Yet Another Liquid Motor

[gidge348]

Just had another thought on how to seal this easily.



If you just cut a 45deg chamfer on the inside of the tube and slide an "O" ring on the outside of plug section.



Then you will have an seal that should not move?

Just make sure you have a little bit of crush on the O ring.

Regards
Ian

[stevep]

Leak testing, part II

When I ordered the new gasket material, I really had no idea what to get since I had no idea of the Durometer rating of the material I was using (EPDM, left over from a roofing project). My o-rings have always been 70A (I have no idea if folks outside the US use this rating or not--sorry) and they seemed harder than the EPDM, so I went with that.

I'm now quite deft at cutting gaskets and reassembling the motor, especially with the help of this nifty handle for the allen wrench:



I've used these handles for other sizes and found them very handy for spinning in a bunch of screws; I then use the "L" handle for final torquing. Anyway, when I ordered the gasket material I also got handles for the tiny wrenches I seem to be using for this project and I'm mighty glad I did.

The first thing I noticed when reassembling the motor is that 70A is not all that much harder than what I was using. Nevertheless I did the pressure test and noticed an improvement, but it is clear that I would need much harder material than 70A. The next step would be to order harder material and try again, but between these experiments and Ian's excellent summary of the issues, I've decided to abandon the gasket approach since I have no reason to pursue it further (see next entry).


While I knew I was cutting it close when I chose 1/4" (6mm) wall thickness for the chamber (see post of 11 July 2013, page 5), I certainly didn't expect this many problems with gaskets. Looking back I think I got suckered into thinking sealing would be straightforward with this type of injector because of how easy it was to seal the test injectors against 125 psi (8+ bar) shop air. That, and because o-rings seal so well if you get the gland dimensions right (and I have tons of experience doing that for my solid motors), that it just didn't seem like there was much to the problem of sealing in general.

I have two, 2" (50mm) thick binders full of notes--there are sections covering combustion chemistry, hydraulics, heat transfer, valves, plumbing, chamber design, injector design, and more--but none on the subject of seal design. Looks like there should have been one more section in one of those notebooks...and if there had been, I think in red letters somewhere there'd have been a note that read "avoid gaskets like the plague".

--Steve

[stevep]

Combustion chamber 2.0

(Sorry, Ian, events have overtaken you :-))

While I was waiting for the harder gasket material to come in, I made a new chamber:



Ironically, this was one of the first chamber designs I came up with several years ago, but when it came time to actually make one, I chose the drilled version over this design because I thought milling would be too slow! While it's true that, had I stuck with my original 0.125" by 0.125" (3mm x 3mm) channels, it would have taken a lot longer because the channels would have been deeper (requiring more passes) and, because a 1/8" (3mm) end mill is quite a bit more delicate than a 1/4" (6mm)one, the feed rate would have been pretty low. Even so, had I known how long it was going to take to drill the channels vs. milling them, I'd have gone with this design from the beginning.

As you can see, the basic idea is that a jacket fits snugly over the chamber thereby forming the outside of the cooling channels which are 0.05" deep and 0.25" wide (1.3mm x 6mm) . The jacket is sealed against the inner portion of the chamber by means of the two o-rings at the ends; this keeps the coolant from leaking out the ends of the jacket. There is not, of course, a leak proof seal between adjacent cooling channels but with only a few thou of gap between the walls of the channels and the cooling jacket there shouldn't be much "short-circuiting" of the coolant flow. (There is a body of belief/experience that suggests that if the gap is small enough, the temperature difference between the inner wall of the chamber and the jacket will cause the chamber to expand just enough more than the jacket that the gap will close, thus sealing the channels. Maybe...maybe not; we'll see....)

The (air cooled) graphite nozzle will attach just as it did to the previous chamber: machine screws threaded into the end of the chamber (it will also have an o-ring, just haven't cut the groove for it yet). The injector is similar to the previous injector in that the fuel flows from the cooling channel through a hole in the wall of the chamber into a groove in the injector. It then flows out into the chamber through a bunch of (as yet undrilled) tiny holes in the groove. However, in this injector this idea was taken one step further: the fuel now flows from the injector *into* the channels through a second set of holes in the wall of the chamber.



