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

[Diving Gecko]

The slider should be installed in the muzzle with a slight tension on the O-ring. This will be enough to ensure the spear exits with sufficient speed for hydrostabilization at the point where the spear's center of gravity leaves the muzzle. It's good that the slider is imparted with speed before the spear strikes. Recoil adjustment is also achieved by shifting the speargun's center of gravity. To do this, you need to create a weight for the receiver and experimentally move it from the handle to the muzzle in the water. A Cobra float on the receiver also helps. The Archimedes force also plays a role in the speargun's recoil when there is no reaction force between the muzzle and the spear!

Yes, I am planning to add a little bit of a belly (cobra/gecko/cuttlefish) in foam, sand it to shape and then cover it with a sleeve of carbon fiber. It's the same as what Carbozzi used to do/does. And then of course, I would need to add a bit of lead somewhere in the gun.
But ultimately the dream is still to make that shape as a one piece reservoir and not with foam - but at least, I have a firmer understanding of how it should aid in "calming" the gun down when aiming.

But all that will be for later - I am heading back home to Europe for a month or so.

 

[tromic]

But this is a vacuum muzzle
I was just wondering if it's friction or the tiny bit of water at the end that pushes the shaft out. But I don't think it's a big deal if the slider stays put as long as it seems to have done in that shot.
If the slider gets pushed out right away, I don't think the shaft will stay centered and the tail end will probably hit the edge of the shock absorber.

The slider should be installed in the muzzle with a slight tension on the O-ring. This will be enough to ensure the spear exits with sufficient speed for hydrostabilization at the point where the spear's center of gravity leaves the muzzle. It's good that the slider is imparted with speed before the spear strikes. Recoil adjustment is also achieved by shifting the speargun's center of gravity. To do this, you need to create a weight for the receiver and experimentally move it from the handle to the muzzle in the water. A Cobra float on the receiver also helps. The Archimedes force also plays a role in the speargun's recoil when there is no reaction force between the muzzle and the spear!

I would be interested to see the same shot of the slide separating if you were to load the gun above water so that there is no water in the head..
The cloud seen in the picture is quite large. It looks like it is made up of water and air. Some guns, such as the Taimen, have an O-ring in the mouth of the head that holds the slide in place better and longer. A similar effect is present in the Tomba kit

 

[Zahar]

But this is a vacuum muzzle
I was just wondering if it's friction or the tiny bit of water at the end that pushes the shaft out. But I don't think it's a big deal if the slider stays put as long as it seems to have done in that shot.
If the slider gets pushed out right away, I don't think the shaft will stay centered and the tail end will probably hit the edge of the shock absorber. A small amount of water enters the barrel during loading, and this is enough for the vacuum created in the barrel to form a mist of vapor along with a small amount of water. When the gun is fired, the vapor doesn't have time to condense back into a liquid state. This is why a gas cloud is visible when firing from the muzzle!

 

[Zahar]

So!

 

[Ultimatepro2021]

Thanks for the link, some really useful info there.

Although work commitments have been keeping me from spending much time with my 3d printer, I have done a few prototypes which I'm keen to take further.

Still stuffing around with speargun handles... I made this first version which fits my hand really well but feedback from other guys are the support flare is bit aggressive:

View attachment 55312

View attachment 55313

View attachment 55314

So I knocked up another design which is more generic:

View attachment 55316

View attachment 55315

I'm still not 100% happy with it so may end up with with something in between the two. In terms of strengthening the handle I initially used your method of sucking resin into the void of the handle under vacuum. I was unable to get it completely filled so I'm now sticking to creating two sides filling each with resin and then joining the two halves.

View attachment 55318

Also designed a roller head which I'm keen to retrofit on one of my broken wooden roller guns. Printed with wet filament so the quality isn't great but the proof of concept is positive so after refining the design a bit I'll move ahead with installing it. The idea is fit the head and then wrap everything in carbon fiber.

View attachment 55319

Hey,
I’m a little late to the conversation but I am working on making a speargun, would be be able to send the link to the roller barrel stl if you can?

