After having loads of fun with the Wildman – Intimidator 5 it is time for a new experimental rocket. A minimum diameter rocket for some serious altitude. Based on the very succesful KILO 6G motor, with a higher total impuls, more grains and updated with a light weight 86x3mm motor casing. Rocket will be made “sustainer-ready” similar like de DECA IV for possible use as a second stage rocket.
Keep coming back for regular updates.
- Rocket – KISS design, aluminium, true minimum diameter, experimental rocket.
- Motor – experimental KILO 12G KNSB “lite” motor (86mm).
- Flight control – RDAS Tiny.
- Onboard camera – Mobius mini (pics or it never happened).
- Tracking – Talky GPS by LD, 433mHz AM emergency beacon.
- Diameter – 86mm.
- Length – Approx 3000mm.
- Fully loaded – 21.700gr.
- KNSB Propellant – approx. 9.806gr.
- Cd – to be determined*.
- Motor actuals – to be determined*.
Note: * – to be updated according to as-built values
- Successfully fly a KILO 12G motor (12 BATES grains) with a “lite” motor casing (86×3 vs. 90x5mm) – even though the grain number is doubled from 6 to 12 grains.
- As a personal first: Go MACH 1+, breaking the speed of sound. Estimated maximum velocity MACH 1,5+.
- Validate the modified NERO design through-the-motor-casing fin brackets and 3mm fins in a supersonic flight.
- Safely recover the rocket with the use of high altitude suitable deployment cannons. Derived from the “high altitude deployment” article by Jim Jarvis and info from the Australian Rocketry Forum.
- Fly as a “sustainer ready” minimum diameter rocket.
- Explore the use of 3D printed parts for rockets.
“Flight 1”: Flights of Experimental Rockets (LRE) / Drawsko Pomorskie, April 2017. She “flew” for 10m after which a CATO abruptly ended the flight.
Design & Construction
KISS design – Form Follows Material, the non-optimised rocket geometry follows from matching, readily available materials.
UPDATE 20170907: Due to the KNER propellant misbehaving during casting, I opted not to invest any resources in this KILO 12G KNER project anymore.
What I noticed during casting KNER in the casting tube is that the solid KN particles settled towards the bottom of the grain. Leaving 25mm of plain ER at the top. With the normal spring-tensioned plunger type casting set-up the ER quickly “froze” which resulted in large air pockets (not bubbles). This phenomenon was also experienced by SF. The easy fix would be to overcast the grain / casting tube with KNER and cut 25mm of the top. This would result in and oxidizer rich KNER propellant which would have hurt the Isp. Since I had already cut the casting tube to the correct 135mm length and I didn’t feel cutting 12 casting tube with 25mm over length and cutting KNER grains I abandoned the KILO 12G KNER motor.
New motor hardware to have another go at 12 grains. So what is different with this motor over the original KILO 12G KNSB motor:
Velocity based erosive burning - As a rule of thumb I used a very conservative port-to-throat area ratio of 1.5 in all my motors designs over the past 10 years. For the ill-fated KILO 12G this resulted in a rather high average Kn of 700 and a MEOP of 100-130bar. In hind sight this 1.5 ratio is outdated and according to online literature "Erosive burning design criteria - Charles Rogers" values between 1.1 - 1.36 are acceptable. By increasing the nozzle throat diameter from 20 to 23mm and keeping the core diameter the original 25mm a new Ap/Ath ratio of 1.17 was realized. This increases the chance of velocity based erosive burning at ignition but the advantages of lowering the overall Kn outweigh this risk. As a direct result of increasing the nozzle throat diameter the average Kn was lowered from 700 to 530. This lower Kn in combination with the KNER propellant also results in a better manageable MEOP 50bar.
Mass flux based erosive burning - By chosing a slower burning KNER propellant at lower Kn the theoretical mass flux was cut in half from 6.87gr/s-mm2 to 3.43gr/s-mm2. The KNER will be hard to ignite so we are anticipating a delayed ignition with the regular BP/Mg booster charge.