The injector therefore has two grooves for fuel, and three o-rings. It is all one piece so there are no face seals to cause trouble as in the previous injector. Nitrous flows down the center of the injector (just as it did in the previous one), fuel flows down a second hole (with the AN fitting), then out through a radial hole into the forward groove where it then flows into the "down" channel through the forward ring of holes in the chamber wall. The fuel emerges from the "up" channel through the aft ring of holes into the aft groove and then through the fuel orifices into the chamber. Hopefully fuel flow won't be too unbalanced because of the lack of symmetry (some channels are much closer to the supply than the others) which will cause a small pressure imbalance.

This chamber is *much* easier to build than the previous one. The hardest part is getting the jacket (which is 2" (50mm) aluminum tube with a 0.035" (.9mm) wall) to fit snugly over the chamber--that takes some care and quite a bit of time on the lathe with a file removing miniscule bits of metal until the jacket slides on (and off!) with only muscle power. Sort of like fitting a bearing on a shaft, except that this is one heck of a wide "bearing" :-) I went at it quite cautiously, not wanting to overcut; it probably took a couple of hours from raw stock to having a jacket/chamber combo that fit. [If anyone has a great method for preventing aluminum from clogging a file, feel free to chime in!] It then took just a bit more time to fit the o-rings--the grooves are about 0.080" (2.03mm) deep and the o-rings are 1-series meaning they're about 0.103" (2.6mm) in thickness. The tubing for the jacket did have to be selected and handled carefully--it's easy to get it out of round.

I cut the channels with my toolpost-mounted drill and a 0.25" (6mm) endmill. Depth of cut was 0.010" (2.5mm) and feed rate was fairly brisk, though certainly not the fastest possible. All twelve channels were cut and joined up at the aft-end in only 30 minutes. They were then cleaned up a bit with sandpaper. The cuts are clearly not as smooth as they'd be with a mill, but I think they'll be fine for coolant. (Just rechecked the hydraulic calcs for the effects of a "really rough" wall, and even if I double the friction coefficient, the pressure drop, while large at 250 psi (17 bar), is still small enough to allow a decent pressure drop across the injector orifice given the fuel's initial 600-800 psi (42-55 bar) pressure. On the other hand, I haven't allowed for all the turns in the passages, so maybe I'm in trouble and just don't know it.)

Because the channels are accessible from the exterior of the chamber, drilling the holes through the wall is trivial--with the previous chamber I had to drill from the inside and that quickly got difficult because I had to angle the bit fairly steeply (I ended up using a 1/8" (3mm) endmill in that case, but it wouldn't have been long enough to reach farther into the chamber for a second row of holes).

The next question was would it hold water? Compared to the previous chamber, it was a snap to assemble and get filled with water. Took it out to the test stand and it held 850 psi (59 bar) first time! There were a few drops here and there from the jacket, and, if you look closely at the photo, you can see the jacked has bowed out slightly from the pressure (the ends are smaller than the rest of the jacket). However, there was no spray or real leak, just the odd drop. I'll want to fix that, of course, but it's clear that this approach is superior in fabrication and sealing. The bowing will, however, increase the clearance between the channel walls and the jacket--I may therefore need to go to a heavier-walled tubing.

So far I'm much happier with this design than the previous one, assuming there isn't any noticeable short-circuiting of coolant. I'm now trying to devise a way to evaluate whether the short-circuiting is happening or not. I'm also going to make a dummy injector that will have a port to sample coolant channel pressure so I can see what the pressure drop really is. And, I suppose, I really ought to do a pull-test on those 4-40 machine screws to see if aluminum threads will hold 75 pounds (34 kg).

--Steve

[stevep]

New jacket

Well the reason the jacket turned into an experiment in hydroforming (it expanded to 2.25" from 2"!) is that it was 5052-O (annealed) which has a yield strength of only 9000 psi (62 MPa). At 850 psi the strain is about 24,000 psi (165 MPa), so clearly it wasn't the right material for the job. I ended up using it only because I had it on hand and initially was just interested in seeing what it would take to fit it tightly to the chamber. Once I got it fitted, one thing led to another and I ended up making the whole darn motor.

So today I went to the metal yard and picked up some 6061 T651 (with the same 0.035" wall as the previous jacket) which is what I'd normally use for this. It's yield strength is 35,000 psi (241 MPa) so in theory it might bulge a little but it should at least return to its original size once the pressure is released. Of course, that's at room temperature so I may run into trouble when I'm running hot fuel through the cooling channels.