 

[Diving Gecko]

Radio Silence - But Good Stuff Will Be Happening...!
So, life happened. I was back home in Europe on a long overdue trip but now back on the island. Also, I had - and will have to - devote a lot more time to the freediving product I am developing. And to make that easier, I have decided to put some effort and funds into an overall upgrade to the workshop and its capabilities. But the idea is that they will also benefit the speargun projects

May the Fourth Be With You
First of, this will be coming:

This is some early design work on a 4th axis rotary for the CNC. These are super frikkin expensive if you want them stiff, precise and repeatable enough to do proper work. Even on a small machine like mine.
The off-the-shelf one that comes closest is a MicroARC 4 from Tormach which sells for +USD3500...!

I have looked into what the MicroARC uses for parts and sourced mine secondhand. The real meat and bones is a Japanese harmonic drive (gear box) from Nidec Shimpo which is probably USD +1500 from new, but I sourced a used one for less than 100 bucks. I also went up a size so mine should be stiffer, unless it's completely worn out. (Supposedly cycloidal gear boxes are better than harmonic drives but also more costly).
But the MicroARC has a cast iron housing whereas mine will just have a 7075 to start with so I will loose stiffness there. Same with the adapter/face plate. But I just need to get started - the housing and face plate can be upgraded later on. Also, since I will just be machining small alu parts to start with I think it will work OK.
The rest of the parts on this expensive MicroARC are ridicously cheap (a super cheap stepper motor/driver and a cheap chuck). I will upgrade to a 400w Delta servo motor and also make sure to design the face plate so that it can accept a lot of accessories (regular chuck, self-centering CNC vise, ER collet chuck and tombstones). And yet, I think I can build the whole 4th axis setup for less than USD 350. So, 10 times cheaper... (Being able to source cheaply is the main reason, I have been able to slowly build my "machine park").

In the render above, the 4th axis is sketched out with a custom tombstone for the freediving product as that will really help speed up small production runs.
But where this 4th axis will really come in handy in terms of speargun parts is when being used as regular rotating axis when I will be making my own handles.
So imagine this picture below being me machining an aluminium handle:

It means work holding will become a whole lot easier. E.g. when I first thought about how to machine an alu handle, it looked like I had to do at least four different work holding setups, maybe even five. And some of them would be a bit janky. Same if I want to machine custom grips. With the 4th axis, I am down to two setups and a whole lot fewer tool changes, too. This means, I can machine the part a whole lot faster and with a lot less challenges and overall just have less frustrations. Which also means, I will be more inclined to iterate on the design and make more parts.

One last thing on the 4th axis, I made sure to get a "hollow bore gear box" so that I can pass stock through it. The motor will be off set, too. I did this so I can insert long tubes and e.g. machine the air transfer cut outs and transverse pin bores on custom shooting barrels. And since I am planning on going to 14/15x18mm stainless barrels, that would come in handy.

Time To Start Cooking...
Next up is something I have been pondering getting into for a few years by now...:

It's an anodizing setup. I am limited in size by the fact that I want it to be in my small bathroon in the workshop. Yet, I want it to be almost semi-professional. And... I am hoping I will be able to do Type III/hard coating anodization.

Hard coating is thicker, harder and more corrosion resistant than the regular anodization you most often see. It's however more costly and can't be dyed in fancy colors. But hard coating is ideal for marine parts so I don't get why we never really see it in speargun parts. (Well, I guess I answered the question already - no fancy colors and it's more expensive...). The aluminium allow I am using will supposedly turn a dark grey when hard coated, which is actually a color I was aiming for anyways. So, fingers crossed.

The tricky thing is that hard coating ano baths need to be kept at 0-5C whereas regular Type II just needs to be around 20C. So, I have gone into a very deep refrigeration rabbit hole and let me just say, it's not simple. E.g. the "industrial chiller" I sourced may not like getting down that cold whereas it should happily chooch along at 20C. But for 5C, I may need a whole other compressor and coil diameters to match, etc, etc. Anyways, I will learn as I go.

