The main issue I see is that the transmitter does not use a rechargeable chemistry cell. If it did, the it would open up a whole new world of possibilities.
As for using inductance to charge, while the concept is a novel one, the needed hardware is too much. Too much as in physical real estate, not price.
Pogo pin style contacts (DIY friendly for prototyping stages)would be another option to charge if there was a charging management circuit. Or simply a micro USB port with a rubber plug. (Simplest option and allows the wearer to keep the sensor in use while charging too as long as software protocols allow this.)
Lastly, since you will most likely be familiar with it, why not use a hot knife to do the case separation? It would be a much finer tool than a grinder and have a minimal footprint. It would also allow you to do a full reseal of the casing once finished since the material is not destroyed completely by the process. Just a thought. :)
That is probably what I looked at when I was first looking into them a few years ago. Out of curiosity, how much are the Medtronic batteries? I'm guessing the transmitter is inside them as well?
As much as I would love things to go right the first time, they almost never do. It may take a couple of tries to get a good method to opening them up, and documenting it. So it will definitely not hurt to have a couple on hand.
So Terry's sensor arrived the other day. I wanted to get right to it and at least make a quick CAD test model that I could have 3D printed. That way I would have something to work with/test. I just finished one that I think has a relatively accurate base where it clips in. Can anyone confirm, only the bottom piece is used to secure the transmitter? The top is just some random organic shape to fit the circuitry and battery perfectly, I'm assuming. I've seen some pictures of smaller transmitters online and the appear to have the same clip design, but shrunk the top of the casing by a lot. I think I will probably buy a pack of sensors the do some test fitting with the 3D printed part.
Here is what I've got so far.
To compare to the original:
I also noticed that they machine the bottom of these transmitters. Probably to expose the contact leads and give a perfectly flat bottom to the transmitter. You can see the milling marks in it.
Hey hodginsa. Awesome stuff you're doing. I'm super encouraged and grateful that you and others are working on these types of projects. Keep up the good work!
I've been using my current transmitter for about 10 months, so it should be dying relatively soon. I'll definitely let you know when it does and send it to you if you're still in need.
Lastly, if you want, I can mail you my used sensors as I go through them, rather than you having to buy a box (they're pretty pricey). I'm sure some others on here would be willing to as well.
PS - Just a heads up - I noticed a couple times where you mixed up "sensor" and "transmitter", such as the TL;DR section of your original post.
Thanks Mark, Its great to hear that this is encouraging to other people as well. Helps with the motivation for sure.
I think the more transmitters I have, the more successful this project will be. After discovering a second circuit board underneath the battery, the casing removal is more difficult than initially thought. I'm looking into chemical removal. The issue with removing the material is that it sometimes removes components as well, that generally breaks them. So sending me yours would be really helpful to me.
As I was designing the CAD model I looked up the sensors and discovered the slimmer one today, apparently they were able to do so because they reduced the number of components on the circuit board. I took that as, they made it a single circuit board inside, which would be interesting to find out. But I will stick with figuring out the thicker ones for now, since they are all backwards compatible. Also probably wont be able to get a slim one for some time.
Some sensors would be great, even just one really, so I can test my 3d printed model with fitment. Also I am looking into how they actually take the reading from the sensor, and the possibilities of making a fully open source unit while still using Dexcoms sensors. That to me would be really cool.
Well, I've definitely got my work cut out for me. My first mistake was thinking this was a plastic, its definitely an epoxy. I tried to solvents with no luck(what I normally use), then went to my rework station which pumps out really hot air. Somehow managed to get most of it off. It turns brittle at a certain temp and comes off in chunks, but it seems if you go over the temp it gets even harder. Lost one component though, I know this transmitter won't work now. But this is good, I now know the dimensions of the board that is in there, and the basic layout. A new slimmer transmitter may actually be easier to open if it only has one PCB but Ill think about that later. Its still sitting in some acetone overnight, might it will surprise me and get the rest of the stuff off. Its some tough stuff, and doesn't smell great when you heat it. I'm sure there is some chemical out there designed to take this stuff off while leaving the circuitry intact.
Are these the first pictures of the circuitry of a transmitter that is still intact? Maybe!
Bonus points if you can spot where the small chip came off.
Might I suggest that you use a scanner and get a circuit path layout put into Eagle along with a part legend? This would be only for posterity. It is just very interesting to see what all is used and perhaps even how it can be improved upon with higher budget/tolerance parts. For a majority of them I would assume that they are ghosted or unreadable so an oscilloscope would be needed.
I could definitely map the layout of the board(if its only 2 layer). But as far as the components identification, that could be really difficult. The 0201 resistors and caps would have to each be individually measured once off the board, and they're tiny. I think most of the ICs have labels, but the one that got ripped off did not, and it was a very small BGA package chip, so there are a few. I don't know how much more epoxy will come off without ruining the board, but I could just try heating it up again and removing components and ordering them in the way they were removed or label them.
The top is a known CC2511 RF chip(first picture), and the circuitry up there is mostly just for RF.
I'm really interested in the type of measurement they are getting from the sensor. Is it resistance, capacitance, etc?
Yep. Those are the batteries. You can see in these pictures how the one battery sits between the connections for the board. Interesting design.
This was as clean as I could basically get it, lost a few small components as well. I'm wondering if baking a good one in an oven at around 120C would cause it to be brittle enough to come apart. I did this all with a heat gun. The tough part is not getting the epoxy to hot or it starts to harden. There is a specific range around 100C-120C where its brittle and gummy and pops right off cleanly.
