Understanding OpenAPS Compatibility: Devices and Sensors for a Closed-Loop System

OpenAPS (Open Artificial Pancreas System) is a community-driven, open-source initiative that enables people with insulin‑dependent diabetes to automate insulin delivery. By connecting a continuous glucose monitor (CGM), an insulin pump, and a small controller (typically a single‑board computer), OpenAPS can adjust basal insulin rates in real time based on glucose data. Compatibility among these components is the cornerstone of a safe and effective system. This guide details the devices and sensors you need to assemble a reliable OpenAPS rig, along with practical considerations for sourcing, configuring, and maintaining your setup.

Core Components of an OpenAPS System

Every OpenAPS loop requires four primary hardware elements: an insulin pump, a CGM, a controller (microcomputer), and a radio communication interface. Each component must be compatible with the open‑source software and, in many cases, with one another at the hardware level. Below we break down each category and highlight the most commonly used options.

Insulin Pumps

The insulin pump is the actuator in the loop, responsible for delivering micro‑boluses and adjusting basal rates. Not all pumps are suitable. OpenAPS works exclusively with pumps that have a history of reliable, low‑level radio communication and are supported by the open‑source community. The most widely adopted pumps include:

  • Medtronic Paradigm Series (5xx, 7xx) – These are the gold standard for OpenAPS. Models such as the 515, 715, 522, 722, 523, and 723 use the 916 MHz radio frequency and are fully documented. Older Paradigm pumps (511, 712) often lack the necessary radio commands. The newer Medtronic 630G and 670G are not supported due to encryption and proprietary communication protocols.
  • Medtronic Revel (x23/x23M) – Minor firmware differences exist between Revel and Paradigm models, but the vast majority are compatible with OpenAPS after proper setup.
  • Roche Accu‑Chek Combo – This pump communicates via Bluetooth and has growing community support, though it requires different controller hardware and configuration steps. The Accu‑Chek Insight is not supported.
  • Omnipod / Omnipod Dash – The original Omnipod (with the Eros pods) can be driven through the Omnipod implementation in OpenAPS using a Rileylink or similar radio board. The Dash pods use Bluetooth and are not currently supported (though AndroidAPS has partial support for Dash).
  • Tandem t:slim X2 – While not natively supported in OpenAPS, Tandem has released a commercial closed‑loop system (Control‑IQ) that shares lineage with the open‑source approach. For DIY looping, many users prefer to stick with Medtronic or Omnipod Eros.

Before purchasing a pump, consult the latest compatibility lists on the OpenAPS documentation site and the community forums. Older pumps may require a battery door magnet or a specific firmware version to enable remote commands.

Continuous Glucose Monitors (CGMs)

The CGM provides the sensor data that drives the algorithm. Real‑time glucose readings are essential; any latency or gap in transmission can cause the loop to operate blindly. The following sensors are currently supported:

  • Dexcom G6 – The most popular choice for OpenAPS. It provides readings every 5 minutes via a dedicated receiver or smartphone app, and it does not require finger‑stick calibration (though one can still calibrate). OpenAPS can ingest data from the Dexcom G6 through the Share API (for G6) or via a Bluetooth bridge such as xdrip+. The G6 has a 10‑day wear time and is factory‑calibrated.
  • Dexcom G5 – Still supported in many OpenAPS installations. It requires twice‑daily calibration but offers similar real‑time data. The G5 transmitter uses Bluetooth Low Energy, making it easy to connect to the controller. Note that as of 2025, Dexcom has officially discontinued the G5, though transmitters may still be found.
  • Dexcom G4 (with Share or 505 firmware) – An earlier model that requires a separate radio bridge (e.g., a CareLink USB stick) to communicate with the controller. It still functions well for many users who have existing hardware.
  • Abbott Libre 14‑day / Libre 2 – These are flash glucose monitors, not continuous, but with third‑party transmitters (such as MiaoMiao, Bubble, or BluCon) they can be converted to near‑real‑time CGMs. The Libre 2 in some regions has Bluetooth that can be accessed with xdrip+ and a compatible transmitter. Compatibility varies by firmware version and region, so careful research is needed. The Libre 3 (fully Bluetooth) is not yet supported in OpenAPS due to lack of an accessible data stream, though community efforts are ongoing.
  • Medtronic Enlite / Guardian Sensor 3 – These sensors are intended for use with Medtronic pumps and require the Medtronic MiniMed Connect or a CareLink USB stick to relay data. Integration with OpenAPS is possible but more complex, and the sensors have higher calibration requirements. Most users prefer Dexcom for its ease of use.

