subdermal temperature sensor - short term
  • wolfiewolfie March 14
    Hey all,
    so I've been stalking this forum for a couple months now but haven't made an account until now, figured it was time to interact with you.
    An idea has been floating around my head for a while, and I started to flesh it out just recently. Wanted some constructive criticism or just a 'that won't work', either is cool.
    So high blood pressure runs in my family, and I was envisioning a subdermal device that sits on your lower forearm and just logs data, transmitting over Bluetooth LE for alerts (VERY similar to the circada, but ideally a good bit smaller). I'm not planning on actually implanting it anytime soon, just running prototypes (although if anyone knows of an implanter in the Pittsburgh area, please let me know). Due to the difficulty in actually measuring blood pressure, as a proof of concept i'd just like a temp sensor, reasonably accurate,  to work over the course of a couple of weeks/ a month.

    My vestigial parts list so far is:
    ATtiny85 - possibly with a breakout board like the Trinket
    an ultrathin 45 mAh battery (3*11*16) or similar
    a temperature sensor (along the lines of the Sparkfun TMP36
    Qi compliant wireless charging reciever, maybe like the one on DFrobot, or TI's interpretation (25.0 * 15.6 * 1.0 mm and 1mm radially respectivly)
    an LED, just for basic interfacing.

    The goal here is to just create a proof of concept, and maybe display it in my biomedical engineering class. 
    If anyone has suggestions, comments, gaping holes in my plan, please let me know

    *// I'll discuss power issues and coatings in a later post //*
  • wolfiewolfie March 14
    POWER - 

    I've worked with LiPos in hobbyist applications before, and i'm reasonably confident in my ability to keep a single cell stable and healthy, but with something that close to my carotid artery, safety is of course the priority. if anyone else has a plan for power (i.e. feasible supercaps that I can actually order) i'm still taking suggestions. 
    A 45mA battery will bring total dimensions to ~4mm thick, with 16mm wide and ~38 mm long (is that small enough? without coating.
    unfortunately, thats like 6 hours of power, even w/o having the microcontroller always active. Unless i can creatively code and get it to sleep 'deeper' or something, I'm basically stuck with wearing my charging armband all the time. Suggestions?

    Also, i have NO plan for coating besides just making a little battery box from aluminum and injection-molding a silicone bar around it. :) Long way to go. 
  • ThomasEgiThomasEgi March 15
    The attiny85 comes with a buildin temperature sensor which, typically, has slightly better accuracy than the TMP36.  Check the Attiny85 datasheet, chapter 17.12 Temperature Measurement on page 133. Neither are horribly accurate tho. it's 1 or 2°C accuracy at best.

    Also, you only need batteries if you have high current demands in your circuit (which you don't). And you only need Qi compliant charging if you have VERY high demand on charging current (which you very certainly don't have).

    The LED is nice for show but wouldn't really serve any function other than showing that the circuit is alive and eating a lot of current. The LED would eat somewhere between 5 and 20mA while the rest of your circuit will probably be happy with 200μA. So you really want to think about using that led since it eat like 99% of all energy you supply to the system.

    As a starting point I'd recommend you go and have fun with this circuit: http://scanlime.org/2008/09/using-an-avr-as-an-rfid-tag/ and adding components as needed. Maybe a bigger coil, and buffering caps, a zener diode to prevent overvoltage.

    As for coatings there is no easy and field tested sollution for the DIY community yet. You can get implant-grade platinum curing silicones , which are a bit pricy but still affordable (read ~250 USD from www.appliedsilicone.com ). There are also a number of uv-curing and 2 component epoxy mixtures which are short-term implant grade (means they are ok for less than 31 days of implantation).
    When searching for coating materials be sure to rely on the ISO 10993 and not USP Class VI. And for the ISO 10993 you need to make sure you the product certifies for the right category: here's the related chart for checking https://www.namsa.com/bio/ . Your use case is implant device=>bone/tissue=>B/C depending on the duration you plan to use. Type C materials will be really hard to find. Type B is way more common and easier to get hands on. If you plan to use the implant fore more than 31 days and can't get hands on Type C materials, don't panic, chances are the damage you suffer from continued use of a type B material are small compared to using something totaly non medical.

    If you have more questions, ask here, or drop by on irc for a chat.
  • wolfiewolfie March 15
    okay, so.
    thanks for saving me the redundancy of 2 temp sensors, that's super helpful.
    rfid power delivery is a fantastic idea, but I don't understand how that could translate to always-on logging? wouldn't I just touch my phone to my arm to take a reading then? or am I missing something

    essentially, my two (initial) prototypes will probably consist of an uLP module with purely RFID power and the onboard temp logger, and a larger one with a Bluetooth LE- enabled chip and some other charging solution. you say qi is overkill?
    I'll order some parts, start playing with ideas.
  • ThomasEgiThomasEgi March 15
    I doubt that a Phone would work directly, the mechanism described on the website uses a low frequency reader instead. To have it always-on you'd need to constantly supply power (more or less). Given you can put the microcontroller into idle mode with occasional watchdog wakeups you get like 20 to 30μA current demand on average so you can potentially run it off a small cap for quite a while. Using something like a 5V 220mF gold cap (10mm diameter and 5mm thick) Numbers suggest you get a runtime of 1 to 3 hours. That'd be enough power to read the temperature every 20 seconds and store it in ram. You'd have to read your values out of ram every 2 hours to make room for more data and charge. Which is not too bad for a circuit that consists of no more than 4 parts.

    If you want bluetooth you'r in an entirely different realm. For this you most likely need a lithium based battery, or a  nimh one (they require less circuitry to operate safely but are usualy only suited for low current charge/discharge given the small sizes, might still be an option for you to take). But even for this "high" current application (reading as up to 100mA peak current) Qi charging would be total overkill. Qi was designed for power transfer from 5W up to kW. It comes with controll mechanism to not overpower your device. If you run on a tiny lithium battery you'll hardly ever need more than 500mW. So going with a homebrew inductive charging circuit would be a real alternative (if done right).
  • wolfiewolfie March 15
    Custom reader is fine, I have no deadline for this really so I can put in plenty of R&D.
    Ideally, the run time would be ~1 day/12 hours with readings every minute or so, but less is fine, this is purely prototyping. Bluetooth would also be nice, just for the 'mainstream adoption' use case of presenting numbers and readings as more than just a text file, but I think for v. 0.1 not having a battery in the implant might take precedence.
    do you have any links you could pass me in reference to the custom wound induction charging? I'm not super familiar with the concept (besides just 'put the coils together and voila')
    thanks for the help :)
  • ThomasEgiThomasEgi March 15
    well .. basically.. you put 2 coils together and voila. Ideally you want to have an LC circuit to boost your range, but the basic idea remains the same. You may want to measure the actual power you transmit on the sender side, so you can tune power down if you hit dangerously high levels but if you stay within reasonable parameters to begin with there's little to be afraid of. 
  • wolfiewolfie March 15
    okay, great. I'll keep you updated.
    Side note, what kind of surgical skill/implanter/tattoo artist would it take to implant something like this ? (1.6*1*5mm)
  • ThomasEgiThomasEgi March 15
    That's something you can do yourself if you stick with a proper procedure. That's like a really small object to begin with.
  • wolfiewolfie March 15
    alright, thanks
  • wolfiewolfie March 15
    wait I totally messed up the dimensions , it'd be like 10*20*5mm
    :P
  • ThomasEgiThomasEgi March 15
    in that case you may want to look for someone to be able to stitch up a 17-20mm wide cut.