There are a bunch of ICs that don't come up on google when I put in their label. I'm wondering if the TI chip is doing the sensor measurement or not. It doesn't seem like it is, but rather only getting a value and sending it off.
How many milliamp hours are the batteries? The very low duty cycle of the transmission (a millisecond or two every five minutes) enable efficient use of the battery power.
The ICs may be proprietary design and not commodity chips.
Regarding measurement from the sensors, I saw a highly relevant post on Facebook today in the "Dexcom" user group. You might want to join the group and share your story there. There are almost 5k members, all who use a Dexcom. The group's description is, "This is a group for people who use (or know people who use) Dexcom⢠Continuous Glucose Monitoring technology. This is not run by Dexcom and is no way affiliated with the company."
Anyway, here is what someone posted to that group below:
The theories of high bgs causing the enzyme to get used up faster are correct. From a scholarly article:
"The operating principle of amperometric sensors is the measurement of the current flowing from an oxidation reaction, at a working electrode, to a reduction reaction, at a counter electrode [33]. To this purpose, a potential is applied between the working electrode and a reference electrode. Three electrodes are thus needed (working, counter and reference electrodes), although some sensors use a two-electrode configuration (working and counter-reference electrode), combining the counter and reference electrodes. Medtronic and Abbott use three-electrode configurations. DexCom uses a two-electrode configuration. In the case of glucose sensing, GOx is immobilized at the working electrode. GOx catalyses the oxidation of glucose to gluconolactone. "
So, the higher the BG, the more of the enzyme will be used up in the oxidation reaction, resulting in a shorter lived sensor.
And there's the science behind it. If anyone is interested in the article, it's linked below. It's from a few years ago and talks about the Seven+, but the principle is the same as the G4.
I couldn't tell you the exact mAh of these batteries, but I from the dimensions 11.6 x 2.05mm they seem to be ~60mAh. Says Google. They are very efficient in their design that is for sure. Of course the latest BLE can have Bluetooth devices last a year on a coin cell without too much work. But as MarkR's post suggests, they are possibly using current to get the measurement, so that is probably where lots of mA go as well.
The design of the antenna is just a simple "chip antenna", which is the big white piece on the one board. I'm sure there are a few figures online. From that DexDrip project he uses a Wixel to communicate. Which uses the CC2511F32 https://www.pololu.com/product/1337 - The chip on this board is a CC2510F32, which is most likely exactly the same.
There are definitely some proprietary things going on. Though I might be able to hand it to some of my more knowledgeable friends and have them identify it. If they are measuring current, then a lot of the circuitry could be there to measure the uA at a very precise level.There is also the same ocillator on each of the boards, which I'm assuming is for Low power, which would suggest that there is two microcontrollers. Where to go from here, I'm not sure. But If I get a couple more transmitters I now know where to be careful when removing the coating. I might even just be able to get to the spots where the batteries are connected and power the device myself and see if I can talk to it, and trick it into taking measurements from an old sensor.
Also, as a side note, I've been thinking of designing a stand alone "watch" that gives your real-time glucose measurement. It would be using the DexDrip project where it doesn't interfere with the receiver but just listens to the transmitter. The difference is you wouldn't need an android device or anything, could just put this small watch on, or stick it in your pocket/key chain, and have the reciever in your bag doing the graphs and still taking readings. Could be good for people when they exercise, or are doing active things, just to glance at every so often but still have the receiver near by.
The creator of the DexDrip did this exact thing already. That is why many users have ported things over to work with smartwatches now as well. ;) There are even some parents that use the Pebble smartwatch to do this for two T1D children and have split screen sotification for monitoring using NightScout. It's pretty cool!
Yes I have read a bit about the smart watches displaying the dexdrip data. Do donât you still need to have an android device to recieve the data from the Wixel and send it to the watch? I think it works great for those who have an android and a smart watch. That sounds like a great tool for parents, totally cool.
Iâm actually talking about something different; a stand alone device, it would replace the Wixel, smart watch, and android device. I would design the Wixel into a circuit board with an oled display. It would be an inexpensive replacement to those who donât have android phones and canât afford or donât want to buy an expensive smart watch. Basically a Wixel with a screen and battery/charger without all the bulkyness of a real Wixel.
THe Pebble and other smart watches are mostly all Android OS based so no need to have a phone and the watch itself if not wanted. If I recall correctly, the watches have WiFi capability as well so they can update through the cloud service that NightScout uses. It is a bit cumbersome, but seemingly working well enough for most.
I like the idea of a stand alone piece very much. For me, the only thing that would make or break it would be form factor. If you could somehow manage to do a circuit layout that used to spec SMD parts on a dual layer board (possibly with a daughter board for the OLED) that would give it its best chance of success.
Then you get into whether you want to have touchscreen capability or micro switches, which again adds to the complexity or the part count not to mention additional controllers for the touch screen.
I definitely think we could do something like this though. I still have my contacts in China for mass producing PCBs.
We could create our own stencils or have them laser cut (Likely the better option for materials and the ability to re-use)
I also have a good friend that works for NASA who used my 3D printer to develop his new pick n place robot. A PnP bot is mine for the asking there as well so if it would help?
The other remaining issues would be screen choice and form factor. We could get assistance in coding if needed as well. I have only dealt in design and hardware and some electronics engineering, I have never done any coding unless some basic C years ago counts. ;)
Either way, let me know if you would like me to help or not. I am just happy to see someone following their passion and trying to make the world a better place at the same time.
