15G works!

15G worked successfully until the paraffin regressed to its maximum diameter and flamed out. While the grain's core was initially 3/8", its very likely that attempts A-F made it regress slightly. This resulted in 1.7seconds of sustained combustion for 15G.

Tests #13h-j unsuccessful

Refilled the nitrous tank, and used the new DAQ controller functionality.

• Test 'H' failed since the DAQ's flow and ignition lines were wired backwards
• Test 'I' used a 0.25s delay between burn and flow. Did not catch.
• Test 'J' used a 0.5s delay, the augmentation certainly lit but paraffin did not catch.

We think there may be some issues regarding driving the igniter, solenoid, and lamps all off of a single 12v 4.5Ah battery. Despite it being recharged immediately prior to this round of testing. As a preventative measure, it has been decided to split everything into at least two, possibly three batteries. We have four in total, complements of the old WARRIORS I 50v capacitor charging system known as 'Sparky'.

Its likely these igniters are not working well; time to order some 30AWG nichrome and build a new batch - with plenty of spares for just igniter testing so we can build up some confidence. We might need more pyrogen too.

Transistor board & DAQ controlling

The DAQ's digital outputs, being driven by Nick's LabView program, are now providing the timing for the motor. This means the Linux-based laptop and serial port arrangement is obsolete.

Since the outputs are very much current limited, a transistor board was assembled. A NPN, setup as a NOT gate is chained to a PNP setup as a switch. This provides a common ground, separated positive voltage needed to go into the launch controller. This setup also drives some LEDs on the board. Pictures and schematic to follow, but its based off of http://www.kpsec.freeuk.com/trancirc.htm

Test #13 - Computer Control

Computer Control Problem
The computer control malfunction is more unfortunate, and actually has nothing to do with the computer. Originally it was thought the code was somehow malfunctioning, and despite nothing being found it is currently in a state of rework to add additional functionality. This included burn durations that are not integer values (2.5seconds, for example).

I took the opportnity earlier in the week to completely rewire the controller. It was an awful mess of wiring which made it impossible to debug. It also used a lot of high gauge solid conductor wiring, which is rather fragile and easy to break. In fact, the manual ignition firing button was having an intermittent problem likely attributed to a break in the wiring.

So, it was time to gut the controller and rewire it with lower gauge stranded, and minimize all the back-and-forth between the terminal strip and the console buttons by wiring the console buttons up directly to each other. Electrically speaking it would be identical to the old configuration.

Or that was the plan. While everything is working on manual control, the computer control is experiencing some faults. One problem is that when the flow is armed, but ignition is not, the flow is engaged. The second problem is that when the computer is commanding both to engage, as it should be during the ignition/flow startup sequence, neither event is actually on. This results in a very short ignition event, a pause, and then a flow event. This clearly wouldn't result in a good start. The exact circuit error has not been uncovered yet, but will clearly have to be resolved before the next test attempts, currently slated for Wednesday of next week.

Manual control
So the first three attempts were done with a faulty automatic sequence, and the last four were done under manual control. This was rather reminiscent of the tests last summer, when there was no computer control. As such, attempt 'D' was ignition, and then flow as soon as combustion was seen. Failing that, attempt 'E' was simultaneous flow and ignition. Attempt 'F', done this morning, was preflow, and it literally blew the igniter out of the motor. Attempt 'G' was another preflow attempt, with the igniter held secure with some tape.

Test #13 - Nitrous issues

Nitrous Oxide Problem
We've been running low on our nitrous tank for some time, and while we're still getting a line pressure of ~760psi its likely whatever liquid that's left in the tank is bellow the tank's stem. So only gas ends up in the line, causing the motor to deflagrate instead of getting a good burn.

So we need to refill the tank. We just had a very large tank of nitrous delivered; however the tank is gaseous only (it doesn't have a stem). Even if you set that aside, the transfer of nitrous from one tank to another is rather complicated and requires equipment we don't have, nor have the budget (or time) to invest in.

So the large nitrous tank is useless. Our small tank will have to be brought to the auto-racing store to be refilled.

Unlucky Test #13!

Over the past two days there have been 7 unsuccessful attempts for Test #13. Is the number cursed? Our experience seems to indicate that it indeed may be! This is our first test of an entirely new 10lbf paraffin motor.


We've uncovered two major problems:

• Only getting gaseous nitrous oxide; no liquid
• Computer control malfunction

There's so much here that those two major problems are going to get their own entries. For now, here's some of our secondary issues:

Igniters:
We are out of igniters, and don't have all the supplies (namely nichrome) to make new ones. Our pyrogen is also running low. Thankfully the guy that made the igniters for WARRIORS II is still around, and is willing to make us a new batch when we have everything.

Paraffin:
This batch of paraffin didn't have any specifications associated with it as it is sold for arts&crafts purposes. We already have a large quantity of high-grade paraffin ordered and are just waiting for it to arrive.

