Aug 21 and Aug 25, 2001 Meeting Notes

 

In Attendance:

 

John Carmack

Phil Eaton

Russ Blink

Bob Norwood

Neil Milburn

Joseph LaGrave

 

Plumbing

 

Russ got a new tank manifold cut with a half inch straight through port. He originally tried it in stainless, but couldn’t cut the 1/2" thread with a tap, so we went back to aluminum. We now have aluminum manifolds on the little lander, the big lander, and the test stand, with no more NOS bottle manifolds in service.

 

Each manifold has a fill port quick connect, a 1000 psi pressure gauge, a pressure transducer for the computers, and a burst disc or relief valve. We are still waiting for our larger burst discs, so we are moving the small NOS ones around right now.

 

We made a length of –10 AN Teflon hose for the test stand. It has a slight leak at low pressure, so we should probably cut that end off and try again, or just get a pre-assembled one from McMaster.

 

I have ordered a 3 gpm DC powered Teflon diaphragm pump from PEP-Plastics. This can pump into 30 psi of pressure, so we won’t need to draw vacuums in the tanks, and will generally make things a lot faster now that we are doing tests with over a gallon at a time. It was $2500.

 

I have put down the deposit for 5000 pounds of peroxide from X-L. We are getting it at 90% instead of 98%, so we will be able to skip the dilution stage in the future.

 

 

Large motor testing

 

On Tuesday, we first tested with an existing solenoid, and it didn’t make any more power than the smaller motors, clearly showing they are the limiting factor. It probably flowed more peroxide into a lower chamber pressure, but the additional over expansion more than compensated.

 

We then hooked up the big motorized half inch ball valve and made a run at 400 psi. We got a bit of a surprise:

 

http://media.armadilloaerospace.com/misc/LiftingTestStand.mpg

 

The data collection turned out to be clamping at half scale, so we just saw it maxing out at > 300 pounds thrust for most of the burn.

 

On Saturday, we installed concrete anchors for the test stand, moved up to the 3000 pound load cell, and ran it again at 400 psi, then at 600 psi.

 

media.armadilloaerospace.com/2001_08_25/TestStand.jpg

 

media.armadilloaerospace.com/2001_08_25/BigMotor.xls

 

This is a lot rougher than we would like, and somewhat below our estimated thrust. Now that we are out of the micro-motor range, Russ is going to try and cut some cavitating venturiis and see how they work for us. They will be designed to slip into a 1/2" union fitting, and we will probably start out with a 1/4" throat, which should still be good for 400 pounds of thrust.

 

New Data Acquisition

 

I finally have everything weaned off of AC power, so we can move everything to the test range when needed. I have a Dataq DI-195B with a load cell signal conditioning module, and (on order) another straight voltage module for the pressure transducer mounted in a box with our solid state relay that starts the test runs.

 

I have run into some difficulties with this setup. On my laptop, I get jittering values in the three least significant bits, but when hooked up to my desktop, it only has LSB noise. The laptop only puts 5v on the serial port, vs 10v for the desktop, but it is odd that it works at all, and just degrades the values. Phil is going to make an RS232 buffer and see if that fixes the problem.

 

One really nice feature of the new setup is that I wired it up so I can control the test stand solenoid with the serial DTR line, instead of having to run a separate parallel port cable from the computer.

 

Big Frame Hop

 

We securely glued the foam pads on, so we don’t have to rely on duct-tape any more, and Bob cleaned up a lot of the wire and hose routing, so it is starting to look really good.

 

I changed the electronics box so that it can control a single engine in the center with a solenoid signal. We will eventually be putting the throttled ball valve in the center, but this allowed us to stick a 70 pound thrust motor in the center to effectively just lighten the lander mass for the attitude engines. We might wind up doing this for the early manned flights as well – jet the big motor so that it makes enough power to almost lift the vehicle without a pilot, and just treat it as an on-off motor, letting the fast acting attitude engines do all the throttling just like on the small vehicle. That lets me dodge the motor-response time control law issue for a while.

 

media.armadilloaerospace.com/2001_08_25/CentralMotor.jpg

 

Bob is going to be building mounting plates for the big motor and adding a shelf for the computer next week (the computer is currently bolted where the pilot is supposed to stand), at which point we will be just about ready to strap in our first “pilot” a heavy punching bag donated by Christian Antkow.

 

We loaded up six liters of peroxide and pressurized to 500 psi, which should be good for around six seconds of flight. Gain was set at 0.010.

 

The central lifting engine worked fine, allowing it to lift off with a lot less throttle than with just the side engines. I jostled it off the ground a bit, then let it fly up to a reasonable height. I let it come down a bit harder than I wanted, and the computer died. Resetting did bring it back, so it was a transient effect. From the accelerometer logs I was able to tell that the computer was ok when it hit the ground, but there was a fairly severe +/- 4G acceleration at impact. The closed cell foam isn’t nearly as soft as the open cell foam on the small lander, so things don’t get quite as soft of a landing.

 

We have video of this, but it will be a few days before I can get it on the web.

 

We opened the electronics box up and checked things out, looking for something that could have caused a temporary reset. Twisting the PC104 stack is a possibility, but one of the power leads to the CPU board screw terminals was relatively loose again, so that is our leading candidate. Russ pulled the little terminal block off the CPU board and directly soldered wires to the board, and I have it all back together in that form now. I tried to move our relay driver board off of the top of the PC104 stack and onto the box lid to reduce twisting on the stack, but things didn’t quite work out, so it will probably have to wait for electronics box 3.0 and a major reorganization.

 

If I had been a little more delicate with the throttle, this would have been flyable for the proper amount of time. However, we are still planning on testing my new asynchronous flight control code next week, which will hopefully make it smoother and more responsive.

 

The back-and-forth roughness is proportional to the sensor latency plus the propulsion system latency times the engine control authority. The small lander’s engine arrangement lets it fly smoother, because the engines have less of a lever arm, and the landing gear is extended out past them. We may want to consider moving the engines on the big platform, or testing again with the short form after we work out all the control system issues.