670G and Dexcom

Has anyone tried to have a Dexcom CGM communicate with the Medtronic 670G?

My big misgiving about the 670G is that by all accounts Medtronic’s CGM is inferior to Dexcom’s. I know that the 670G is designed only to communicate with Medtronic’s own CGM. Is anyone aware of an attempt to “unlock” the 670G? I certainly don’t have the technical expertise to do it, but it seems like the kind of thing someone might be able to do. For example have a Dexcom G5 communicate with an iPhone and have the iPhone send out a signal that mimics the Medtronic CGM? Clearly a lot easier said than done, but just wondering if there is anything out there in the works among the technical elite of the We Are Not Waiting crowd.

I’m intrigued by the Loop system but wary of something so DIY and, from a quick internet search, the older Medtronic pumps that it uses are very expensive to get.

Automated insulin dosing systems like Loop are not for everyone. People part of the #WeAreNotWaiting movement believe that progress toward better treatments for diabetes has been much too slow. Here is a statement of their beliefs taken from the Looped Facebook page:

Looped is an an open and transparent effort to make safe and effective basic closed loop technology widely available to reduce the burden of Type 1 diabetes. The members are not willing to wait for an approved solution and instead are using Loop. The system is not something that you can purchase from a manufacturer but a system that requires work to install, support to use and equipment to purchase. If you want a perfect system, this is not for you. Support and troubleshooting is available here for OpenAPS, AndroidAPS and Loop system users.

I use the Loop automated insulin dosing system and follow a few online groups populated with people like me. I don’t know the answer to your question about reverse engineering the communication protocol of the 670G. I think this new technology, unlike the older pumps, is likely locked down well.

The do-it-yourself open source movement is grassroots and does not have a budget, at all. All the current DIY automated insulin dosing platforms came about due to the altruistic efforts of talented people. They made a strategic decision several years ago to spend their volunteer time on the “low hanging fruit.” That’s why the systems are usually based on older pumps.

By the way, the 670G sensor, the Guardian 3, by all reports I’ve read, is a very good sensor, comparable to the Dexcom G5.

Your wariness of the DIY open source automated insulin dosing systems is rational but many of us who have adopted this technology believe that the current diabetes treatment status quo is riskier than the thoughtful, vigilant, and intelligent use of these DIY systems.

I see that you just joined our community today. Would you like to share anything about your diabetes story or lessons learned from living with diabetes?

1 Like

Thanks so much for your reply, Terry. I am in awe of the people behind Looped – their inventiveness and their altruism. I hope that my reference to the cost of an older Medtronic pump did not imply that the inventors of Looped were making any money off of their incredible work. It just seems like there aren’t a lot old Medtronic pumps out there for sale and the ones that are command a high price.

You are probably right about the 670G being difficult or impossible to unlock. It’s just lack of knowledge combined with wishful thinking that made me bring it up.

About me, I’ve had type 1 diabetes for 25 years and Animas pumps for about the last 15 years. I’ve used Dexcom CGM for about 6 years.
Because Animas is defunct, I am considering what my next pump will be and the prospect of partially closed loops systems is tantalizingly close (and, of course, a reality for Loopers like you.) I am in warranty so don’t have an easy path to the 670G. (My understanding is that Medtronic is offering 630s to Animas users who are in warranty and want to switch.)
My negative take on the Guardian sensor comes from my endocrinologist who said that people who go from Enlite to Guardian are happy but people who go from Dexcom find it frustrating. Of course, with the supply issues, the amount of experience with Guardian may be somewhat limited. I also like the idea that the G6 is on its way and will be easier to insert. (My least favorite part of the CGM experience.)

Thank you, again, for taking the time to reply to my post.

One pump you might want to consider is the Dana pump made by the South Korean company, SOOIL. It is not yet available in the US but has recently become more available in Europe. It can directly communicate with an app running on a smart phone. The company is not hiding it’s communication protocol. This might be a good pump for someone who wants to DIY loop with a pump that’s under warranty.

Welcome to TuD!

1 Like

Dexcom will not interface with the Medtronic 670G. As a former Dexcom user, I can say I love the 670G and find it to be about as accurate as the Dexcom G5 for 6 of 7 days.

Not by my account. Switching from Dexcom to Guardian 3 was my biggest misgiving about moving to the 670G. My first week I wore both sensors, and if anything the Guardians were slightly more accurate. I’ve now been using the system for about 3 months and I’m really not seeing any difference from the kind of accuracy I had with Dexcom. The Guardians are not Enlites.

1 Like

Thanks for the update regarding accuracy of guardian, as it is one of my concerns.

I watched you tube video showing someone doing a guardian insert one handed, on their other arm. Seemed like a lot more steps with the inserter and taping, compared to dexcom. Have you gotten used to that ?.

How many days do you usually get per sensor ? I recall having to recharge the transmitter when I used Sof-sets.

I think I will hold out for tandem, dexcom combo, or whatever else comes out in the next 2 years.

Yes, the taping is a pain. I’ve got better at it, but avoiding having to change the darn things is the main reason I roll over to a second session every time.

I usually get two full sessions, i.e., 12-13 days. Some people recharge when they do the roll-over but I haven’t figured out how to do that without pulling the whole thing out and it hasn’t been necessary. But I’ve gone back to manual mode. I think Auto uses more battery because of how often it needs to “talk” to the pump.