I fitted it up and tested it out. There were no visible leaks, and it swelled maybe 2-3 thou at 800 psi, so I think I'm fine (barring excessive loss of strength due to heat when the motor is running).

--Steve

[stevep]

Pull-out testing

At the end of an earlier post I said I felt that I needed to do a pull-out test to see just how much force the 4-40 machine screws in the ends of the chamber could take. I had no fear that the screws themselves would break (or even stretch), but just that the threads in the aluminum chamber would strip. At that small size, they *seem* fragile and I just don't have any experience with things that small, so I felt I ought to have a better idea of what they could handle before I subject them to real chamber pressure.

Speaking of which, the design pressure is 200-250 psi(13-17 bar). The area of the injector is about 1.7 in**2 (1100 mm**2), so there's 425 lbf (193 kgf) trying to blow it off; this is resisted by the 6 screws so they're holding 70 lbf (32 kgf) each which isn't a lot, but if you look at the threads, they're *tiny*, so who knows?

At this point, I kinda feel like taking a poll to see what people think, just for the helluvit, but I won't.

I happen to have a rough test jig for this sort of thing, left over from when I was testing the tensile and bearing strength of some carbon fiber layups I was doing:



Here's a closeup of how the test article is mounted:



Because of the leverage, I can apply anything from about 10 lbf to over 1000 lbf (454 kgf). I started with 5 threads engaged and 50 lbf just to be sure I wasn't delusional about what loads could be handled. Then I cranked it up slowly to over 600 lbs (272 kgf) with no sign of any problems.

Curious about how many threads needed to be engaged, I backed it out to 4 threads, then 3 threads (with 600 lbs each time) with no problems. Finally, at 2 threads, the threads stripped.

The chamber is designed to have 5+ threads engaged, so at 600+ lbs x 6 screws I think I've got more than enough holding power!

(I'm out of town for a couple of weeks, so updates won't happen for a while)

--Steve

[cursorkeys]

I don't have any technical expertise in this area but I wanted to let you you that I'm greatly enjoying reading your updates, just so I'm not a totaly silent audience

Your experimental data for the thread engagement is very interesting. I remember reading something once that force is sometimes not distributed evenly amongst engaged threads and interesting cascading failures can occur as the acual load-bearing threads fail and then give impulse loads to the next threads that become actualy loaded in the joint. Fascinating stuff; I wish I'd done mechanical engineering rather than electronic sometimes.

[stevep]

Hey Cursorkeys,

Thanks for letting me know you're following along--I assume some people are, but it's nice to hear from folks.

As far as thread-holding goes, I did some reading--not a lot, just enough to realize that calculations on this are not very reliable. A lot of variables are involved, few of which one knows (outside of the lab) with any degree of accuracy. One widely-held belief is that only the first couple of threads do any real work, but counter to that is the idea that in at least some materials the threads deform under extreme load such that the remaining threads start becoming involved with the result that the fastener is more effective than one would predict. Then, as you mention, there are cases of cascading failures. Given all that, I figured it was better just to run the experiment :-)

--Steve

[stevep]

Flow testing the motor

After I got back into town, things started to move ahead quite nicely. The fuel tank I've been using for my testing had only been tested with shop air at 125 psi (8.6 bar), but it will need to handle pressures higher than that, so I spent some time pressure testing it up to about 1200 psi. As expected, this went smoothly with no leaks.

Then I repeated some flow measurements with the old prototype injector I was using for drop size experiments. A long time ago I'd measured the flow at 90 psi, but I was curious if higher pressures would cause the Cd factor to change at all, and, of course, there's nothing like having real measurements as opposed to calculated results. I pressurized things with CO2 (800 psi, roughly) and boy does that make the water flow! However, this process didn't go smoothly because the orifices were 0.013" (0.33 mm) and there was a lot of junk in the tank and plumbing lines that kept clogging the orifices. Visited a kitchen supply store (of all places) to get some stiff-bristled brushes of the right diameters for the tank and other parts and I practically oxygen-cleaned the whole setup. That's also when I decided I was tired of cleaning the orifices using those delicate drill bits, one of which I dropped onto the shop floor and broke, despite it landing in a pile of sawdust.

