So, I'm working on a project right now to implement a large electrode sensor array for measuring bio-potential (128 x 128 sensors) that are in close proximity to each other (~0.06cm along horizontal and vertical). The signals themselves are relatively small, being on the order of tens to hundreds of micro Volts.

I want to select an arbitrary 64 sensor subset of these signals and route them to an amplifier circuit and then to an ADC circuit sampled at 250Hz (each selected signal has its own respective amplifier/adc path. Thus 64 amps and 64 ADCs, one of each for each selected signal). The configuration of selected signals only changes once every half minute or so, thus switching speed is not of paramount importance.

That said, I'm not sure how to go about selecting the signals to use properly. It doesn't seem practical to directly select 1 signal from over ~16,000 possible sources by having an enormous number of inputs to a multiplexer(s). I imagine (and vaguely remember reading some approaches like this) I could leverage the grid pattern to select from the analog inputs without massive amounts of logic, but I'm not entirely sure how I could go about doing that.

I would supply some datasheets to help out, but I'm still looking at this from a very top-down angle in order to get it conceptually straight in my head before hammering out specific components and such.

Thanks in advance for any help you can provide.

  • 3
    \$\begingroup\$ You'll need to supply a lot more information. What are the sensors? Links to datasheets? What voltage? What response time to multiplexing power, for example? \$\endgroup\$
    – Transistor
    Commented Sep 9, 2017 at 22:18
  • 2
    \$\begingroup\$ Do your sensors have an output enable feature? \$\endgroup\$ Commented Sep 9, 2017 at 22:30
  • 1
    \$\begingroup\$ "Arbitrary subset" is awfully vague, too. Do mean an 8x8 tile of adjacent sensors, or something else? \$\endgroup\$
    – Dave Tweed
    Commented Sep 9, 2017 at 22:38
  • \$\begingroup\$ @DaveTweed Not necessarily 8x8. By "arbitrary" I mean that they could be anywhere in the grid without pattern. Vladimir Cravero: No they do not, but that's because I'm trying to devise the sensor select scheme on my own. Maybe that's a possible solution though? I'll try and add a bit more detail to the question, I'm amending it now. \$\endgroup\$
    – Scorch
    Commented Sep 9, 2017 at 22:44
  • 1
    \$\begingroup\$ You are still not telling us what the sensors are and this matters. If they are two-wire such as RTDs it might be easy. If they are three wire (V+, GND and output) then it becomes rather different. "V" or "volt" as per SI units. Units name after a person are capitalised in abbreviated form but lowercase when spelled out. \$\endgroup\$
    – Transistor
    Commented Sep 9, 2017 at 23:52

2 Answers 2


If your biosensors have low low Zout, you can simply scan the contacts, one contact at a time with confidence the multiplexor input capacitance will be fully charged by the biosensor. Your scan time would be 1/(250Hz * 16) or 4,000Hz or 250uSeconds.

If your multiplexors (perhaps X and Y) and amplifier Cin total 30pF, you can have 10Meg Ohm biosensor Rout, producing 300uS Tau.

So, are your biosensor well described by 10Meg Ohm Rout, able to charge 30pF thru the cell wall to that 10mV or 100mV activation potential?

Note a single TAU of settling has 37% error. You need 90% (2 tau) or 3 or 4 tau, if you need 1% data measurement.

At 4 Tau, at 30pF, for 60 microsecond Tau, the biosensor R can only be 2 Mega Ohms.


After accepting the initial answer, here's another approach (it's actually the paper I mentioned vaguely remembering in the question). It's not quite an "arbitrary" selection like I'd put in the question, but it's pretty close. It goes up to pretty high density blocks (23x23 at maximum), but otherwise you can choose the input arbitrarily. Link is here for anybody who's curious about that method.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.