|
 And this came! Now to scan
|
#
?
Mar 6, 2021 18:56
|
|
|
|
| # ? May 1, 2024 19:11 |
|
|
If I were being careful, I'd incorporate a few different scans to make sure I got al the features so I could reverse engineer properly. Instead I got impatient. Again sort of a rough scan but I got the inside and outside surfaces  And there's the moneyshot!
|
|
#
?
Mar 7, 2021 06:14
|
|
|
I might have missed it but what are you using to scan those in?
|
|
#
?
Mar 7, 2021 12:42
|
|
|
NitroSpazzz posted:I might have missed it but what are you using to scan those in? I'm using an Asus Xtion which is one of the Xbox 360 Kinect style sensors, and a software package called Recfusion. There may be better sensors now (Recfusion supports a lot of the Intel Realsense sensors) but why not work with what I have?
|
|
#
?
Mar 7, 2021 16:04
|
|
|
How accurate do you think it is or could be with enough fiddling?
|
|
#
?
Mar 7, 2021 19:07
|
|
|
honda whisperer posted:How accurate do you think it is or could be with enough fiddling? Well, that's a really good question. Most honest answer is I have no idea yet. I did a very basic tape measure sanity check and it seemed dead on but, well, tape measure on one dimension. I'm going to try some dimensional test stuff this week I think; getting potentially usable results from this setup is pretty new to me and I just wanted to start off by trying some cooler stuff because OMG it worked and it's cool. 
|
|
#
?
Mar 7, 2021 19:30
|
|
|
Cool, keep us posted. Even if it's fairly rough it seems it would be really helpful for say engine/trans packaging in the chassis.
|
|
#
?
Mar 7, 2021 20:36
|
|
|
Yeah, this is exactly what I need to figure out what kind of room I actually have for my turbo manifold. This is super cool.
|
|
#
?
Mar 7, 2021 21:03
|
|
|
 Not an in-depth examination of the question but I mean, the data's all there. 
|
|
#
?
Mar 7, 2021 22:41
|
|
|
Are you going to go try and fit splines over all of those points? I use Solidworks daily and that picture made me shudder.
|
|
#
?
Mar 8, 2021 15:10
|
|
|
Rectal Placenta posted:Are you going to go try and fit splines over all of those points? I use Solidworks daily and that picture made me shudder. Oh heck no. So once I get a good clean scan, my intent is to use the splined 3d scan automatically made at that section (what's in those images) as a guide for a manually created 2d sketch; same sort of way that you'd use a 2d scanned image as a guide. If I had a good point cloud of all sides, I could do sections at various locations to repeat that... like... the section I created is, once turned into a good 2d sketch, revolved, another one perpendicular to that could give guide curves for the spokes and stuff, etc. Or, honestly, looking at that nasty spline and seeing it doesn't interfere with stuff honestly gets me the actual answer I need.
|
|
#
?
Mar 8, 2021 15:55
|
|
|
Perfect is the enemy of good enough, my constant nemesis, and I suspect yours too. I think your "get a 2d spline and spin it around to check for contact" plan is good enough.
|
|
#
?
Mar 8, 2021 18:31
|
|
|
Oh yeah - for this yeah, what I have looks like it fit my needs already. But there's some other stuff I want to do more complete reverse engineering on to produce patterns for cast parts.
|
|
#
?
Mar 8, 2021 19:53
|
|
|
 La de dah, what have we here?
|
|
#
?
Mar 21, 2021 04:38
|
|
|
Brap? Looks like brap to me.
|
|
#
?
Mar 21, 2021 04:39
|
|
|
I mean yes it'll brap but will it brap enough? Mazdatrix Turbo 2 half bridge pattern template  Scribed onto the RB j-bridge template; notice the later closing. I figure the half bridge template is a bit more radical on the closing since it's the EFI irons where the primary (center) ports are smaller... so I'm pretty sure 12A irons can go that big if you're careful, ergo, gently caress it, moar brap.
|
|
#
?
