GeckoSub Mirage Evo - And Adventures in 3D Printing Speargun Parts

[rollerGR]

There has been some research on speargun dynamics, biggest paper on it was by Spiller (he posted about it here some years ago) I think. It showed that one of the most important factors isn't just force (m*a) but rather the work done throughout the barrel. This charges the shaft with KE which is what fights drag and allows the shaft to maintain speed (and thus momentum) better. It's why rollers/inverts shoot more powerfully with half the rubbers (total kg on the shaft/mech) and can use lighter shafts. I don't use pneumatics but my understanding is that they're like rollers.

Other tests from Rob Allen (very heavy shafts) and another guy from Greece have been made but that's mainly for the acceleration curve. Maybe that's what you mean by chronograph? See pics

Rollers/inverts accelerate more slowly and the shaft spends much more time and distance under peak or near-peak force which allows for incredibly more energy charge. But because there is a speed limit (rubbers themselves contract at about 50 m/s without load) you can only barely squeeze out some more velocity... but the energy can go up! This is the biggest takeaway and it's rather counter-intuitive. So you can get shafts with very similar maximum speeds but very large energy differences. Momentum also scales with KE
p=√(2m⋅KE )
This is just one of the various equations that are in play. Here are a few others
dv/dt = - (k / m) v²
(1/2) m v₀² = ∫₀ᴰ (k v(x)²) dx
(1/2) * ρ * C_d * A * v²
And so on, for calculating different things. I didn't expect it to be as complex when getting into it but it's actually easy to understand after you get the whole picture. There is no paper (or study) of these out there yet as far as I know. Spiller was the only one to get into KE up until the muzzle. And up until then -2021- nobody really had any idea how to calculate it, people just assumed velocity is the key (it's not).
Then there are other factors like less jerk which results in smaller energy loss during firing.

For shaft mass there is an approximate table. See pic
In the Med, 370gr are enough to hit a fast-moving 3-4kg fish that's 5m away from the tip. Heavier shafts fired with the same power will travel farther away but will reach the target less fast. So it really depends on what you're hunting and from what distance.

 

[Diving Gecko]

There has been some research on speargun dynamics, biggest paper on it was by Spiller (he posted about it here some years ago) I think. It showed that one of the most important factors isn't just force (m*a) but rather the work done throughout the barrel. This charges the shaft with KE which is what fights drag and allows the shaft to maintain speed (and thus momentum) better. It's why rollers/inverts shoot more powerfully with half the rubbers (total kg on the shaft/mech) and can use lighter shafts. I don't use pneumatics but my understanding is that they're like rollers.

Other tests from Rob Allen (very heavy shafts) and another guy from Greece have been made but that's mainly for the acceleration curve. Maybe that's what you mean by chronograph? See pics

Rollers/inverts accelerate more slowly and the shaft spends much more time and distance under peak or near-peak force which allows for incredibly more energy charge. But because there is a speed limit (rubbers themselves contract at about 50 m/s without load) you can only barely squeeze out some more velocity... but the energy can go up! This is the biggest takeaway and it's rather counter-intuitive. So you can get shafts with very similar maximum speeds but very large energy differences. Momentum also scales with KE
p=√(2m⋅KE )
This is just one of the various equations that are in play. Here are a few others
dv/dt = - (k / m) v²
(1/2) m v₀² = ∫₀ᴰ (k v(x)²) dx
(1/2) * ρ * C_d * A * v²
And so on, for calculating different things. I didn't expect it to be as complex when getting into it but it's actually easy to understand after you get the whole picture. There is no paper (or study) of these out there yet as far as I know. Spiller was the only one to get into KE up until the muzzle. And up until then -2021- nobody really had any idea how to calculate it, people just assumed velocity is the key (it's not).
Then there are other factors like less jerk which results in smaller energy loss during firing.

For shaft mass there is an approximate table. See pic
In the Med, 370gr are enough to hit a fast-moving 3-4kg fish that's 5m away from the tip. Heavier shafts fired with the same power will travel farther away but will reach the target less fast. So it really depends on what you're hunting and from what distance.

Will read more carefully later on but I think I did see some posts about Spiller's work back then - but I took some years off of the forum and the hardcore nerding out so never really went into depth on it. So, I am kinda slowly catching up these days. But it's great to have more proper academic thoughts on speargun performance as I realized early on that a lot of performance claims are never backed up by any kind of data. People would make statements but often they were based on feel (bias) or rather crude testing.

