February 7 and 9, 2002 Meeting Notes

 

In attendance:

 

John Carmack

Russ Blink

Phil Eaton (sat)

Bob Norwood (sat)

Joseph LaGrave (sat)

 

 

Catalyst Pack Work

 

We finally figured some things out. All of this work has been with a 0.25” throat motor with a 1” diameter by 2” long catalyst pack chamber, with up to 0.6” of spacers supporting the shorter packs. All tests have alternated 20 mesh stainless steel screens with each silver screen.

 

Last Tuesday, we had a screen based engine that seemed to be working well. We couldn’t get it clear with 40 silver screens at even very low flow rates, but adding 10 more silver screens let it produce nearly 40 pounds of thrust at 800 psi. There was a hint of cloud in the exhaust, but it didn’t seem to be deteriorating. We ended the day with a very long run that still seemed to be working well. All of the runs were among the smoothest we have ever seen.

 

Last Saturday, the engine was cloudy through everything we tried. Our theory is that the long final run on Tuesday had been enough to heat soak the brass anti-channel rings and compress them against the chamber, so after it cooled off they were no longer a tight fit, and the engine channeled. When we took it apart, the rings and the final dozen or so screens were indeed a looser fit than when they were put in.

 

This Tuesday, we had two new things to try. We have a sheet of 80 mesh silver screen that McMaster just started carrying. It is extremely overpriced at $350 / square foot, but we will be able to compare three different mesh sizes when our other order from India comes in. The cut discs look more like cloth than screen. Note that McMaster informed me that their catalog is in error, the wire diameter is 0.0045”, not 0.0055” as listed. The open area is 31.4%, down somewhat from the 46.2% of 20 mesh screen.

 

The other thing we were trying is using stainless steel gapless spiral retaining rings (McMaster 92370A161) as anti-channel rings. The hope is that these will retain their spring and make a good seal even after thermal cycling. They are a bit of a hassle to get in the bore, but they make a very tight seal. We decided to put anti-channel rings every 10 silver + 10 stainless discs, instead of just near the top of the pack, because it seems like the last quarter of the pack is where much of our activity takes place, and that has show shrinkage of the screens after long runs.

 

We built an engine with 10 x 80 mesh (alternated with stainless, like always) + ring, then three sets of 10 x 20 mesh + ring, for a total of 40 silver screens. We hoped that the fine screens would reduce the total required, but the run was cloudy. We ran it a couple more times to see if it needed to bed in at all, but it didn’t clear up.

 

We added another set of 10 x 20 mesh (plus stainless) + ring, which brought us up to fifty total silver screens, the same as the pack from the previous week that ran well. It was still cloudy, and didn’t improve over a couple runs.

 

We had no theory for this. We took the engine apart, and everything looked good. The anti-channel rings were very difficult to press out, and sprung back to their full size without any show of taking a set. We were about to try building an engine with nothing by 80 mesh screens, but we decided to do some tests on the catalyst first. This was a Very Good Thing.

 

We laid out samples of all the things we have used as catalysts, and dripped peroxide on them for comparison.

 

The silver plated foam instantly catalyzed the drops.

The pure silver foam instantly catalyzed the drops.

Juan Lozano’s silver plated screens almost instantly catalyzed the drops.

The used 20 mesh pure silver screens sizzled and slowly catalyzed the peroxide.

The fresh 20 mesh pure silver screen was initially unreactive, but eventually cooked it off, becoming more reactive later.

To our shock, the fresh 80 mesh pure silver screen did NOTHING. The peroxide just sat on it.

 

We wondered if it might not actually be silver, but we decided to try washing the screens. We washed one in Tide and hot water, and when retested it was much more active, but didn’t seem to stay that way for long. We then tried washing one in Tide, then rinsing with isopropal alcohol (making damn sure that all the alcohol was gone before testing with peroxide!). The screen was now almost as reactive as the foam, causing drops of peroxide to skate around on their own decomposing gas bubble. We tested this several ways to make sure of the exact process. Cleaning just with alcohol didn’t improve fresh screens at all. Cleaning just with Tide helps some, but doing both makes a huge difference.

 

So, we have a clear lesson:

 

CLEAN YOUR SCREENS BEFORE PACKING THEM!

 

That explained the 50 screens not working, because the first 10 were completely unreactive.