As for the rest of the semi-pro setup, in case anyone is curious:

• I will have eductors circulating the ano bath, maybe in the other tanks, too. Eductors are small "jet nozzles" that pull in the surrounding liquid in order to mix the chemicals and keep the liquid moving around. If this is not enough, I will add aeration with air pumps though I am hoping I can get by with eductors only as they produce less bubbles and fumes and bubbles
• Heating in all baths that need it, active chilling in the ano bath
• Temperature controllers in all baths that need it (secondhand industrial temp controllers run through solid state relays)
• Titanium cooling coil in the ano bath (I may later run refrigerant straight into a titanium coil in an external flow tank and circulate the acid through there to keep the coil out of the bath itself)
• A powerful 120V/30A power supply for hard coating (and the odd titanium ano job)
• That said, I will likely start with LCD type currents (Low Density Current) to have less heat going into the bath, less fumes, less need for agitation and perhaps ramp up towards more industry level currents as I get to know the setup
• Insulated tanks (ice chests as you'd see on a camping trip) - these are also benefitial as I have limited amperage in the workshop and dont want to trip breakers in the middle of a run
• 1060 "pure" alu cathodes, NO lead cathode in the tank and no alu alloy containing metals that can contaminate the ano bath
• pH meter to keep a check on the dyes (for Type II ano)
• No tap water, ever. Will be using distilled drinking water in big bottles to start with (my tanks are small) and rinse often and carefully
• Also, I will have a titration setup so I can keep an eye on the dissolved aluminium levels and "free acid" in the ano bath

This could defintely be considered overkill but I know myself well enough that I need to plan this carefully and aim a bit high. I am a tad messy by nature and that's not cool with chemicals so having a dedicated setup with proper tanks should help in that regard. Also, I have been able to source a lot of this cheap or secondhand so I can actually aim for a nice setup with very little investment. E.g. a chiller, proper acid-proof pumps, titanium cooling coils, pH meter, etc.
Furthermore, it will be used for a product that I am selling so I need the consistency plus it will pay for itself fairly fast.

CNC Electronics
Oh... Also, I will finally fully retrofit the electrical and control side of the CNC to run LinuxCNC. It will be super daunting but in the end should be worth it as I am really running into the limitations of the standalone, cheap, cannibalized Chinese controller now. In fact, it wouldn't even be able to run the 4th axis but there are many other reasons to make the change.
LinuxCNC is the opposite of limited, it's an open playground that can rival industrial controllers in its features but that's also part of why it's a scary undertaking - there are just a gazillion things that can be customized. But they have a nice, helpful forum and I may have to lean on some AI, too to get me up and running.

Fixture Plate
I'd love to have a fixture plate with proper locating features on the CNC. Especially since I will be taking the 4th axis on and off and also, for the freediving product I will have quite a few fixtures mounted on the table. Would be really nice to have a super repeatable way to put them back on when needed. The problem is that to make one that takes full advantage of the size of my table, I would need an even bigger CNC. Whilst the stock plate material itself is actually affordable, I am not sure out sourcing the machining would be. Will have to see if I can pull a favor somewhere in China for this one...

Lathes...
My mini lathe also needs a final push to be fully done - just a new controller box for the ELS and tidying up the wiring now that the 1st janky version has proven itself worthy and stable.
And then sometime in the future, I really need to do a restoration on that little mini lathe-sized CNC lathe I have sitting around. I think I have most of the parts for that project, too. (It will also end up running on LinuxCNC). What's keeping me back is that I'd love to have it somewhat enclosed so I can run flood coolant on it as I hope to be making titanium and steel parts on it. So, will have to design and build a proper stand for it that I can enclose and drain.

So, yeah, lots of stuff to do in the next few months.

 

[FlotsamBob]

Following your work. I am funnily enough in the process of brainstorming a decent spec makerspace with a few printers and CNC machines... Big fan of budget solutions!