As for chronograph, it's an instrument that measures the speed of a projectile. So, it would be a proper real life data point. Shooters who tinker with ammunition use them. Even some PCP airgunners and possibly archers, too. A chrono test is often included in gun reviews and tests. The instrument is pretty simple and not that costly anylonger - but that's for on land use and not sure if anyone has made something similar work underwater.
As you post proves, a lot of the best work done on this subject in our hobby is theoretically or with fairly simple pool testing where it often ends up "only" being comparative which was why I wondered about how to actually measure this in the real world, if possible

 

[tromic]

Thoughts On Practical Speed and Energy Measurements
Admittedly, the math and physics of all this was never my strong side but yes, that's the conundrum as I always understood it: A lighter shaft with a faster shot that will taper off in energy vs. a heavier, slower shaft which may carry more energy further out.
I guess without proper data, it's one of those points that can be argued either way and has been for decades.

This is just speculation but given that airgun shafts are rather short and thus don't have much mass combined with the fact that we often have quite a lot of energy stored, perhaps the rule of thumb should be to keep the shaft as long/big as possible. Especially on a Mirage-style gun with even more energy potential.

From time to time, I do think about underwater chronographs and ways to measure the energy of the impact. But I have way too many other projects before I can get to that point. I think the best I can do is to take a known good, tuned bandgun and put my airgun up against it. We wouldn't get specific numbers but we would see comparisons.

If I was to make a chronograph some day - and I may have picked this idea up from some of Tomi's older work - a way to do it would be to use contact microphones (you could probably use simple piezo resistors) and mount one on the gun handle to sense the trigger moment and one on the target to sense the impact and then record the two signals into either a laptop or audio recorder on the pool side. Then looking for the two audio peaks from the impacts in editing software would give us the time between them and thus, the average speed could easily be calculated if we were good at keeping to a known, precise distance. You would possibly need to amplify the signal but that can probably be done really cheap with some off-the shelf analog amplifier modules.
But for average speed, just counting video frames off of a modern actioncam in slow motion would be precise enough, too.
But would this combined with knowing the mass of the shaft could give us the energy of the impact? Maybe not as we don't actually have the speed at the point of impact but only an average speed?

I think gun chronographs which do read the speed of the bullet at one point only - as opposed to an average speed - work with optical sensors but whether they can be made to work underwater, I don't know. If they could, you could have two - one right at the muzzle and one right at the target. I guess that would give us the best data.

Practical energy measurements would be harder. I have thought about mounting load cells on the target - maybe on 3-4 points of a plate and then averaging the readings. But I don't know if load cells and their "drivers" can sample fast enough to actually measure the peak of the impact.

Once you start going down this road, you quickly realize why Majd, back when he used to relentlessly test settled on "just" comparative penetration testing between various guns but always on the same target material - in combination, of course, with accuracy tests.

It's been years since I last went into this rabbithole and at the time, noone had made an UW chronograph work to my knowledge but maybe it's been done by now?

This would be perhaps the most easy to accomplish and the most obvious: " comparative penetration testing"

 

[tromic]

This would be perhaps the most easy to accomplish and the most obvious: " comparative penetration testing"

This was my indirect comparative energy ("penetration") testing equipment. It worked very well.

 

[Diving Gecko]

This would be perhaps the most easy to accomplish and the most obvious: " comparative penetration testing"

True.
Actually, it's a little bit embarrasing that I haven't gotten much of that done yet since I have very good access to a pool on this island and could test pretty much as often as I would want/need. But just too much other stuff to do. What I don't really have is a known, perfectly good benchmark. Some years back, when I was based in China, the problem was the other way around. I had a friend with a perfectly tuned Abellan and other good guns but we were not in the same city and getting pool time was an issue.

But of course, I can still do comparative testing against one of my own guns so I can validate my itirations, development and setup changes that way. I just wouldn't be able to compare it to anyone not in the same pool, almost.
I do that sometimes in terms of picking a shaft thickness but you could use it for many things. It just takes a long time, longer than most people think. But let's see if next year I will have more time on my hand for this kinda stuff.

Actually, let's say we did find out how to build an underwater chronograph and impact energy sensor - the idea would be that we could compare "worldwide" but then we would likely start running into issues where people would question if each test setup is calbrated properly. I see that happening in the world of engine tuning. If people don't trust one guy's results or statements, they start blaming the dyno setup, haha. So, yeah - maybe just keeping it simple is good enough for a long while.

