MCP6N16 Instrumentation amplifier - differential input impedance too low

[mrmp17]

I'm working with Microchip MCP6N16-001 instrumentation amplifier to amplify a signal of 0-1.5V with high input impedance by a factor of two. The configuration is very basic, as per datasheet typical circuits. Picture of the schematic below.

The problem is the amplifier seems to have a way lower differential input impedance (Everything is DC, so call it resistance) then expected - datasheets specifies 10^13 Ohms (table 1-1) and input bias / offset currents in the range of a couple 100pA. What I'm measuring is about 3.3Mohm and a current into the inputs around 0.2uA, depending on the input voltage. That's about 1000x discrepancy, looking at currents!

How I'm measuring this? I supply 1V from an (isolated / floating in relation to GND and VDD) adjustable power supply through a 1M resistor into the input pins. The output in this case is at about 1.4V instead of the expected 2V. Shorting the 1M resistor, output jumps to the expected 2V. I've done the input impedance measurement a few different ways, such as measuring the voltage drop on the 1M resistor (taking into account the multimeters 10M internal impedance) and calculating from there. I've also done some measurements with a SMU, looking at the differential input current. The measured input impedance always seems to end up at 3.3Mohm.

I've also done a common mode input impedance measurements (see schematic) which comes out as expected - I'm measuring about 160pA, using multimeter's 10M impedance as a shunt at about 3V of common mode voltage. Not sure how reliable this measurement is, but seems to be in the right range. The common mode leakage is low throughout the whole range, until the protection diodes kick in, above the VDD.

I'm measuring the same differential input impedance (within a few %) on all 6 channels on my board. I've also soldered up a dead-bug style adapter to test a fresh, never reflowed amplifier on a breadboard with minimal external components (see schematic note) - exactly the same results.

Flux residue leakage and esd / overheat damage come to mind but damaging 7 different ICs to have exatly the same differential input impedance? With one IC being soldered in a totaly different way? I don't think so.

IC's are from Mouser, so they should be legit.

I've opened a Support case at Microchip about this - their latest response (after about three weeks of waiting...) was that their internal team has not yet had the time to check this out, but the support guy I'm talking to did some tests and got similar results to mine. Their recomendation is to "work around the experimentally determined value".

Any ideas what's going on?

[Alex Nikitin]

Welcome to the somewhat unpredictable and amusing world of chopper stabilized amplifiers. In this particular case it looks like the amplifier dislikes high resistances on the inputs as chopper spikes create a problem. Here is what the datasheet says:

Make the resistances seen by the inputs small and
equal. This minimizes the output offset caused by the
input bias currents.
The inputs should see a resistance on the order of 10Ω
to 1 kΩ at high frequencies (i.e., above 1 MHz). This
helps minimize the impact of switching glitches, which
are very fast, on overall performance.

On the other hand

The capacitances seen by the inputs should be small.
Large input capacitances and source resistances,
together with high gain, can lead to positive feedback
and instability

You might try to search "chopper amplifiers" here on the forum, and find some enlightenment (or the opposite of it ) . 

Cheers

Alex

[mrmp17]

Interesting!

The input filter was inspired by exactly the two datasheet sections you've quoted - keep the impedance at high frequencies somewhat low, but not add capacitance directly to inputs to avoid instability. It is still somewhat higher due to the 4.7k resistors. I can try shorting those out to get the impedance at high frequencies lower and see if that has an effect on input currents.

But still, even with low input impedance, the problem remains. Connecting a SMU (a low impedance source) set at 1V to the inputs, I measure about 0.3uA. This is basically measuring the bias current of the positive input, isn't it? I can't find anything in the datasheet indicating much higher bias currents for some specific conditions.

The evaluation board schematic does show a 0.1uF cap directly on the input pins, which is interesting.

Update: Shorted the two 4k7 resistors and placed a 0.1uF cap directly on input pins. Nothing changes, bias current stays the same...

[MasterT]

I got similar results regarding diff input impedance, about  R = 468 kOhm mcp6N1-10-100.
Interesting, that microchip  is not alone who is chiting :
ada8237 ~ 1.2 MOhm.
"Zero-drift" is likely to be the cause,  charge injection even < 1 pQ at 100kHz would define input R in about that range

[mrmp17]

For anyone with similar problems in the future: the cause for lowe than expected differential input impedance indeed seems to be the switching of the chopper amp. See scope screenshot below, clearly showing the switching glitches at about the frequency specified in the datasheet. This was done with 1.5V at 500kOhm impedance on the input.

Calculating the average input current of the chopping from the frequency and specified input capacitance also produces results that are very similar to what I'm measuring.