For all CGMs, ensure that the data can be transmitted to the controller without relying on a proprietary cloud service that may introduce lag. The community strongly recommends using a local‑first solution such as xdrip+ or Nightscout.

Microcomputer (Controller)

The controller runs the OpenAPS software and orchestrates the loop logic. The most common choices are single‑board computers that can run Linux and communicate via serial, USB, or Bluetooth with the pump and CGM radios.

  • Raspberry Pi (3B, 3B+, 4B) – The workhorse of OpenAPS. The Pi 3B/3B+ offers built‑in Wi‑Fi, Bluetooth, and sufficient GPIO pins for attaching a radio transceiver (such as the CC1111 or Rileylink). The Pi 4 is also supported but consumes more power; many users prefer portability with the Pi Zero W.
  • Raspberry Pi Zero W – A compact, low‑power option ideal for portable rigs. It lacks Ethernet but has Wi‑Fi and Bluetooth. Its single‑core processor is adequate for OpenAPS, but some users report slower compilation times during setup. For daily operation, the performance is fine.
  • Intel Edison (with Arduino breakout board) – Used in earlier generations of OpenAPS. The Edison is no longer in production, but many rigs still run reliably on it. It has built‑in Bluetooth and Wi‑Fi, plus GPIO for radio modules.
  • UDOO Neo / BeagleBone Black – Occasionally used, but community support is thinner. The Raspberry Pi is recommended for new builds due to the wealth of documentation.

Regardless of the board, the OpenAPS software is installed using a custom image (such as the oref0 distribution) that includes the loop logic, communication drivers, and a web interface. The controller must be able to run autonomously 24/7 without crashing. A dedicated power supply (e.g., a USB battery pack) is recommended for portable use.

Radio Communication Interface

To talk to older Medtronic pumps (which use 916 MHz frequency) and some CGMs (e.g., Dexcom G4 with Share), a radio transmitter/receiver is needed. The controller itself does not have a 916 MHz radio, so an external module bridges the gap.

  • CareLink USB Stick – Originally designed for Medtronic’s own software, this stick can be flashed with open‑source firmware (e.g., via the mmeowlink project) to enable communication with Paradigm pumps. It is inexpensive but requires Windows for initial setup and is limited to Medtronic devices.
  • Rileylink – A custom radio board designed for the open‑source community. It supports both 916 MHz (Medtronic, Omnipod Eros) and Bluetooth (for some CGMs and the Omnipod Dash, though Dash support is experimental). The Rileylink connects to the Raspberry Pi via USB and does not require tinkering with a CareLink stick.
  • CC1111 USB Stick – An alternative to the CareLink stick, the CC1111 can be reflashed with mmeowlink firmware. It works well with Medtronic pumps but is less commonly used due to the availability of Rileylink.
  • Bluetooth Modules (for CGMs and newer pumps) – Many CGMs (Dexcom G5/G6, Libre with third‑party transmitters) already use Bluetooth, so the controller’s built‑in Bluetooth is sufficient. For the Accu‑Chek Combo pump, a Bluetooth dongle may be needed if the controller lacks native Bluetooth (e.g., older Pi models).

When selecting a radio interface, consider the ecosystem: if your pump is Medtronic, a CareLink or Rileylink is mandatory. If you are building for Omnipod Eros, a Rileylink is required. For future‑proofing, the Rileylink is the most versatile option.

Compatibility Checklist for Sensors and Devices

To ensure a smooth build, verify each component against the OpenAPS hardware compatibility page. Below is a summary of what works with the current stable release (oref0 and oref1).

  • Pump: Medtronic 5xx/7xx (excluding 511, 712, and 670G), Accu‑Chek Combo, Omnipod Eros (requires Rileylink).
  • CGM: Dexcom G6 (native Bluetooth), Dexcom G5 (Bluetooth), Dexcom G4 with Share (requires CareLink stick or Rileylink), Libre with MiaoMiao/Bubble/BluCon (compatible with xdrip+).
  • Controller: Raspberry Pi 3/4/Zero W, Intel Edison (deprecated).
  • Radio Interface: Rileylink, CareLink USB stick (flashed with mmeowlink), or CC1111 stick.
  • Software: oref0/oref1 (latest release), Nightscout for remote monitoring, xdrip+ (for CGM data on Android).

Always cross‑reference with the OpenAPS official site and the community spreadsheet maintained at the oref0 wiki on GitHub.

