Home / News

News Archive

Nov 01, 2000 meeting notes


Location: Norwood Autocraft


In attendance:


John Carmack

Phil Eaton

Russ Blink

Neil Milburn

Bob Norwood


Next meeting at Long Range, unless notified.


We tested a small motor tonight, and learned a few things.


I took some pictures, which I will get on a web site sometime soon.


I brought a checklist for the procedure this time:


Fill and Fire Checklist


Everyone put on goggles and gloves.

Prepare buckets of water for emergencies.

Set the test stand up on the other side of a concrete wall from the controls.

Position the observation mirror for a view of the test stand.

Load the peroxide into the fill container.

Set the nitrogen regulator to 450 psi.

Open and close the launch valve. (it should have been left open)

Connect the fill cart to the flight tank.

Open the vacuum valve. (it should have been left open)

Start the vacuum pump.

Wait until the vacuum is drawn.

Stop the vacuum pump.

Close the vacuum valve.

Ensure the peroxide hose is properly submerged.

Hold the peroxide switch until all the peroxide is loaded.

Watch the tank pressure gauge for any unexpected increase due to


Hold the nitrogen switch until pressure is equalized.

Disconnect the fill cart and move it away from the test stand.

TEMP UNTIL WE HAVE A CATCH CAN: disconnect the vacuum line from the pump

Open the vacuum valve to release pressure in the fill system and leave it

in a safe state.

Start logging the load cell readings on the computer.

Optionally pulse the launch valve open and closed to warm the catalyst.

Open the launch valve and let the engine fire until the tank is drained.


After the engine has cooled, make sure it is drained of all liquid.

Should we always wash the engines with water?

Put a plastic bag over the engine before putting it away.


There was temptation to overlap different parts, like drawing the vacuum while someone is getting water.  I think we need to be more disciplined about following exact procedures, because we certainly aren't doing everything flawlessly while winging it.


The other checklist lesson learned is that we must acknowledge each checklist item before proceeding on.  "Start logging" was called out, but not acknowledged, and we didn't get a data log.


I wrote a little utility last week that simplifies the logging of load cell data for our testing.  We used to capture the meter readings in hyperterminal, then clip out the relevant sections in an editor and paste it into excel.  The utility lets us start logging from the command line, and closes the file when you hit escape, which lets you avoid having lots of extra junk in the file.  It automatically increments file names, so you don't need to worry about stomping on anything (which I did once when saving from hyperterminal), and it also automatically puts the file on the windows clipboard.


This was the fourth or fifth time we had set everything up for filling and testing, but it is still a huge hassle arranging power, nitrogen, tanks, and the vacuum pump.  We need to bump finishing the fill cart up on the priority list.  We should not (try to) fire an engine again until all the fill apparatus is completely nailed down.


Related to that is the load cell meter, which currently takes 120V.  We should try and work something out so we can run that without an extension cord.


We were testing a small 30mm motor that Russ made to the same basic design as the large motor we fired successfully last week.  This is the basic size we are planning for each of the four lifting engines on the VTVL platform. We had another pint or so of hand distilled peroxide.


Positioning the mirror for viewing the test stand was more of a hassle this time because of the limited space to move around behind the stand.  The wall and door at long range was more convenient.


There was a miscommunication about the catalyst warming pulse.  I had expected we were going to look at how the pulse went before deciding how to proceed, but Phil fired the motor quickly after the pulse (which is what we did last time).


The motor didn't work.  It rapidly jetted out all the peroxide as a foam, with almost no catalyzed gas.


The peroxide seemed good, as it was clearly decomposing on the concrete.


The catalyst pack was cut from the same foam as the last motor, and there was obviously some amount of decomposition going on inside the motor, so it wasn't completely bad catalyst.


We finally decided that we were just flooding the engine with too much peroxide for it to catalyze.


We had catalyst surface area proportional to the throat size of the bigger (successful) motor, but the feed system was still the same size.


We had been assuming that the feed system could be oversize without causing any problems, thinking that a greater peroxide flow would result in a higher chamber pressure, which would lower the peroxide flow, and everything would balance out with the smaller motors having a smaller pressure differential between the chamber and feed system.


That may or may not be true in steady state, but the lesson learned was that when you are starting at zero chamber pressure, a high feed pressure can push the peroxide through fast enough that you never get a chance to build up chamber pressure to slow the feed down.


We ran water through the little engine, and we were surprised at how strong a jet blasted out, compared to the gurgling flow we got out of the big engines.


With computer control of valve pulses it would probably be possible to build up the chamber pressure in a controlled manner, but the general lesson is that the feed system should be restricted so that it doesn't flow more than you need.


