Friday, May 09, 2008

A Glimpse of scary little details...

I have not reduced the data from last weekends first static fire, but I know that we had a main valve calibration problem. Tonight I brought up the flight computer to work on the main valve calibration and I got prodigious quantities of magic smoke. (All electronics runs on magic smoke, if you let the magic smoke out it stops working.) This is scary on a number of levels. If the main flight computer dies in flight the vehicle is lost, so I start asking did I make a design error, did I have an assembly error? What caused the system to let out the smoke?  So I remove the computer from the vehicle and take it into my office/lab in the house and hook it up to the current limited power supply…. no smoke. The magic smoke happened inside a closed case so everything in the case smells so no clues there.

One by one I test the sub systems… Main CPU, OK,Vane control OK, GPS OK, IMU interface OK, Telemetry Radio OK, Servo drivers OK Pressure sensor power error not OK. So I examine the little sealed DC to DC converter component that provides this power and it is failed shorted with smoke /char marks under it. The good news is that this failure would not have caused a crash, a loss of  pressure transducer data, but not a crash.

This little power supply adds to the head room for the two wire pressure transducers. I use 2 wire 4 to 20ma transducers because of the reduced wiring, from the three wire transducers. The transducers need 10V minimum, since I run the into a resistor that makes 0–>5V from 0 to 20ma I need a 15V supply. I’d been running 11.1V to the Big tonegowa servos so I added a 5V isolated DC to DC to bring the voltage up to 16.1V. In past setups I’d added one very small sensor battery,  but this time I though I’d remove one more service item. The sad part is that this is not really necessary anymore. This time around the Vane Actuators need ~18V  so I already have a voltage rail, 18V that could drive these. The smoking power supply has been voted off the project.

I wired a sacrificial 1/8W carbon 20 ohm resistor between the sensor circuit and the 18V supply so I’m now good to go. If any of the transducers wiring gets melted and shorts to ground the 20 Ohm carbon resistor will act as a  fuse. Protecting the flight critical power rails.

We also learned another lesson tonight. With LOX you can be 100% sure its gone two days later, not so with peroxide. We ran the vehicle until there was noting but nitrogen coming out of the motor. Apparently this is not enough. When my son  removed the main valve it dumped peroxide on his jeans.  No harm done, but it was exciting for a second when he leaps up removes his pants and runs out of the garage heading for the garden hose… on the way by saying “make sure we there's no fire”  Fortunately none of the many fire extinguishers were necessary.  a few teaspoons of peroxide  makes for some excitement but in this case not much else.

This does say that our run tank and main valve is clean and properly passivated.


Anonymous said...

thank you for the humor :)

Anonymous said...

For what its worth, my experience with those sealed dc/dc converters is that they are very fragile - if it says 1 watt on the tin they will be reliable at 0.9W and toast at 1.05W. I think they have a short circuit fault tolerance of 1 second. Its best to give them plenty of slack or better still, leave them out.

Thad said...

Hey, my niece works for Flowmetrics, and she's having the time of her life down there. If you run into Judy Emaus, say hi for me!

It seems like an ideal company to work for -- small, but with incredibly capable people and state-of-the-art equipment doing a huge variety of exciting projects.

Keep up the good work, and thanks for educating people about the magic smoke.

Thad Beier