Proposed hardware components

Hi everyone,

since the Southpaw space is pretty much empty, i figured it would be nice to set up a list where we collect parts which may be used for the device.

for all parts suggested it would be good to have:
-Short describtion (like 3axis digital compas with I²C interface)
-Part Number
-Reason why you think it would be suited (maybe pointing out advantages over previously suggested parts)
-Link to the datasheet
-available packing options (x y z dimensions + pincount , because size matters)
-estimated power-consumtion for operation and idle (because we most likely run on some kind of battery)
-a supplier to buy the part from (maybe one for the bare part, one for the part on  a breadboard for easier prototyping)
-anything else that seems relevant, like number of external parts required for operation.

so i'll start with some core components:

Atmega8HVA Microcontroller  
-well known microcontroller, good crossplatform toolchain, widely used, low power, ADC and I²C to interface with analog/digital compass, lithium-battery-management buildin, 1.8 - 9 V operation range. plenty IO ports.
-available in tsop 28 (1x8x14mm ) and LGA 36 (0.5x3.5x6.5mm)
-draws 800µA under normal operation @ 1MHz , 270µA idle, 25µA powersave
-listed at digikey but currently not in stock ~6$
-no external parts needed. can provide a 3.3V output (with automatic linear-down or step-up with very few external parts). LGA packing may be hard to solder by hand but first choice when keeping size down.

HMC5843 3-axis Digital compass with I²C interface
- i²c interface so no mesing with analog stuff, connects nicely to the atmega, 2.5-3.3V
- 16 pin LCC (4x4x1.3 mm)
- draws 160µA when sampling at 2 Hz
- breakout boards and chips  available at digikey for about ~20$
- requires 2 capacitors: 0.22µF and 4.7µF , 2 pullup resistors required for i²c (may be possible to use atmega's internal but that needs to be checked)

VL-3032 Rechargeable Lithium Battery
- 100mAh capacity , 3V,  4mA charge current , small encapsuled coin-cell, Ø27x3.5mm, easy to solder, integrates well with the atmeg's li-battery management.
- 5 days of operation for 1000µA load, 15 days for 300µA . recommended continous load : 200µA
- using this battery requires agressive power-saving on the controller/compass. only allows "slow" charging.

additions and improvements are highly welcome.

small addition: Attiny 85V would also do great, it can be clocked significanntly slower which means.. a lot less power, has a smaller footprint. it lacks the battery management functions, but those can re-implemented easily. and.. .it's available for purchase and a lot easier to solder.


  • edited January 2011
    What's stopping us from building this now? is it just the code we'd need to write for the microcontroller?

    If finances are a concern, maybe we should start thinking about setting up a Kickstarter.
  • neuroelectrodes are. aside from that the parts are quite small, so soldering is impossible without a PCB. then we'd need concider mechanical issues and then make pcb-layouts and have them produced (or producing yourself using direct toner method). the charging circuit is also missing (altho it's comparably easy). 

    there are no big show-stoppers but it requires careful planning, the pcb , and time (since my exams are in 2 weeks i wont be able to build that stuff right now). 
    i guess it would be quite possible to create a prototype pcb just to see if it's working as such. if charging works, to see the actual powerconsumtion , doing relyability and failsave tests etc. after that, the final pcb would have to be designed/produced.

    oh and signaling those neuroelectrodes... yet another thing :P most information i found was purely theoretical.

