2009-11-04

Wow, lots to talk about - haven't tested the new motor yet but it's getting much closer. One thing that has changed is the tanks on the test stand. I proof tested the stainless steel fire extinguisher tanks and all was fine but when I was cleaning them out in preparation for the cold water tests, I noticed about a one inch long shard of metal sticking up inside near where the bottom fitting was welded on. I'm guessing the hole didn't get deburred before the fitting got welded on. I managed to break off and extract the shard but then I started looking more closely at the inside of the tank and that's when I started getting nervous. The various weld seams of the tank are just lap joints and there is a little bit of an edge sticking out along the whole length of each weld. I could imagine all sorts of debris getting caught in those overlapping joints that would be difficult to clean because the lap joint geometry changes slightly as it is pressurized. That in conjunction with a general worry about fatigue life of the tank led me to change to a different tank. I previously had good luck with the E-sized aluminum oxygen cylinders so I decided to go with something similar. I found that a 15 lb CO2 cylinder would give me about the same internal volume (2.6 gal) as the 2.5 gal fire extinguishers and it is rated to 1800 psi. The thick bottom makes it easy to drill and tap a hole for a threaded fitting. This particular cylinder I'm using is the B15 model from Catalina Cylinders but I got them from BeverageFactory.com for about $90 each. The only real downside of using a thicker tank on the test stand is it will take a lot longer to chill down the LOX tank because it weighs 16 lb empty. I've posted a CAD drawing of a fitting for the CO2 cylinder on the Test Stand page.

The new purge system seems to work properly. Each propellant operates separately with its own solenoid valve and check valve. The check valve is adjusted so as the pressure in the feed line to the engine drops more than 40 psi below the tank pressure, the purge gas (helium, same as the tank) will blow out the propellant line and *hopefully* allow me to run back to back tests without having to take everything apart and clean it. When I checked out the solenoid valves on the test stand for the first time, I ran into an interesting problem. Whenever I would actuate the fuel purge solenoid, all the solenoids on the test stand would de-energize. I assumed it was some sort of software problem like writing a whole word instead of a bit but I couldn't find anything obvious. Then, I noticed that the indicator on my screen for the 24V "power good" went out whenever I energized that particular solenoid. It turns out that solenoid had an internal short to case on the pin I happened to pick for the positive side of the power supply. I took it apart and never actually found anything obvious but the problem was occurring 100% of the time and now I'm not seeing it at all anymore. I went ahead and swapped the solenoid wires internally just to be safe as a short of the negative terminal to ground won't cause any adverse effects. I also noticed the other solenoid valve (LOX side) didn't seem to be putting out as much flow as the fuel solenoid valve. When I energized the coil, it didn't make the same strong click as the other one. I took it apart, cleaned it, and reassembled it with some help from Circle Seal for the adjustments. They emailed me a service and test procedure for the SV series valves so if anybody needs it, just let me know. Both of these valves came from eBay and they looked to be in nearly new condition when I received them. In any event, both of them seem to be working properly now.

I re-calibrated the load cell up to 140 lb - that's all the weights I had. Plus, since these aren't slotted weights that you would typically use in this application, I had to lift the entire amount each time I added more weight on and my poor back wouldn't take much more. I was curious to see how much static friction the linear slide had since that was a concern and it was so much less than the cable/pulley friction that I didn't even notice it. However, I did measure the load cell during one of the cold water tests and the flex hoses generated about 2 lb of force on the motor when they were pressurized.

I've completed the cold water flow vs. pressure tests went but I ran into another interesting problem. I was seeing strange zero shifts on the fuel pressure transducer about about 10 psi. I initially suspected a grounding problem because whenever I would bump the test stand, the pressure would jump all around. It turned out to be a bad crimp in one of the excitation pins for that particular transducer. I've been using that connector for several years so I guess it's better to fail now than during a test. I'm debating whether to go and inspect all the crimped connections on the test stand and maybe put a dab of solder on the connector just to be safe.

