September 12 and 15, 2001 Meeting Notes

 

In attendance:

 

John Carmack

Phil Eaton

Russ Blink

Joseph LaGrave

Bob Norwood (Tuesday)

Neil Milburn (Saturday)

 

 

On Tuesday, we had a couple disappointing motor firings.

 

We tested the hybrid motor in monoprop form twice, but we couldn’t get it to run smoothly. We fired it with a grain again, and it lit promptly, but it also ran roughly, as expected. This was always a quick hack job motor, so we will be replacing it with a new one that we can get a good monoprop run out of before we put a hybrid grain in it. We will also have a proper nozzle size in the copper heat sink area, so we don’t have to mess with phenolic nozzle inserts. We may add a water jacket so we can just leave a hose running over it for longer rins.

 

We were planning on mapping the thrust curve for various fractional openings of the ball valve on the big motor. At 20% reading from the potentiometer, the valve is just beginning to crack open. We set up a run like this, but it was incredibly rough, actually banging the motor back and forth on the test stand.

 

We realized that the plumbing setup that we were using to try and get the smoothest possible flow to the big engine was causing serious problems for a minor throttle opening. We had the ball valve at the bottom of the tank manifold, then a four foot length of 1/2" hose going down to a T fitting that had our pressure transducer and a long brace bar to the back of the test stand. With the throttle just cracked open, it probably never filled the hose at all, just letting a rivulet of peroxide roll into the engine.

 

On Saturday, we rebuilt everything so the valve is right by the engine again. We put another 90 degree fitting back in to brace against the test stand, so we know to expect a slight drop in flow rate. We put the pressure transducer back on the tank manifold, instead of at the engine entry, because the T fitting could trap an air pocket if it was left upright.

 

The drilled ball valve on the test stand has become slightly leaky, giving a wisp of peroxide smoke wile pressurizing. We must have scuffed the seal with metal chips from the ball drilling. We haven’t tested the other ball valve since it was drilled out.

 

media.armadilloaerospace.com/2001_09_15/BigMotor.xls

 

We did two firings of the large motor with new orifice jets. Each run was two liters of peroxide at around 400 psi initial tank pressure. The first one had a 0.187” diameter orifice, and the second one had a 0.2” orifice. The runs were acceptably smooth, and scaled as expected.

 

When an engine is flow limited by a restriction, mass flow (and power, discounting nozzle expansion issues) scales with the area of the restriction and the square root of the pressure. An unrestricted motor, like the big one with 1/2" plumbing the entire way, scales linearly with pressure, but will be much more likely to run roughly.

 

This motor’s nozzle is grossly over expanded when run at these power levels. We still need to get a chamber pressure tap, but I suspect it is only running about 100 psi with this much restriction, and a 4x expansion ratio will be putting that will below ambient. 426 pound seconds of thrust (1900 NS) from two liters of 85% peroxide gives a whopping 76 Isp. Previously, we have measured 110 Isp on a motor operating at its design pressures. With 90% peroxide, no restrictions, and a 600 psi tank pressure, we should see about 130 Isp and over six hundred pounds thrust on this motor.

 

media.armadilloaerospace.com/2001_09_15/SmallMotor.xls

 

We made four test runs with the small engine to test some new configurations. All runs were with one liter of peroxide at around 400 psi initial tank pressure.

 

The first run was with a new engine top that was nearly flat, without the dish present in the original motors. There was also a single disc of non-plated nickel placed on top of the micro-etched plate, which filled the entire area. We ran this with the huge ball valve plumbing, and it was wildly rough.

 

We then put a solenoid in front of the engine to give it a restriction, and it ran very smoothly, but didn’t make a lot of power. Ideally, we would probably want a 0.1” metering jet on the motor with the big valve, but we used the solenoid for the rest of the tests.

 

We then tried a dished top closure, but filled with shrinking catalyst discs, like we used to smooth out the large motor. This also ran smoothly, and made slightly more power at a given tank pressure than the previous motor.

 

We then ran a motor in the original configuration, with an empty dome top. It was rough like the original tests.

 

We can draw a firm conclusion from the last couple weeks of tests – having a large space above the spreading plate is bad, it should either be filled in with catalyst, or have the spreading plate moved closer. Cutting different sized discs is a pain, and they may shift around, so future engines will probably have flat tops just a very small amount away from the spreading plate. I would like to test the flat top motor without the nickel foam disc at the top, because that may have restricted the spreading, giving the slightly lower performance per tank psi.

 

Phil figured out something interesting about many of our runs recently. There is a pretty consistent little disruption in the graphs about halfway through the small engine runs. It looks like that is the point where the fluid in the tank is running out, and the last 500 ml is contained completely in the half inch hose and plumbing going to the engines. There seems to be a little disruption as the dregs of the tank swirl into the manifold.

 

We have a new plumbing setup for the big lander that makes things worlds easier. Previously, the engine, valve, manifold, and tank were all screwed together with pipe fittings, so the entire stack had to be lowered down onto the mounting studs together, which was a huge pain. We now have the metering jet cut to replace the top of a –10 AN fitting that goes into the engine, and a swivel AN female fitting coming off the valve. This lets us bolt the engine in by itself, then lower everything else down on top of it and tighten it up. The engine jet looks exactly like the NOS nitrous jet from hell – nitrous metering jets are designed to fit on top of –3 AN fittings.

 

One of the adjustable braces for the big lander galled while we were adjusting it, so we are going to have to get it to Bob to fix before we can fly again. We figured out why we have had so many problems with them: when they powder coated the tubes, a lot of sand blasting grit was left inside them, which found its way down to the threads.

 

The reason we weren’t following up on the mapping of partially opened ball valves today is that I managed to burn out our motor drive board somehow while working on our master cutoff computer. I still have no idea how it happened, but we had been planning on replacing it with a custom board soon anyway, so this just made it a priority. The boards I have been using communicate over a serial port, and offer a lot of features, like variable speeds, that we just don’t need.

 

Russ was able to give the board a lobotomy, so that the driving transistors could be controlled by two binary voltage lines instead of with the serial protocol. This was exactly the interface that I wanted – forward, backwards, and break, controlled by two bits, taking effect instantly, and running completely isolated from the CPU power bus. I reprogrammed the flight computer to use this interface, and we plugged it in to test. It opened the valve, then closed the valve, but when I commanded it to open to 50%, it sat there for a couple seconds, then smoke and fire started coming off the driver board (more fire from our electronics than our rocket engines!). Phil got the power yanked out, and the post-mortem showed a diode going first, then the transistors. I only had a five millisecond minimum delay between direction changes, so it may have oscillated too fast for it around the fractional position.

 

Russ and Phil are going to pick up some high quality components and build a new driver board next week, and I am going to modify the code to just remove the variable seeking strategy. I am going to set it up for a single enable and disable per adjustment, and not care about the minor overshoots.