So I ordered up some lengths of small diameter piano wire, waited for that to arrive, and started using that for cleaning the orifices--much better! Also, I decided I was tired of holding the injector up to various light sources at just the right angle to see if the orifices were clogged, so I futzed around and made a more usable light source:



What an improvement! (If you wonder why rocketry takes time, it isn't building the rocket that takes forever, it's finding/building all the rest of this stuff that slows you down.)

Once I got rid of all the clogging, the next obstacle was reading the pressure gauge and accurately setting the pressure. Running tests at just 400 psi were emptying the tank in ten seconds or so, and between watching the gauge and my watch while also being ready to shut off the CO2, it was difficult to see that the pressure was right. At higher pressures things happened too fast to get reliable results, so I ordered up some inexpensive but easy-to-read gauges; once they arrived I found things went much more smoothly, but it still took almost half a day to get a set of measurements I was happy with in terms of accuracy and consistency.

These measurements showed that the Cd was pretty constant up to 400 psi or so (the limit of my testing). The next question was what effect would the cooling channels have on the pressure, but to test that, I needed the real injector which still needed to have its holes drilled to 0.013". To drill the test article had taken a lot of time and effort and I was always running the risk of breaking drill bits, so I ordered up another ten bits, just in case.

As it happened, I had to turn a small part for a different project, got fed up with the backlash in my cross-slide, and took a few days to make an "Evanut" bbs.homeshopmachinist.net/threads/43645-Making-Acetal-leadscrew-nuts-the-easy-way out of acetal. Wow, what a difference! Zero (and I mean zero) backlash. Micrometer-like smoothness, repeatable accuracy, and it really cut down on the chatter during parting operations. If you don't have one, make one--you'll love it! At the same time I finally decided to spring for a gallon of Vactra #2 way oil--that, too, really makes a difference--wish I'd done that a few years ago :-)

Then I got fed up with not having a real carriage stop--I had been using a clamp and steel bar--so I took the time to make one of those. Then I got ticked off with one of the chuck keys for my 4-jaw and ordered a new one--it was oversize so I had to carefully grind it down to fit.

Then I really succumbed to temptation. I've been drilling those super-tiny precision holes on the drill press, but it's hard to do and the sensible way is to just chuck the cylindrical injector in the lathe and drill them there. But my drilling arrangement there is even less rigid/precise than the drill press, so I've never even attempted it. I've long known that what I should have is a decent toolpost-mounted spindle which I could use for both drilling and light milling for things like the cooling channels. Trouble is, I've never seen one that I could afford, or that seemed robust enough, or that would mount the way I wanted it to, etc. I've also casually considered making a spindle, but all the build threads I've seen seem to have been done by folks with way more skill than I believe I have (not to mention probably more accurate lathes), so I've not proceeded down that path. As a result, I've just kept on using the clamped-drill arrangement or the drill press.

But I guess my frustration level must have risen to some critical, threshold level, because I found myself rereading a bunch of spindle building threads and I just decided to give it a go. I didn't need super accuracy or super-high speed, so I figured that would make it easier and cheaper. Long story short, I built one out of $14 deep-groove (not AC) bearings, a cheap ($20) ER 16 straight shank collet chuck, and and a brushless DC RC motor and associated electronics. Total cost was less than $200 for the spindle itself. The toolpost mount (a big chunk of aluminum) cost me another $30.

I spent the better part of 3 weeks standing in front of the lathe for about 4-6 hours per day--thereby making it the most complex, most precise, time-consuming bit of machining I've ever done. But miracle of miracles, it works!

And it works far better than I had any right to expect, given the low cost of the parts and my inexperience at this sort of thing: it has a few tenths runout 1/2" from the nose of the chuck. I can mill a 1/4" wide, .050 deep slot in aluminum and drill those tiny 0.013" holes.

Here it is, set up for radial drilling:



And here it is, "flopped" for axial drilling:




The really cool thing is that I can just place it on the tool post and it's perfectly aligned, doesn't matter which orientation. To change orientations with the clamped drill, I had to remove the drill from the clamp, turn it around, rotate the tool holder, reclamp the drill and realign it (in both pitch and yaw) which was never fun. This setup is *so* much more convenient--as long as my compound is accurately aligned (which it is nearly all the time), the spindle goes on perfectly aligned. Changing orientations just involves turning it over and rotating it--the spindle doesn't have to be removed from the mount and no realignment is necessary.