Hardware - No big changes here. A static test casing of 90x5mm and a flight casing of 86x3mm were machined. Since I cut the casings before I decided to increase the nozzle throat diameter I had to change the convergent angle of the nozzle from 45 to 38,5° to fit the nozzle geometry in the same length. Nozzle material is the same tough EK20 graphite from the SGL Group. It's very aggressive on the tools and a headache to machine. For this test I decided to keep the material the same but some new fine isomolded graphite has meanwhile been ordered.
" order_by="filename" order_direction="ASC" returns="included" maximum_entity_count="500"]
Final thoughts on the CATO / overpressurization of the KILO 12G KNSB motor at the LRE 2017 launch:
- The motor had a very conservative port-to-throat area ratio of 1.5 to prevent velocity based erosive burning. This resulted in a higher than necessary, Knaverage of 700 and thus higher Maximum Expected Operating Pressure (MEOP).
- Mass flux based erosive burning. Little has been described about this phenomenon and KNSB or "candy" propellants although it's widely known among the APCP community. There is an informative article about both velocity and mass flux based erosive burning called: "Erosive burning design criteria - Charles Rogers". By modifying the SRM 2014 excel file I was able to calculate the mass flux for the KILO 12G KNSB motor at:
- KNSB coarse: 5,15gr/s-mm2
- KNSB fine: 6,87gr/s-mm2
- As an example, maximum mass flux for APCP at 55-96bar is approximately 1,41-2,11 gr/s-mm2
Although my gut feeling says that KNSB and candy propellants in general are not very susceptible to mass flux based erosive burning, the above mentioned figures might be a bit on the aggressive side.
- It appears the "KNSB coarse" propellant burn rate in SRM 2014 software isn't as well documented as the "KNSB fine" propellant. Although it provided accurate simulations on the KILO 6G motor, I suspect it isn't properly characterized at a Kn of 700 / chamber pressure of 100bar. Keeping the same motor configuration but only changing the KNSB propellant from coarse to fine resulted in an increase of the MEOP from 98 to 130bar. Since the casing (theoretically) starts to yield at 110bar and bursts at 160bar. I suspect a higher than 110bar chamber pressure, shortly after ignition, caused the motor to overpressurize.
- Collapse of the bottom grain / casting tube due to the high G-forces causing nozzle blockage thereby overpressurizing the motor.
- (micro) Fractures in the propellant due to the <0°C temperatures at night thereby increasing the burning area.
- In any of the above mentioned cases, the safety margin on the design pressure and the yield pressure of the casing, without static testing, proved to be too small.
- As a note: the grains were some of the nicest grains I cast with a very high density (average density ratio of 0,983) due to the spring-loaded mold design. Motor assemble went without a hick-up, nothing was forced and the liner fitted perfectly with no buckling. With confidence I can say the failure is not related to any of these items. Furthermore the casing ruptured in a typical hoop failure manner. Initiation appeared to be about 2 grains from the forward closure side of the motor. Since the casing failed it is expected that there was a chamber pressure exceeding 150-160bar. The forward closure remained in place with no blow-by validating both the forward bulkhead retention, as well as the o-ring design.
I will machine another nearly identical motor, make some minor alterations and static test this motor in September 2017.
The GIGA rocket has been designed as a true minimum diameter rocket with fin brackets mounted directly onto the casing. This is not a particular KISS approach but rather a result from the “sustainer ready” requirement. Alternatives considered were:
- Welded aluminium fincan. However due to the lack of availability of matching off-the-shelf materials (it is nearly impossible to find two commercially available tubes to accurately slide over each other) and machining the OD of the casing with the 100bar MEOP was consider not an option.
- Composite fincan. Downside was time.
- Both fincan designs suffered an additional challenge – how to retain such a fincan half way up the casing.
Hence the proven NERO through-the-wall fin bracket design was chosen and modified. Changes consisted of:
- Matching the 86mm OD casing
- Lengthening the bracket to meet the fin design
- Increased fin thickness from 2 to 3mm.