 

[tromic]

How did I use this rubber tubing testing?
I would shoot an arrow with a tested slide and/or a tested string (of various types and thicknesses) underwater. It is enough that only the rifle is underwater, you do not have to dive and do not have to shoot at the target. The length of the rubber elongation when fired is recorded. The measurement is repeated several times for each setup being tested. A higher average elongation means that the arrow has a higher momentum or kinetic energy at the measured distance. It is also possible to obtain more precise data using this simple equipment, such as the kinetic energy of the arrow at the measured distance, momentum, speed, etc.

 

[Diving Gecko]

The Downside Of Low Profile Nose Cones
This gun has a custom nose cone CNC machined out of aluminium. It seems to work well but I would like to slim it down even further but that's for later.

Here's a refresher:

Notice the two small threaded holes on the sides - we will get to them in a bit.

"Historically" we call this part the nose cone. I assume it comes from the more or less cone shaped protrusion forward of the reservoir. Like this original Mirage nose cone:

But my alu nose cone is missing the whole cone part. I have nothing really sticking out. Which is fine, I have grown to really like this more utilitarian look which also gives you more reservoir length for the same overall length but there are two downsides. The nose cone normally has a sight on it which I will now have to glue onto the reservoir itself but that gets in the way during assembly and disassembly as it sit exactly where you'd normally grab the reservoir. Which leads us to the next and bigger downside, there's nothing to hold onto on the nose cone itself during disassembly as the cone is gone... It's actually pretty much impossible to get a good enough hold on it to pull it out of the reservoir.

And that's where the two M3 holes come in:

I obviously planned this a while back since it's in the CAD and I machined it with the rest of the features but it took a while for the eyebolts to get here and I only got around to actually using them a few days back. The idea was always that the eyebolts would somehow help in pulling the nose cone out of the tube (it's a tight fit with quite a bit of compresson on the o-rings so it doesn't just slide out by itself). But honestly I wasn't sure exactly how I was going to use them. I thought of just having some dyneema between them that I could grab onto and pull but I went a different way. The black Delrin plug to the right is there to protect the shooting barrel when I use a screwdriver inserted into the eyelets to pry the nose cone out:

Swap between each side a few times and the nose cone comes out pretty easily now:

I actually don't want to travel with a flathead screwdriver or any screwdriver as I don't have gear that needs it. So, my idea is to make a multitool that does everything needed to service this gun. But that's for later. Come to think about it, I could probably use the tip of the shaft to do the prying. Also, I don't want to carry around a lot of extra weight in spares but those two eyelets and the Delrin plug weighs next to nothing, so I am OK with those.

 

[Diving Gecko]

Servicing Pressure Gauges
This wont be a great post as I didn't shoot proper before and after pics. But I had a pressure gauge where the glycerin inside had gone so yellow and dark I couldn't read it anylonger. I always suspected you could put in new oil as there's a small rubber grommet on the back of these gauges so I looked around and found some silicone oil meant for this exact thing.

The only pic I have is this one. The yellowed one is still way better than the one I serviced which is the super clear one on the right:

The top one has had its oil drained but there's a lot of goo and dirt on the inside that I am still trying to get out. The one on the right didn't have that problem so after draining the old oil, I swooshed around some alcohol a few times and then refilled with new, clean silicone oil and put the grommet back in. I definitely put more in than the factory does as I only have this small bubble left:

I should probably drain a bit of it out...

 

[Diving Gecko]

Still Leaking...!

I haven't mentioned this in the past few posts but my two CF reservoirs are still leaking. I also haven't had time to work on them until now since I last tried to fix this. But what what's strange is the reservoirs would leak from the same point - both of them near the trigger guard of the handle even though they seem really smooth on the inside now. And what was worse, if I turned the tube 180 degrees and tried again, it would still leak near the trigger guard.

Which led me to believe that perhaps the handle itself was actually where the leak was. But even though it's pretty beat up on the outside, the o-ring groove looks smooth and unharmed:

I even tried prying the flange with some alu rod to see if there was a crack I hadn't noticed but it didn't seem like it. But to be sure, I had to make a quick end cap and dummy shooting barrel (as I don't have an alu barrel that matches the length of this handle and any of my shooting barrels:

I made it so it just covers the o-ring and leaves the rest of the boss that normally inserts into the reservoir "naked" so that I could check if the leak was in that area. But of course nothing. Not a single bubble escaping. Which on one hand is good because a cracked, leaking plastic handle would not be easy to fix but on the other hand it's perplexing as I have no clue why the CF tubes leak near the trigger guard no matter how I turn them in relation to the handle...

