9 Days of 38mm Mach-busting

[Adrian A]

This is the story of my last week and a half of rocketry, that ended yesterday with a J single-stage record (unofficial, but official application in progress) There were 2 bad, a 4-day rebuild, a partial failure, and one complete success. A couple of the following posts will be cut and pasted from the 2024 NSL event thread, where I recorded some of what was happening at the time.

I started out with two complete 38mm sustainer airframes, an additional complete nosecone, a forward shell of a nosecone, and a complete booster. Here is what I have left:



The Violent Agreement booster is the oldest component, built in 2010 or 2011. It had a hard fall with a full motor on one of my Balls 3-stage attempts and needed 2 new fins, which I made a few weeks ago. It's thick, heavy, and solid. Unfortunately it's too short for the Loki 38-1200 motors I want to fly at Balls, but it's a good size for smaller J motors. I made a new fiberglass RF window for GPS tracking just before NSL this year, and that worked great. The booster innards include (left to right) an anchor for the shock cord and the motor, made with a 1/4" aluminum threaded spacer that's screwed from both sides, an av-bay with a tracker and 1 Blue Raven bolted to a chute holder that is 2 sections of 38mm coupler tube cut into 2 L-shaped pieces. Everything but the anchor gets ejected at booster apogee.

My second oldest part is the Violent Agreement sustainer, which I have had since 2016 or so. The nosecone in front of it is one I made about 2 years ago. My strategy for the sustainer at the time was to make the fins small, to avoid moving the stacked CP too far forward, and stabilize with a lot of nose weight. The VA sustainer is also too short for the Loki 38-1200 case.

I made StratoSpear to handle the Loki 38-1200 case, and before I built it I figured out that a better strategy for a sustainer for >100,000 foot flights is to keep it as light as possible, and use highly swept fins which are more efficient at high Mach numbers, and less effective at low speeds. This is a win-win because having less-effective fins at low speed is good for a sustainer. The driving case for stability margin for the full stack is when it first leaves the tower, at its lowest speed and with an angle of attack from cross-winds that further reduce the stack stability margin. Less effective fins on the sustainer in this case only helps, since they are forward of the CG. When the sustainer goes fast enough for the stability margin to be reduced to be the new driving case (Mach 3+), highly swept fins are more effective and have lower drag than trapezoidal ones like Violent Agreement uses.

The build thread for StratoSpear is mostly here. Both StratoSpear and Violent Agreement had forward-facing chute cannons for deployments

Going into NSL, I had 2 38mm sustainers and a booster. I was hoping to set the J and K multi-stage altitude records. After Saturday was blown out, on Sunday morning I flew StratoSpear on a Loki J1026:




The Google Earth view above was taken from the ground station (GS) log, which gave an apogee altitude of 26,126 feet. Unfortunately, the nosecone got stuck at apogee:




I had to cut off the end of the motor case to salvage the fins. The chute was also repairable. The rest was a total loss. I felt terrible for letting one of my rockets come in ballistic. One day a there will be a much worse consequence for a ballistic rocket than a hole in the ground, and I don't want that to be my rocket.

The apogee altitude showed that my fins did their job. They were untouched despite the high speed ascent and descent. RASAero had predicted 27,070 feet with smooth paint and biconic section fins, and I got 26,126, so that was a nice confirmation. I think 27,000 feet is about the upper limit for potential altitude with existing J motors.

The chute cannon has some big advantages, including a single airframe break that is far from the tip, enabling a very smooth nosecone and laminar flow for a long length. The main chute deployment out of the cannon is also robust and reliable. But the inherent problem with this design is that the nosecone has to slide past a lot of taped-down harness, and also harness that is not taped down because it's unavoidable slack to get the nosecone over the cannon. This slack can go anywhere when putting the nosecone on, and I think in this case it got wedged against the end of the cannon.

 

[Adrian A]

Here is a pre-flight photo of the prepped cannon:


In this latest generation, I used a 29mm chute cannon tube so that I would have the volume to fit the main chute harness inside the tube and I wouldn't have to run it down the outside of the cannon.

Here's what I wrote in the NSL 2024 thread on that Saturday evening:

<start snip>

Sorry About the bad luck, Adrian.

Did the charge fully burn? 26k AGL when ground level is 8k ASL makes the air at apogee super thin.