 

[Diving Gecko]

A New Nose Cone - Part 1
Since the CNC will be out of commision for a while - whilst I redo the electronics and controller - I have to hurry up and machine few things I might need.
One is a new nose cone. The previous one has been pretty flawless, just that I made the pumping barrel bore such that it will only ever accept a 7mm shaft. But I want to play around with 7.5-8mm shafts now.

The nose cone looks almost the same, only one addition to the front:

It's a small pocket with a threaded M2.5 hole. I plan on making a separate "line hook" and the little pocket should help locate it so it doesn't twist from side to side. It reduced the weight-saving cutout a little bit - maybe I gain 1-2g but it's ok as the cutouts on the rear are much deeper now and some walls have been thinned, too so overall this version is almost 10g or 25% lighter than the previous one:

These cutouts are 15mm deep and since they have tight corners, they call for small endmills. So, I will be using a 2.5mm cutter with a very long reach.

But let's look at some of the machining.
In the first picture, I am pretty far along already. I have machined the outer contours, the bores and halfway done on the side cutouts. These weight saving pockets are also tapered to try to give the thin, long endmills a chance of surviving. I used two different sizes of tools and the left pocket has been finished with the smaller endmill already with tighter stepdowns to smoothe out the stair stepping and get a bit further into the corners - the right pocket is still awaiting that.

I think both @Zahar and @tromic recommended I try more generous lead-in angles on the chamfers where o-rings need to pass through and finally, here we go - a 20d chamfer tool (40d included):

Last time, I did this with a tiny ball end mill but it's a whole lot faster using a dedicated tool.

Overall, I see a little bit of chatter all over the part today. I expected it on the inner walls of the cutouts because of the long, long thin tools but even the outside of the whole part which was done with a much stiffer 8mm tool has some "texture". Maybe the bearings in the old spindle that came with the machine are finally going - I meant to change it out long time ago but there will be another upgrade on that front soon enough.
Anyhow, I think it's mostly cosmetic. As long as the faces where the o-rings seal are OK, it's not an issue.

I checked the shooting barrel bore with a gauge pin:

To get it to slide in, I had to redo the bore - it was too tight so perhaps the tool was undersized or was deflecting.

But here's a good example of one of the many reasons, I want to upgrade my CNC controller: As it is now, to make that bore slightly larger I have to go back to the laptop and redo the CAM, export the G-code, load it on a USB stick, then go to the machine and load it onto the controller and then - finally - run the program. All that for a cut that takes mere seconds... It gets frustrating pretty fast. But once I get my controller upgrade to LinuxCNC done, I will be able to do this on the controller itself (e.g. by using tool wear compensation). So, with a few button presses and without ever leaving the controller, I can creep up to size on important features like these.

Also, these gauge pins are cheap and super handy to use, so I think I will order some more. I will go up in size a bit to allow for a slightly larger gap plus once I start anodizing, the bores will shrink a little bit, too. About 0.05mm on the diameter for hard coating but a little less for normal anodizing according to my research. So, if I cut a bore to be 18.15mm and hard anodize the part, the bore should end up being ~18.1mm. One could argue that you should just allow a bit more of a generous gap between the parts but when using small section o-rings under pressure, you risk extruding the o-ring into the gap which is why I aim for pretty small gaps.

Next up, I cut the o-ring grooves with another looong neck tool taking tiny cuts to limit deflection and chatter:

This op took close to 10 mins for just the two o-rings grooves! It would be so much faster and easier on the lathe. That said, had the grooves not been sitting so far from the top face, I could have used a much shorter, stouter tool and pushed it way harder. Or spent a whole lot more money buying a T-slot cutter with a larger diameter shank which deflect less. But the grooves ended up in this position as a result of me wanting to get more weight out of the part - the further forward they sit, the deeper I can make the rear weight saving pockets.