 

We made another pack with just cleaned 80 mesh screens, trying 40 screens to see if the now highly reactive 80 mesh screens would work with only 40. The pack warmed up almost instantly, a huge difference from the previous screen packs, but it was still not clear.

 

On Saturday, we added another ring and ten more 80 mesh screens (intermixed with 10 stainless screens). The pack took longer to warm up than it had on Tuesday (but still faster than our earlier screen packs), which we will have to track on future days, but it cleared rapidly, and has stayed that way since. We made a total of 19 runs today, including a 110 second final run. All but the last run were with 500 – 700 ml of peroxide, for a total of over four gallons (400+ seconds) of peroxide through a 0.25” throat engine. It was running every bit as clear at the end as the start. This seems Good.

 

Some lessons that we seem to have learned:

 

We have seen on four occasions now that there is a very non-linear function at work with the amount of silver screen required. 40 silver screens doesn’t seem to be able to make a clear run even if only 5 pounds of thrust is made, but 50 screens can make nearly 40 pounds of thrust. We will probably make some larger nozzles so we can see just how much we can flow through the 50 screen pack before it starts getting cloudy.

 

The stainless steel screens give the pack good physical characteristics. Halfway through testing today, we added five more stainless screens to the bottom to get more pressure on the pack, but it isn’t clear if we needed to do that. This is in huge contrast to a pack of just silver screens, which gets more compressed every time you run it.

 

The spiral anti-channel rings are working great.

 

The copper gaskets between the flanges are working great.

 

This motor had a fairly significant volume ( 0.75” diameter by 1.5” long ) after the catalyst pack. There was some speculation that post-pack volume would allow more thermal decomposition, possibly helping avoid some cloudy runs. We don’t have back-to-back tests with a smaller volume, but we had plenty of cloudy runs with it, so I don’t think it has much value for peroxide engines. This seems to make sense – if 35% of the peroxide was not decomposed by the pack, there isn’t enough energy in the peroxide that decomposed to even vaporize all the remaining peroxide. It would be like running 65% concentration peroxide. Even vaporizing the peroxide isn’t guaranteed to cause it to decompose, so I doubt that any size post-pack chamber can make up for more than a 10% or 15% pack leakage, which isn’t all that much. With biprops and hybrids, the combustion chamber is high enough for full thermal decomposition.

 

Raising pressure never seemed to clear up a cloudy engine, either. If it was clear, it was clear at 200, 400, 600, and 800 psi. If it was cloudy, it was cloudy at every pressure we tried.

 

Smoothness

 

We still do not seem to have the magic formula for smooth running. Some of our screen packs have been the smoothest engines we have ever run. The first run today was very smooth, but it got worse over testing, at one point being nearly “maximum roughness”, bouncing around the entire level of thrust.

 

We tried adding more inert screens so that the pack had to be slowly drawn closed with the screws, giving it a lot of static pressure. Didn’t help.

 

We tried changing back down to smaller peroxide jets to give more pressure drop into the engine. Thrust went down, but it didn’t smooth out.

 

We tried running at higher pressures. Thrust went up, but id didn’t smooth out.

 

Then a few runs later as we were messing around with fuel jets, the monoprop runs started getting smoother again by themselves.

 

There are a few things we could investigate:

 

Try putting the spreading plate back at the top of the engine. These have been working fine just relying on the screen to spread the peroxide, but it might be a contributor (although the previous, ultra-smooth engine didn’t have a spreader).

 

This engine had a silver screen at the very top, which would be best for low latency in an attitude engine, but there might be something to leading with a couple stainless screens before it hits silver.

 

We might experiment with mixing different mesh silver screens.

Possibly the pack compression force is pickier than we think, but that would make it tough to deal with even a slightly relaxing pack.

 

 

Biprop Work

 

Our current test motor makes for a very, very poor biprop, but we went ahead and did a bunch of work with it. We had good enough success that we will be addressing the issues and pursuing the work. The main problems are:

There isn’t nearly enough chamber volume. The chamber is 0.75” by 1.5”, and the injection point is a third of the way down. That makes for an L* of only18, which is certainly not enough. Huzel & Huang lists L* for peroxide / RP1 as 60-70, counting the catalyst pack. That works out to be a chamber of basically the same size as the catalyst pack, or 1” by 2” in our case.

