Grim Reefer
- Mary Dertinger (RIT Student)
- Richard Sommers (RIT Student)
Members: @Mary Dertinger (RIT Student) @Alex Moczarski (RIT Student) @Sam Larson @Richard Sommers (RIT Student) @Louis Fleisher (RIT Student) @Dean O'Brien (RIT Student) @Jonathan Godfrey (RIT Student) @Braxton L. Crandall (RIT Student)
Attachments: Grim Reefer Schematics, Grim Reefer Manual (test and launch instructions), Payload data Analysis (data from Grim Reefer, RRC3 and Radiosondes IREC), L1 test flight data (fast.csv, slow.csv), L1 test flight data (L1 graphs), IREC data (merged.csv)
Status: Completed; successful READ THE FAILURE SECTION BEFORE YOU COPY ANYTHING!!
Project: IREC 2023-24 payload “Ritchie” mission was to record flight data and footage of reefing event.
Grim Reefer Requirements:
Detect boost and landing
Read loadcell output
Control camera power
Ability to arm on pad
Component Overview
Requirement | Component |
|---|---|
Processing power | STM32 |
USBc to UART | CP2102N |
Debugging | SWD connector, reset header, debugging header |
Board status indicators | LEDs and Buzzer |
Store data locally | Flash mem → W25Q512JV |
Voltage rail sensor | INA260 |
Temp and humidity | BME280 |
Altimeter | MS5611 |
IMU | LSM6DL |
ADC | MCP3561R |
Loadcell reading | Fully Differential Amplifier → LMH5401 |
Loadcell | 50kg ATO Tension and Compression |
7.4V Camera Power controller | Load Switch → TPS22810 |
3.3V rail for Grim Reefer | Buck → TPS565208DDCT |
5V for load cell | LDO → TPS7B8450QWDRBRQ1 |
Cameras | RunCam Split v4 |
Camera connectors | 2 pin JSTXH |
Battery connector | 2 pin screw terminal Link |
Arm on pad | FingerTech screw switch and buzzer (to note if grim is flight ready) |
Load cell connector | 4 pin screw terminal |
Power Delivery
Powered by a 2S Li Ion battery at 5000mAh. We needed three Voltage rails, excitation, grim reefer and cameras.
5V for Load Cell Excitation
Needs to be stable because load cell measures voltage as it undergoes compression/tension. Resistance chances during compression/tension therefore measuring a different voltage (classic V=IR) so excitation voltage for load cell must be stable throughout entire measurement to get best results.
For highest input to output ratio the excitation voltage should be run near max, which was recommended on the datasheet as 10V. We had concerns with the 10V supply being unstable and getting too hot. Temperature also changes resistance so we would have had to correspond what temperature the load cell was throughout flight and compensate for that in our analysis. We decided that 5V was acceptable for this application especially since we picked a 50kg loadcell instead of a 100kg loadcell.
7.4V for 2x Cameras (only on during flight → enables load switch once boost is detected and disables after 24 minutes)
Camera's voltage ranges from 5V-25V but need to be aware of power consumption and temperature. It is worth noting that these cameras were power hungry, and we would have been cutting it close with battery life if not for boost detection.
3.3V for Grim Reefer
Needs to be relatively stable to support sensor suite and STM32.
We have a simple LC filter on this rail see below in figure labeled 3.3V LC Filter.
Voltage Rail | Component | Rational |
|---|---|---|
7.4V (Cameras) | Load switch TPS22810 | Cameras could take any voltage between 5-25V and were only going to be powered on during flight so we didn't have to worry too much about power consumption. |
3.3V (Grim Reefer) | Buck from 7.4V (Battery) to 3.3V TPS565208DDCT | Grim reefer needs relatively stable voltage for sensors. LDO has lower noise floor but is less efficient - put a simple LC filter to help mitigate noise on Buck. LDO vs Buck |
5V (load cell excitation) | LDO from 7.4V (Battery) to 5V TPS7B8450QWDRBRQ1 | Load cell excitation needs to be stable - fluctuations in excitation will affect the measurement. |
Boost/Landing Detection
Two ways:
Acceleration based: 5G for 1/4 of a second
Altitude based: 500ft gain in altitude over 5 seconds.
The acceleration-based boost detection was the mechanism that triggered first for the flight and the L1 test flight.
Load Cell reading
Requirement: Read a 10ms event with a CALCULATED maximum force of 27kg.
50kg Load Cell Load Cell ATO Link → we had a granularity of about 6 grams.
Fully Differential Amplifier was picked was the LMH5401. This chip was definitely over speced (sorry Louis this one kinda broke da bank) but it worked well for our application. Has ability to implement certain gain - we did 12dB.
