Jan 11 meeting notes:

Location: Norwood Autocraft, then moving to Long Range Systems

In attendance:

John Carmack
Phil Eaton
Russ Blink
Bob Norwood

Darin brought the accelerometer sample board on Tuesday, but it turns out that the one we got was a vibration accelerometer, so we can’t use it for inertial navigation purposes.  I am hoping to get some other people interested in pursuing a 6DOF rate sensor for multiple groups’ use, but we will probably still pursue a single linear accelerometer ourselves in the meantime.

The VTVL frame is done.  The frame itself, without tanks or engines, is about ten pounds.  We certainly could have saved mass by making it out of composites, but this should work out fine for our phase one development plan.

There are some more pics at:
media.armadilloaerospace.com/misc/vtvl2.jpg
media.armadilloaerospace.com/misc/vtvl3.jpg

The changes from vtvl1.jpg are the wire tie points between the legs, the addition of the slip fit leg mounting, and the mounted engine.

The cross bar is 1 1/8” .060 wall 4130 tubing, and all the other tubes are 3/8” .035 wall 4130 tubing.  The legs slip fit into tubing one size larger that is welded through the ends of the crossbar, and is held in place with a cotter pin.  The legs are tied together with wire to keep it from bending at the very top, but there is still lots of spring in it.

There is a thin plate of 4130 welded to the top that we will be mounting our electronics box on top of.

Each engine mounting tab is welded to the crossbar, and each engine solenoid screws directly to a matching tab that allows us to cant each engine individually for the fixed engine roll control algorithm.

It disassembles and transports well.

It can be dropped repeatedly from a foot or two with no problems.  We dropped it from three feet, and there was a slight bend in one of the legs at the wire attach point, which was simply bent back.  We may need to replace the wire with something with a little more give, because we are probably going to have ten more pounds on the platform when it has all the engines, plumbing, and electronics, as well as some peroxide in the tank.

Russ is going to make a custom manifold that screws directly onto the tank, and has eight 1/8” NPT ports around the side, and a flat bottom.  We will mount a protective plate and a rubber bump stop on the bottom, so it should be structurally sound even if it comes down hard enough to bend all the legs out.

We picked up another nitrous tank at Bob’s to leave with our test stand, so we don’t need to disassemble the VTVL every time we want to do an engine test.  We picked a small 5lb bottle, which turns out to be a lot more convenient that the 12.5lb carbon wrapped bottle we have been using, because it draws a vacuum quicker, takes less nitrogen, and takes less time to vent.

We got our second batch of foam silver plated, so we have plenty of catalyst material.  We still haven’t received the silver screen from India.

We did three engine tests once we got back to Long Range.  Our primary goal now is to get the thrust up to where we need it for the VTVL: 7kg of thrust.

The engine expansion section has been cut down to an appropriate value now, so the nozzle should hit 7kg with a 300 psi chamber pressure.  We really need to get a pressure transducer in the chamber, because we are just theorizing about it now.

We started out with about a third of the pack from last session removed, and we took more out with each test.  We got more thrust with each test, but we are down to a pack well under an inch thick, and we are thinking we should just build another pack and not compress it so hard.

We know that if it isn’t compressed to some degree, that we will get a wet run, but it seems we have it far too dense right now.  A cavitating venturi would almost certainly be a big help, but we are probably going to see if we can get where we want to be by just balancing the pressure drop.

The test runs are graphed on: media.armadilloaerospace.com/misc/jan11.xls

All runs are with 200ml of peroxide of around 88% mass concentration (170ml of 98% plus 30ml of distilled water), and 400 psi of tank pressure.

The first run didn’t have a noticeable discrete warm up pulse on the graph.  We had made a shorter feed line tonight, so 200 msec of warm up pushed enough peroxide into the engine to make thrust until the main shot came on.  I shortened the warm up pulse to 100 msec for the later runs.

I believe the spikes at the end of the runs are related to the increasing chamber volume before the catalyst pack as we shortened it.  We didn’t have anything to shim it from the bottom, so there was a pretty fair volume between the inlet and the pack.  When the peroxide in the tank runs out, nitrogen will start blowing past the good sized puddle of peroxide in the engine, instead of forcing it through the pack.  I suspect that the puddle catalyzes on that side of the pack, giving a higher chamber pressure than we get through the pack.  That is just a guess, but the effect was increasing as we shortened the pack.  If we had a vertical test stand, I suspect we wouldn’t see it.

The engine screw closures started leaking a bit tonight, and got worse each time we pulled it apart to change the pack.  Russ is going to make some copper gaskets for it this weekend.

We ran into a new problem for the first time tonight.  After our first run, the solenoid did not close properly.  We noted that it was odd that we didn’t have residual pressure when we were getting ready for the second run, but we chalked it up to the smaller tank.  Because we have a check valve, nothing odd happened while drawing a vacuum and loading the peroxide, but as soon as we pumped a little nitrogen in, everything came out the engine.

The solenoid was definitely off, so we wondered if we had somehow worn it out.  I cycled it off and on a couple times, and it was back to working fine.  We still don’t know exactly what happened, but I added a double tap at the end of the controlling script that may help it seat better in the future.

We are holding off on more engine testing until we get the gaskets for the engine closures made, and hopefully something we can use to support shorter catalyst packs (washers?)..

Since the last update, I finally understand why I was confused about magnetometers giving three degrees of attitude sensing.  A magnetometer just gives you a vector, which is two degrees of freedom.  Some satellite systems can get the third degree by collecting data over entire orbits, but that obviously isn’t any help to us.

So, a magnetometer can’t solve all our problems.  We might still use it for some of the sensing, or at least some calibration, but we are going to need something else as well.

I integrated support for the Gyration GyroMouse into the simulator, and it looks like it works well enough for us to use it for our initial hover tests.  It is only two degrees of freedom, so we won’t have roll control, but we can probably lift off with it.

The Precision Navigation magnetometer / inclinometer compass arrived this week, and I have it working with the simulator.  Using it in it’s full compass mode gives exactly the three degrees of freedom we need, but the inclinometer is extremely sensitive to vibration, so it isn’t going to be workable.  I will be experimenting with just using the magnetometer data to get two degrees of freedom.

The Billingsly magnetometer and interface board from JP Aerospace also arrived, but I haven’t hooked it up yet.


The single board computer that we will be using for the flight computer also arrived today.  I will be experimenting with it this weekend, probably trying to do a small linux install to the 8mb disk-on-chip flash chip I got with it.  Darin (or anyone): do you have any tips or pointers for making a tiny embedded linux install?


Things to get/do:

Friction tape for the bars holding the tank in the VTVL
Garden hose to screw onto the sink at Long Range
More 5 gallon buckets
Medium size pipe clamps for securing the engine to the test stand
Hazmat stickers for our peroxide stuff
The PWM program should also report the log file name when you exit
The first line recorded by PWM is often slightly malformed