Mar 21, 2021 23:13
|
|
|
mekilljoydammit posted:I mean yes it'll brap but will it brap enough?  The answer is always more brap.
|
|
#
?
Mar 22, 2021 03:31
|
|
|
The big question will be "can I go that big without hitting water or breaking side seals" I'm sure it'll be fine.
|
|
#
?
Mar 22, 2021 04:05
|
|
|
mekilljoydammit posted:I'm sure it'll be fine.  Sorry.
|
|
#
?
Mar 22, 2021 12:29
|
|
|
mekilljoydammit posted:The big question will be "can I go that big without hitting water or breaking side seals" I'm going 4mm taller than the pineapple template just fine.
|
|
#
?
Mar 22, 2021 20:17
|
|
Have you seen my apex seals? I seem to have lost them.
|
Aeka 2.0 posted:I'm going 4mm taller than the pineapple template just fine. I'm using 12A irons though.  To the template worked; not a clean job, just doing a lot of quick metal removal to see if I hit water. Will have to remember to bring a degree wheel and see where that closing edge is now in degrees; I think it's probably enough.
|
|
#
?
Mar 22, 2021 22:36
|
|
|
I wish i knew how different 12a irons are. Looks insane from that pic though.
|
|
#
?
Mar 23, 2021 01:24
|
|
|
One of the 12A variants has the biggest port runners on the primaries of anything; no injectors among other things. S4 and S5 T2 end irons actually have the most available meat; IIRC the basic casting was shared with the NA irons with the 6 ports. Even not opened up, I think the secondary runners are quite a bit bigger. I'll take a turbo iron over to the shop tomorrow for a comparison pic. In a world where I was starting from scratch I'd go T2 irons, mill o-ring grooves into a 12A center iron and use T2 housings. But in the world I'm in, I have at least 3 engines worth of 12A irons and GSL-SE rotor housings.
|
|
#
?
Mar 23, 2021 01:39
|
|
|
Port talk! J-bridge roughly carved onto 12A iron; you can kinda see where the main port is kinda feathered in and see an idea where the water jacket is on the other side. More iron might be able to come out, I'll check, but you get the idea.  Bridge template I used for the main port on the J-bridge 12A stuff laid against a streetported T2 iron I got a hold of that was damaged. You can see this has already been ported about to that template, but can kind of get an idea how the port is a lot more vertical near the closing edge.  You can see how from that closing edge you can go basically straight down without risk of hitting water.  From the manifold side - the 12A is on the right and has been lightly worked on from this view - you can see how much bigger the runner is on the 13B T2 iron, though I think there's meat enough to get there.  Template against dead 12A center iron, which actually close about the same place as 12A end irons. You can get an idea how much bigger the port is going to be, timing-wise. I think for what I'm doing 12A irons will be fine, and god knows I have enough of them.
|
|
#
?
Mar 24, 2021 23:40
|
|
|
OK. I figured out how to get a few things figured out which means once I put some time in I can get my bodywork designed up in CAD. Sitting and chewing it over, I have a hypothesis that a CNC'd female mold would be easier to work with than "make plug, pull female molds" especially since I can engineer in parting features in the mold. There's a few different ways to do it, so I need a test piece to figure out what works. Enter test piece. Not that I need NACA ducts desperately or anything but it's cool shaped, kinda complicated but not too bad, needs a 2 part mold, and not too big.  Female cavity generated in CAD  Sliced into 2 inch sections, with alignment features for rods and separated where it'll have to in order for the mold to come apart. First hypothesis on construction is "CNC rout out of pink styrofoam"; so cut slices on the CNC router, glue together, and gelcoat the tooling surfaces. Maybe lay up glass reinforcement too; the whole point of this is figuring out what's necessary. Also to see if the whole mess will stand up to 250F where the prepreg cures at.
|
|
#
?