Additional Considerations for a Reliable Setup

Firmware and Software Updates

OpenAPS is continuously evolving. The development branch (oref1) introduces advanced features like dynamic ISF and carb‑sensitivity adjustments, but it is considered less stable than the master branch. When building your system, choose a release that matches your comfort level with risk. Always test firmware updates on a non‑lifestyle‑critical rig (e.g., during a time of low activity) to verify that radio communication and loop decisions remain safe.

Safety and Redundancy

An OpenAPS system is a medical device, even if it is DIY. Build in multiple layers of safety:

  • Low‑glucose suspend thresholds – Configure the algorithm to stop insulin delivery when glucose falls below a user‑defined level.
  • Maximum basal rate limits – Hard‑code a cap to prevent pump‑induced hypoglycemia.
  • Backup glucose meter – Finger‑stick checks remain necessary to calibrate CGMs (except Dexcom G6) and to confirm readings when the loop or CGM fails.
  • Independent alarm system – Nightscout or xdrip+ can send alerts to your phone. Do not rely solely on the controller’s screen.

The community emphasizes that OpenAPS is not a medical device approved by the FDA; it is a tool for individuals who accept the responsibility of managing their own care. Read the OpenAPS risks page before proceeding.

Power Management

A portable rig needs reliable power. A Raspberry Pi Zero W draws about 0.7 watts; a Pi 4 with Rileylink may draw 3–5 watts. Use a high‑capacity USB battery bank (10,000 mAh or larger) that supports pass‑through charging so you can recharge it while the rig runs. Ensure the controller’s microSD card is a high‑endurance model to avoid corruption from unexpected shutdowns.

Community Support and Documentation

The OpenAPS community is one of the strongest pillars of the project. Forums, such as the OpenAPS Facebook group and the #openaps channel on the Nightscout website, provide real‑time help. The official documentation is thorough but can be dense; many users start with the “Getting Started” guide on the OpenAPS site and then move to the GitHub wiki for device‑specific steps.

Building Your First OpenAPS System: A High‑Level Overview

While we avoid step‑by‑step instructions here, the general process is as follows:

  1. Gather hardware – Acquire a compatible pump, CGM, Raspberry Pi, and radio interface. Check serial numbers against the compatibility list.
  2. Set up the controller – Flash the OpenAPS image to the Pi’s SD card, connect the radio interface, and boot.
  3. Install Nightscout – Set up a cloud‑hosted Nightscout site (e.g., via Heroku or Azure) to receive CGM data and upload pump status.
  4. Configure the CGM – Connect the Dexcom or Libre transmitter to xdrip+ (or other uploader) and verify data flows to Nightscout.
  5. Test pump communication – Use the command‑line tools (e.g., oref0‑test‑pump) to confirm the Pi can talk to the pump.
  6. Run in open‑loop mode – Let the algorithm suggest basal changes but do not automate yet. Manually confirm the suggestions are reasonable.
  7. Switch to closed‑loop – Enable automated adjustments, starting with a conservative target range and low‑glucose threshold.
  8. Monitor and iterate – Review logs daily until you are confident the system is behaving as expected.

This process typically takes several days to a week. Do not rush. The OpenAPS Getting Started guide is the authoritative resource.

Future Developments in Hardware Compatibility

The open‑source diabetes ecosystem is evolving quickly. Projects like AndroidAPS (which runs on Android phones) are expanding the range of compatible hardware, including Bluetooth‑only pumps (e.g., Dana RS, Dana-i). While AndroidAPS and OpenAPS share similar algorithms, their device compatibility differs. Users considering a new pump should research both ecosystems. Meanwhile, commercial closed‑loop systems (Tandem Control‑IQ, Medtronic 780G) are becoming more sophisticated, but they remain closed to DIY customisation. For those who value control and data transparency, OpenAPS remains a powerful option.

Conclusion

Device and sensor compatibility is the foundation of a successful OpenAPS implementation. By selecting a supported insulin pump (Medtronic Paradigm/Revel, Accu‑Chek Combo, or Omnipod Eros), a reliable CGM (Dexcom G6 or Libre with a third‑party transmitter), and appropriate controller and radio hardware, you can build a closed‑loop system that significantly reduces the burden of diabetes management. Always cross‑check the latest community guidelines, prioritise safety, and expect a learning curve. OpenAPS is not a plug‑and‑play product—it is a tool that rewards patience and attention to detail. With the right components and a supportive community, you can achieve time‑in‑range numbers that are difficult to match with manual therapy alone.