It sure took a bite out of my rocket-building time, but it is a great joy to use, so it was well worth it. And I did finally get my injector holes drilled....didn't break a single bit!



Now it was time to run the flow test with water. Naturally nothing went smoothly. First, it took a zillion runs to get all the orifice-clogging crap out of the motor. I'd cleaned it with a toothbrush and soap/water, but there was a lot of stuff hiding in the crevices that found its way into the orifices. Finally got a few runs in, but ran out of CO2. Went to my usual supplier but found that they'd shut down my local outlet, so I ended up spending a day scouting around for a supplier that was nearby but didn't require opening a corporate account.

Found one, but he only supplies aluminum tanks which have a much larger diameter neck than the steel tank I made my valve for, so I'll have to rebuild the clamp for that. While I was there I asked about nitrous--all he supplies is 50 lb cylinders but the price is pretty good: less than $3/lb but the tank will cost me $260 and it will weigh 100 lbs when full which makes it a pain to deal with. Gotta do some more looking, I guess.

Anyway, went back to flow testing but now the "fuel" (water) tank started leaking for some reason. Chased that for a couple of days, then went back through the test-until-the-orifices-stop-clogging routine. Finally after a couple of days of actual testing I got a set of data I think is reliable.

Seems like my cooling channels plus the injector only create about a 100 psi (7 bar) pressure drop which is somewhat less than I'd calculated (always good). I wanted a drop of about 100 psi across the injector itself, so that plus a chamber pressure of 250 psi (17 bar) means that I need a pressure of about 400-450 psi (27-31 bar) on the tank itself. Given that CO2 at room temp is about 800 psi, I have a decent margin to play with.

OK, that's with water--my fuel is going to be diesel #2 meant for vehicles here in the States--I have no idea what it's called elsewhere. Since it could have flow characteristics different from water, it was time to buy some and start testing with it. Testing with water tended to result in water all over the place (spills while filling the tank, spills/splash as the motor ejects the water, etc.) and I wasn't looking forward to having diesel fuel all over the place, so I spread some plastic sheeting around to contain as much of it as I could, and used larger vessels to capture output and any spills during tank filling--both of which helped. Anyway, here's the results:



The X-axis is pressure (psi) just before the motor, the Y-axis is g/sec of fluid (water-blue, diesel-red) that flowed through the motor. A couple of observations: first, the diesel is about 80% as dense as water and it appears that the difference between the lines is about this magnitude. That tells me that the diesel is flowing just about the same as the water did, which is cool, since it means I can keep testing with water and just correct for the difference in density (less mess, less smell).

Second, as I indicated above, I'm getting plenty of fuel flowing through the motor. If I go with a 75/25 nitrous:fuel ratio, I need 32g of diesel/second and if I go with 65/35, then I'll need 50g/sec. In the case of 32g/second, a tank pressure of 450 psi will suffice and for 50g/sec I'll need 600 psi in the tank (see small arrows on the chart).

At this point I'm a happy camper. I have an injector that works, cooling channels that don't leak and don't create a huge pressure drop, verification that the real fuel will flow adequately, and a fuel tank that seems to work, along with a way to pressurize it. And a new toy for my lathe :-)

Now I really must finish up the electronics....

--Steve

[shade]

Steve, very interesting thread, I have been thinking of doing a similar project, I have been researching myself. My thoughts were to use hypergolic fuels; I am a chemist by profession so the hazards associated with those fuel choices do not bother me as much as other options. I am intrigued by your choice of fuels. I have bulk diesel here on the farm so that is always a benefit.

I have not read your whole thread, working on it but the white on orange is tough on the eyes. But from what I have read you combustion chamber is 6061 aluminum. My initial design were to use high nickel stainless like 316/316L. I dismissed aluminum, as I was thinking it could not stand the heat in the combustion chamber. Clearly you do not have that concern, would you comment? Also as far as injector designs, "Liquid Rocket Thrust Chambers," Yang et al. American Institute of Aeronautics and Astronautics, Inc., 2004 is an excellent resource.

Have you had a chance to test fire the engine yet?