Apr 1, 2021 19:33
|
|
|
Your approach should work, I've done similar before. You might want something other than styrofoam for the plug due to temps, but there are tooling foams that might be a little more costly but stand up to multiple parts. If you're doing it in slices I'd recommend adding locating features for the slices and leaving ~0.010" for hand finishing for good transitions before you gel coat. The biggest potential issue is probably going to be mold life for multiple parts and getting good consolidation for multiple plies - vac bagging that shape is gonna be a pain and doing bladder molding can be a headache to get good pressure.
|
|
#
?
Apr 1, 2021 21:45
|
|
|
Those holes towards the bottom of the sliced mold are intended to have piece of conduit or rod or something slipped into them for locating. One of the other material hypotheses is gelcoated MDF but for something the size of a car fender that's going to end up heavier and costlier than styrofoam.
|
|
#
?
Apr 1, 2021 21:59
|
|
|
I think insulation/styrofoam is gonna max out around 175, basic tooling foam (high density machining foam, renshape, etc) will get you to 250 I think without needing to go to high temp options. You can also do a positive master and cast plaster molds for each part if you don't want to make a glass negative, but that's gonna be way more work than your current plan, it just gets you around the temp requirements. Other option is wet layup but that's its own beast and I've only done pre-preg or resin transfer moulding so no experience there.
|
|
#
?
Apr 1, 2021 23:03
|
|
|
There's a lot of specialized stuff that would be appropriate for this that wouldn't be relevant to stuff sized for body panels - I've done wet layup stuff but haven't worked with prepreg so wanted a place to learn and try some stuff. If I can't get a good vacuum bag seal on this part, I might just leave off the back half of the mold for proof of concept. It's just more interesting than "oh, here's a carbon fiber square" or whatever. Doing a bit of poking around, I'm finding surplus renshape board cut to size advertised at $0.20/in3 which is ... kind of a lot. Making "just use a shitload of MDF" seem pretty attractive.
|
|
#
?
Apr 1, 2021 23:31
|
|
|
Are you planning on autoclaving these parts or STP cure? Prepreg or wet layup?
|
|
#
?
Apr 1, 2021 23:37
|
|
|
I found some relatively low-cost prepreg that says it's compatible with a vacuum bag cure - obviously still at elevated temperature but without an autoclave. For that matter I've heard from a constructor of hillclimb cars in the UK that even structural stuff can be done with vacuum bag tooling. What I'd like the ability to do is go from CAD to female patterns that are usable for prepreg; slicing the mold like this is one thought on how to do it that I want to test. Plan D (or so; this may change if I get more ideas) is "just use wet layup on male plug" which I know is workable, just has extra steps that I'd like to see if I can avoid. Plan D+1 is "do wet layup in that" which I 100% know is doable, but I want to find out if prepreg stuff is as less-awful as it seems.
|
|
#
?
Apr 2, 2021 00:01
|
|
|
Yeah I feel ya on the not doing a flat square and mdf is probably going to be your cheapest option tbh. Getting a good vac seal is totally doable, I think your main challenge is going to be patterns for your vac bags so you don't get wrinkling or bridging, you're gonna have to do some weird cone sorta jawn but 100% doable - paper pattern templates will be your friend here.
|
|
#
?
Apr 2, 2021 00:04
|
|
|
Yeah; that's kind of what I'm looking forwards to also. Like, the whole thing is a capability-stretch for me without being huge... I think most of the actual parts I'll be doing will be simpler to bag up. Another thing I'm thinking is a big positive with MDF is it just seems a lot easier to make work for bolting together mold sections accurately and repeatably... and it's not like I'm going to be doing production-scale work (I'm not planning on it anyway) but it would be nice to have stuff where the molds hold up. Further funny realization - CNC drag knives aren't terribly hard to setup, it's not impossible I could get to a point where I can scan paper templates and maintain a database of cutouts to make various parts. "Oh, I need another fender, better get the pattern cut out..."
|
|
#
?