    i guess i could pull of that stunt with enough time and some equipment upgrades. producing the pcb, soldering these tiny things and getting the neuroelectrodes are the most difficult parts.
  • I'd chance a dead-bug build for a second prototype, if at all possible; the implant package needs to be as small as possible.
  • if everything works out as planned, it should be about the size of 2x one or two pound coins which are either stacked ontop , or arranged side-by-side in a semi-flexible way. which would be about as small as you can get without producing custom chips.
  • about the electrodes: I was supposed to write here as opposed to the comment section, right?
    So, as a pretext: my knowledge is about one year old, when I had a course of neurobiology including some experiments with an earthworm and some locusts. With the earthworm we actually induced signals (electronically) and recorded the reaction (as I remembered today..) when with the locusts we were only recording.
    I can't find my protocolls, but will look for them. Right now I read about the current that is necessary to induce a single ap ( You can also see it here:
    it is about 20 mV. (using dc works just fine)
    One short peak for about 2 miliseconds is enough (for what I remember)
    However you have to consider getting through the skin and other tissue. Plus: you don't only want one ap but some more in a row. But as said earlier: I will read up about the details of the experimments with the earthworm, as soon as I find them.

    As for the electrodes: we were using some aluminium foil or something compareable.  Can't really be sure about the kind of metal, but I think it was aluminium. Not, that it matters. We put the earthworm on the foil and put a current on it. The other pole of the dc was ground, for what I recall now.
    What that means for Southpaw: as mentioned in the comments a simple uncoated
    loop of metal (platinum is bioproof) should suffice.
    Again, this should be tested thoroughly for the details...  For one, because Lepht is not a worm and we not only care about her safety, no, we want to spare her pain and she actually could tell us if something really hurt...
    If I had the equipment (or access to the lab once again...) I would test it myself, but I don't so, I just write down how I would do it (and how it SHOULD work) so if anyone got what is necessary...
    Best place for this test are the fingertips, basically because they got a lot of nerves. So they are easier to hit even though there is skin in between. Then start with a small current of 20mV dc. Turn the power on and off and see what it feels like. Then put the current up by steps of 5 (or if there is no feeling at all 10 to 15)mV. Also small variations in duration are possible. Also variations in the setup (both poles on the skin, variants in the distance between them, only one pole and the other one is grounded, such stuff). I mean even if in the original there was only one pole, we have to get both of them inside the skin...

    This really should be safe (dc plus very small currents), however I suggest only someone knowing what he (/she/it) is doing should do it. There are mistakes, such as choosing the wrong current (pain) or the wrong electricity (ac, could cause serious injuries).

    As I said, as soon as I know the original setup I will be able to provide further information, but the electrodes themselves won't be a problem. They are just some bioproof metal.

  • edited January 2011
    @Ben If you can write me a list of the exact supplies I'll need to test your hypothesis, I'll guinea pig and film the test.

    There's a mechanical shop I'm friendly with nearby, I can see about scrounging parts.
  • My idea for testing was to grab a cheap TENS [Transcutaneous Electrical Nervous Stimulation] unit (Maplin had 'em a month or two ago) and to interface with that.
  • Ok, what I was writing before did NOT work out...
    Just found some of the recordings and we used WAY higher currents. 3V and higher. Plus I remember having seen a bet of someone testing 12V batteries by touching them with his tounge (though I still would not recommend that. Won't harm, but hurt a litte)
    Sry, I will have to find the instructions, as soon as I remember where I put them...
    Not too sure about the dc now either, as I remember touching batteries and not getting shocked...

    How much was that TENS?
  • Less than 50 quid; can't remember exactly now. There's one on Ebay, "Buy It Now" for £29.99 at the moment.

    Trust me, you lick a 12V battery, you will definitely feel it. Hell, it's possible to shock yourself by chewing on aluminium foil, if you have the right dental work.
  • Dry skin's a good insulator; there's a _world_ of difference between touching and licking a battery.
  • @ben small electrical note:  3V is voltage, not current. current is messured in A (ampere) and currents should remain minimal here.

    i just finished to rape my soundcard since no function generator was around.
    so i put 2 needles under my skin, parallel, about 1cm deep and 4mm distance and checked with various signals and voltages. since there was no isolation towards the muscles below i sometimes wasnt sure if its the muscle or the skin responding,
    from my tests i started to feel something with about 400mVpp for square and pulse-shapes. with the sking pulled away from the muscles it took up to 2Vpp.
    the whole setup was a bit unreliable cause i used some old wires and a soundcard and i taped the stuff onto the skin which sometimes resulted more unpleasance caused by the tape than by the voltage. oh and... dont try this with a music player running at full volume,, it'll cause major discomfort.