[stevep]

I'm still alive, and what's more, still working on this project :-)

It's been really tough to find big blocks of time to spend on it this summer and I've learned the hard way that trying to debug electronics with only an hour here, an hour there doesn't make for much progress.

Lately, though I managed to eke out enough time to get the A/D board working, at least in it's basic operation. It wasn't easy and involved buying a USB-type logic analyzer which in turn required that I buy a new PC because my decade-old box wasn't up to the task. Looks like I'm also in the market for a new 'scope, too, since my venerable TEK 2232 has only one channel still working and even that one is acting flaky.

The first task was to get the 8-channel A/D chip to recognize commands being sent from the uC. My first problem was that I didn't think through my 5-volt-to-3.6 volt interface very well and that caused all sorts of problems. It took me a while to figure out what was going on there and to create a temporary fix so I could get on with debugging. The real fix will involve a small auxilliary board that I'll have to add to the A/D board. Sigh...

At that point the A/D chip seemed to be running but not responding correctly to the commands. I spent hours/days chasing that and finally decided that I must have fried some of the digital inputs while I was getting the 5/3.6 volt issue chased down, so I ordered a new chip, waited a few days for that to arrive, snipped off the old one, soldered on the new one, and fired it up.

Now the chip responded properly to commands, but it was giving totally flaky results for the things I was asking it to measure. Rechecked everything--software, hardware, all ground pins, etc etc and still couldn't figure it out. Finally I decided to check all voltages at the pins themselves and that's when I found that a trace that I'd run *under* the chip (and therefore
couldn't see once I'd soldered the chip in place) was not at the correct voltage. Dunno exactly what's wrong with it because I checked all the traces before I soldered the first chip in place and they all *looked* OK after I pulled the first chip off. Anyway I soldered yet another "patch" wire in place and things worked properly right away.

Next step is to try it with the pressure transducers and make sure that all works. Assuming it does, then I can populate the board with the small amps necessary to deal with the load cells and test that out. Hopefully that will go somewhat faster.

--Steve

[stevep]

Hi Shade,
Aluminum is fine if you get the regenerative cooling right. A nice amateur example is Robert's motor here
and he's using LOX.

I have not fired the engine yet--still working on the electronics to run the test/measure the results. That's turning into be a major project in itself, but seems to be unavoidable if you really want to test safely and get enough in the way of results to know what the heck is going on.

I look forward to reading your thread.

--Steve

[stevep]

Finishing the A/D board....

I connected up a pressure transducer and after a bit of fiddling I managed to get the connections right and even got the calibration test to work. Here's a view of the test setup:




The pressure transducer is the cylindrical object at the bottom of the photo. It has a calibration feature which is activated by shorting a couple of wires and the result is a voltage signifying 0.75 of full scale. That's handy to let the micro know that the pressure transducer(s) are connected and working (it will be part of the power-on-self-test that the controller does).

The small rectangular silver-ish object at the lower-right corner of the breadboard is a load cell. Once I got the pressure transducer working, I added the parts for load cell data acquisition and tested that out. There's a small potentiometer for each load cell that can be used to adjust the amplifier gain so as to maximize the sensitivity for each load cell.



Here's a closeup look at the A/D board itself:



You can see all the blue "patch" wires I had to add to get things working. I need to clean those up, and also find a better location for a small through-hole resistor that I had to add to get the calibration to work.

The "daughter" board at the lower left of the A/D board is the DC/DC converter (12v to 28v) I built a while back--it's necessary to boost the voltage to match the pressure transducer's required 28 volts.

Having gotten rid of all my stupid mistakes, the thing seems to work pretty well and I'm pleased with it so far. There's still some mechanical work to do: add the DB connector for the load cells, add the rest of the circular connectors for the pressure transducers, and make a retaining mechanism for the pressure transducer connectors so they don't fall out during the test.

Now it's back to the finishing the valve for the nitrous tank and getting a board made to control that, but I'm heading out of town for a couple of weeks so there won't be an update for a while...

--Steve

[stevep]

A/D Board and Nitrous valve motor

I got back into town and attacked the connector portion of the A/D board:



First up was to cut a hole for the DB-25 connector which was a pain since I don't have a mill. I chain drilled and filed it out--tedious to say the least. Also, before I wired everything up, I fabricated a retaining plate that is intended to hold the male ends of the circular connectors in place (see below).

