As I wrote about a week or so ago, there is now a reference design from JoernL for a bridge that allows the Libre sensor to be used like a Dexcom CGM. It can be found here. http://joernl.github.io/LimiTTer/ In the spirit of #wearenotwaiting, I set out to build myself one to see how it worked!
After guessing at the BoM, based on the limited information available in the website, my parts turned up last week. I got the opportunity to put most of them together yesterday evening. This initial test rig is not chargeable as the LiPo charger had not arrived, but it’s also not set up to fixed yet.
Anyway, doing it in a stepwise process, I followed the instructions provided by Joern, firstly programming the Arduino. My Pro-Mini board from Cool Components has a slightly different layout from that used in the reference design document, but as I had ordered a set of jumper leads when I did my bulk buy, this wasn’t really an issue. It meant programming the Arduino remained relatively straightforward and brings me to tip one. Buy jumper leads for the programming phase. They cost very little but really make life a lot easier, and much more flexible, especially when you find you have a different layout.
With the Arduino programmed, is was time to move on to the build. Note that I am not building the design to fit within the small box that Joern has used, my aim is to wrap the circuit in a neoprene skin and wear it like an armband. I’ve therefore got a lot more wire going on in my circuitry to distribute the components within the band.
Next up was cutting the lengths of wire, preparing the ends, a dull, if not particularly long job, and then getting on to the soldering. My approach was to prepare the bluetooth module (an HM-10) and the NFC module first then attach these to the Arduino board.
This is the most fiddly bit. It’s bad enough with all the wires in place in my take on the design. Trying to get all of these in place in the reference, with much reduced spacing between components could be a nightmare. I think there is a big opportunity, if you plan to box it up, to make a ready made PCB to do this.
I’m also using a higher capacity battery than the reference document prescribes as I really don’t want to be charging it too regularly. Mine’s 850mAh. In this process I discovered that the terminal layout on my board had a GND terminal in a different location from the one in the reference documents. Not really an issue. Just a “use a different layout” diversion.
The final product from last night’s build looks like this:
This is it sitting on the Neoprene that will (eventually) become the sleeve. I need adding the LiPo charger and a more robust battery connection, but from this point, the BLE module connected to the phone. I’m not sure whether the circuit is working correctly, as I have the default HMSoft BLE link from the HM10, but that will be confirmed by the developer as it’s raised in the GitHub chat.
I’ve started a Libre sensor, and told xDrip to start the sensor, so now we await the LiPo charger and switch to complete the circuit, and then it will be time to glue the armband closed. and try wearing it.
So far so good, and all that. An interesting project that hopefully will result in an interesting and cheaper form of CGM.
Building this has highlighted a couple of things.
- For an armband, this really needs a flexible circuit that the components can be attached to. These can be made, so it would simply require enough interest to make it worth getting a batch made.
- Use of an integrated Arduino/Bluetooth board would reduce the componentry significantly. These are available in a mini form factor so I think it’s worth looking at how that might work (although it would increase the cost quite a bit).