The test stand control program also got a makeover this summer. I never was really happy with the sequencer and I wanted to set it up to do a real countdown so I completely rewrote the section that handles the digital I/O and at the same time, ported it from NI-DAQ to NI-DAQmx. There were also some issues with the IRIG card that were bugging me. When I did my first uncooled tests in 2004, I was using a borrowed CompactPCI IRIG card that went into the same chassis as the DAQ card. For all the subsequent tests, I had to use my eBay special PCI IRIG card that goes into the PC a couple hundred feet away from the MIO card. Normally the two boards would share a 10 MHz reference clock in addition to a start trigger but I've been just connecting the start trigger assuming that the reference clocks were close enough. Well, it turns that the DAQ master clock (NI PXI-6030E) has a clock accuracy of +/- 0.01 % while the bc635PCI IRIG card specs its drift at 2 ms/hr. In actual use, I was seeing about 33 ms of skew over 700 seconds which is about 2 video frames. The whole purpose of using the IRIG card and video overlay is to allow correlation of the data with the video and if there is a large skew, then there's really no point. So, I'm going to run the 10 MHz reference clock from the output of the IRIG card to the PXI chassis CLK10 in connector on the back. I went this route instead of connecting directly to the DAQ card because if the clock goes away for some reason, the PXI chassis will automatically switch back to its internal clock and the DAQ card won't just freeze as it would if the reference clock disappears during acquisition. Finally, I changed the output module to write direct to CSV instead of the binary LabVIEW WDT format. By the way, I really hate the way LabVIEW tries to be clever and automatically display times in the local time zone. In general, I'm tired of fighting with LabVIEW and I'm close to just ditching it and rewriting the whole thing in C/C++. Maybe it's just my lack of experience with it but some things that should be simple just seem to be hard to do with LabVIEW sometimes.

I also spent some time playing around with an ADIS16350. This is a triple axis gyroscope/accelerometer inertial sensing system that I may put in the flight vehicle for basic pitch and roll guidance. I didn't get very far with it - just got it to talk to a 68HC11 eval board I was using.

2009-04-30

Good news with the burst discs! I talked with one of the Oseco engineers and they thought that my clamp force might be too high which would cause the discs to fail in shear instead of tension on the dome. Sure enough, when I checked my torque values on the cap screws, it turns out that I was using torque values appropriate for stainless steel instead of aluminum discs. I had been using about 50 in-lb on each of the #10-32 fasteners and I should have been using somewhere around 15 in-lb per fastener. I didn't have a torque wrench that went that low so I just used my calibrated fingers to see what would happen. Sure enough, one failed right about where it should have, on the domed area right around 580-590 psi. I wish I would have figured that out sooner before I wasted 6 discs! I also updated the drawings on the Test Stand page to show a radius instead of a chamber on the outlet of the Burst Disc Holder Cap.

I wasn't having much luck with the grease guns so I bit the bullet and bought a "real" hydro tester (the Dudgeon Model 4A-30.1 listed below) and wow, it works great. The action and pressure rampup is super smooth and there is no pressure fallback when you relax the handle.

Next up is to hydro test the propellant tanks and associated plumbing. After I get some new burst discs, I can proceed with the cold water tests and checkout of the new purge system.

2009-04-12

All the new plumbing for the test stand is complete and I'm just about ready to start cold water testing to check out the purge system. As part of my plan to use more accurate burst discs this time, I ordered some burst discs from Oseco. I got 0.5 inch aluminum FSTD discs for 600 psi and they ended up being $25 each in quantity of 8 (the minimum order was $200). The cert sheet that came with my order said the mfg. range was 546 to 600 psi (I had assumed the tolerance was +/- but apparently it was only -). The "rated" burst pressure was 583 psig at 72 degF and the two they tested burst at 590 and 576 psig.

Everything looked good with the discs so I got set up to do some hydro testing to verify my new burst disc holder (drawings are on the Test Stand page). For the hydrostatic testing, I tried a grease gun since they are readily available and others have reported good results using them for simple hydro testing. I tried using just water in the grease gun and I had a hard time getting it to pump up the pressure. I guess the water is just not viscous enough for the cheap seals in the grease gun. Also, the check valve built in to the gun leaked pretty bad so I had to pull a Swagelok check valve off of the test stand to get it to hold pressure. Even though, plain water didn't work too well so I filled up the grease gun with liquid soap. That seemed to help a lot better but the pressure was hard to control. I was trying to gradually creep up on the burst disc pressure and each pump ended up being about 100 psi with a lot of dropoff afterwards.

The first disc I checked burst at about 300 psi (way early compared to the ~580 or so it was supposed to be). There was a slight burr on the hole in the holder so I smoothed that out and the next three burst at about 500 psi, still quite a bit short of the 580 the vendor got in their test samples. Here is a picture of one of the ruptured discs and it looks like the disc is failing right where I'm gripping it in the holder instead of on the domed portion. I have a theory on why this might be happening: The vendor called me and asked me what outer diameter (OD) to cut the disc to so maybe they did their testing with a larger OD that may have had more grip area. I'm going to ask them if they tested it as shipped or with some other fixture. Maybe the way I'm clamping it is causing it to fail at a lower pressure than the way it was intended to be clamped. Another thing I thought of is that maybe that due to the pressure swings, the response time of the disc was faster than the needle on the pressure gauge. However, I still think the disc is rupturing in the wrong place. The whole idea of using my own holder is that later I'll be able to integrate it into my flight vehicle and save both weight and volume by building it into the tank bulkhead. The holder I constructed initially is to verify the feasibility of using a customer holder on the test stand.