 

The fuel injection isn’t atomized at all. Because of the way the plumbing was set up, the fuel metering jet is upstream of the solenoid, so the flow into the chamber is just going to slop in. Ideally, we would have two impinging jets, but I think that we will still be much improved by arranging the plumbing so the jet is right at the chamber, causing a high velocity stream to shoot across the chamber underneath the cat pack. Using gaseous ethane will remove the entire issue.

 

It doesn’t really effect performance, but having the fuel jet on the pressurized side of the plumbing means that we have to close off the tank and leak some fuel whenever we change a jet. The oxidizer jet is in an NPT fitting, rather than an AN fitting, which is also more of a hassle to change. The pressurization is by two separate tanks in blow-down mode, which means we probably never get the exact same conditions twice, and loading the fuel is sort of a hassle. We are going to move the fuel to a different tank that is dual ended, and experiment with using regulated pressure common to both propellants (with check valves, of course). We will be limited to around 200 psi with that fuel tank, but that will work out ok for us.

 

Last Saturday, we did all the necessary modifications to the test stand and plumbing, and we tried to light it a few times, but because our catalyst pack wasn’t working well, it never did light.

 

Today, we had absolutely perfect decomposition, so we were expecting things to go well. We had gasoline in the fuel tank, and started off with a 0.060” peroxide jet, a 0.020” fuel jet, 250 psi in the fuel tank, and 200 psi in the peroxide tank. The procedure was to let the computer fire the peroxide valve, and manually pulse the fuel solenoid to see how it reacted. The peroxide fired and very quickly went to perfect decomposition, but pulsing the fuel solenoid just left a mist of fuel coming out of the engine with no combustion.

 

We tried a 0.040” fuel jet, and it had basically the same behavior. We went and got some kerosene to replace the gasoline, and running again with the same parameters, we saw our first signs of combustion. Kerosene definitely lights easier than gasoline.

 

We did a lot of runs chasing down mixture ratios that worked reasonably. Running the fuel valve manually was interesting. Sometimes during a single peroxide run I would try the fuel and it would fail to light even if I held it down a long time, but then releasing and trying again a couple seconds later would let it “catch”. Our best results were with an 0.060” peroxide jet with pressure at 350 psi, and an 0.024” kerosene jet with pressure at 300 psi. The kerosene is injecting after the cat pack, so it has a larger delta-P for a given tank pressure (we have check valves on all the lines).

 

Most of the runs we did with just a few seconds of hot firing interspersed in the 15+ second long monoprop runs. The chamber walls were 0.625” thick, so there should be a pretty good margin for heat sink cooling. There were noticeably different thermal effects on the engine after a run with some hot fire time, but everything held together fine, and we didn’t even wreck the screw threads. There was no throat erosion whatsoever. For the last two runs, we set up a hose to just pour water on the engine while it was operating, and we let it hot fire for ten seconds without a problem. We will probably make a water jacket cooled chamber, but as a near term solution, I am going to buy some machine tool cooling spigots that we can use to direct water at different parts of the engine.

 

We will probably fix up all of our issues and make some healthy mach diamonds in the coming weeks.

 

Getting a biprop to light was a lot trickier than lighting a hybrid.

 

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

 

 

Pending Stuff

 

We are waiting for lots of stuff to show up:

 

Laser altimeter from Laser Atlanta. I had problems ordering the ALTM-400, and wound up getting this one instead, because it was in stock.

 

Sintered silver plated nickel foam. This is similar to the foam we have been using, but the sintering should make it last a lot longer. We will likely repack the main lifting engine with this and fly test flights with the altimeter before we consider building a new screen based engine. If the new foam still wears our rapidly, we will make a new engine with a pack deep enough for screens.

 

Lots of 20 and 32 mesh pure silver screen.

 

Pair of 13’ x 8” chord rotor blades.

 

Flange mount tapered roller bearings for the rotor hub.

 

A tank of ethane. ($250, not too bad)

 

Furfuryl alcohol, propargyl alcohol, manganese acetate, and samarium nitrate from city chemicals.

 

Joseph is going to pour us a concrete pad out at my 100 acre lot soon. We will do the high RPM rotor testing out there, and our first untethered flight of the seated lander.

 

I am making progress on a big reworking of most of our software infrastructure.