    but overall i'd guess it is quite possible to use a method like this. iridium or platinum wires/pads insulated with medical-grade silicone should work. testing indicates that the available battery voltage of 2.5 to 3V may already be enough. what worries me more is how exactly to determine the right parameters. there are lots of potential influences like dimensions of the contacts, conductivity of the skin etc. since the difference between a noticeable sensation and major discomfort seems to be rather small it would be vital to get more detailed information about what is the requirements and infulences.
  • hm.. apparently the atemga8hva does not come with a hardware i²c . why didnt i read the specs (the 8L does)... can be implemented in software tho. it does come with the usual SPI, maybe there are other compas modules which support that.
  • First of all: Hi.

    I just tested what effect 5V and 3A would show ... and i have to say it's sort of tingly ...

    I recently found some subdermal electrodes for animal tests that could work if one could get hold of some.

    I'm working on a little projekt myself at the moment, to get a wifi detection module working as a subdermal device.
    The problem is, i'm sort of a wuss when it comes to cutting myself, so i only test the signals on my tongue.
    And i have to wait for a new detector for further inspection, because i could not find any usefull documentation on those things on the net, except in phrack.

    So long ...
  • edited January 2011
    @nibot. with 5V and 3A there would be nerves left to stimulate. currents usualy are very tiny. if you checked witih your tongue, than you propably had a tongue-resistance of maybe 200kOhm which would let about 25µA through. 

    the smallest successfully printed line is 50µm width (mil2). line spacing is 200µm. using 250µm between wire-edges gives good,relyable prints.the smallest parts i intended to work with have 280µm space between pad-edges so (the out-of-focus structure on the left side of the image).. IF i can manage to get those prints etched that way (which i have not yet tried) the results may be very pleasing.

    i should mention the printer has only 600dpi resolution which would be ~42µm
  • I like that project, nibot!  Very cool and useful.
  • edited January 2011
    This is a lot of very technical discussion (not a bad thing). Those grinders out there, like me, who are not nearly so knowledgeable could benefit from a very basic explanation of some of the material presented here. I think there's plenty out there in our community who would like to contribute to Southpaw but don't know where to start.

    If you have a suggestion for how us laymen can contribute, please share. I have a number of resources I can draw on here but I need to know what, very precisely, are the challenges facing Southpaw's construction before I can reach out. Hackers to engineers to auto technicians, if you can tell me what they can do to make this real, I can bring them in. 
  • if you could make the logistics company of this planet to deliver stuff on saturdays... that'd be a great help. aside from that i dont really know. it's a technical device so it's a technical discussion. we'r still lacking detailed information when it comes to the output parameters for the electrodes. aswell as electrodes itself. most other things is just a matter of searching the right parts, ordering them, manufactor the pcb's , soldering, programming. the usual stuff so to say.
    searching the web for a source of electrodes (i think the medical term for that type is "epimysial electrode", surface-muscle electrodes for the rest of mankind). having a source of medical-grade silicone would be great,too. sugur may be better than hot glue, but i personally would prefer something that got at least some medical certification. aside from that, medical stuff usualy is designed to be molded into some form, so it's more "liquid" which allows thinner coating.
  • @ThomasEgi Possible to just talk directly to a hospital's administration and ask them where they purchase their silicone and surface-muscle electrodes from?
  • ...or we could ask Google.
    I've got a link to some electrode suppliers as well, but it's in the mythical "somewhere." I'll see if I can find it.
  • today i found . selling all sorts of electrodes to reasonable prices aswell as nice material suchas teflon-coated wires. the company is not far from a railroadstation i pass by every few month. as far as i can see they ship to germany only. but that shouldnt be much of a problem, if i can get hold on some of the stuff i could aswell send it to whereever people are in need of them.
    not long a go i also stubled over a webside with medical-grade silicone which sold experimentation kits in small quantities. didnt bookmark, but i'll find it again.
  • First of all: shame on me...
    (not for the not writing, but for the confusing current and voltage...for not writing as well, of course)
    I know quite some people who would slap me if they could read what I wrote there, I couldn't blame them.. It is even in the unit! (V for Voltage)