KvdL kindly offered to CNC these parts. Especially the 43mm radius is difficult to machine on a conventional mill. Needless to say: the fin brackets came out great! Subsequently FdB offered the use of his lathe/mill combo to accurately, parallel drill the M2 mounting holes in the casing. Finally the fins were cut with a sheer and matched in size on belt sander. ERGO 4203 hydraulic thread sealant was used to mount the M2 bolts into the casing.
" order_by="filename" order_direction="ASC" returns="included" maximum_entity_count="500"]
After some struggles with the high L/D ratio EDPM liner I managed to find an approach which worked well both inserting the liner into the casing as wells as inserting the grains into the liner. Below sequence seemed to work best:
- Talc both the casing and liner (in & out), remove excess talc.
- Insert both Full Length Coupler Tubes (FLCT) into liner. Measure liner length: 1629mm.
- Insert liner and FLTC assembly into casing, overlap on the side with loose flap facing bottom.
- Remove FLCT’s.
- Dry fit nozzle (no o-rings) and secure retainer to position liner into casing. Press liner against nozzle.
- Insert grains one by one (except last one) into casing from top. Use PVC dowel to push the grains in. Use a rolled A4 office paper as a sleeve to insert the grains in the liner.
- Remove nozzle, grease new o-rings and nozzle. Clean casing nozzle end from talc, add RTV inside casing + top convergent nozzle and grease. Insert nozzle into casing and secure retainer.
- Press liner and grains from top into RTV bead.
- Measure clearance liner to forward end casing and thereby verifying the liner is butt up against the nozzle.
- Grease new o-rings and forward bulkhead. Clean casing forward end, add RTV and grease.
- Place motor vertical and insert forward bulkhead and secure.
" order_by="filename" order_direction="ASC" returns="included" maximum_entity_count="500"]
Boring post but to be used as personal reference
Liner dimensions:1629 x 300mm.
- Width of liner: In general the liner is constructed with a 20% overlap. Hence: pi x ID of the casing x 1,2 = 301mm = approximate 300mm wide.
- Length of liner: Depending on chamber length but chamber length usually is grain stack + 10mm. In this case 12 x 135 + 10 = 1630mm. Liner is kept 1mm shorter to prevent it from inadvertently compressing / collapsing ergo: 12 x 135 + 10 - 1 = 1629mm.
- Liner is made from 1mm thick EDPM rubber bought at the local Intratuin. The EDPM is cut with x-acto utility knife along a 2m long sturdy angle bar. An appropriately sized piece of plywood is used for marking the 90 degree corner(s).
The typical roll and tuck method for rolling liners over a full length of casting tube works good up to a KILO 6G motor. Any larger and the liner fabrication became cumbersome to make with a high possibility of error and requiring multiple persons to keep it perpendicular. Hence a new method needed to be found. A more mathematical approach was selected. Goal was to check whether the average circumference between the inner diameter of the casing and the outer diameter of the casting tube would result in a workable liner.
Pi x 80 (ID motor casing) = 251mm
Pi x 76 (OD casting tube) = 239mm - average = 245mm.
A “ sanity check” was made by rolling a smaller test piece of EDPM liner over an actual casting tube which resulted in 242-243mm width of the EDPM liner.
- Before start marking I cleaned the EDPM liner from chalk etc at the appropriate place with some alcohol so the tape would stick to the EDPM.
- Using a steel ruler every 200mm (starting 20mm from the side of the liner) I measured 244mm with the ruler from the side and marked with a felt tip marker the head end of the ruler with a 10-15mm line. This line is about 1-1,5mm thick. So the middle of the marked line would be 245mm. Subsequently connected the lines to form a polyline out of 200mm lengths each along the length of the liner.
- Length wise I marked the liner every 100mm again starting 20mm from the end. Did this perpendicular thus intersecting the polyline and at the edge of the free overlap part of the liner so that when they were folded over they would match up for alignment.
- The liner was turned upside down and the flaps were folded over to align to the polyline and the perpendicular marks. Starting at the middle and working to one end of the liner, the flaps were taped in place at the 100mm marked points with some 7-10mm wide aluminium tapes.
- With all tapes in placed a single length of full width aluminium tape was taped over the overlap joint.
- The result was a liner which is easy to insert into the casing and the grains were easily loaded into the liner. During both operations plenty of chalk was used.