Apr 2, 2021 00:14
|
|
|
Prepreg is a LOT less miserable. We used mostly prepreg at TF when I was there, with a few assemblies done wet where we had to for various reasons (complicated, extremely thick, or a combination of materials that we didn't have on hand in prepreg). As I recall we did oven baked vac bag for most parts. Long as you get your material well laid and pushed into the corners to avoid air pockets and bridging, it works alright. Especially since it sounds like most of your parts aren't going to be structural.
|
|
#
?
Apr 2, 2021 00:21
|
|
|
I know we used pink foam for toolpath testing, but I can't remember if we used it for any actual baked parts. We mostly used MDF molds coated with epoxy to reduce vacuum loss, with some more critical parts done in full CF molds where they were large enough that MDF wouldn't work for dimensional stability or vacuum loss reasons.
|
|
#
?
Apr 2, 2021 00:25
|
|
|
You're definitely going to want to keep either a physical or digital archive of patterns as you go along, you can even do pattern development digitally for tweaking as you get down the line and you start making smaller changes. Even without the drag knife printing templates and spray mounting to light cardstock before cutting with a razor knife works pretty good. No idea what your experience level is with this stuff, but vacuum sealing putty (we called it monkey poo poo) is going to be your best friend for weird vacuum bag shapes and an ultrasonic leak detector can save you endless frustration trying to get it to loving seal. And don't be afraid to give in and just bag your whole tool in an envelope bag if you're getting vac loss through the mold or it won't seal to the mold surface. I don't know poo poo about cars but I've been enjoying the thread so it's cool to be able to give some (hopefully helpful) input.
|
|
#
?
Apr 2, 2021 00:39
|
|
|
I've done some not-too-complicated wet layup stuff, my dad has done a bunch of wet layup vacuum bagged stuff. I know that I'm probably going to ruin a bunch of stuff learning even if it's theoretically right, but I've done a bunch of reading on the basics so hopefully I'm not too far out in left field theoretically. I definitely, 100% appreciate input from people who have more experience with this stuff or indeed anything I'm touching on; I know a bit about a lot of stuff, so every partially-baked idea that someone more knowledgeable can sanity check is a bit I don't have to learn by trial and error.
|
|
#
?
Apr 2, 2021 00:48
|
|
|
This is all awesome work man. I did some experimenting with Kinect based 3d scanning back in 2015, without much success. I was using Meshlab and ReconstructMe to try and get things into CAD. I managed to get a few things in a form where I could use them for mock up (See M52b28 below). I'll have to check out the software you mentioned and see how much things have improved the last 5 years. 
|
|
#
?
Apr 3, 2021 02:40
|
|
|
So what's working best for me is export from recfusion to a point cloud instead of mesh - solidwork's "import scan" functionality seems to work better with point clouds than polygons. I got distracted by other stuff and haven't tried for a "this is a CAD model based off of a scanned thing" yet - what I envisage is basically using point cloud features as guides and checks to building stuff manually. Like, think of reverse engineering something with a caliper - except you have a bunch of points already in CAD that you can measure in software and interpret the features of. Which is going to be a decent amount of work still but IDK, maybe someone who knows better will chip in, this is another one of those things where I scratched the surface enough to have useful-to-me results but I'm willing to bet there's all sorts of better methods out there somewhere.
|
|
#
?
Apr 3, 2021 03:22
|
|
|
|
| # ? May 1, 2024 19:11 |
|
|
It's very expensive but I've used geomagic wrap with their now defunct scanner at work and it'll happily generate surfaces/cylinders/holes whatever from point clouds that you can tweak and verify yourself, very powerful, very easy to use. There must be dozens of software sets like this available but trialling most of them is difficult or impossible without committing to their hardware too.
|
|
#
?
Apr 3, 2021 10:48
|
|