Anyhow, it seems I need to coat the inside ends again. But on that task, I was also behind on the cutoffs that I still intend to blow up. So, I did those first.
I have mentioned that years back, I would normally use a really thin layer of fiberglass and I finally have some again:

The idea is that resin wetted out in a fabric has a harder time orange peeling - which the resin I have on hand is particularly prone to.
But before I get to coating, I put the cutoffs on the lathe and evened out the last round of coating - which had some pretty bad orange peeling:

The electronic leadscrew and DRO works a charm now. I don't know how I ever made do without an ELS for this long

As usual, once the epoxy is out, I don't snap many pics, but here's one after the coating:

It's always a neat trick to cut off extra fabric once the resin has starting curing but before it gets too hard. But with this 0.05mm thick fabric, I didn't bother. I just cut it the next day without any problems. There was a bit of a line where the wrap overlapped but I sanded it all down and it looks pretty good now. And even though the layer is thin, I could sand with #220, #400 and #800 without having to be too careful about not going through it - which was a nice surprise. (Once the sanding "water" turns from white/light greyish to dark you know you have gone too far and are now into the carbon fiber...

Also, I left these two pieces as is. I didn't bother to add the servo back to the lathe to turn it into a rotisserie as I did some posts back. But I still want to spin the real reservoirs once I get to coating those.
And that leads to the next little project...

 

[Diving Gecko]

A New (Standalone) Rotisserie

The "lathe rotisserie" from a few months back was made possible by swapping out an encoder for a stepper motor and adding an external stepper driver and power supply.
Here's a reminder of that setup:

But I don't want to have to swap parts out like this anylonger as the lathe is running really well again and I use it almost daily.
Back when I added the servo to turn the lathe into a rotisserie, I actually did place an order for some cheap 12V DC geared motors to build a standalone rotisserie. They arrived a while back and I thought, since I now have them and since I don't wanna mess with the lathe, I might as well put them to good use.

The motors and the speed controller were cheap. I bought three motors with different gearings/speed but had I just gotten one motor, the total cost, incl. the speed controller, would have been less than USD 7.
I also had some 2020 aluminium extrusion left over from another project and the rest of the parts, except some small ball bearing wheels are 3D-printed.

My FDM 3D-printer is really crappy these days. I could spend quite a lot of time getting it to run better - but it's old tech, slow and unenclosed so I think I will just upgrade to a (used) enclosed one next time I have some money come in.

Anyhow, the pics will be pretty self explanatory so here it is almost fully done:

It probably looks bigger than it is - the reservoir is only about 870mm long. But of course, the alu extrusions make it a bit over-built. But it's what I had on hand and they are easy to work with.

To actually get the tube to spin, I plan on keeping it simple: I will 3D-print a custom pulley for the motor and use an o-ring as a belt. The o-ring will just go over the tube. If I need more grip, I will first try some painters tape wrapped around the tube. The motor can slide up and down on the legs to ajust the tension on the "drive belt". Hopefully it will work just fine.

The only thing needed to get this to work is that 3D-printed pulley which I will make tomorrow and then a little enclosure for the electronics.

P.S.
I may have mentioned this before, but the perfect solution to a lathe working as a rotisserie would be to change the BLDC headstock motor out for a beefy servo motor which would be able to run 1-3000RPM at full torque which means, I could easily go from rotisserie speeds to proper turning speeds just by turning the potentiometer for the RPMs. That was actually the plan but I spent a bit too much money lately on other gear so no servo motor for the lathe just yet. Also, with the ELS I can thread at pretty high speeds where the BLDC has plenty of torque which again reduces the need for a servo motor.

 

[tromic]

A New (Standalone) Rotisserie

The "lathe rotisserie" from a few months back was made possible by swapping out an encoder for a stepper motor and adding an external stepper driver and power supply.
Here's a reminder of that setup:

View attachment 61286

But I don't want to have to swap parts out like this anylonger as the lathe is running really well again and I use it almost daily.
Back when I added the servo to turn the lathe into a rotisserie, I actually did place an order for some cheap 12V DC geared motors to build a standalone rotisserie. They arrived a while back and I thought, since I now have them and since I don't wanna mess with the lathe, I might as well put them to good use.

View attachment 61287The motors and the speed controller were cheap. I bought three motors with different gearings/speed but had I just gotten one motor, the total cost, incl. the speed controller, would have been less than USD 7.
I also had some 2020 aluminium extrusion left over from another project and the rest of the parts, except some small ball bearing wheels are 3D-printed.
View attachment 61288

My FDM 3D-printer is really crappy these days. I could spend quite a lot of time getting it to run better - but it's old tech, slow and unenclosed so I think I will just upgrade to a (used) enclosed one next time I have some money come in.