Hoping for better luck with your planned two-stage flight tomorrow!

The charge is contained within a .25" ID steel piston, and the ematch fires downward into the powder, so there's really nowhere for the powder to go until the nosecone moves over an inch. The inside of the piston has grey soot like it does after the ground tests.

Steve Lubliner and I did a post-mortem on the remains, and our best theory is that the nosecone just got wedged onto the tightly-packed harness. The space around the nosecone ejection piston is completely full of reefed harness, and there is unavoidably some uncontrolled slack that has to go somewhere unknown when the nosecone is moved into final position. When everything is right, the top edge of the 29mm chute cannon is in contact with the inner surface of the cone, and the end of the nosecone ejection piston is in contact with the harness attachment hardware that is cast into the tip. But if some harness got trapped between the edge of the chute cannon and the cone, it could have really wedged the cone into place during the 120G boost. The lower edge of the cone was unsupported because it slips over the motor case, and the aft airframe is just friction fit onto the middle of the case.

The sustainer I'm going to use tomorrow was an earlier iteration of the same chute cannon concept. Last night I tried and failed to prep it with a high-performance slender nosecone, again because of this harness jamming problem, even though I flew it successfully a couple of weeks ago. For tomorrow's shot I'm not going to use that nosecone, but use a more conventional cone that isn't a custom long L/D layup but is instead a normal commercial 5:1 VK cone that is extended on the bottom with some commercial FG tubing. I think the supersonic drag performance won't be as good, but this cone has the full 1.5" diameter cylinder over the whole overlap with the chute cannon, so when I put that cone on, there are no points of contact except the coupler section of the aft airframe. This nosecone goes on easily and had very vigorous ejection during a ground test I did this afternoon. This sustainer has the nosecone slipping onto a section of coupler tube that is glued into the lower airframe, and so the bottom edge of the nosecone has a solid stop.

I'm confident that I won't have the same problem I had this morning, and hopefully I can avoid the other potential problems too.

<end snip>

Here is the ground test I did on Saturday afternoon. Very energetic apogee charge:

View attachment 1716743574035.mov

Here's a photo from Sunday morning prepping at the hotel:


I was as confident as I could be that the apogee deployment was going to work.

 

[kjhambrick]

Adrian --

Persist is a great name for a rocket !

-- kjh

 

[Adrian A]

I was confident I would finally have my first fully-successful 2-stage flight. And yet:....



Above is from the ground station log. Booster in blue. Sustainer in red. Apogee 28,197. The distant flights near the horizon are last fall's F record flight from Hartsel and over the horizon, the later G and J record flights from NCR.

The new map feature of the Featherweight UI app really helps with a rocket search. I went to where the last packet of the sustainer was, in purple, in the middle of a plowed field, and couldn't find it. I walked downwind back along the track in case the main deployment managed to get the nosecone off and it drifted backwards, but no luck. So then I started at the last location and walked back and forth down-track, using the app to show me where I had been. The actual location is where the green landmark is, below:




I still have no idea why the apogee deployment didn't happen. I managed to dig out the nosecone ejection piston from the dirt, and I confirmed the charge went off. The best theory that I have is that the boost caused the taped-down shock cord along the side of the chute cannon to slide down and jam, in a way that couldn't be tested on the ground test. Or maybe the blue tape I added to tighten up the nosecone joint became a problem.

Whatever the cause, I was beside myself with frustration. I had done everything I could think of to have a successful deployment, the ground test was great, and it still didn't come apart. That Monday and Monday evening I was seriously considering giving up on 38mm high performance rockets for a while.

When I'm faced with a disappointing failure or technical problem, I can't help but console myself dreaming up alternative solutions. So that night I came up with a new design and decided to build my way out of this problem. The long slide of the nosecone over the chute cannon seemed to be at the root of my issues, so I decided to flip the avionics and chute cannon around, and have the chute fire out of the bottom of the nosecone.

 

[Adrian A]

I had a new VK FG nosecone shell that I had previously purchased, and I was figuring out how to attach the innards to this cone. My plan was to add some features that would lock the cannon into place inside the cone after giving it a twist. I found that a 38mm coupler thickness mostly fills in the diameter difference between a 29mm airframe tube and a 38mm coupler, so I used my tile saw to slice a section of coupler into 5 pieces, and then I would glue 2 pairs of pieces to the inside of the nosecone coupler and the outside of the chute cannon:


This basically worked, though once the cannon is twisted into place, there wasn't any kind of locking feature to keep it from rotating back and coming out again, so I added 2 small external countersunk screws that go into nuts I had JB-welded to the sides of the chute cannon:


These screws don't take any deployment loads directly; they're just there to keep the cannon from twisting out.