We changed the plumbing around so that we could use NOS nitrous jets in the feed line before the engine.  A 45 thousandths jet gave a water flow that looked similar to what we got with the big motors.  If we had more peroxide, I bet it would have worked.


Once we have engines firing reliably, we need to tap a port for the pressure transducer so we can see what chamber pressure to feed pressure deltas we are actually getting.


Something that we didn't consider is that we might want to try compressing the foam catalyst in the engines.  We were feeling good about the very low pressure drop across the catalyst bed, but it might turn out that a significant pressure drop actually makes it easier to start the engines. It certainly isn't crucial, because last weeks engine started well, but it is something to think about.


We have the basics down well enough now that we need to begin doing scientific development work.  To start efficiently accomplishing things, we need permanent facilities and a good supply of peroxide.


I am hopefully going to be looking at some places this week for us to set up permanent shop at.  I talked with the mayor about the various regulations, and it looks like the most serious issue for us will be the noise.  There are regulations regarding the maximum db at the edge of the property line.  We may wind up in an area outside city incorporation limits.


We need to have a large enough supply of peroxide that we can run several experiments in an evening.  If we had had more tonight, we probably could have had more successful runs after installing the restrictor jet.


The home-brew seems to work well enough, but I don't like the extra uncertainty that it adds to our results, and I am still nervous about us doing it in any quantity.


The 30 gallon drum of 90% from FMC is $1200 plus about $1500 in shipping, but they won't sell it to us until we have a concrete containment for the drum.  They sell gallon samples for $300, which Russ is pretty indignant about, but I think we should go ahead and get one.  That would only be a single full length firing of the large motor, but we can do a lot of work with the small motors with a gallon.


Dumb rocket update:


Neil has started picking parts for the dumb peroxide rocket.  It will be a simple 7" diameter rocket with a single stage recovery system.  Mounting the flight tank will be the primary challenge.  We should probably try to include my GPS telemetry system, which is currently contained in a 2' x 4" tube section.


We will make an adapter for either Russ's big motor or Juan's motor so that it fits in a standard 98mm motor mount, allowing us to test fly the rocket with water loaded into the tank on a certified motor to make sure all the recovery systems work and that the rocket is stable before running it with a peroxide motor.


It's going to be heavy.  At a minimum, we are going to have 12lb of dry hardware in the tank, manifold, launch valve, and engine.  More likely, it is going to be closer to 20lb unless Russ replaces the manifold and launch valve with smaller custom cut parts.


With 10lb of peroxide loaded, plus the normal rocket mass for airframe, fins, and recovery, it is clear that a 100lb thrust motor won't get it going fast enough to be stable off the rail.


Juan's motor was speced so that we could open the throat up and get 170lb thrust when we want to, and Russ's motor is even larger, but our current launch valve is not even flowing 100lb/s worth of peroxide based on the length of last week's engine test.


We have quite a bit of room for increasing tank pressure, which will help some.  We have been testing with 450-500psi, but the tank is designed for use at 1100+ psi (it is DOT approved, so the burst pressure is probably 3x that).  Bob has some 2000psi N2 regulators, but we need to pick one up for the fill cart.


We need to do more exact tests, but I am nearly certain that the primary restriction is our launch valve.  We are using an NOS remote bottle valve, which is functionally perfect for our needs (two position latching, current only required to change positions), but it doesn't flow all that well.  It is clearly more than sufficient to flood the 10lb engines, but the large engines seem to become feed restricted around 50lb thrust or so.


We know that the large nitrous solenoid flows significantly better than the bottle valve, but there would be two drawbacks to using it.  We would need to carry a battery on the rocket to keep the valve open, and when the battery wore out it would leave the tank closed instead of vented.


We could go to a large normally open solenoid for the launch valve, which would solve both the issues with the normal solenoid, but that has some significant safety issues during fill.  If the battery on the ground wore down during filling, the rocket would launch by itself.  That would be Bad.


Russ and Phil are investigating machining our own pyro-operated slide valve.  I would like to avoid pyro if we can, in the name of simpler turnaround.


I want to bring back up my earlier idea:  use a decent sized, pneumatically actuated valve, but put a check valve on the air inlet.  Connect an air hose to the check valve with a push-on barb.  To launch, pressurize the air hose, which will open the valve, which will launch the rocket, which will pull away from the air hose, but the check valve keeps the valve control pressurized anyway.  Russ: consider using an air hose instead of pyro in your design.


Both the pyro and pneumatic valves would prevent us from pulsing the engine to warm the catalyst, which may be an issue.