Anyhow, the pics will be pretty self explanatory so here it is almost fully done:
View attachment 61289

View attachment 61290
It probably looks bigger than it is - the reservoir is only about 870mm long. But of course, the alu extrusions make it a bit over-built. But it's what I had on hand and they are easy to work with.
View attachment 61291

To actually get the tube to spin, I plan on keeping it simple: I will 3D-print a custom pulley for the motor and use an o-ring as a belt. The o-ring will just go over the tube. If I need more grip, I will first try some painters tape wrapped around the tube. The motor can slide up and down on the legs to ajust the tension on the "drive belt". Hopefully it will work just fine.

The only thing needed to get this to work is that 3D-printed pulley which I will make tomorrow and then a little enclosure for the electronics.

P.S.
I may have mentioned this before, but the perfect solution to a lathe working as a rotisserie would be to change the BLDC headstock motor out for a beefy servo motor which would be able to run 1-3000RPM at full torque which means, I could easily go from rotisserie speeds to proper turning speeds just by turning the potentiometer for the RPMs. That was actually the plan but I spent a bit too much money lately on other gear so no servo motor for the lathe just yet. Also, with the ELS I can thread at pretty high speeds where the BLDC has plenty of torque which again reduces the need for a servo motor.

From few days ago I can not see images on this forum...?

 

[Diving Gecko]

From few days ago I can not see images on this forum...?

Me neither, I hope it gets fixed and that we don't lose the pics already uploaded...

 

[Diving Gecko]

Getting Ready To Blow Some Sh*t Up
It's been a long time coming and it will take a few more days, but I am getting close to blowing up some carbon fiber reservoirs (offcuts).

Here are the parts I had to make for the destructive tests ahead:

Two end caps, a connecting tube and then to hold it together, I am just using an old Salvimar Vuoto muzzle. And I plugged the tube with a dummy brass piston (and there's a vent hole in the middle of the tube, too:

Lots of o-ring grooves. A few of them because I had some chatter which messed up the surface and other times I am trying different o-ring thicknesses and/or compression ratios. Also, three grooves on the end caps to increase my chances that one of them would end up in a place on the tube with a smooth surface. But it turns out that on the end caps, the tubes sealed fine with just a single of my go-to 1.5mm o-rings with ~25% compression.

I have four offcuts I intend to blow up but since I have not had much luck lately with epoxy coating these tubes, I only tried my latest method on two of them - that method is using a thin layer of fiberglass to better hold the epoxy in place. And surprisingly, with just a fairly quick round of sanding, both tubes are leak-free:

I thought of adding a pressure gauge onto one of the endcaps but I opted against it. So, I will just have the one of the scuba tank adapter. That one will show a higher pressure than what's really in the reservoir as it "includes" the opening force of the inlet valve, too. But I will check how much that actually is and then we will just subtract it from the gauge reading to get the real burst pressure.

Also, I actually wanted to go on record already and put some numbers down for what I think these tubes will break at. But I honestly don't have much of an idea. It'd be nice to see a safety factor of 2 which would mean they have to hold up to around 60 bar... Let's see.
The whole idea is that we actually get some numbers - and then I can adjust the specs for the next few tubes I order if need be.

Even though I pretty much could blow these up tomorrow - I just need to borrow a scuba tank - I think I will coat the last two now that I know that the latest (and old) method works. That way, I can blow all four up in one session a few days from now.

Stay tuned, it'll be fun

 

[Diving Gecko]

BREAKING A GLUED-IN SPEARGUN REEL DOVETAIL CONNECTOR
So, years back I thought I was going to use this cheap Chinese reel on my big gun. But I am not sure whether I actually got around to using it or not. I feel like I didn't. The reels are cheap and with one modification they are actually not too bad (they need this mod otherwise they risk locking up). I have used one as a belt reel for a long while, though.

Here's the original post from 2020:

Glue!
Since I am pretty much treating this gun as a platform for various experiments, I thought I would try something that in theory might work but may very well not work in reality: Glueing a reel onto an airgun...