Core diameter:
Our current grains have a 1/2 inch diameter bore. To speed up the nitrous velocity, we would need to reduce that to around 0.33 - 0.38 inches. To slow down the nitrous velocity, we would have to increase it, probably in the 0.6-0.7 range. Would changing it improve the motor's ability to ignite, and at what penalty to overall thrust?

Grain augmentation:

• Is it needed? (can the igniter do it on its own)
  
• Where should it be applied? (core, mixing chamber)
  
• What should it be? (kno3, standard AP solid composite, pyrogen, sorbitol/potassium nitrate, etc)

Additives:
Right now all our tests have been done with 2% (by mass) of lamp black / charcoal. This slightly suppresses the combustion. Should our tests next week still use this, straight paraffin, or have an accelerant added?

Timing:
Right now we have no idea what the best ignition sequence is for paraffin. The three options are:

1. ignition/delay/flow
2. simultaneous ignition & flow
3. preflow prior to ignition

Data Acquisition
On the plus side, the LabView interface is working great now. No buffer overruns, data can be displayed in realtime with high refresh rate, not all data has to be stored to the hard drive, etc. Its really slick now, and getting to the point where it may replace the linux laptop & perl script for the actual motor control. No, the outputs from the controller for event recording hasn't been completed yet -- I've been kinda busy with higher priorities.

We certainly have our work cut out for us!

Test 11 and 12 video

50N paraffin motor parts machined

We are nearing the end of manufacturing the new parts for the paraffin ~50N motor. We decided not to go for testing on Friday 12/14. A few more parts have yet to be made, such as the mount for the motor, the nozzle insert which will be milled, and the combustion chamber needs the retention holes to be milled. The team also needs to work on the grain manufacturing. We need to do some density testing as well as structural testing. Right now we are also looking for paper tubes with an OD of 2".
Here are some photos of the completed parts.

Nozzle holder, and Injector

Injector and nozzle holder

How the motor should be when it goes together, this is not the combustion chamber in the middle but a part of the grain mold.

Design and Machining in hopes for a test at the end of the week.

So wondering what we are up to?

Well, there have been countless nights with little sleep getting ready for new testing and a nicer motor. At the moment the motor is mostly completed... I will be able to post many more photos and more detailed analysis information. But here are the general concept photos:

Here is the most updated CAD model. 3" OD with 2" ID. Most parts are aluminum where the nozzle inset is Graphite.

Here is the nozzle holder being machined. Not a great photo but I'm usually too busy to take photos. In the photo is a ~3x3x2 in billet being machined by a 3/4 End mill in a HASS VF4. We are only too lucky to have a fantastic machine shop at our disposal. At the moment I had just figured out how to fix a problem having large chord heights on the circular parts. I have never had this problem in the VM3 that I normally use, however I had made some errors in Gibbs CAM. The parts all look much better now.

Batteries - we eat them like candy!

We did not have a burn on Friday since the new test stand is being rebuilt to provide more reliable load cell readings. The team has been working quite hard on the paraffin design in an effort to get the first paraffin burn before term end.

So, in the meantime, I thought it'd be fun to talk about what electrically powers everything we have. That means batteries. Lots of batteries. So here's the inventory:

Hybrid Launch Controller

• 2x 9v batteries in series - drives lamps and 5v regulator for control logic output board.

DAQ interface Box

• 1x 9v battery - drives 5v regulator and other circuitry for signal conditioning.
• 1x 28v bench supply - not a battery, but provides the full operating range for the pressure transducers for maximum resolution.

Relay Box

• 2x 9v batteries in parallel - energizes the relays. Recently added trickle-recharge function.
• 1v 12v lead acid - powers nitrous solenoid, pyrogen igniter, dump solenoid (when implemented), and indicator lamps. Also trickle-recharges the 2x 9v batteries when connected.

Sum it all up and we have 5 9v batteries and a 12v lead acid. All these have to be periodically replaced or recharged. Some of these much faster than others! Plus a bench supply and two laptops that we run off of normal AC. How are we possibly going to run all this in the field?

Controller / DAQ interface

Above is a partial schematic of the hybrid controller and the relay box. I'd like to discuss some of the gritty technical details behind the black box. It serves as a central interface between human input (arming, manual control), the control computer (timing of ignition & flow), the relay box (and in turn the nitrous solenoid and pyrogen igniter) and finally the data acquisition unit and its computer.

Its this last interface (controller/DAQ) that has been recently added and was giving us trouble on #9 and #10. It is already on its third iteration. The first was a parallel hookup identical to the lines going to the relay box. This proved problematic since this interface betweens float (logic low) and ground (logic high). While it may be possible for the DAQ to sense this, it isn't ideal since it adds the potential for a malfunction in the DAQ or the associated wiring to trigger an event.

The second iteration provides that needed layer of isolation through the use of a quad NAND gate (CD4011). It also provides readable voltages that is easier for the DAQ to handle. This is the 5v logic low and 0v logic high mentioned in the previous post. The oscillations shown in that graph are because the operating voltage of the NAND gate (5v) is lower than the operating voltage of the relay box (9v). The NAND operating voltage has to be higher, so the easy thing is to up the voltage to 12v or even the 18v that the controller's lamps run on.

