For accelerometer Freescale and Analog Devices provide some good options.

I don't own any accelerometer based altimeters, though I have built two of my own to experiment with.

Does anyone know what is the range used in the commercially available units? The chips can range from 1.xg to well over 50g per axis.

I'm tempted to settle on the +/-18G per axis, knowing that a small amount of data will be off the scale on certain flights. I'm also wondering it that's enough sensitivity.

I've looked with lust upon the 6DOF all-chip inertial sensor system at Sparkfun.com, but it maxes out at 6G's. If those parts were say 40G's with 10 bit resolution, we'd be talking some serious fun... That would get you .02G resolution if the sensore behavior is relatively linear. The crux is it would have to be at the absolute CG of the rocket unless you wanted to write a whole bunch of code to correct for it's location in the bird and the non-balanced moment arm. I believe the accelerometers are Freescale chips and perhaps they make a plug or pin compatible 40G version. The gyro chip does 500 degrees/sec so it would have a problem with rapid rates, but for a stable rocket with limited spin rate, it would be good enough.

This link gives pretty good references: http://www.edn.com/article/CA472836.html

The blacksky I'll use again for the N drag race at Oktoberfest goes to 50 g's, and when I put it back on the board, I'll get the chip id and--hopefully--find a data sheet online.

Does anyone know what is the range used in the commercially available units? The chips can range from 1.xg to well over 50g per axis.

I'm tempted to settle on the +/-18G per axis, knowing that a small amount of data will be off the scale on certain flights. I'm also wondering it that's enough sensitivity.

The compromise that I'm happiest with is a 2-axis accel by Linear ADXL22286, available at Digi-key, that has 70 g's in one axis and 35 g's in another. In the Parrot's implementation, with 3.6V going into the sensor, and a 13-bit A/D operating over a +/-2V range, I get about 0.025 g's in the axial direction, and .0125 Gs laterally. The lateral measurements aren't nearly as useful as the axial measurement, but it's still sometimes interesting for checking out the deployment/impact dynamics, adjusting for board or sensor mis-alignment (in 1 direction) , and estimating spin rate. When there's a cross-wind you can see the axial sensor oscillate at the frequency of the spin, until the rocket weathercocks enough to eliminate the angle of attack. Without a significant cross-wind, or after weathercocking is done, you'll get a constant signal that's proportional to the spin rate and the lateral distance from the CG to the sensor. I saw this sort of signal in Chris's cirrus dart flight data, and was wondering what could be causing an apparent constant lateral acceleration that was proportional to axial velocity, until I realized it was from the spin.

By the way, I'd be more tempted to go higher than 70 g's for the sensor range, rather than lower, because as long as the sample rate is high enough (>=200 Hz), and the range is high enough, you can get a lot of useful information about deployment dynamics (good for troubleshooting), and still have plenty of sensitivity for the thrust phase of flight. The best reason to have high sensitvity at the low end, in my opinion, is for getting good Cd data after the burnout, but 1/40th of a g is still good for that.

If cost, board space and A/D inputs weren't a factor, it would be great to use a +/-250 g single-axis accel, plus a +/- 15g 3-axis chip.

The biggest tip that I can pass along is to have a plan to mitigate temperature effects. Each piece of the analog chain for each sensor is affected by temperature, either in scale, offset, or both. And in my experience, these are pretty signifiant effects. On Chris's flight, the accel that was calibrated at room temperature was showing 1.14 gs while it was sitting on the pad, because it was in a black rocket on a hot day and got up to 128 degrees F. (I know because the microcontroller I'm using has a built-in temperature sensor) I'm planning to calibrate each sensor end-to-end, at two different temperatures, and do the math on board to do at least linear compensation for temperature.

We chose to go with the MMA2202 50+/- accel sensor on our flight computer. It is already conditioned for noise and works very well at a 12 bit resolution: http://mbrocketry.com/projects/elec/flight_comp/flightcomp.html
We just finished a production board run and I plan to have a couple of these units (along with the MiniAlt) with me at Oktoberfest to test fly and or purchase.

Warren, Mike,

We're all on the same 'wavelength'. I picked up a MaxStream development kit as well. With my homebrew GPS tracker in the front seat it was pretty cool to watch the wife's car move around the neighborhood on "MS Streets and Trips" on my home office PC.

I also built a talking GPS/baro altimeter that will transmit MP3s, liftoff, count out altitiude, and then range and bearing of the rocket upon descent. I built this for the Saturn but did not fly it as it was only hand wired and was a bit too fragile. Also, broadcasting over a FRS radio is not allowed if you're playing by the rules (which I should do as I don't want to lose my HAM license)

The current project on the bench will use the MaxStream for the telemetry and then have the ground station responsible for the narration.

Mike I saw a post that you were selling a couple of Maxstream modules. Is that still the case? If this project works I may want a couple of more modules.

I was selling them, now I need them again. I have a total of 6. I am using two, I have two off with Brad to integrate into the RMCS flight computer so that leaves two if you need to borrow a couple.

Does anyone out there use the full version of Eagle for their PCB layouts?

I am using the freeware version and am limited to a 4" long board. I would like to make one a couple of inches longer to accomodate the Maxstream but I'm a bit limited (in more ways than one). Perhaps there is someone who can do the final layout for me.

Here's a pic from Eagle 3D

We used PCB Express because their software is free and the demo board runs are only 60 for three boards. We just did a fifty board run for about $500 which gives us 50 of each the minialt and full blown flight computer and 50 serial interface boards. I will post a picture of the board run later tonight.

I use Orcad--version 9, and used to use it at StorageTek to design boards. Haven't used it in years though. Wish I could help. We used to send our boards to Rigiflex; very quick turnaround.