I am glueing the reel mount/adapter directly onto the CF reservoir and the reason why it could work is that high end epoxy adhesives like the 3M-DP460 I am using has strengths of up to 30MPa which means that in ideal conditions it can hold a load of app. 300kgf/cm² which is kinda nuts! But that's with a perfectly prepared surface and a strong material. On weaker materials, the substrate (the parts being glued) will fail before the glue itself. 3M says when glueing fiber reinforced plastic it fails around 70kgf/cm². I would think the surface of the adapter is around 4-6cm², so that would give me +280kgf of max theoretic hold if the surface prep is perfectly done.

Now, in reality perhaps an off-angle knock might undo the whole thing right away, but as said, it's an experiment

I'll let the pics do the talking from now on:

Many epoxies, this one included, likes a post cure at elavated temps so I left it in the tropical sun with the rest of the gun covered by some wet towel and clothing:

The reel is a super cheap Chinese offering. Not too shabby for the price once deburred.
"Pro tip": The "wheel" is a two-part thing which is a bad design as the line on it will often expand once it gets wet which will force the wheel apart leading it to easily jam. Glueing them together solves that - I have one I use as a belt reel, hence why I know.
And yes, I probably overloaded the reel and will cut off a bit of line after an outing or two:

I'll also tie a piece of "safety lanyard" from the reel to the trigger guard just so I don't loose the whole thing if the glue joint fails.
Lastly, my tinker OCD is really hurt by the gap between the guard and the reel - but I wanted to be able to pop the reel on and off without having to depressurize the gun and pull off the handle.

Anyways, I have since cut this long reservoir down to a size more needed for the island I am now on - and that left the glued-in dovetail on the cutoff. So, I thought I might as well test how much for it took to pull it off of the reservoir (before I blow the cutoff up in an upcoming test).
Also, someone back then asked if I had actually used this. Perhaps @marco15499, @Kodama, @landshark sa or someone else of the regulars from way back when.

I got to use my old test rig and setting it all up was pretty easy. I used one of my recently made endcaps for the "blowup test", an old very short Mirage barrel that a previous owner had cut down, a Predathor bulkhead for centering the barrel, a Vuoto nose cone and some dyneema:

I even made a video this time but in short, I ripped the dovetail off at 62/47kg depending on how generous you wanna be:

There was an earlier peak of 62kg and then the breakage seemed to be 47kg. I can see two explanations for the discrepancy: The scale doesn't seem to refresh/sample that often so we may have missed the peak at breakage or the 62kg peak did some early damage to the glue joint so that it failed at a lower force.

Anyways, more thoughts and findings in the video:


Also, strangely, dozens or hundreds of the older pics aren't showing anylonger in the earlier pages of this thread - but the pics in the old quoted post show when I am editing this post quoted post but not when actually published ... Weird, but it seems they are still on the server somewhere.

 

[Diving Gecko]

A New (Standalone) Rotisserie - Part 2
A few days back, I mentioned I only needed to make an electronics box and a pulley to finish this. Well, there are just not enough hours or days but I just got the pulley printed and took the rotisserie for a test spin.

I made two versions of the pulley. One with two smooth grooves and another with a single groove but with some ribs. Here they just finished printing:

Supports pulled off, excess resin cleaned off and now ready for post UV-curing:

They both warped a bit - but I also went much lighter on the supports than I normally do. Still, I think I stand by my conclusion that while resin printing has very high resolution, it's not easy to get accurate large parts as warping is a real issue. Maybe other resins do better in this regard though.

And here's the final pulley mounted on the little geared motor. The UV-curing yellows the resin a bit but it goes back to white after a little while - except when you handle the parts with dirty "workshop fingers"

And we even have a video for this little thing, too:

 

[Diving Gecko]

Epoxy Coating (the inside) of Carbon Fiber Reservoirs
This probably will sound like I am stuck in a loop but anyhow, this will be about my latest experiments leak-proofing carbon tubes.

I thought I would try using mylar on the inside of the tube, against the thin layer of fiberglass. The idea comes from the RC glider guys who use mylar over the outer skins of wings and it leaves a glass-like super smooth surface when done properly.
I actually had some mylar already from years back so I cut it up into strips:

The strips are rolled up and taped "shut" in preparation for inserting into the tube:

This picture is in the middle of the process but it shows the idea - and the potential issues. Where the mylar goes dark, resin is pressed against it. This is good, but we want that all over. The light pockets will end up as voids, or "valleys" in the resin.
My first attempt was pretty bad. Where it was good, it was perfect, but I had some pretty nasty voids. So, the next step was to try to force the mylar against the inner wall of the tube by using a balloon:

I think this idea has some merit. But one tip, you have to not have the mylar protrude past the edge of the tube. Otherwise the balloon will press on the extended mylar which will lift it off of the inside wall. So, the prep should look like this - before inserting the balloon:

It's really hard to photograph the resulting surface. But in this picture, everything not marked in red is actually a super even, incredibly smooth surface:

I think the different shades of carbon has to do with the sanding and how the fibers reflect the light. Trust me, the surface is like glass. Except here:

And you can see some small pin holes scattered around. Also, those voids - where air was trapped under the mylar - whilst big enough to cause a leak are way smaller than when I didn't use the balloon.