Of course, the solution creates a new problem. The DAQ can only accept 0-5v, so the output of the NAND gate has to be bumped down somehow. The solution is a op amp configured as a comparator (LM339). Combined with a resistor bridge (voltage divider) across the NAND's power source, the op amp will check whether the op amp's input is higher or lower than the divider's halfway point. Its output will thus be either 0v or 5v accordingly.

Phew! And that's what is necessary to safely get the controller's logic into the DAQ.

Good burns, but w/out data did it really happen?

Tuesday had two successful burns that premiered a few upgrades. They are:

• Chamber Pressure Transducer (~62psi)
• Long burn (10s apiece)
  
• Metal braided nitrous oxide delivery line
• Controller state output to DAQ (partial)
• Improved computer control code (advanced settings menu)

As seems to be the case as of late, we had a difficult time igniting the first burn of the day. Tests #9a and #9b both used the 0.15s delay between ignition and nitrous flow that worked so well for #8. It was decided to remove & inspect the grain before #9c. The two attempts didn't even get the augmented fuel lit. The augmentation was moved slightly so that it would be closer to the igniter. We also changed the delay timing back to 0.25s. #9c lit successfully for the longest continuous burn we have had yet.

Why we have all the ignition trouble is still a bit of a mystery. Best guess is that the pyrogen on the igniters degrade over time, and these are approaching a year old since they were built for WARRIORS II. A new batch is needed, and perhaps this will get us to the point where we don't need the augmentation anymore (which was the case for tests #2-#6).
This is the burn from #10. The white line is the chamber pressure, green is supply pressure. It seems the configuration of the cart is such that it doesn't move freely ("frictionless") so we didn't get good load cell data. The test stand has to be redesigned & rebuilt; this should be ready to debut for #11. More details to follow of course.

The blue line is the flow logic line from the controller. Yes, it was a bit upset w/ me at the time of the tests, and would oscillate between 0v and 5v when logic low, but would stay at 0v when logic high. Yes, that may seem a little backwards to you - 0v for on, 5v for off. I selected this so that you know that the logic line is working before you start the burn. Otherwise what would 0v mean? That the valve was off or that the logic line was disconnected or otherwise fouled?

At any rate, the problem was solved, but it resulted in the output voltage exceeding the 5v that the DAQ can support. So more electronics get crammed into the controller to ramp it back down. It will be ready to support #11 and #12, which are planned for this Friday. The ignition logic will also be included.

Sadly, due to a foulup w/ LabView, that pretty picture above is all the data we have. While we were able to see everything in real time, none of it was saved to the hard disk. We were lucky to still have the window open to take a screenshot of the graph from #10. Nothing survived from #9 at all, although visually the burns were very similar.

10s burns seems to be at the limit of what the 5lbf PVC configuration can handle. The cutaways of the grains (we used a fresh grain for #10) revealed the core diameter nearly matches the diameter of the injector & nozzle.

Recalibrating and DAQ setup

Well work has been continuing since the last test, we have been working on setting up the DAQ so that the control interface is also recorded, such that we will have the time and duration for each event. This way we will know exactly when the solenoid is open and when ignition happens as well as when the solenoid is closed.

The pressure transducers have each been calibrated twice. We noticed that even though during the first round of calibration, the excitation voltage dropped from 28-23 V, we recorded ~ P psi=194.4x-27.34 where X is the voltage output. We recalibrate from a power supply, which will give us a much safer constant voltage. With this setup the transducers were P(psi)=204.1x-4.5 (between the two transducers, the y intercepts are a bit different). This however is fine, and we now have a reliable method of obtaining delivery pressure. Just to mention, we used a helium tank with a regulator and a well calibrated pressure dial indicator for pressure calibration. Not the best possible method, but this will do us well.
The pressure port was drilled today and looks great, we should have minimum pressure loss, though there is some worry that with all the extra volume that we have added that must be pressurized, there must be some way to fill this volume with some non combustible material. One though is krytox, however that would be Very expensive as it would require alot to fill the 90 elbow and lines for the transducer.
Other members machined another fuel grain and we should see two 10 second tests tomorrow around noon!

Video of our most recent testing

Testing and Data = engineers party

First test went off without a hitch.
Here are pics... they do better than words

I'll upload photos of the Data and a detailed description of what is going on.

the control station
Motor in the center of the blast walls. These are 2 sheets of 1/2" plywood with 2" of space between... to be filled with sand.
upper box is the data box and the lower box is the relay controler
relay controler on the left, Data in the center with battery and a pressure transducer on the right.

A bad photo of the new setup.

Thrust Curves

Blog is up, where we are

Well its up and running. We will now be posting daily updates or maybe weekly depending on how easy this seems.

So far we have setup all the hardware and are preparing for the 2nd testing attempt. This should take place on 11/28. This will be the test of two 5 lb motor firings using the PVC grain developed by the summer crew.

Next will be a test of a 25lb PVC motor. And after that, the paraffin fun will begin!

Here is the new motor set up without the mounting plates or motor mount