The 670 auto mode seems to work best for people who are struggling to get their A1C down into the 6’s. For the 9 weeks I tried it, I was running around 6.5 - 6.7, just by the numbers (didn’t actually have a blood draw during that time). Since I was running around 6.0 before going on it that wasn’t making me very happy, and the struggle to get it to do better was aggravating. I finally decided I was working harder for the pump than it was working for me, so I switched back to manual.


Im with DrBB. I find the Guardian sensors far more accurate than Dexcom. And I regularly get 2 weeks out of them.

Dan, How many times per day do you test in order to verify Guardian accuracy? And are we talking accuracy often within 2 points, and nearly always with 10 points? How about accuracy during fast moving bg’s, taking into account a 15 minute delay (ie, test on meter, wait 15 minutes, and see how well the sensor data correlates with the meter reading). Also, what meter(s) are you using? Any model of a Contour?

I’m encouraged to hear that so many of you like the Guardian Sensor.
I also appreciate Terry’s suggestion regarding SOOIL. That’s intriguing.

About the Guardian Sensor, my sense (based not on data but word on the street) is not so much that it is inaccurate, but that it is frustrating – principally that its sensors have more unexplained failures than Dexcom’s do and more generally that it is a little less forgiving. For example, I read somewhere that it just stops showing BGs if you don’t calibrate. I’ve never used one, so forgive me if this is misinformation.

Specifically to Rphil2: could you clarify what you mean when you say it is accurate 6 of 7 days? Do you mean it takes a day to warm up, that the accuracy falls off after 6 days…?

Thanks for all the knowledge and experiences you guys are sharing.

I can comment on MM sensors, Tnyc. You are correct that when it’s time to calibrate, you get no further readings until you do so. Quite an issue for me. I sometimes go more than 10 hours past calibration time, on my dexcom and it is still accurate (I’m sure you already know that!)

Wow, that’s good to know, and would be a major con for me. I’ve definitely gone over a day without calibrating on my Dexcom, and it has remained fairly accurate.

1 Like

Yep, even if I’m around 200, the accuracy is within a few points. Just happened a few minutes ago, AAMOF. 189 on sensor, 194 on meter. I never had that kind of accuracy with the Enlites in more than a year of usage.

It was the same for the MM Sof-sensors, before the enlites. So I tried to do calibrations 3 times a day to avoid times with no data. Sof-sets were nicknamed the harpoon.

@Dave44 With fast moving BG’s, like 2 or 3 arrows, its usually 30 points behind. Normally I test 3-4 times/day. The 670G normally needs at least 2 per day, once every 12 hours. If, while in auto mode, the auto basal is at max for 4 hours, it will require another correction. Kind of like, “Hey, Ive been pumping a boatload of insulin into you for 4 hours. Lets check your BG to make sure Im not killing you” kind of thing. Im using the contour meter that comes with the pump.

To comment on your 10 hours past calibration time…its a different ballgame with the 670G in auto mode. Sure, with a Dex you can go 10 past calibration time, but the Dex isnt making micro bolus calculations on your behalf. Thats the difference.

@Tnyc Yes its true. The Guardian sensors require much more intervention. But think about it. Its making micro-bolus decisions on your behalf, so there really sint room for error. There cant be. I have had an hours worth of micro-boluses as low as .005u and as much as .500u. Thats every 5 minutes. Of course its going to require a BG sanity check so that it doesnt kill you.

1 Like

I have no problem calibrating. I test 4-6 times a day anyway because I like to confirm my readings. I kind of resent engineering that creates an artificial barrier to my seeing the CGMs readings, but that’s minor.

A query. When I first got my Dexcom, I entered my finger stick results every time I did a test. I figured more data equals more accuracy. A call to customer service set me straight. All that data was creating noise and making my Dexcom more inaccurate. I now only calibrate when it’s off by a lot or when it asks for a calibration and that has made my Dexcom more accurate. Does the same principal hold for the Guardian, or does it benefit from extra calibrations?

Finally, to Guardian users, do sensors fail regularly? Do they hold up well with physical activity – sweating or swimming, for example? If you restart a session after seven days is the second session about as good as the first? Better? Worse?

Thank you for all the replies!

Good question. I’ve enjoyed many second weeks where my Dexcom system produced more accurate and consistent results in the the second week than the first.

Hi. I am new to the community. Nice to meet you all. I did a google search to see if someone had figured out how to use Dexcom with the 670.

After having used the 670 for 6 months, I am read to give up on it. The problem is the unreliable and inaccurate CGM. It is junk compared to Dexcom and the closed loop system does not work well due to bad data. The fact that it is not FDA approved for treatment proves to me that it is not as good as Dexcom. I used Dexcom for a prior to Guardian and it was awesome.

I believe if they trusted thir own CGM, the target could be lowered to 100 and the entire system would be much better.

My A1C went from 6.2 to 6.8 after using the 670 for 6 months so I am going back to the 530 + Dexcom. My doctor’s office at University of Washington said this is a common problem with the 670 and numerous people are unhappy with the CGM. Besides it is a disaster to insert and remove (the tapes are sticky and messy). Accessing reports is also terrible and he upload site is buggy. Dexcom is amazingly simple.

Ideally I would get Dexcom to work with the 670 and lower the target to 100. Anyone figured out how to to either of these?


If the guardian was trustworthy like the Dexcom, they would not require you to test after 4 hrs at Max or 2 1/2 hrs at min delivery. I am
Totally frustrated with it. They are so close to building a great system but the CGM is their Achilles Heel.

1 Like