    Anyhow, I couldn't get hold of the instructions from back then and whoever I asked didn''t know the details either. As for the hardware I am almost certain that it was a special signal generator and as such not exactly ment to be sold to private persons (thus rather expensive)
    But I do like the using the soundcard as a signal generator!
    Tried it putting two small needles in my arm, but ok, didn't really work. Was to be expected... Next time deeper, but first: who exactly did you generate the signal? Especially: how did you determine the voltage comming from the audio signal? Did you just measure it and then more or less try and error? (louder, measure etc)

    And  the shop looks good, too. One thing, though. As a quick overview, how much do you guys guess this thing will cost? I remember Lepht saying that with electrodes more than the 8 directions would be nice (maybe I imagined some of that statement, in that case lets calculate with 8).
    Using this electrode:


    4,0mm dia. x 1,0mm disk, 10mm exposed wire

    9,50 €

    which could work fine, depending on the tests in advance, made already a price of roughly 80€  for the electrodes alone. Don't know about the other equipment, though.
    No critics, just asking.
  • for a function generator. I've been using a Sparkfun self-assembly kit for coil testing; depending on what current you want, you might need to throw an amplifier together.
    Also: 10 needle electrodes for $25. They're huge in length, though.
  • life aint cheap. i'd use one or two of those electrodes to to do testing. for the final version i tend to prefer the teflon coated platin-iridium wires. those are expensive, but they also are several meters in length so cutting of a few cm will be cheaper on the long-run (in case there will be more than just one functional device in the end)
    given those electrodes are medical-grade stuff, it's dirt-cheap. like i said, their company is not far from where i live and i may drop by, see if i can get hand on anything worthwhile.

    for the soundcard thing. well. i wired it up,created some wave-forms in audacity, and then i just went for it, increased the volume until i start to feel something. altho it's totaly not a good idea, a tiny bit to much and it'll turn from a slight sensation into major pain. my next try is to use something with proper current-limitation.
  • more research on parts:

    here's a new , short, list:

    another battery appeared, altho comparebly huge, it stores a lot more power than the others with significantly higher charge rates. on the short dimension it's still smaller. so less cutting neccessary

    also, the local supplier for electrodes of all kind/quality/price

    and for the main controller.. looks like the attiny44a fits the bill, cheap (2$), small(4x4mm), i²c and really really low power (if a clock of 128khz does the calculation, we'r down to about 50µA!!! without further sleep/idle/powerdown modes on the controller end), oh and it has plenty of IO ports. it also comes with brown out @2,7V which happens to be the battery discharge-cut-of voltage. so we prevent deep discharge.

    i wanted to get hands on the AK8973 compass since it has more pin-spacing than the sweet honything chip. but it's totaly nowhere to be found. so i'll stick with the first compass recommended.

    the internal pullups should be just about enough for a <100khz speed.
    that leaves us with 3 small smd-capacitors for the compass. a 3x3mm voltage regulator, a zenerdiode to prevent the battery from over-charging.
    and.. anything that might be added due to the output. so far the battery is the biggest part. the rest would fit on a fingernail.

    without the output, the total current demand ranges around 250µA (mathematically). which would mean.. around 12 days of operation without a charge, day and night. however there are lots of optimisations possible. like putting the device into sleep mode when the leg is horizontal (hooray for 3 axis sensor). which , depending on the sleep duration gives more battery time. it may also be possible to make the microcontroller operation vastly interrupt driven. which may or may not save another 20 to 30µA. so far it looks quite promising. with that much load the battery should survive +10years with a small charge once every few days.
  • I didn't know about the compass' 3rd axis; I suggested an SMD tilt switch to cut off measurement while taking a step and lying down (anything that takes the module through an angle that would make the compass' measurement less than sensible, basically). It would mean that you could switch off the chip entirely when lying down... Dunno if that's necessary. Sorry, I haven't had time to get my head around the new design.