Photos will soon be inserted
- Liner dimensions
- Cleaned edges
- Liner ruler
- Folded over not taped
- Folded taped
- Fully taped
Need a good reason to thoroughly clean your lathe? Try machining some graphite. A good shop vacuum cleaner such as a wet /dry Karcher helped a lot in keeping the mess to a minimum. However it showed over time that it was unavoidable that, after machining a full graphite nozzle, the entire lathe was covered in graphite particles.
I had to cut the outer diameter of the nozzle to 79,60mm for easy inserting into the casing. The casing end I had in mind for the nozzle had a small mystery "ding" to it. Average inner diameter of the casing was 79,85mm but a minimum ID reading was as small as 79,70mm and the nozzle with the designed 79,7mm didn't fit. Being a bit carefull with the fragile graphite I opted for a 79,6mm OD. I also sanded away the inside of the small "ding" as it was less than a 0,15mm. The entire nozzle (of a fairly simple design) took me about 4 hours of machining.
The o-ring grooves came out at 70,7mm for the secondary o-ring groove and 70,8mm for the primary o-ring groove. Where the design called for a 70,9mm o-ring groove which served us well with the previous KILO motors. It will help inserting the nozzle but I hope it will hold up to 100bar. Even at the second try for the o-ring groove I didn't get it right. It was difficult getting to the correct depth as the dull tool didn't cut the final graphite cut but pushed the work piece. Mental note for myself: for the next graphite nozzle keep in mind below tips 1-3.
Drilling the center hole was easy, center drilled, 8mm drill through all, a 18mm drill through all and finishing to 20mm was accurately done with a boring tool.
Tool wear on the carbide inserts was limited and both the parallel turning / facing insert as well as the boring insert became dull (for machining metal) but it was not required to replace the inserts before finishing. Cutting the o-ring groove with a HSS tool was different and would required 1-2 sharpening sessions on a bench grinder for a single o-ring groove. It was a pain to get an accurate depth o-ring groove as a dull parting tool didn't cut the graphite but started pushing the workpiece. Also uneven wear on the tool bit caused different depths.
- Go back and forth in the groove with a dull parting tool to accurately (and actually) remove the graphite instead of pushing the work piece away.
- Re-sharpen to the tool just prior to the last cut, get a diameter reading (leave 0,1-0,2mm left) and remove the final material by going parallel to the work piece for an even finish of the o-ring groove depth.
- Take a wet stone in absence of a bench grinder to keep te tool wear even / parallel to the work piece. Better yet, use a carbide insert cutting tool.
One week, one part. The nozzle retainer will hold the full graphite nozzle in place. It is a variation on the simplest form of a retaining ring as used earlier on the KILO 6G motor. The variation consists of:
- the nozzle retainer has a divergent section where it lengthens the actual divergent section of the graphite nozzle thereby increasing Isp.
- Secondly the nozzle retainer acts as a boattail and has the same OD as the casing. The 15° boattail - simple chosen to match the divergent angle - will help to reduce drag and, since the GIGA is designed as “sustainer ready”, will help keeping the rocket aligned. When used as a sustainer I'm planning a none binding taper in the interstage coupler. This will perfectly center the rocket on the booster and eliminating any radial movement. Down side of this design are the blind holes required for the M5 threads which are tedious to make.
Finished the casing today. It is 1780mm long and will hold 12 BATES grains. Important factor in long rockets is that all parts being properly aligned. To facilitate this all the tubes (casing, airframe & nose cone) rest on square faces. As I didn't want to rely on the squareness of the band saw cut I had to square off the ends of the casing. However I don't have a steady rest which can accommodate the 86mm tube and didn't want to spent the time (yet) on a DIY steady rest. Therefore I outsourced this to the machine shop at my work. They had a lathe with a 100mm ID spindle to directly clamp the tube it in and machined it to length. Now it is perfectly square.