Here is the attachment of 2 Blue Ravens to the top of the cannon, with wires for main and apogee deployments.


I also soldered wires onto the top Blue Raven's third channel for a backup apogee charge (the first time I have left room for a backup charge) and to the 4th channel of the bottom raven for an airstart ignition.

Here's the result I was after:


This is what has to slip into the coupler at the end of the aft airframe. There's much less distance and less stuff to get hung up.

 

[Adrian A]

My other problem was that the aft airframe with the fins is too short. I wanted to save the fins, so I glued on an airframe extension. The tube I had available was a bit bigger ID and OD than the rest of the airframe, so I dremeled out a shallow taper on the inside of the new tube, and sanded a shallow taper onto the outside of the old tube, and then just JB Welded them together with the motor case holding them straight. There was about 1/4" of overlap, which is pretty good for this joint that doesn't have any bending loads, since the motor case will always be inside. It's not as aerodynamic as it could be, though.

After 3 ground tests with increasing BP each time, I was ready for a flight on Saturday, at Northern Colorado Rocketry's Mile High Mayhem.




I loaded a Loki K627, which is the highest impulse 38mm commercial motor. RASAero predicted could go Mach 3 and get to 31,000 feet. The current K record, on a 54mm rocket, is 29k and change from Jim Jarvis.

The boost was straight, but the altitude was disappointing at about 25,000 feet. And based on the descent rate, the main came out at apogee. But at least the apogee deployment worked well enough for it to come down slowly.





The rocket looked ok at first when I got to it, but then I saw that the nosecone extension tube had zippered.

 

[Adrian A]

My other problem was that the aft airframe with the fins is too short. I wanted to save the fins, so I glued on an airframe extension. The tube I had available was a bit bigger ID and OD than the rest of the airframe, so I dremeled out a shallow taper on the inside of the new tube, and sanded a shallow taper onto the outside of the old tube, and then just JB Welded them together with the motor case holding them straight. There was about 1/4" of overlap, which is pretty good for this joint that doesn't have any bending loads, since the motor case will always be inside. It's not as aerodynamic as it could be, though.

After 3 ground tests with increasing BP each time, I was ready for a flight on Saturday, at Northern Colorado Rocketry's Mile High Mayhem.




I loaded a Loki K627, which is the highest impulse 38mm commercial motor. RASAero predicted could go Mach 3 and get to 31,000 feet. The current K record, on a 54mm rocket, is 29k and change from Jim Jarvis.

The boost was straight, but the altitude was disappointing at about 25,000 feet. And based on the descent rate, the main came out at apogee. But at least the apogee deployment worked well enough for it to come down slowly.



The rocket looked ok at first when I got to it, but then I saw that the nosecone extension tube had zippered.

 

[pugachu]

Whatever the cause, I was beside myself with frustration. I had done everything I could think of to have a successful deployment, the ground test was great, and it still didn't come apart. That Monday and Monday evening I was seriously considering giving up on 38mm high performance rockets for a while.

Adrian,

I'd really hate for you to give up on these high performance flights. What you are trying to do is hard, so it shouldn't be surprising that it isn't easy

I'm glad you convinced yourself to trying again. I am working my way up to pushing my own personal records, which are nowhere close to the records you are setting.

Keep trying! I know how frustrating it is to lose a rocket, and we all benefit from the electronics development that you have been doing to push the boundary.

Thanks for the updates so far, I am looking forward to the J record update.