Anti-Aliasing filter, fairly simple and standard to implement in these types of applications. If you are an EE/MicroE you will learn all about it in Linear Systems, if not read about it here Anti-aliasing link. Pretty much you need to filter out unwanted frequencies that you could potentially misread. Cut off frequency should be 1/2 of max sampling rate (thanks Nyquist) so ours was 5kHz because we have to sample at 10 kHz to get enough data for a 10ms event. Simulation done in LTspice (spice is life) - shown below in figure Filter Sim.
Load cell excitation was run at 5V to GND. Usually load cells have +/- excitation voltage shown in Typical Load Cell Wiring diagram but instead of having a negative excitation we ran it to GND. You lose part of your input/output range (differential centered around 0 with +/- excitation vs differential centered around half of your excitation), but it was simpler to use GND and worked for our application.
Load cell calibration @Richard Sommers (RIT Student) and @Mary Dertinger (RIT Student) hung a bunch of known masses on the load cell then in the same order put on, they were taken off -hysteresis- shown below in Load Cell Calibration/Data.
Control Camera Power
We used the load switch TPS22810 from TI. Pretty simple implementation just don't mess up a 10 ohm for a 10 kohm resistor. CAM_EN goes to GPIO on STM. JSTXH were used to connect to cameras. Cameras must record for 12 minutes to properly save file (it will be corrupted if not) so we had the camera power run for 24 minutes to make sure we got everything. Upward facing camera video & Scenic Camera video
Successes
This was a successful project - it had a few hiccups, but it completed its objectives.
Success | Rationale | Would implement again? |
|---|---|---|
Boosting/landing detection | Very helpful - eliminates worry for excessive power drain on pad. See Grim Reefer Instructions/github for more details on SW implementation | YES |
Controlling camera power | Eliminates worry for excessive power drain on pad | YES |
Load cell reading (sorta) | Mission requirements but could use schematics/grim again for similar projects | Yes - as needed |
Sensor suite | Mission requirement for boost detection and helped us understand flight events especially when flight is not nominal. | Yes, but fix MS5611 mistake and I2C lockup issue. We got lots of helpful flight data for rocket and payload - see Payload Data summary attachment |
L1 test flight | Wanted to test boost/landing and overall functionality of Grim. | YES, super helpful and great to add in Tech Report - see L1 test flight data attachment |
Battery/power scheme (see Payload Analysis for voltage readings throughout flight) | Had to meet component voltage supply requirements. | Yes, would use batteries / setup again. |
Failures
Every project has its failures… here is a comprehensive list of ours (any figure/schematic explicitly mentioned are in the Failure Figure expand below)
Failure | Solution | Future Mitigation |
|---|---|---|
Wrong transistor on buzzer which was copied off of Auto Pilot Module (Q1 on schematics). Originally used AO3400A, a N channel mosfet see schematic below, Buzzer Schematic Failure. Need a resistor on this implementation. | Removed and replaced with a digital npn but could also have replaced with a sub-logic mosfet. Digital transistor is not a discrete transistor, it has a pull-down resistor and a current limiting resistor built in. It is simpler but can be either fully on or fully off, hence digital. This is ideal for our buzzer application. | Check schematics that are copied from other boards, especially when they have not been fully tested. We need a deeper level schematic check. |
MS5611 Schematics issue - SCK wired in I2C mode which was copied off of Sensor Module (U1 on schematics, see MS5611 Schematic Failure below) | Remove MS5611 and use BME280 for altitude (slower refresh rate but still suitable). | Check schematics that are copied from other boards, especially when they have not been fully tested. We need a deeper level schematic check. |
10 ohm resistor from 7.4 V → GND on load switch (R26 on schematics, see 10 ohm Resistor Schematic Failure) | Took off resistor - several had already completely popped off | We need a deeper level schematic check. |
MCP3561R was oriented wrong on breakout (See MCP3561R Incorrect Orientation Photo) | De-soldered and re-soldered chip with help from @Jim Heaney (See MCP3561R Correct Orientation Photo) | May have been JLC issue but if not - always check the component on JLC especially the ones we import components that are not made ourselves. |
I2C lock up issue | WIP (same as sensor mod) | WIP |
Issue with INA260 on camera rail ? | WIP | WIP |
Load cell sensor breaking - load cell underwent excess forces, but it read the initial snatch force then -8000 for rest of flight - see payload analysis attachment | Still determining failure mode (Likely due to ~7.5 sec free fall after deployment causing an unexpectedly high snatch force) so unable to pinpoint exact solution. We could have gone up to a 100kg load cell, but issue would have likely still occurred. | Likely more of a recovery failure, but make sure the static line to the parachute bag is strong enough, and assume there will be very high forces, and it's better to have less sensitive data than none at all.
|
Camera Exposure adjustment caused the first few seconds of video to be very overexposed, giving us less information immediately following separation (Minor, not sure if this belongs here) | None taken | Add some bright LED’s next to the cameras to lessen the difference between the inside of the tube and the sky. |
Measuring the acceleration of the buzzer (LOL see graph - Buzzer Accel - below. For more detail see Payload Analysis pdf. | WIP | WIP |