Then I crimped wires onto the pins that go into the circular connectors, then soldered the other end of the wires to the board. Then I soldered up the load cell wires to the DB-25 connector. Thatsa lotta wires :-)

Here's the connector side of the mounting plate, with the retaining plate separated:



Once the pressure transducer connectors are in place, it looks like this:




Next up was the valve motor for the nitrous tank. I had this all assembled and working some time ago, but didn't have the multi-turn potentiometer attached. Then I went and got a fresh tank of CO2 which turned out to be aluminum (not steel like the previous tanks had all been) and therefore had a larger neck. Coincidentally, the valve stem was different, too. That meant the mounting bracket and valve handle both had to be re-fabricated, so I messed with that for a day (more filing):





Then I set to work on the pot (the blue thing):




It was a little fiddly because the pot needs a nut to hold it in place, but the gear on the motor is practically flush with the mounting plate, so I had to use a second mounting plate for the pot so I could recess the nut:



Anyway, it's physically all together now, the handle turns, and the gears mesh nicely.

I did a bit of experimentation to make sure that the current draw is compatible with the electronics I'm planning to use. As long as the motor is turning the valve, the current draw is less than a couple of amps which is easily handled by the chip I'll use to control the motor. However, when the valve fully closes (or fully opens) and won't rotate any more, the motor is stalled which tends to draw quite a bit more current. That worries me some, but I think the micro can catch it quickly enough to prevent burnout--it just has to watch the pot and when the value stops changing it shuts off the motor right that instant.

That's the next step: get the motor control board built. It's virtually identical to the board I built for the little motors/valves, but the motor control chip is a beefier version (same electrically, just a different package) and I'll use bigger diodes. That means a tiny bit of rework on the PCB, but it shouldn't take long.

I've also been thinking through the testing procedure and making sure I understand all the failure modes. There's starting to be quite a list of things to check and fail-safe procedures to put together. You turbine guys who have designed/built FADECs have a good idea of what I'm talking about....anyway, more on that in a future post.

--Steve

[stevep]

Nitrous valve controller

I was able to jump on this right away after my previous post and made good progress for the first week or so. I laid out the board using as a starting point the board I built eons ago to control the throttle valves. The major changes were that I used the higher power configuration of the L298 motor controller chip (multiwatt instead of surface mount), added some beefier diodes, and added some circuitry to control an emergency shutoff valve that I'll talk more about in a future post. I also added a big resistor to keep the current through the motor and the motor controller within the safe limits.

Got the board printed, etched, and the parts soldered in place without too many headaches:




Wrote some quickie test software that revealed a couple of minor soldering problems, but which took a while to chase down. As a software guy I'm used to trusting that the hardware is good, but I keep forgetting the hardware I make myself isn't quite the same as hardware produced by real manufacturers, LOL. As a result I sometimes spend too much time chasing what I assume are software problems only to find out later that it was something I screwed up with the hardware. Anyway, I got the minor problems sorted, then started testing with the CO2 tank:




Found a few more problems having to do with timing which required that I write more elaborate test software to debug, but after a week or so of futzing I finally got the board to open the valve fully, wait one second, then close the valve fully. All without burning out the motor or the board :-)

In case some foolhardy soul cares to build a board that plays with valves I'll describe the issue briefly: the intent is that the micro watches the pot to determine when the valve has opened or closed--if the pot value isn't changing then the valve isn't turning any more (meaning that the valve has either opened fully or closed fully) so the micro should stop supplying power to the motor. However, it may be that the motor never started turning in the first place because something was jammed or the valve was closed too tightly for the motor to turn, etc. So first the micro watches the pot to see that it IS changing (meaning the valve has started turning), then it watches to make sure it stops. However, it may be that there's a some fault with the pot (or its wiring) so I put a time limit on both the watch-to-see-that-it-starts and watch-to-see-that-it-stops operations. If the time limit is ever exceeded, it means there's something wrong and the micro stops supplying power to the motor. That timing took a bit of fiddling to get right, but once I gave things enough time to happen, the valve opened and closed as if by magic.

Now onto some fail-safe stuff....
--Steve

[Johansson]

Keep up the good work!

/Johansson the Rocket-lurker