We are going to have to get our own launch insurance, since Tripoli's won't cover liquid propellant rockets.  I am pinging John Powell of JPA about their flight insurance provider.  Theirs is pretty expensive, so I hope it is pro-rated based on our far lower altitude.


VTVL platform


We have the four small solenoids for the engines now.  I took two and Russ took two.  I ordered a couple solid-state relays that should be able to drive the solenoids, but Russ is also building a custom circuit for them.


The fittings we need should be here any day now.  The engines, solenoids, fittings, and jets connect together into a nice rigid assembly for each corner of the platform.


I wrote code that runs multiple channels of PWM off of my laptop parallel port.  I tested it with LED's, but it should be ready to hook to the engines as soon as we get driver circuits.


How well PWM modulation will work is our big uncertainty right now.  We should be able to find out in the next couple weeks.  If we have to change to something based on proportioning valves (probably driven by stepper motors), I am concerned about finding something that will be small and accurate enough for the 10lb motors.  The full-flow jet orifice is already pretty tiny.


The square aluminum tubing I brought looks like it is going to work out fine.  A 3' square platform should be good for us.


Phil had a very good idea for the initial testing of the VTVL platform. Set up four posts around the testing area and hang the platform between all the posts with shock cord or cable.   Post and cord lengths can be selected so that the platform is hanging off the ground, and has several feet of free movement, but positively cannot hit anything or tip over.


If we start with this setup, we won't need landing gear immediately, we just need the frame for four engines and support for the tank and electronics.


I have a basic flight console program working.  It is designed to work with a modern joystick with twist for yaw and a separate throttle.  I wrote and tested it on a desktop system, but I have a USB joystick on the way that will work on my laptop.


The current "flight logic" is dead simple, direct control.  The four primary lifting engines are set to the throttle value modulated by the X and Y joystick positions.  The twist yaw action controls two additional engine levels (we don't plan on adding the yaw engines until we are forced to).


The engine levels are used to control PWM outputs on the parallel port.  I am defaulting to 10hz for the PWM, which is pretty marginal, but we will be experimenting to find out how fast we can go with our solenoids.


Theoretically, we could hook this up and fly with it.  In reality, it is a fair bet that a person will not be able to control it at all, let alone do anything intelligent with it.  One of the first things I will do to make it more controllable is limit the joystick modulation values to the minimum needed to tip the platform, instead of the full 0% to 100% range it currently is.


If it turns out that it is manually controllable, we will build landing gear and give it a longer leash.


If it isn't controllable, we will need to install the attitude sensor and go to a closed loop flight control system.  I think that taking joystick position as a small range (say, +-5 degrees) of desired pitch and roll values and letting the flight computer try to maintain that will be sufficient control to let a human control the craft.  Manual throttle control and indirect positional control by tipping the platform should be within the realm of what a person can manage.


Later, we may want to investigate fully automated landing and flight paths, but that isn't an overriding goal for me if remote piloting is working out ok.


Once we get it out from the posts, we still won't be flying it high enough for a parachute to do much good in the early stages, so there is a lot to be said for giving it a foam or inner tube base and flying it over water.


When it is hung between the posts, the direct control system could just be wired directly to the laptop parallel port with a long cable, or we can do a very simple one-way telemetry system.  While we decide what "real" flight computer we want to use, I can do crude telemetry PWM control by having a basic stamp just copy the serial byte stream from a radio modem onto the solenoid driver bits.  That would only allow 120 switching points a second, but the times could be dithered, and it would probably work out ok.


This would be raw FSK data, not packet.  To be legal, we would need to stay under one watt of power in the amateur bands.


Once we put a sensor on board, we are going to want a full duplex telemetry link with two radios.  The flight control logic will eventually migrate to the on-board computer, but development will be a heck of a lot easier if we can treat the platform like a device, reading its sensors and writing its actuators from the flight console.


I am getting the information for the serial output magnetometer board, which sounds perfect for our attitude sensing needs.


Once we let the platform off the leash completely, we will want a GPS on the platform.


Landing will be a lot easier if we have a laser rangefinder.  I have found a couple with serial outputs, but they are $2000 surveying units.


So, our real flight system will probably need four serial inputs (radio in, magnetometer, gps, range finder), one serial output (radio out), and at least six digital outputs (PWM engine drivers).


We are currently debating a very wide range of possible flight computers.


At one end, Russ is looking at tiny microcontrollers.  He is going to give me one of the dev kits to look at next week.


Terry Parks is suggesting something like the 68332 board available from:


or the MPC555 board available from:



And I am looking at using a WinCE palmtop as the flight computer with all the peripherals hung off the USB bus.



© 2001-2011 Armadillo Aerospace, LLC. All rights reserved.