I also ran a double edge chamfer tool on all four edges of the two o-ring grooves - the cool thing about this tool is it can reach the underside edges, too:

So far, work holding has been really easy on this part. I just stuck a piece of round stock into a v-block in the vise. But for the next operation when the part gets flipped, I will have a very hard time holding and indicating it. The solution I came up with is to use this simple jig:

Besides the generous amount of Loctite, which will mostly be squeezed out as the fit is pretty good, it has locking screws on the sides - but I also have a small brass insert that the screw is pressing on. So, the softer brass and not the hard steel screw is what jams onto the alu nose cone:

I forgot to snap a pic, but the last step here was to indicate the jig which was pretty easy. I just ran a dial indicator back and forth along the back face of the jig and tapped it in place until the indicator ran true.
The idea is that tomorrow, I can flip the assembly, put it in the vise and it should line up pretty nicely.
There are other ways to do this, of course, but it worked well the last time I did it so hopefully it will again.

But another way to do it - after the controller upgrade is done - would be to use a 3D probe to locate the two bores in relation to each other after the part has been flipped (and the stock that used to be held in the V-block has been removed to liberate those two bores). Even if the part ends up being rotated a bit, the controller will know that and implement an angular offset for all the remaining code.

The front machining should be easier as all the important features were done from the backside and extend through to the front. Also the cutouts on the front are much easier to machine as they are pretty shallow so I wont need those extra long, chattery endmills.
Worst case if I am little bit off on the indexing of the jig, the chamfers on the front will just not be even. But tomorrow will tell

 

[Diving Gecko]

Following your work. I am funnily enough in the process of brainstorming a decent spec makerspace with a few printers and CNC machines... Big fan of budget solutions!

Sweet!
We actually do have a makerspace in the nearest provincial capital, just 30mins from me. They have a full sheet CNC - great for plywood but sadly it's been out of operation for a while. No machine tools for metal but a good bunch of printers.
They also have an older 3D scanner with a turn table that I used to scan a few of my custom grips. Unfortunately scan files are still super tricky to work with in Fusion but the idea is to adapt my favorite custom grip to my alu handle design and then machine a grip out of hardwood - when I get that far...
And then as I hopefully migrate all my guns to my own handles, I can have the exact same CNC'ed - or printed - grip on each of my guns.

 

[Diving Gecko]

A New Nose Cone - Part 2
Let's start with the finished product:

It turned out rather nice if I was to say it myself and as the engraving on the bottom alludes to, it will accept shafts up to 8mm - whereas the previous nose cone could only go up to 7mm.

There are a few machining blemishes but for a small semi-pro machine that may have dying bearings and/or be out of tram, it's alright. I think the important "sealing faces" are smooth enough.
Cosmetically, there's a little stair step halfway down in the pockets but it might be a CAM error, so a me-error, more than the machine itself.
I did manage to break one small 1.5mm doing the tightly radiused insider corners, though. It was a 2 dollar tool, so it's ok. But... the Chinese end mills have more than almost trippled in price lately. Not so much an issue for the small ones, but for the bigger or more special ones, it adds up. I did stock up quite a lot before that happened though, so I should be OK for a while.

On the plus side, I am quite happy with how smooth the line guide fillet ended up. It was rounded nicely with a 2mm ball end mill:

The engravings aren't too bad, either but perhaps deeper than they need to be. They are made with a 0.5mm ball endmill - that's 0.25mm on the radius, so rather small
But I definitely need some new chamfer tools. I think the one I just must me chipped judging from the lines on those chamfered faces.

Also, I did a quick test trying to insert a small piece of a shooting barrel with an o-ring on it and even with a ~27% compression and no grease it was really easy to pop into the bore.
I think that 20d bevel really, really helps. Also, it's so wide it ends up wider than the o-ring sticks out, which probably helps even more. In this next pic, the barrel hasn't even been pushed in yet - the o-rings just sits nicely and fully cupped by the chamfer:

The bore for the pumping barrel sits really close to the shooting barrel and where the edges of the two chamfers meet, there's only 0.175mm:

But seeing how the chamfer for the shooting barrel is wider than it needs to be, I think I could have reduced it a bit and made the one for the pumping barrel a bit wider instead. I will make a note of that for the next nose cone.

