Eventually I'd like to develop two different flight data loggers. The first designed for model rockets, with focus on minimum size, weight, and the ability to place inside an existing kit which lacks a payload bay. The second designed for larger rockets, and acting more like a full-fledged flight computer. This one will focus on extensibility with auxilliary analog and digital inputs, general purpose I/O and the ability to trigger recovery events.

However, before I can do any of this I need to learn about how the individual components work and this requires the construction of a device for component testing. The early stages of this device will be a desktop device, and it will eventually graduate into a simple flight-worthy circuitboard.

When I have the basic modules working, I will move on to designing production-quality boards for use by the general public. Until then, everything on this page is to be considered experimental.

Micro Flight Data Recorder

Current plans call for four sensors - pressure, axial acceleration, axial magnetic field intensity, and temperature. Each sensor output will be sampled with a 12bit ADC with frontend preamplification and offset adjustment where necessary. The needs of each sensor will be determined on a part-by-part basis. I will attempt to incorporate adjustable scaling into the design with small SMT trip pots where possible.

The memory will be a serial EEPROM of at least 1Mbit. This gives enough memory for ~3.6 minutes of data (assuming 100 datapoints per second per sensor and a 6 byte data word). Initially, I was looking for a serial SRAM that I could use as a ring buffer while watching for launch indication due to the ~1Mcycle limitation on EEPROMs. But with 3.6 minutes of data, I can just use the entire EEPROM as a ring buffer and will not likely ever write to it more than a couple times while waiting for a given launch. Once launch is detected, each memory cell is written to at most once. This gives me around 300,000 launches before the EEPROM will go bad, so no problem. No ring buffer RAM necessary.

The MCU currently under consideration is the Cygnal C8051F007. Although it is a bit overkill, it has several very attractive features for this project. Most notable is the incorporateed 4-channel 12bit ADC and temperature sensor. Also, it has hardware support for I2C and UART, making my job very easy. When I advance on to more complex flight computers, the larger cousins of this processor will do nicely and the learning curve will be minimal.

The acceleration sensor is the ADXL150 from Analog Devices. Their datasheet has a nice schematic for a good two-pole post filter with trimpot adjustments for offset and scale factor. In this configuration, the acceleration sensor portion of the circuit will require 10 passives, 2 trimpots, 2 opamps, and the ADXL150 itself. This makes it by far the most part-intensive portion of the entire circuit, but its data is also the most useful.

The pressure transducer is the MPXA4115A6U from Motorola. The response time for this sensor is 1ms, meaning that it won't need a filter for any pressure oscillations above 500Hz. The amount of low-frequency audio present in a rocket launch, I'm guessing, is relatively minimal. A small RC lowpass filter may be needed here but I may be able to skip the front end signal conditioning alltogether with this sensor.

The magnetometer is the KMZ51 from Philips. This was chosen over the HMC1001 by Honeywell because the HMC1001 is only available in SIP and is difficult to orient the sensitive axis in the plane of the circuit board. These magnetoresistance sensors have compensation and set/reset coils on board for getting accurate field measurements. However, for rocketry I think the only information we're interested in that the magnetometer can provide is roll-rate information. In which case it might be unnecessary to implement the coil driver circuitry at all. I need to do some extensive testing with this part to determine whether this is true. This is the sensor about which I know the least because very few people have used them in their rocketry projects. They are very expensive and their utility for roll-rate determination has not yet been shown.

If the magnetometer ends up needing an opamp or two, I may go ahead and use the OP496 from Analog Devices, which is 4 opamps in a 14-pin SOIC. If I only need the two opamps I'll either go with the OP296 or two separate LMC7101 ICs from National Semiconductor.