I'm also looking at a "real" hydrostatic tester instead of the grease gun. Even after adding a separate check valve it still is very intermittent - some pumps have no effect on the pressure and in general it is very inconsistent. I tried two different grease guns and neither of them worked very well. The one I'm looking at is the Model 4A-30.1 at www.dudgeonjacks.com. This particular model is $235 and the kit with hoses, bleed valves, etc. is $87. But first I'm going to call them and ask how smooth the pressure output is. When I re-test the propellant tanks to 600 psi (and which fail at 700 psi), I want to be really careful I don't overpressure them.

2009-01-17

I made some new fittings for the top of the new propellant tanks. A photo is posted here and the drawing is posted here. The fittings mate with common 2.5 gallon stainless steel pressurized water fire extinguishers. I kept the original nut that is used to hold the fitting in place. The original fitting used what looked like some sort of square cross-section o-ring and relied on the amount of torque on the nut. My new fitting is just an o-ring in a groove that fits down inside the neck and does not depend on the nut to be torqued down. I used brass for the fittings because of good LOX compatibility and ease of fabrication. Stainless would have been a better choice to keep the dimensions consistent over the temperature range the LOX tank experiences but given the required tolerances for an o-ring seal, the brass fittings should work fine.

Actually, the fabrication of these seemingly simple fittings is an interesting story. I went out the shop last weekend to do what was supposed to only take a few hours but turned out to be an expensive lesson. I was drilling the thru hole in the fitting on the lathe and as the drill poked out the back side, taper decided to pop out of the tailstock. I really haven't worked with brass that much and it turns out that normal cutting drill geometry doesn't work too well with brass. It tends to grab really bad and wants to pull the drill toward the work. Since the taper is only held in the tailstock by friction, a large drill (for 3/8 inch NPT fitting) grabbing in brass was enough to pop out the taper. I surveyed the damage and discovered that the MT3 taper galled up the tailstock pretty bad so that a taper wouldn't fit in solidly anymore. After searching lots of discussion boards on the web, I decided to order a MT3 finishing reamer to try and fix the tailstock myself. After a $75 order to MSC, I was able to repair the tailstock taper with no indication of the original damage. I left the tailstock barrel in the tailstock assembly and used a dead center in the headstock to keep the reamer centered. A few careful turns with a crescent wrench cleaned it up nicely. Since my taper adapter and drill chuck were also trashed, I used the opportunity to get a nice Albrecht keyless chuck with an integral shank. For tapping the NPT thread in the fitting, I used a special "high hook" tap which is recommended for brass. I could definitely tell the difference between the new high hook HSS tap and the carbon steel one I was using.

This time, I'm not going to make my own burst discs for the tanks. These stainless tanks fail around 750 psi so I'm going to use some Oseco burst discs, probably set at a conservative 600 psi just to be on the safe side. I talked to their engineering department and although they wouldn't provide me with the dimensions of their holders, they were very helpful and told me what I needed to know to fabricate my own burst disc holder for their 1/2 inch STD discs.

2009-01-03

I've been working on retrofitting the test stand for the new 250 lb engine. I've completed the new load cell mount which should be good enough for the new engine and with some minor mods, provide room for growth. Although the linear slide has 1-2 lb of breakaway friction, this new design takes away the side load and moment worry I had with the previous design. The bearing block and rail (THK SHS 35LC) are quite a bit bigger than what I originally wanted but someone outbid me at the last second on eBay. I was a bit dismayed at first to find the bearing assembly had about 5 lbf of breakaway friction but after I removed the the dust wipers, it dropped down to a manageable 1-2 lbf. I analyzed the buckling load on the 1/2-20 B7 threaded rod and came up with over 35 kips assuming a formula of 4*pi^2*EI/L^2 for two fixed ends. The weakest parts of the design are the 1/2 inch right angle aluminum brackets. If I had aluminum welding capability, even some small gussets would dramatically increase the capacity of the load cell bracket. I suppose I could machine some gussets and bolt them on but the 1/2 inch thick bracket by itself will work for now.

I don't have pictures yet but I've integrated some 2.5 gal stainless steel fire extinguisher tanks into the test stand as the new propellant tanks and I'm just waiting on some additional Swagelok fittings to complete that work. I've been told that fire extinguisher tanks will fail at around 700-750 psi along the longitudinal weld line. These particular tanks have had a 3/8 NPT fitting welded onto the bottom and were hydro tested to 600 psi (thanks Carl!). Based on some suggestions, I'm also implementing a cascaded purge system that will blow out the propellant lines to the engine before and after each run to keep it nice and clean. I'll update the test stand drawing at some point to show the new configuration.