As for the weight, I did end up managing to reduce it by almost exactly 25% or 10g. The previous version was 40.4g and this one is 30.44g:

As a reference, the original nylon(?) Mirage nose cone is ~54g:

(Also, the little endmill I snapped in the background).
But the Mirage nose cone has some displacement as it protrudes forward of the reservoir which mine hardly doesn't. So, you could argue that it "weighs" less in the water. But on the other hand, my shorter nose cone allowed me to make the reservoir itself longer, so in the end, I guess the alu nose cone does end up lighter.
That said, it doesn't take long before a 30g nose cone starts feeling "heavy" in the hand and I already want to shave more weight off, haha.

OK, let's move on to a few machining pics:
After I flipped the part, the left over stock that was used to hold the part in the V-blocks in the first op needs to be removed:

It's about 35mm tall and honestly, it's a total waste just machining it into chips. Had my mini lathe been up and running or if I had a bandsaw, I could have parted it off and saved the offcut for something else. But I had to just let an end mill do its work. Well, I guess 10 mins with a hacksaw would have done the job, too.

Here's a nice moment in the machining - when the bores are "liberated" as I called it yesterday:

I had made sure I cut the bores deeper than needed in the first operation so I wouldn't have to do anything to them from this side apart from chamfering them.
Also, the heat from the machining, the alcohol in the mist coolant or both softened the Loctite but the part still stayed in place.

Since I had to drill out the bores for the two threaded M3 holes that I use to help pull the nose cone out during disassembly, I went ahead and Swiss cheesed the weight saving pockets, too. Drills are good at removing material fast so it helped the later machining of the pockets:

In this last pic, I am test fitting those M3 threads:

The thread milling left a small burr - I will try to remember to run the chamfer tool after the thread mill in the future. And I have said this before but thread milling is such a cool thing. I have thread milled anything from tiny M1.6 to M8 by now with zero drama and not a single broken tool and all the threads have come out great.
I cut the threads a bit bigger than last time in case I end up anodizing this part in the future. It's an easy adjustment to make in the CAM, just one value you need to change.

After this pic, I only had to thread mill the smaller M2.5 in the little pocket for the line hook, round the edge of the line guide hole and do some engraving and then the part was done. Well, I had to put the jig in a vise, take a heat gun to it and then gently tap the part out once the Loctite had softened.

Tomorrow I will pick up some stainless tubes I ordered a while back. They are supposed to be really smooth on the inside, so fingers crossed as the idea is to use them for shooting barrels. But it will make the gun a sinker once again. So, if I like the stainless barrels, I will borrow a trick from Carbozzie and add a small foam belly to the reservoir and wrap it in a carbon fiber sleeve. But that's in the future... I can test it as a sinker first.

 

[Diving Gecko]

Barrels - Chasing My Tail

These are 316L BA tubes in 14x18mm and 15x18mm:

BA stands for bright annealed and though it took a while for these to arrive I seem to recall that the vendor said they would be as smooth on the inside as they are on the outside and surprise, surprise - they are not.
Taking pics of reflective surfaces is difficult, sometimes it looks better or worse than real life, but let's have a look:

At least they measure out alright, the ID14mm tube is ~14.02mm.

I did a quick test with my weird grinding/polishing thingey:

And there may be hope - the next pic is hard to decipher, though:

It did polish it up some, but also left some white'ish "pock marks"? I don't think they are deep, but they do stand out visually. Not sure what that's about but perhaps they will go away with subsequent polishing. I have some cotton "bits" that can be loaded up with diamond paste so will probably try that.

At the very least, they are better than the titanium tubes from a while back. Also, there's still one level up in terms of surface finish - you can get electropolished tubes, too. But last I looked, I don't think I found a suitable size, but perhaps I will look again. Or try a different vendor of BA tubes.
That said, supposedly a super mirror finish is not ideal. Something to do with having just a tiny bit of texture holding onto the oil film being preferable.

[EDIT] I looked into the surface finish question a bit more and for dynamic o-ring use, something like ~0.4µm Ra is the ideal target for a piston type application (not an engine piston!). Supposedly, you don't actually want a mirror finish which is what an electropolished tube would get me.
I have no way of measuring surface finish but supposedly, I can get to that kind of finish with some 800-1200 grit hones followed by 3-6um diamond paste.

