August 7, 2001 Meeting Notes

 

In attendance:

 

John Carmack

Phil Eaton

Russ Blink

 

Electronics Box Electrical Noise

 

Russ took a look at the electronics box with an oscilloscope today, and it wasn’t pretty. There were 50mV signals at 100 Mhz all over the box, which nicely explains the seven bits of noise I see on the A/D channels.

 

We were planning on trying out some single pole resistor / capacitor filters on the A/D breakout board, but it doesn’t look like that would help much, given that there is 18” of ribbon cable between the breakout board and the A/D board.

 

If we move away from the PCMCIA 802.11b card to a separate base station that communicates with the pc104 stack by Ethernet, we can enclose the entire pc104 stack in a shielded box, with just a few slots for ribbon cables to exit. There is still noise in the A/D board, but that should prevent a lot of the noise that gets picked up by all the other wires in the box.

 

The longer-term solution is to move the A/D out of the pc104 stack completely, and send the results over either USB or Ethernet. I’m not sure of the perfect product for this yet. I need proper documentation to access it from Linux. It needs to be very low latency, not just high bandwidth, which may not be the case for communication over USB or Ethernet. I want at least 32 channels, with the ability to individually pair them for differential measurements.

 

 

Engine Tests

 

media.armadilloaerospace.com/2001_08_07/SmoothnessTests.xls

 

We tried several things to smooth out the roughness in the 70 pound thrust attitude engines, with mixed results.

 

All tests at 500 psi initial tank pressure, with one liter of 80/20 peroxide, except test 3, which had only 500 ml of peroxide.

 

Test 1: reduced the catalyst pack thickness from 21 uncompressed discs to 13 uncompressed discs, adding a spacer underneath the retaining plate to keep the pack at the top.

 

This was smoother than the runs with the full catalyst pack, and produced a small amount more thrust, but it got a little rougher towards the end.

 

Test 2: We ran the exact same parameters again, to see if the roughness that showed up at the end was permanent.

 

It ran at least as good as the first time, so the reduced pack height seems to be a solid improvement.

 

Test 3: We have never determined what the lower limit for cat pack size is, so we cut the pack down to 7 uncompressed discs for a test run. Because we thought it likely that this would be a wet run, we only used 500 ml.

 

It still made a decent amount of thrust, but it was clearly failing to catalyze everything, giving us the first cloudy plume we have seen in a long, long time.

 

Test 4: We have always noticed a preferential wearing of the cat packs in the center, even with our micro-etched spreading pates. We put a strip of metal over the center of the spreading plate, so no peroxide would directly flow out of the fitting and into the middle of the pack. We used 13 uncompressed discs.

 

It was fairly smooth, but we were somewhat surprised to see a bit of cloudy exhaust. We are led to believe that pushing the peroxide away from the center may cause some channeling. It made only slightly less thrust, so it was probably only channeling a couple percent, but it was definitely visible.

 

Test 5: We took the injector blocking strip off and ran it again, which should be identical conditions to runs 1 and 2.

 

It was still showing some cloudiness in the exhaust, and thrust was down a bit. Sigh.

 

Test 6: We compressed 13 discs to the height of 7.

 

It had a slightly different pattern of noise, a slight drop in peak thrust, and it also showed an effect we have seen before: compressed packs have less thrust variance as tank pressure drops. We aren’t sure what to make of this. I favor compressing the packs somewhat, because they will compress themselves during operation, and doing it ahead of time should cut down on run-in variation.

 

Test 7: We ran one of the other engines from Saturday with the 21 disc packs for direct comparison.

 

It was definitely rougher across the board, so we are going to go to 13 disc packs for all the engines.

 

We are unsure of the lifetime of these cat packs. The engines in the lander are still working well after over two minutes of firing times on the various hops. The second batch of foam we had plated used slightly different parameters, which we think might be allowing the silver to be worn off faster.

 

On the last run, we also ran into the valve problem we had seen a few weeks ago – the NOS Pro-Race solenoid stuck on. It does have a warning about not sealing bubble-tight below 450 psi, but sticking open (until you tap it with a wrench) is damn poor behavior. If it had just happened with one valve, we were willing to let it go as a bad valve, but now that we have seen it on two different valves, we aren’t going to use these for the manned vehicle. We may still hop the frame unmanned with them (as far as we know, it only shows up as a failure to close after venting the tank down to low pressure after all the firing is done), but I am now actively looking for more valves again.

 

Our required specs are:

 

303/316 stainless steel body

Viton or Teflon seals

12V, direct acting solenoid drawing less than 15 amps current

Able to open and close within 10 msec, and capable of surviving continuous pulsing

As much flow as we can get

 

We can move down one step on the NOS solenoids to a model that flows a little less, but hasn’t given us problems, but this is probably a good time to shop for some higher end valves anyway.