 

[Adrian A]

Ok, continuing the story:

Out at the K627 landing spot, while the data was downloading to my phone, I looked at the flight summary and I could see that the third channel of the top Blue Raven, which had the backup apogee charge, fired at about 20 seconds, which is well before apogee, and before any of the other charges fired. And that the bottom Blue Raven fired the main charge before either altimeter fired the apogee charge. So something went wrong about 10 seconds before I was expecting apogee, which was consistent with the tube zippering. But still, I was happy to get the whole rocket back and to finally be getting data from the GPS and 2 Blue Ravens on a high-performance flight. And I was super excited that there was no damage to the fins, at all. I mean you can't tell that they didn't take their (what was supposed to be) Mach 3 trip in the backseat of my car. This was in contrast to the two high-performance flights last year, one of which destroyed the fins from fluttering and the other of which ripped off the metal foil I thought I needed to protect the fin leading edges. I just used some 4461 Cotronics epoxy for the fin construction and final finish, and it was totally fine.

I also figured that the early main deployment was the main reason why the rocket didn't get to 31,000 feet like RASAero said it would. The damaged tube would be easily replaced if I were home, and I could diagnose and fix the premature main deployment, and then have a good flight on Sunday. So that's what I did. I told my friends what was up, left my EZ-up and table and launch tower at the site but took everything else back with me for the 2.5 hour trip back home.

I got home, cleaned out the motor, got the batteries recharging, and took a closer look at my deployment settings. These two altimeters were ones that I had been using to test out and verify the deployment settings. They were sitting on my workbench, and I installed them without carefully checking or simulating the deployment settings I had been using. I erroneously assumed that they had default settings. Instead, they had some 2nd stage ignition examples for both the 3rd channel on one and the main channel on the other, in the secondary deployment settings. I should add verifying the deployment settings to my launch checklist.

The repair was straightforward. I removed the zippered tube with a Dremel and some prying, and then cut a new section of tube and squared it off using my Ryobi benchtop belt/disc sander, which is really my primary rocketry construction tool. Here's how I verify that the disc sander part of it squared off the tube


Looks square!

While I was at it, I took the time to add a magnetic switch to my tracker, so that I could button it up and complete the prep at home, and then just turn it on at the launch site:

The magnetic switch mounting holes are trimmed back on one side to provide more clearance to the GPS antenna (which helps its performance)

Just before the hot melt glue:


With the repairs and upgrades and prep, it was about 9 PM before I would have been ready to pack up and go back to camp, so I decided to go the next morning instead.

Here's some data from the K627 flight:



The main chute deployment is clear at 20 seconds in, which was about 10 seconds before apogee. The GPS regained lock about 15 seconds before apogee would have been. Once I did the lookup from balloon sounding data from 6 AM that morning in Dodge City, KS, inertial altitude and Baro altitudes all agree well with the GPS altitude, which is the most accurate reading.

The bad deployment settings explained some of why the rocket was lower than expected, but when I looked at just before the event, the rocket was going about 500 feet/second at about 23,500 feet. When I adjusted my RASAero sim to get it to match those conditions, the apogee was only going to be about 27,000 rather than the possible 31,000. This was disappointing and suggested to me that the replacement nosecone I was using had a lot more drag than my scratch-built 7:1 VK. Running the same sim but with a J1026 the prediction was only about 21,400 feet, well below the Tripoli record of 23,725 feet.

 

[Adrian A]

It was quite a straight boost:

Here are the accelerations:

Zooming in on the motor burn, it looks pretty normal:


The change from noisier data to smoother data happens when the accelerations are low enough for the Blue Raven to switch from the +/-400 G rang accelerometers to the +/- 32G range ones. The little stair steps are maybe when different grains burn out?

Here are the velocities:


There is a good match with the GPS velocity once it locks up again. But the max velocity of 2765 feet/second is well below the RASAero expected max velocity of 3262 feet/second, even after increasing the simulated drag to match the conditions right before the premature deployment.

My best guess now, after comparing the Cd vs velocity to Sunday's record-breaking J1026 flight is that the motor didn't perform as expected. This particular motor had been through a lot. It was out of its package and glued into the liner about 2 years ago with no protection from humidity, and it had been through a flight or two before this one, including at least one failed 2nd stage ignition and hard landing. It also required one of Conway's special thermite-blend igniters to get it going; an igniter with a big blob of Magnelite didn't cut it. Part of the reason why I chose to fly it last Saturday was because it was a reload that I wasn't very confident in because of its rough history, and I wanted to use it to get out any kinks there might be in my new rocket. I'd like to try this again with a fresh motor; I think it could set the K record.

 

[Adrian A]

The chute strength was better than I could have expected for this high-speed deployment.


