Ultra-high input impedance DAQ

[Kryan9]

Hi!

As a small project I've decided to undertake during my break, I'm working on creating a very high input impedance, 8-channel differential DAQ to work on monitoring neural cells and their impulses (a group I'm associated with is trying to get them to create basic digital logic circuits). As such, I've been trying to figure out how to create as light of a load as possible on the cells, because I don't know how many will be in each "node", and I need some advice. My aim is for the DAQ to draw sub-picoamps as well as having sub-picofarad input capacitance.

My current plan is to incorporate an AD8639 op-amp as the front-end buffer in a follower configuration, or a instrumentation amplifier configuration, given +/- 5V supplies. Based on the datasheets, it should be doing only about ~100-500 fA at room temperature. I would have a trace from the negative input of each op-amp acting as a shield for the non-inverting input, going back to the pin of the input cable connector.

However, it does appear to have several picofarads of input capacitance, which seems like a no-go when the cells can only contain so much individual charge, even though the resulting rise-times are more than acceptable (the bandwidth is about 1 kilohertz tops). Is this reasonable to think it may be a problem?

I've been looking into bootstrapping the supplies using BJT current mirrors, and then using another op-amp to buffer the shield and supplies to reduce the capacitance. Does anyone think this is overkill?

Full schematics and board design to follow.

[SArepairman]

good luck on the sub picofarad probe. This would require an active device.

Active oscope probes are 2pF maybe, good ones.

 You are essentially looking for an electrometer on the tip of a needle I believe.

I think these types of probes are very expensive (typically used for high bandwidth testing).

Your daq can at best read the output of several of these probes. You can't have any reasonable length of wire for sub pF loading

If it maybe possible to monitor some other change in the cell that is associated with electrical transmission, as not to need a multitude of expensive active devices?

[SArepairman]

maybe something like this?
http://www.eevblog.com/2012/04/01/eevblog-263-extech-lp100-laser-test-probes/

[Kryan9]

maybe something like this?
http://www.eevblog.com/2012/04/01/eevblog-263-extech-lp100-laser-test-probes/

Are you being serious?

My original plan was to use a mix of shielded cable for each differential input running from the probe tips on the cell culture to the input connector, but you have a good point - if the input leads are more than a few inches long, it really doesn't matter what my AFE is, as the capacitance in the wires will dominate. I might need to rethink how I'm probing this and consider making it a two-part system, with the main DAQ system having a connector that streams power and ground to an active buffer that is inside the probe itself.

[Vgkid]

Would it be possible to use an lmp7721 as the buffer, 3fa input bias.

[krivx]

Could you use a FET buffer with the gate leg as a probe? You can achieve very high input impedances and commercial probes using this topology can get ~1pF input capacitance.

If it's possible I would try to borrow (or rent?) an active scope probe for a few hours and see what signal integrity is like.

[Kryan9]

Would it be possible to use an lmp7721 as the buffer, 3fa input bias.

I think it would work, but it wouldn't simplify any of the work. The datasheet claims a 10 pF input capacitance, so it would still require bootstrapping of some sort, even though the bias current is less.