For indexing the countersunk M5 bolts I reverted to my indexing technique by fixing a small indexing head to my lathe and an expandable mandrel clamped into the hollow spindle of my lathe. I have a laser cut custom tool, made out of 10mm steel plate, for an electric drill with a 43mm collet which mounts to my quick change tool holder. On the the lathe:
- Center drill,
- 4mm tapping drill (approx. M5)
On the column drilling machine on the slowest speed possible the counter sinking was done, lubricated with plenty of ethanol.
Finished the combined retainer ring / coupler for the KILO 12G "lite" motor. The coupler was made from 80x5mm tube and almost a direct fit into the 86x3mm casing. It is a bit beefy. However lets first see how she copes with the 96bar MEOP and the countersinking of the 86x3mm casing which will probably extend into the retaining ring / coupler effectively reducing the threaded area. Also I didn't have a boring bar long enough, or even bothered, to bore out a significant weight saving from the inside. KISS. After a succesful flight it can potentially be optimised to a 80x3mm coupler.
The nose cone avionics arrangement went through quite a few iterations as it was more or less designed around the 4000mAh battery. However it proved to be quite a challenge mounting the battery on the centerline of the rocket / nose cone and still being able to fit the other components. Although the 86mm 5:1 Von Karman nosecone is quite spacious it does require a central M6 threaded rod. When tightened, the aluminium nose cone tip and bulkhead securely clamp the nose cone in between the two parts but basically divides the internal space in 2. The reason I didn’t want to place the battery back off-center was because of the (expected) inertia roll coupling. Actually, a relatively heavy battery pack mounted in the very top of the rocket is the worst place to position such an off-center object. With the traditional mechanical mounting methods (which use a combination of bolts, nuts, threaded rods, angle bars, u-profile and G10 plates) I wasn’t getting a satisfactory design. So in the end I drew up a “dumb” (very simple) 3D print which perfectly met all requirements.
Nose cone avionics consisting of:
- 3D printed battery holder.
- 3D printed pull pin dual switch array.
- Main battery pack: Turnigy 4000mAh 2S intelligent transmitter lipo pack.
- 1W Talky GPS by LD. Featuring both APRS position messages as well as live audio feedback of altitude and position. A stepdown converter set to 4.2V sits between the battery and Talky GPS as the tracker is designed to work on a single lipo cell only.
- 433mHz AM emergency beacon by LD on a separate 850mAh 1S lipo.
" order_by="filename" order_direction="ASC" returns="included" maximum_entity_count="500"]
A 3D printer camera shroud for the Mobius Mini was ordered from Additive Aerospace. However upon closer inspection I noticed the design could be optimized a bit for the GIGA rocket being:
- The Additive Aerospace camera shroud used four screws positioned in a square to secure the shroud to the airframe. This is particularly hard to accurately drill in an aluminium tube as the holes are not radial / perpendicular to the centerline. A skewed camera shroud could induce undesired roll to the rocket. Slender rockets, like the GIGA with an L/D of 34:1, are highly susceptible to Inertia Roll coupling. Taking the roll out of the equation – as much as practically possible - is one of the mitigating actions which can help to reduce the possibility of this phenomena.
- Improvement: Removed all mounting holes and added only two holes on the centerline of the part. These holes accept M3 socket cap screws. The bolt head tightly fits into the 3D printed part thereby centering the camera shroud and absorbing the loads from the 3D print into the airframe.
- We / I approach the rocket for activating the camera and arming from the same side. This is where the pull pins are also conveniently located. However the status indicator LED of the Mobius Mini is at the exact opposite when it is mounted in its normal orientation. Not a big problem for a single stage rocket where you can walk around the rocket but when it is used as a sustainer and it’s 5m in the air it would preferable be on the same side as the ladder and pull pin. Hence the camera needs to be mounted inverted in the shroud and due to recess for the protruding push buttons this is currently not possible.
- Improvement: the recess for the push buttons inside the camera shroud was mirrored vertically to accept an upside down Mobius Mini. Furthermore the camera was programmed to record the image inverted.
Results: the camera is now accurately aligned to the airframe by means of two M3 bolts and the status LED is now facing the operator arming the electronics.