The third channel popped off the nosecone and then about a second later, the main chute charge fired. When the main chute inflated, it put over 300 Gs on the nosecone, which weighs about 1/2 lbm. So the chute provided about 150 lbs of drag without tearing or breaking a shroud line.

 

[kjhambrick]

got home, cleaned out the motor, got the batteries recharging, and took a closer look at my deployment settings. These two altimeters were ones that I had been using to test out and verify the deployment settings. They were sitting on my workbench, and I installed them without carefully checking or simulating the deployment settings I had been using. I erroneously assumed that they had default settings. Instead, they had some 2nd stage ignition examples for both the 3rd channel on one and the main channel on the other, in the secondary deployment settings. I should add verifying the deployment settings to my launch checklist.

Thanks for that, Adrian.

I play with the Blue Raven Simulation feature pretty often and part of my process is to reset the Blue Raven configs after the sim,

But I have never REALLY checked my deployment settings on-site, just before a flight.

I will do that from now on -- it is now on my check list !

Wow !

The rocket was transonic and supersonic for 10 -to- 12 secs !

The tilt data looks really good.

How did the rocket roll ?

Your GPS velocity and integrated inertial velocity are spot on when the rocket slowed down during the coast phase.

But I don't see an inflection in the accelerometer data when the rocket coasted down into the subsonic range.

Is it visible at higher axial accelerometer resolution ?

Or maybe in the gyro data ?

My other problem was that the aft airframe with the fins is too short. I wanted to save the fins, so I glued on an airframe extension. The tube I had available was a bit bigger ID and OD than the rest of the airframe, so I dremeled out a shallow taper on the inside of the new tube, and sanded a shallow taper onto the outside of the old tube, and then just JB Welded them together with the motor case holding them straight. There was about 1/4" of overlap, which is pretty good for this joint that doesn't have any bending loads, since the motor case will always be inside. It's not as aerodynamic as it could be, though.

Did you re-sim the flight with the slightly larger diameter airframe extension and the 'transition' ?

I wonder if the transition might have robbed some ot your altitude ?

Last Q for now ... How do the two altimeter data sets compare ?

It sure looks like there is a lot to learn from your flights !

Thanks for sharing !!

-- kjh

 

[Adrian A]

But I don't see an inflection in the accelerometer data when the rocket coasted down into the subsonic range.

Is it visible at higher axial accelerometer resolution ?

Here's the acceleration, zoomed in:

The roll rate pretty good IMO considering the speed and the diameter of the airframe:

I'm quite pleased with how my sustainer fins turned out.

Or maybe in the gyro data ?


Did you re-sim the flight with the slightly larger diameter airframe extension and the 'transition' ?

I did. The fin cross section choice is one of the biggest drivers, and with one of the most optimistic ones I was getting a good match on the Cd:

 

[kjhambrick]

Wow ! Very nice, Adrian !

NOW I see the transonic region in the acceleration data

Thank you !

Your acceleration and roll gyro show a very clean, slippery coast phase !

And your CD -vs- Mach Number curves for the two flights are GREAT !

Any ideas about the blip around Mach 1.45 and again at Mach 1.9 in the K627 CD -vs- Mach data ?

Thanks again.

-- kjh

p.s. I also meant to comment before that your integrated axial velocity and your GPS velocity converged VERY well when the rocket coasted out to 15 secs or so.

It is so beautiful when aero theory works in real-world data sets

 

[Adrian A]

Any ideas about the blip around Mach 1.45 and again at Mach 1.9 in the K627 CD -vs- Mach data ?

It is so beautiful when aero theory works in real-world data sets

I think that the blip at Mach 1.45 is from something like the smoke grain giving a little extra oomph or maybe it's some propellant that hadn't burned before. There's nothing like it in the J1026 flight.

I'm really curious about the step changes in Cd at M1.9 in the K627 flight and at M2.3 and M1.62 in the J1026 flight. They look quite similar to each other but at significantly different Mach numbers. One of the only differences in the airframe for the 2 flights is that on my replacement nosecone extension that I used for the J1026 flight, I have an accidental extra hole. I wonder if these Cd increases with Mach number are related to these holes tripping the boundary layer turbulent at that location or something like that.

 

[Adrian A]

Sunday's J1026 record flight is here, and then I followed that up with my G record rocket Half Fast, described here.

 

[BABAR]

Congratulations!