Data and Power

Forking from to start the discussion about optical data transmission, and close-coupled inductive power.

To answer gareth's question outside irc for everyone: it's charging the LED in reverse, measuring the time it takes to discharge via the photo-induced current. It's very slow, papers talk about ~250Bit/s

Next faster option would be to use something like the MAX3120. a integrated frontend IC , all you need is a buffer cap and a pair of led+photodiode. It gets you up to 115kBaud, with very small footprint and super-low power. With a suited communication protocol and syncing you can get down well below 1μA standby current demand.

Using a reasonably small circuit (like 4 or 5 transistors , some diodes, resistors and an op-amp) you can get to about half 500kBaud.

Faster stuff tends to be more difficult. AVAGO TECHNOLOGIES HFBR-2406Z has a great datasheet with a variety of circuit examples for different speeds.

Most reasonable pick in my opinion would be MAX3120. That's about 36MB/h or 25GB/month. Not much but even if you need to fill a 4GB microSD, it's not entirely out of question (patience given).

As for power transfer the suggestion remains:
Basically have a U shaped core, with a flat piece in the implant to close the magnetic circuit. The design goal would be to increase the area of each end piece of the U and flat piece as large as reasonably possible, the airgap small, and the ends spaced enough.


  • Have you experimented with any of these options before? I have half a mind to order some of the IC and play around with them.

  • I feel like most data applications would be low volume, like daily temperature flux, firmware, so this is plenty of data. It's not like we're planning on storing a movie subdermally, right?

    As for inductive data transfer, Qi already specifies data transfer, but it's only for the proprietary data codes, power transfer, battery state, that kind of stuff. I'd say take their specification and expand on it. Would be fairly simple to add a sensor in the loop to read the data coming in and out, and it would allow you to use the info they're transmitting for whatever.
  • edited November 2015
    Also, there's a lot of transceivers on the market, just take a look at mouser :
    Why the MAX3120 specifically, is there a specific reason?

    Example, look at this one

    6 mm x 3.1 mm x 1.9 mm
    That's tiny! Easily within space constraints for any system, with low power too.
  • "It's not like we're planning on storing a movie subdermally, right?"

    Speak for yourself........
  • @ChrisBot optical free space communication systems are the reason i got into electronics in first place. I've been experimenting and building those things for years now. Ranging from the size of a matchbox to the size of a old TV, with datarates from about 1bit/minute up to 10Mbit. That's all without integrated circuits, just bare transistors etc. Only difference to this sort of systems, mine have to work with signals hundret and thousand times weaker due to the distance of several hundret, up to several kilometers.

    @ElectricFeel The MAX3120 specifically because it allows you to pick your own transmitter and receiver diodes. It's true there are tons of tiny premade packages, but they are all invisible light and the wavelength may not be optimal for penetrating skin. So with the MAX3120 you can pick red led's and a matching photodiode. Which gives you free visual debug output,too! You can drive the LED whatever way you want after all. Besides, it makes alignment a bit easier if you know exactly where your transmitter is.
    Another argument is, you can space the transmitter and receiver diodes apart. With the tiny packages you'll 100% find yourself receiving whatever you send under the skin cause light just scatters the few mm from transmitter to receiver. Having your own led and photodiode spaced apart enough you may be able to get away without that problem, enabling full duplex operation without a need for a protocol to fix the crosstalk problem.

    What I like best about those systems, except from the super low power demand when not in use. They are simple and robust. Unlike Bluetooth or wifi there's no layers, no os, no software, _no antenna_, virtually no source of error or point of failure. Bits go in, bits go out. And many microcontroller's uarts can be configured to output and accept IrDA conform pulses.
  • Well actually, between 700 and 900nm is ideal for skin penetration, so these modules are optimal for that. I would say add this in to say, the NorthStar. It shouldn't be such a stretch.
    There will be scatter under any distance, but I don't know if it's serious. That would be something to research.

    IrDA is very robust and a very mature technology. It's very primitive as far as communications go.
  • If the wavelength is matching... all that'd be left to do would be to implement a bit of a protocol to prevent receiving and transmitting at the same time.
    This should include a clever method to keep the implant's optical frontend powered down most of the time. Basically having the base-station spam out syncs, with the implant listening to such a few times per second only, remaining in power-down mode for most of the time.
    Once both devices detected each others presence, time slots for communicating can be exchanged and then half-duplex fun begins.

    Effectively that'd limit bandwith to 56k, but that's still plenty for a terminal. With default settings of 9600 we'd get away without buffering more than just a single byte.
  • I think we should buy a pair of these and test them through pigskin. I think that we might be able to still operate duplex.
  • You can probably find a whole bunch of those in older electronic devices. Mobile phones, palms, pretty much every not-quite-as-recent mobile/portable device should be equipped with IrDA transceivers. Spending 5 bucks to get them new would work too, i guess.

    I'd bet the last 10 bucks on my bank account against full duplex, but I'd love to be proven wrong again as it would greatly reduce the overhead.
    In the end, some of those units are so tiny and low on power, we could easily stuff 2 of them onto an implant at different edges and still get away with it.
  • edited November 2015
    To anybody with a valid .edu email account, it looks like Maxium Intergrated will ship you a free sample of the chip, or at least the MAX3120. 

    I am sure that you guys are clever enough to social engineer a few free chips out of these guys to test.

    This is the link that they sent me in the email.

    At the time of writing the page says that it is currently under maintenance (???).
    It may just be broken
    If so.... here is the to link to their contact form. 

    Unfortunately, my hichschool doesn't provide me with an educational email, so I would be interested to hear what sort of luck you may have. 

  • Guess it's easiest to simply order them. And just for the fun of it, there are faster IrDA chips around. Like 4Mbit/s are available. They do require a bit more standby current, but if kept in shutdown mode as imagined by protocol, the extra current is well worth the 40x bandwith. Energy per data is certainly better with the 4Mbit version.
    VISHAY TFDU6300-TR1 would be one of the candidates.
  • That part is obsolete, no longer manufactured.

    I went ahead and grabbed a sample of the Maxim chips.
  • edited November 2015
    @ElectricFeel you and your fancy college email address... 

    I plan on making a digikey order sometime next week so I will pick some up then.

    Let me know how it goes!
  • @ElectricFeel, did you ever get those sample chips? Did you get them working?
  • Nope, never heard back from Maxim. Might try again, my order might have just been dropped through the cracks.
Sign In or Register to comment.