 

[Zahar]

A honed surface works well in cylinders with metal sealing rings, like those found in internal combustion engines! When working with O-ring seals, a honed surface will cause rapid wear of the O-rings when they break free from static friction. A piston in the muzzle—static friction until the gun is loaded. A piston in the sear—static friction until the shot is fired! Rubber sticks to the roughness of the barrel, and during the initial stroke, rubber particles are torn off the O-ring. A mirror-like surface of the barrel is necessary for the piston O-ring to work properly!

 

[Diving Gecko]

A honed surface works well in cylinders with metal sealing rings, like those found in internal combustion engines! When working with O-ring seals, a honed surface will cause rapid wear of the O-rings when they break free from static friction. A piston in the muzzle—static friction until the gun is loaded. A piston in the sear—static friction until the shot is fired! Rubber sticks to the roughness of the barrel, and during the initial stroke, rubber particles are torn off the O-ring. A mirror-like surface of the barrel is necessary for the piston O-ring to work properly!

Perhaps when I talked about honing you expected something rougher. I think when I said "piston type application" - by which I meant an axial movement in a bore - you might have read it as "engine piston".
It wont be anything like as rough as an engine cylinder nor was it ever intended to be.

Now, as to needing a mirror-like surface, you might be right but Parker o-rings disagree so here follows Parker's long standing recommendation:

They state 16µin RMS which I think is a tad under 0.4µm RA. A true mirror finish is said to start around 0.1µm RA and go lower.

But 0.4µm RA is the number that pops up a lot when I researched dynamic reciprocating seals. It is also mentioned in Aerospace Standard AS4716C which sounds rather proper. (And it's also the value I mentioned I would aim for in my previous earlier post).
Many of these sources state that a true mirror finish wears out seals, too as it wont hold the oil film. The seal will wipe the oil away and the o-ring ends up running over bare metal rather than oil. Granted, a very smooth metal but not as slippery as an ever so slightly "rougher" surface which holds onto the lubrication.

That said, I kept looking and then I found this:

They reference the "older" industry experience of sticking to around ~0.4µm RA but say that you can go lower:

Their recommendation is to go 2-4 times as fine and get very close to a mirror finish or just a bit above it.

Now, where it gets really confusing is trying to relate Ra to e.g. sandpaper or grinding grit - it just comes out way too rough. But at least the equavalent to 0.4µm RA in diamond paste is about a 3000 mesh and I know from experience that's getting really smooth, close to a mirror - just a bit of a dull mirror

So, to bring this post full circle . Like I said earlier, I will be polishing up to at least 3000 mesh diamond paste, perhaps 6000.
In fact 3000 - 8000 mesh diamond paste is considered a finishing grade which is supposed to leave a satin or cloudy surface - just short of a true mirror finish. This does does sound in line with what the literature says; stay just above mirror but make sure you don't go too rough either.

Another way to go about this would be to source an electropolished (EP) barrel which should be mirror smooth - and then according to the above, dull it just a little bit with some very fine diamond paste. The issue is, the EP tubes are 2.5 times more expensive than the BA tubes, I already received - so, I will give the polishing a go first and see how it goes.

Now, besides surface finish, reducing the compression on the piston o-ring is supposedly can reduce the friction a whole lot, too. In most of my other parts, I tend to go on the higher end of the allowable compression but for pistons, it could pay off to be just skirting the limit of minimal compression, just before a leak. I know UBL likes to do it this way for the highest performance but of course mass produced pistons are playing it safer with higher sqeeze on the ring. Same with the number of o-rings. So, a piston with only one lightly compressed o-ring should be the best performance wise - but also the least reliable.

I guess it comes down to personal preference. In the seach for optimum performance, are you ok with possibly having to change out the o-ring on the piston, perhaps even whilst on a trip - or would you rather never really have to worry about it.
Given that the Mirage is one of the few guns with plenty of power on tap, perhaps one could trade a bit of it for more compression of the piston o-ring(s) and more peace of mind.