Parts for reading Bioelectrical signals

while searching for a bunch of amplifiers to do non-implantable ecg reading i stumbled across a nice part that qualifies for implantable stuff, too.
it is pretty much a "everything in one chip" sollution to get feedback from nerves into a microcontroller. that means. electrodes on one end, μc on the other, done. it contains the entire analog frontend that would usualy fill up a box.
this friend comes with 1 channel input, has a bunch of bigger brothers that can handle 2, 4 and 8 channels.
it comes in packages that are small, but still possible to solder without expensive reflow equipment.
they also are reasonably low power, theoretically run several days on i tiny rechargeable lithium battery.
in short. i am hyped up about those things, they could be the ticket to real interaction with active-implants. and up to 8 channels would be quite a lot of data you can use.

i ordered a whole bunch of samples to build some test boards. they wont ship cause they are backordered. but i hope to get them end of january.
comments or previous expericences are welcome :)


  • Don't know if this is any help, but I know that amplification of these signals can be tricky and on this years ccc I heard about this little thingy:
    Could be worth being checked out
  • @Ben , the chip includes all the frontend, including programmable amplification. all it needs to operate is a bunch of capacitors, and 2 resistors to provide protection of the human body.
    i made a small breakout board for experimentation but it is at the limits of what i can manufactor ( given the tiny size). so i might need to have it made in a real pcb factory. so far, the part is very promising, it's pinout is really nicely done too. given an industrial production, an implant with that chip would be quite tiny. even DIY boards fit on very small space.
  • @ThomasEgi: if you decide to get a board fabbed, I'll split costs with you for some.
  • @unqualified that would be great, i already have someone to split with, so that would make 3.
    i think will get a premade lab-breakout (16€ :( ) and wire it up, just to see if i can avoid a bunch of parts by using battery supply etc. once i settled on a circuit , i will do a propper layout.
    since the pcb's end up really small. it would be best to order 50 or so at once, the price difference compared to just a few is almost nil. so, we better have a good design to on the first order.
  • Sounds great.
    Only two questions there:
    1. have you actually amplified a real source (like eeg)? I gather the noise can be a real bitch.. not saying you couldn't, just curious
    2. what kind of implant would you build with this? again, not questioning you, just being curious
  • 1. i have not yet soldered the chip anywhere. but it comes with bild-in filtering and right-leg driver and crap to get noise out. when implanted we actually get a pretty clean signal source with little noise, a lot less than you get outside the body.

    2. absolutely no idea. but with that, and some regular electrodes for stimulation you have a basic input output system between you, and your implant. think of something useful that can communicate with a button and buzzer, morse-code being a great example.
  • Perhaps this can be used for a nervous system - nervous system communication system? 
  • communication is pretty much the whole idea behind this thing. once it is working, others can build upon this 'module'.
  • @ThomasEgi:  Would it be possible to use multiplexing to increase the number of channels we can read?  I've seen some multiplexers get pretty small.
  • there are indeed analog multiplexing solutions. altho one problem might be that, if we use existing front-ends they may be optimized for reading one channel, which often includes filters and a sampling rate appropriate for reading only one channel. so if you multiplex with those, you can switch between the input channels, you probably won't be able to read them all at once. it's still a good idea to have multiplexers for a MEA. so you can have like a 64 or 128 electrode MEA, and pick 4 or 8 of those where the signals you are interested in are strong.

    Multiplexing for writing/output is easier, as there are no pulses you can miss. so a single electrode driver can drive a whole bunch of electrodes at the same time, and even a greater range of outputs if you don't need to drive them all at once. there are a couple of good parts available and i already have 2 analog mux/demux in use in the concept of the electrode driver circuit.
  • IanIan
    edited October 2013
    @ThomasEgi: Yeah, that seems to be more or less what Kevin did, though he probably didn't have anything like this front-end IC.  His array had 100 electrodes, and his plan was to use a 25-channel 4-to-1 multiplexer to pull 25 of the signals.  Only 25 could be read at one time, presumably for the same reason you cited.  Of course, he didn't ultimately need that as he ended up only wiring 20 electrodes.

    So let's say that we could fit two of these eight-input ICs on a small enough board, so that we could simultaneously receive and process sixteen signals total.  If we can fit 4-to-1 analog multiplexing on there, that would allow us to read sixteen signals at a time out of a total of 64 of them, which is a pretty significant fraction of the 96 (plus four grounded ones) electrodes that many modern MEAs, including the Utah array, possess.  That's in addition to being able to write to potentially all of them, if we can fit enough multiplexers as well.  How small a board would you say we'd need to fit everything on?
  • having 16 output channels means you'd have to use 16 multiplexers. that'd be rather big. a better alternative would be to use a switching matrix. they tend to be very expensive (but still cheap compared to a utah array).
    in worst case you'd need to pull in a custom made IC. not cheap. but certainly possible.
  • Hm.  A switching matrix would probably do the job, and I remember seeing a company selling switching matrix ICs for $2 or so, though I can't find it offhand.  Another option would be to design our own multiplexer PCB, as they really aren't all that hard to design (if a bit tedious for large numbers of channels).

    As for custom-making an IC, we could theoretically do that for the entire multiplexing + frontend hardware, right?  Although pricey, that would have advantages with regards to miniaturization, I imagine.
  • a 2 usd piece would be awesome. the only switching matrix that's somewhat suitable is for video muxing in airplanes and sells for more than 100 bucks.
  • @ThomasEgi, I wanted to use the ADS1191 chip and need a breakout board for it. How can I find one? Did you have to make one of your own inorder to use the chip?
  • @bioenggal a friend of mine made a tiny breakout for it and put it onto into a piece of unused pcb space when he ordered his own pcb designs. You can get generic brearkout boards for the footprint, there are plenty available on ebay/seeed/etc.
Sign In or Register to comment.