    Any ideas what would assert the interrupt?
  • it's more of  mechanical problem why we need 3axis anyway. a 2axis chip would have to be mounted horizontaly, which would totally be no good since you'd have to mount it 90° to all other components so it would "stick out". so we get the angle-compensation and orientation detection in software for free.
    a physical switch is not really required. the atmega can be put in sleepmode with a watchdog timer that needs only a few µA (like 10 or so).at the same time you can use it to turn the magnetic sensor on or off. so you can simply power on the system once every few seconds to see if the device is still in horizontal position. if not, switch back to regular operation (which would be 2 sensor-readings per second as far as my plans go, more are possible but also require more energy).
  • Ah! A mental model begins to involve.
    I didn't know about that self-interrupt technique; the last interrupt stuff I did that didn't involve a threading library involved 68k assembly and IRQ handlers, so I was focused on it being an external signal.
    Samples per second would need to be worked out; how any lag affects perception would be a good question to have answered before locking that. I was kinda assuming around the same; 1 or 2 a second.
  • microcontrollers offer a wide range of interrupts, the watchdog beeing a quite useful one since it's possible to put the device into a near-zero-power mode.
    for the sample rate.. it's mostly a battery-lifetime issue. the compass module is the part with the highest power demand already. current-demand for only that chip:
    1Hz = 160µA
    2Hz = 200µA
    5Hz = 320µA
    10Hz = 520µA
    you always have to add the rest of the system to it.
    since i'd like to keep the overall consumed currrent around 300µA or below. well... 2Hz would be my suggestion. theoretically even 10hz would be possible but the time between recharges would shrink from around 10 days, to just 2 or 3. and not only will it reduce the time between recharges but also the overall lifetime of the device since the battery has do withstand more charge cycles.
    the updaterate however, can easily be tried out and changed in software. so no restrictions at this point.
  • ok small update after a week of exams. using backery paper, a laser printer with 600dpi resolution and a regular household-iron it is possible to manufacture pcb's fine enough to connect even to the 4x4mm chips.  the process takes some time and a lot of patience so once the design is final and the prototype is working, ordering a industry-quality board is propably best. those go for about 60€ and more, but due to southpaw's size we could fit 10 or more onto a standard-sized board and then resell the single boards to interested people. 
  • has anyone here managed to get an HMC5843 to talk to a microcontroller over I2C? i'm using a couple of these things and i can't get any response out of them at all.

    i've got the I2C bus pulled up with 10k resistors, 10k resistors between the HMC5843 and the ATTiny85 i'm using like the data sheet suggests, and i've tested all the hardware connections and they seem to be fine. i might still have something wired up incorrectly though, so if anyone knows exactly how these things should be connected, that might be useful to know.

    i can't get either of the I2C implementations i've tried to get a response out of them. one Atmel-based one does nothing but give me a USI_TWI_NO_ACK_ON_ADDRESS error, and the one here doesn't seem to be putting any data into the buffer it's given - either that implementation isn't getting any data back either, or i don't know how to use it properly. any other sample I2C code i've found is written for ATMega chips which have hardware I2C, so i don't think there's anything in them that can be adapted to the ATtinys.

    if i had another I2C device i'd be able to at least narrow down whether the problem was caused by hardware or software, and i have some on order, but right now i don't. so, if someone with more than two weeks experience with microcontrollers could help me out, it'd be greatly appreciated.
Sign In or Register to comment.