This is why I say the idea has merit but here's the thing. It turns out that my old technique of only using the 0.05mm layer of fiberglass is actually pretty good, too. It doesn't leave that great a surface so it does need sanding but even the mylar'ed surface needs sanding as even without voids, there's a bit of raised lip where the mylar strip overlaps itself.

I think fiberglass with a surface coating epoxy would be the best overall solution. These epoxies tend to have not only UV-blockers in them but also additives that reduce the resin's surface tension so it self-levels better and doesn't want to orange-peel as much as regular laminating resin often does.

I have some surface coat epoxy on the way but I am getting restless and really want to get in the water so I went ahead and coated the two long, leaking barrels with regular epoxy and fiberglass (the mylar tests were done on the cutoffs). So, I finally got to use the recently finished rotisserie:

(That beefy little lathe in the background is ~130kg of CNC-goodness but I wont have time to retrofit it until next year. But stay tuned for projects on shaft tail ends, vacuum muzzles, slip-tips, etc, etc).

Adding some o-rings at the ends of the tube is enough to make it stay put and not travel off of the rollers:

And then some post curing in nature's tropical oven:

I think I already mentioned that these two tubes are the worst I have ever had to deal with in this regard. I have had others that only needed one thin layer of epoxy and never "complained" since. I think this batch was a tad dry on the inner most layer and once you sand through unidirectional fibers or worse, pull some strands out of the cured laminate, you are in trouble. It could also have been made worse by the fact that the boss of the Evo handle is almost oversized so these inside tube walls has taken quite the abuse with all the (dis)assembly for this long-winded project.

Anyhow, I think I mentioned, that when I order my next tubes, I will ask for them to do one layer of fiberglass or wowen CF (not uni) on the inside and maybe keep it a bit more resin-rich if possible. I actually think it should be pretty easy to get tubes made that doesn't even need this extra level of sealing.

I really, really expect these tubes to not leak after this latest treatment, haha.
And then hopefully for the sake of everyone, I can shut up about coating tubes for a long while now

 

[Diving Gecko]

Still Leaking...! (But I Think I Know Why Now)
So, the one tube I sanded smooth is still leaking at the handle side which is really frustrating:

Previously, I have mentioned that this Mares handle is a bit oversized on the internal boss - the one that protrudes into the reservoir. So, I did something I should have done a long time ago and I measured it more carefully:

And yes, the boss itself is kinda oversized in some areas at around 38.11mm but it's also under in other parts at 37.85mm so it's basically not a proper circle.

What's worse is that the o-ring groove is even less concentric. It ranges from 35.46 down to 35.06mm. That's a change of 0.4mm and it happens within a very short distance - at the bottom near the trigger guard which is where the leak has been this whole time.

Admittedly, I am also at fault. I have inadvertently oversanded the tube a bit at the handle end. And I ovalized it, too. I don't have a proper internal micrometer (will get one at some point) so the ID measurement is a little bit less precise as I am using calipers but it looks like my tube ID ranges from 38.00mm to as much as 38.25mm. I think, but no excuse for this, it's because I used the handle to check that I had sanded enough and then I sanded a tad more for clearance as I had forgotten how crazy tight a fit the handle is on a stock alu reservoir. So, I ended up too large on the ID of my CF tube.

I did the o-ring compression math and with a stock alu reservoir with the ovality of the o-ring groove, we are ranging from 22% to 33% which is why we can keep a seal but with my CF reservoir we can get as low as 16% compression which is already lower than recommended for static seals. And then if you add tolerances on the o-rings where they can be a bit undersized in section size sometimes, we can easily end up even lower than 16%.
I think this is why we have a leak - In one particular area of the CF tube-handle-interface, I never had enough compression on the o-ring. The location of the "deepest" part of the o-ring groove resulting from the ovality also explains why it always leaked at the same spot on the handle.