Consisting of two-piece pin where as one part can remain in the rocket during preparations and a second part containing the “remove before flight” tag can be screwed on/off during assembly. With the Intimidator 5 we noticed that a protruding pull pin, especially in combination with parts / coupler pieces which need to slide into each other, can mean that some electronics are armed during assembly of the rocket. While arming electronics such as a GPS tracker aren’t much of a concern. It’s a different situation when the altimeter is connected to live deployment charges or possible even sustainer ignition. Hence the need for a two-piece design to isolate the pull pin in the rocket. When treading the two parts together the rotation of the internal pin is blocked by 2 flat surfaces filed onto one end of the pin. The “switch array” which accepts the pull pin has corresponding flat surfaces and is made on a 3D printer. More info on the switch array below. The internal pin has a male M3 thread, the external “remove before flight” part has a female M3 thread and an aluminium socket soldered on the other end for easy attachment of the key ring. I found it easier / better to put solder on the pin first and then insert into the socket for a more secure bond. Having the 2 parts slid into each other, heated together and rely on the solder to creep into the crevice didn't quite work out for me.
I adopted this clever idea from LD, to use a 3D printer for creating a switch array part. In the past using the OMRON subminiature switch as an on/off pull pin switch usually consisted of time-consuming, fiddling around with small parts and relatively tight tolerances to make it work reliable. With the 3D printed switch array this is a thing of the past and can be simple bolted with ordinary M3 bolts to the electronics bay. The switches are OMRON SS-5. I ordered the SS-5GL in the past but removed the metal lever for use in the 3D printed switch array. Due to the 74mm ID of the electronics bay and internal width I was limited to 2 switches as I wanted the plunger of the switch facing down along the direction of the G-forces. One switch is used to power on the RDAS Tiny while the second is connected to the safe/arm of the RDAS.
After reading Jim Jarvis’ excellent article on high altitude deployment I designed the electronics bay to integrally hold the 10:1 deployment cannons. I didn’t want a cannon in the airframe parallel to the packed parachute as with one of the DECA flight I experienced a blow by of the straw deployment charge which was positioned next to the packed parachute. Although only required for the apogee charge, to avoid asymetric weight distribution (one of the causes of inertial roll coupling) a second deployment cannon was integrated to hold the main parachute deployment charge.
For the cannons I chose 16x2mm aluminium tube which conveniently accepts M16x1 metric fine thread while still leaving 1mm wall thickness. L/D of the deployment cannon is approx. 10: 1 which results in 120mm internal length. 15mm is taken up by the maximum, initially anticipated 2gr BP charge (pending ground testing of the deployment charges). For simplicity the igniter leads are conventionally fed from the outside of the electronics bay into the cannon which is flush with the bulkhead. The plug which seals the cannon inside the electronics bay is press fitted into the tube, locked in place with a small nail and sealed from the inside of the cannons with a layer of 5min epoxy.
A single RDAS Tiny is used as the altimeter. Deployment charge igniters and sustainer ignition are soldered directly to the wire loom eliminating screw down terminals. The whole electronics bay and deployment charges can be prepped in advance ready to be inserted during rocket assemble.
Recovery notes such as actual flight desent speed will be mentioned below.
Mass empty: Approx. 12.100gr. Lift off 21,7kg – 9,6 KNSB propellant = 12,1kg mass emtpy.
- Desent velocity while drogueless – approx. 30m/s*.
- Desent velocity while under (main) parachute – approx. 6m/s*.
GIGA recovery hardware consisting of:
- 5m - 3/8" tubular kevlar
- Kevlar burito secured to the shock cord at 1m
- 60" Iris Fruity Chute incl swivel.
" order_by="filename" order_direction="ASC" returns="included" maximum_entity_count="500"]
During machine it was noticed that the wet lay up has, as large as 1mm, eccentricity. Even when the alignment jig was used to center the 3D printed plugs in line. Hence, up to 2mm extra outer diameter needed to be laminated so any eccentricity can be removed during the final machining stage. A stepped 5:1 Von Karman shape was machined on the lathe using a parting tool and approximately 130+ 3mm-steps. This rough shape was sanded smooth to the final dimensions. Although the 130+ steps were a bit time consuming it went quite easily with excellent tolerances. An epoxy sanding coat of 24gr epoxy, 2gr glass bubbles, 0,4gr aerosol was applied and again sanded smooth on the lathe. After the final sanding the coupler diameter was machined according to the required outer diameter and the last step was to cut the tip of with the parting tool. The CNC machined aluminium nose cone tip was outsourced.
