Powering sensors from downstream hardware?

zakqwy · June 23, 2016, 9:05pm

Hey folks! I’ve been putzing about with littleBits for several years but haven’t gotten involved in the online forum until now. Seems to be a great community and resource for aspiring makers and experienced hackers alike.

I’m designing an adapter that will allow me to use littleBits sensors with an external neuroscience education platform. Ideally, I’d like to power these sensors through their outputs (i.e. their female bitSnaps), as this method dramatically simplifies wiring for some models (for example, the eye model uses five or more light sensors in ‘dark’ mode to emulate photoreceptors). However, I also want to avoid damaging the the littleBits if the user mistakenly tries to power the boards from both ends.

Any thoughts? Think I need some sort of sensing/cutoff circuit, or just a really good set of instructions? Or should I isolate the VCC rails from the two platforms and just require the user to have two supplies? This method worked well in testing but it’s a bit clunky, and doesn’t solve the extra wiring issue.

JackANDJude · June 23, 2016, 11:38pm

Hiya @zakqwy! Thanks for coming to the forum!!!
Can you share more info about these existing models and draw us diagram(s) of your ideas? We just LOVE diagrams. It will help Bitsters better understand the particulars of your question.

zakqwy · June 24, 2016, 7:15pm

Sure! The platform is called NeuroBytes; I don’t want to run afoul of the Code of Conduct rules so I’ll try to keep this description brief, although to be fair they aren’t really for sale (at least not yet). Essentially, each NeuroBytes board has five input connections (“dendrites”), each of which directly ties into an ADC line on a microcontroller. In certain operating modes, the NeuroBytes will poll the voltage level present on these lines and use it to change internal membrane current, causing the board to fire once a preset threshold is exceeded (and then increase in firing frequency as the membrane current continues to increase). I made a quick video a few weeks ago using a Dimmer and a Bargraph bit to illustrate the behavior that might be helpful to watch, but things get a lot more interesting with light sensors and a bigger neural network.

Currently, I’m isolating the littleBits and NeuroBytes power rails as shown here (3-rod eye model, not showing the downstream neural network for simplicity), so the two platforms only share signal and ground pins (red = VCC, blue = SIG, black = GND):

Ideally, I’d like to eliminate the wire connections along with the littleBits USB power supply on the left; once I add additional photoreceptors and sensors the wiring gets a bit overwhelming in the eye model, and it’s also a pain to use two power supplies. The newest NeuroBytes power rail runs at 6vdc (as each board has a built-in LDO), so I’d add a regulator to the littleBits–>NeuroBytes adaptor. But… I wanted to see if anyone else has run into this problem, how they dealt with it, what kinds of issues they had, that sorta thing. I’m particularly concerned about a user trying to power the littleBits modules from both ends, and the potential damage that could do to the boards.

Thanks for your help, @JackANDJude! Let me know if you need any more info.

zakqwy · June 24, 2016, 7:20pm

Hah, I just realized that the video I posted shows the NeuroBytes board pulling power from the littleBits chain. That’s an option too I suppose.

JackANDJude · June 25, 2016, 5:58am

Based on your video, it looks like at least one NeuroByte can work on 5V. Are they open source, by any chance?

chris101 · June 25, 2016, 9:56am

That’s probably the best way to do it. I have powered bits the wrong way, and they act funny. Try powering a dimmer with a powered arduino input connector. It will be able to read the unpowered dimmer, but the signal will not get all the way to 5 volts - about 3 and a half.

zakqwy · June 27, 2016, 3:52pm

@JackANDJude, yes–the whole project is licensed under GPL v3, although I don’t have the latest design files up yet as the hardware isn’t through testing. Right now the old v0.4 board designs and firmware are available on GitHub, and the newer v0.9 design files are in the repository on the project’s Hackaday.io page. I’m planning to get all the newest stuff into a dedicated GitHub repo in the next few weeks.

@chris101, good to know re: bits acting strangely. I just put in an order for a new adapter design that’s got a switch and a Schottky diode, so (depending on the switch position) the littleBits circuit can power the NeuroBytes (but not the reverse) OR the two setups can be fully isolated. NeuroBytes can be powered by 5VDC although between the LDO and the on-board diode Vdrop the LEDs may dim a bit. In any case, I ordered a bundle of these boards late last week:

By the way, would anyone have any use for a tested KiCad bitSnap footprint? My first iteration got the hole diameters a bit small (oops!) but I think the boards shown above should be pretty good (pending tests when they arrive next week).

quinkennedy · February 24, 2019, 2:40am

@zakqwy yes! I have been trying to find a bitsnap footprint for kicad. Sparkfun has one, but it looks like a SMT connector, not THT. Did you end up posting the footprint anywhere?