voltage ESD could also just arc over the resistor. To try protecting against that, I ran a guard trace between the two pads of the (0805, iirc) resistor, and didn't cover it with solder mask, the idea being that the discharge would then rather arc over to the guard rail. Then I put normal ESD diodes on the guard rails to protect those.

I'm sure I read about a technique like this somewhere, but I can't remember where that was anymore, and I also can't vouch for its effectiveness because this version of the circuit was never ESD tested. However, I thought it may be a good idea to have it here as inspiration :)

Answer 5

Quite old topic but I'll add my thoughts here. I'm using LMP91200 which has a guard net (same potential as input INP pin) so I'm considering adding resistor in series and C0G capacitor between INP and GUARD. Then some ESD diode to GND from GUARD and VCM pin. I was also thinking about gas discharge tubes. Littelfuse claims they have <1pA leakage current.