One possible but slightly hacky solution is to step up one size in o-ring thickness but the ones I have on hand are actually too thick. The next option would be to try to coat the tube one more time(!) and then be very careful sanding to make sure I take off as little as possible. And also aim to get the tube walls to better match the ovality of the handle with some careful measurements as I do the sanding.
I think I will do this but combine it with trying to sand down the high spots on the handle boss, too This way, the handle will actually insert as it should without me having to oversand the tube to allow clearance.

The last and most complicated solution would be to CNC the boss and o-ring groove properly concentric (and then use a thicker o-ring) but that's risky. First of all, because of its shape it would be really hard to mount the handle in the CNC so it's perfectly straight but also holding it sturdily enough so that I don't get chatter could be a challenge, too. And if I get chatter on the o-ring groove surface then forget about sealing and the handle would be trash...

Now, I can't help wonder how you can get a 0.4mm fairly sudden dip on a 35mm circular feature from an CNC machined steel injection mold. Maybe the plastic distorts once it cools down.

Anyways, I should have measured this long time ago. I have been chasing my tail thinking it was the inner wall surface of the tubes that weren't good enough when the reason, most likely, was something completely different... Honestly, even though it's frustrating it's also nice to finally have an explanation.

 

[Diving Gecko]

100 bars and No Cracks aka Trying To Blow Up A Carbon Fiber Reservoir... Part 1
So, a while back I promised I would blow up some of the CF tube offcuts and I just tried, but didn't succeed.

I got to +100 bar but various seals started leaking but the tube didn't crack...!​

I set this up as simply as I could but even though I have a few different camera angles I will just share the handheld clips from the phone - it doesn't make sense to spend time to cut a proper multi-cam video when nothing exciting happened:



The Tentative Conclusion
Had you asked me before the test what I expected, I would have said, I'd almost be OK if we got above 60bar on this size of tube. And it seems we got to around double that and that we are looking at a safety factor of ~4 (if we say 30 bar is the use case). And that is actually really impressive to me.

Safety...
This may look dodgy and dangerous, but the reservoir is not only in water, it's also filled with water and I am hiding around the corner. That said, I didn't fill the high pressure hose with water which would have been even safer.

Next
I have tubes in two different wall thicknesses and the one in this video is 38x40mm - so the nominal, "classic" size and the same as an aluminium reservoir. The second size is a bit thicker and I have two of each. I tried both of the "classics" today but the other one leaked, too before I could blow it up.
I wasn't super disciplined about the size of the o-ring grooves on the test caps. I cut a few different depths for use with a few different sizes of rings since the ID of the offcuts aren't all the same - as I had to epoxy coat and sand them all. It's possible I can mix and match and get a better seal. Though there's not that much I can do about the inlet valve if that leaks. But I may try again in a day or two.

 

[Diving Gecko]

Trying To Blow Up A Carbon Fiber Reservoir... Part 2 (+160bar...!)
I did some more tests with some better matching o-rings and this time we got to +160bar before we had a "catastrophic failure" but yet again, it wasn't a cracked tube:


The two Delrin spacers behind the muzzle gave up and I think that allowed the front end cap ("nose cone") to move forward enough that the o-ring seal gave up. Also, the adapter for the high pressure hose popped out but not sure which one happened first. Either way, the CF reservoir itself survived which is pretty nuts.
[EDIT] I had a look at the slow-mo actioncam footage and it looks like the front end cap fails just before the inlet adapter pops out.

So, 160bar on a 1mm walled no-name T300'ish carbon fiber tube from China... That's a safety factor of more than 5 (30bar use case).

Here are some remnants of the Delrin spacer and even a bit of torn-off alu thread:

Not surprisingly, the threads on the muzzle started feeling a bit "worn". If my math is correct, the end caps were holding about 1850kgf which is more than your average family car weighs...

I then cut off a slice of the tube to do a "burn test":

If you take a torch to a CF laminate you can burn off all the resin and then dissect the CF layer by layer:

By doing this, you can often figure out the "lamination schedule" - how each layer is orientated and how many of each there are, etc. This one had a 50/50 distribution of unidirectional fiber layers in the 0/90 degree orientation - not surprising - and then a 3K twill on the outside.

With the test finally done, I can now order some more tubes and I will ask the vendor to add a layer of fiberglass or a 3K carbon fabric on the inside instead of the normal uni-directional. Hopefully, this will give me a smoother and tougher inner wall and completely eliminate the need for epoxy coating the inside.

 

[Diving Gecko]

A Small Xmas Project
Made a new inlet tool yesterday. Didn't take long as I didn't care too much about the surface finish - it just had to fit nicely.

There's more info in this post