Notes: the aluminium nose cone tip design could have used a shorter shoulder and deeper thread for easier securing the nose cone bulkhead & electronics sled into the nose cone.
I started out making a truncated gore pattern and a spray template out of cardboard. The template can be downloaded from here: 86mm Von Karman nose cone / spray template. The two A4 papers can be easily aligned with tape and secured to the 1mm thick cardboard with spray adhesive. You will have to fabricate two of the same templates. One for the truncated gore pattern and another one for the positive / negative spray template. The cardboard can be easily cut with a sturdy pair of scissors. Furthermore I used a water based felt tip pen to mark the gore contour on the glass fiber and added a centerline for alignment as a waterproof marker line will dissolve in the epoxy blurring the lines. Sprayed the glass fiber edges with a light coat of hair spray to keep the glass fiber from fraying and using the spray template to make sure the rest of the glass fiber was kept clean of hair spray – keep the hair spray coat as light as possible as over doing them will make the glass fibers stiff. Finally I cut out the gores with a pizza style rotary cutter (much better than using scissors or a conventional knife). Used aluminium foil to store the separated gores without much fraying. Indexed the nose cone plug in six parts again using a water based felt pen.
Prior to the lay-up I estimated the epoxy batch size. Totals of the 280gr/m2 twill glass fiber cloth are approximately: 1,53m2 for 24 gores which includes cutting loss (24 gores x l x w = 24 x 0.51x 0.125m). This results in 428gram glass fiber and with a 50:50 glass fiber / epoxy ratio this boils down to an initial batch size of 430gr epoxy. Due to the pot life multiple smaller batches are required and spread over the two - 12 gores - lay-up sessions. Since the above estimate includes the glass fiber cutting loss already no extra epoxy required. Update: 20161108 – Actual used epoxy = 160gr epoxy for 12 gores. 100grams batches are too large. Therefore use 50-70gr batches.
Wetted out the sanded 3d printed plug and coupler, laid down the gores from tip to shoulder using the center line as marked on the glass fiber gore and the nose cone index line. Used a 30mm wide brush and cut down the hairs to approximately 10mm. As from approximately < Ø30mm, the 280gr/m2 twill glass fiber didn’t follow the tip of the nose cone contour so well as the perpendicular gore fibers stand out. The bulging glass fiber became a challenge to handle after 6 out of 12 gores were laminated and the tip laminate was easy to shift / disturb over the plug. So I waited for the epoxy to fully harden, sanded down the laminate with 80 grit sand paper. Surprisingly, the second session of the remaining 12 gores seemed to better grip and little to no sliding / movement was noticed and the wet lay-up stayed nicely in position without bulging. Possible solution for the first 12 gores session would be to laminate a single light wrap of glass fiber around the tip and allow it to reach leather stage. Than continue with the first 12 gores however this would require a 3 stage wet lay-up. Alternatively some sort of heat shrink tubing to force the laminate to follow the nose cone plug is also possible.
The wet lay-up of 12 gores takes about 90 minutes. Hence 3 epoxy batches with a pot life of 30min each are required. The total of 24 gores should make up for a 4,8mm wall thickness which will be machined down to 3mm.
Alternatively to the foam plug I decided to try out a 3D printed plug for increased accuracy of my 86mm Von Karman composite nose cone. Initially this was due to the fact I ran into some inaccuracy / concentricity difficulties of the finished wet lay up of the foam plug (which in hind sight was not related to the use of the foam plug). Furthermore it is interesting to explore the potential uses of a 3D printer in rocketry applications.
So I modeled a 2 part nose cone plug to be 3D printed and ordered it through www.3Dhubs.com for EUR 31,-. This 3D printed plug will be aligned butt-to-butt to the aluminium coupler and once the wet lay up is finished the outer shape of the nose cone, shoulder and outer diameter of the coupler will be machined in the lathe to final dimensions. The 3D printed part will remain in place, locked between the aluminium coupler and the composite nose cone. After a heads up from LD about the lack of concentricity by simply epoxying 3D printed parts together, I decided to make a M10 central spindle with center rings to keep all parts are aligned.
So it is time for a new rocket, a minimum diameter one. With this new rocket the need for a custom nose cone arises. After some consideration and positive builds by “cryoscum” on Australian Rocketry Forum I decided to make a foam nose cone with wet lay up. Just to test this principle as I had made nose cones from negative molds in the past already.
Some design starting points:
The 86x3mm motor casing has an outer diameter of 86mm and I decided to use the same tube as an airframe to keep the design simple with materials already on hand. Furthermore, convenient coupler material was available in the form of 80x2mm aluminium tube. So I needed to make a nose cone with an OD of 86mm and an ID of 80mm, resulting in a wall thickness of 3mm. A bit thick but a compromise to stick with the KISS design.
To start, I needed a profile box so I drew up a equation driven arc in SolidWorks with an offset of 3mm so I got a 80mm OD foam nose cone plug. The .dxf output file was sent to Snijlab.nl and soon after my laser cut 9mm MDF plate arrived. The profile box was quickly glued together with some 5min epoxy. Some rectangular shapes included in the cutting file were glued on the bottom the keep the box nice and square.
As 15mm central spindle made out of central heating pipe was used to keep things concentric. EPS foam plates from the hardware store were cut in 100x100x50 squares with a jig. A 13mm hole (smaller than the 15mm pipe) was drilled roughly in the center. These 100x100 plates were sprayed with EPS suitable spray adhesive on both sides. A pointy head inserted in the 15mm pipe helped stacking the plates on top with each other. A DIY hot wire bow was connected to my LIPO charger which has a hot wire function. The 0,4mm nichrome hot wire settings were calculated with an online calculator and worked like a charm. Important lesson I had read this on the internet, is to pull the hot wire from nose cone tip to shoulder due to sagging of the wire. The reason makes perfect sense when starting hot wire cutting. The foam plug after hot wire cutting looked like a 16-32 segment polygon, a bit of free-hand sanding removed the hard edges. Not to rigorous because it is easy to remove too much.
Note: I do not recommend to take this route for <130mm nose cones. It is just too much work to get right. I would advise to make a wooden plug including shoulder with final dimensions and fabricate a negative mold into which you add the wet lay-up (with the plus that nose cone can be made a lot lighter / thinner) or an undersized positive plug / mold similar to the foam plug onto which you add the wet lay-up.
Some of the down sides: The laser cut hot wire shape box is accurate, the foam plug after hot wire cutting and sanding is not so accurate anymore. Ergo, the outside shape, especially after wet lay up which introduces additional tolerance deviations, results in a not so truly accurate representation of a Von Karman nose cone shape and requires additional machining / sanding. It’s close +/- 0,5mm but not perfect / concentric which one might expect when making a high performance nose cone – or that might be my OCD talking 🙂
Lessons learned: 86mm 5:1 Von Karman nose cone – foam plug difficulties:
However in hind sight I made some assumption mistakes during the wet lay up (expecting I could make a 86mm Von Karman nose cone shape out of a 85,2mm lay-up by shifting the Von Karman curve over the longitudinal axe. This approach does work for conical nose cones but not for a Von Karman nose cone without anticipating this in the design phase. Furthermore I had some concentricity problems when the foam plug based composite nose cone was placed into the lathe for true-ing the shape.
- Foam plug can’t be made with accurate tolerances like 0,1mm.
- Compensate lay up thickness for concentric / eccentric deviation up to 1mm wall thickness or +2mm of the outer diameter.
- Machining VC shape from shoulder to tip – easier
- Cutting foam from tip to shoulder – sagging of the wire
- Use alignment jig for 3d printer parts or design 3d part to be self aligning. M10 thread is not necessarily straight.
- Part NC tip diameter not 0,5mm smaller than NC tip. Initially good idea as can be make bigger and does not result in “hump”. But better to be same as NC tip and solve with paint thickness.