Whilst filling out the BOM spreadsheet to get a guage on pricing, I thought I'd just quickly sketch the reachable area of the SCARA. With a shoulder rotation of 180 degrees, the whole bed would be reachable, though there would need to be a fibre break somewhere in the middle if adding fibre to exceptionally long parts
Even without that rotation, I could have 300mm of X. 298*163mm is the XY of the Photon M3 Max and a more favourable build volume than the Onyx Pro.I think this solution is better than the original idea of the moving gantry as it stays within the outer footprint of the Suspense at all times, allowing it to still be enclosed with doors while printing with continuous fibres. The concept Suspense design also uses a shorter bed than what I'm expecting, so it's likely that the 90 degree reachable area grows closer to matching or exceeding the 320mm of the Onyx Pro.
As I've recently discovered and written about in this log, Desktop Metal seems to use a fibre tape to reinforce their parts. It's likely not as economical and wouldn't be as versatile as Markforge's offering, but I'd rather take "rough but within reach" over "fine line fairytale".
Considering that the SpaceApp, CostApp, PowerApp, TimeToDesignApp and ProgrammingProwessApp (where [X]App = limitation caused by [X]) is lower across the board with the SecSavr Suspence against the SecSavr Sublime, it makes sense to see what high priority features planned for the latter can be implemented in the former in some way. On the top of the list is fibre reinforced parts for the #TEOSS [gd0037] project. It's a shame that continuous fibre printing for hobbyists is still a project seldom anyone these days takes on, so I'd like to see what I can do. I can always get a PCB cheap overseas, but I can't say the same for CNC milled or continuous fibre parts.
The Initial Idea
I'm thinking of something similar to the printer below that has a gantry that can fully move out of the build area so that the next layer can be printed.Another option could be something resembling a SCARA, which may allow fitting the attachment inside the printer's footprint. Prints that use fibre are unlikely to need the full 32cm Z height (and neither Markforged or Ainsoprint have a print height over 21cm anyway). Full Y height would be most beneficial and X would probably be fine considering that it's fairly long already.
The SCARA idea is likely the better route overall because a) I could probably route all 3 axes so that the motors are stationaty and b) may be less intrusive if permantly installed in the printer.
Everything about the printer except the application, collection and washing of resin is more-or-less standard stuff. The exclusions are things that could only really be fully tested when inside a system (imagine trying to find a good FDM 3D printer bed material by using a 3D pen). With only 8 cartridges, 1-bit colour is the only thing I can think of presently that I won't be able to try out.
I'd like to attempt 10 cartridges over 8 so that full 1-bit colour could theoretically be supported. It also allows me the option to leave a 2 or 3 high cartridge installed for print jobs that use a frequently used colour, such as white.
Now knowing this, I plan to use MGN9C for all linear motion. I'd imagine it'll make the BOM slightly less daunting to tackle.
2x300 for the rollers, 4x400 for the Z, 2x500 for the X axis and 4x600 for each set side of cartridges. The cartridges also get 4 additional MGN9C carriages. The roller linear rail length is the only one I've confirmed in CAD for now.
As you can visually see, the design of the rail mount has been changed. Aluminium channel wasn't going to fit in the space available and so I've used 10mm x 10 x 1.5 aluminium angle pieces which is 2.3mm away from the roller surface. The LEDs are 120l/m 2835's. ceil() has been used to make sure that the strip is longer than the X axis and that it is a length that is cuttable (every 25mm I'd assume).
It feels like a while since I've actually done some grind modelling. It's likely because there are so many moving parts to gd0036 that I never really dove deep into a computation since things could change, so most designing has been pseudo conceptual. Modelling this reminds me of the days spent modelling the #SecSavr Skyrise [gd0092].
Anyway, I was hoping that my belt system idea would reduce the roller sides from the estimated 25mm in the concept, but that isn't the case since I needed to be able to mount to the outter carriage holes (the black bolts).
Part of the reason is the rail, for which I first mentally computed a solution that probably woudln't intersect the bed assembly. At least it looks more sleek and actually has a linear rail, unlike the concept model. I'm also trying a new way to do bolts in Fusion. Instead of having a file for each bolt length, have a file for the bolt head and, before importing it into the model, making a "Bolt" component. Then the head can be jointed to a part and a cylinder can be extruded to a specific length. Lastly, the body and the head component are asbuilt jointed.
Whilst computing the solution, I've established that the current order of dependencies is as follows:
It took a few hours yesterday, but I thought of this belt path to be able to move the roller across the Y axis and spin the roller whilst having the motors mounted elsewhere. I was able to think this up because, a few days ago, I got to print on an Ender6 and get to stare at the CoreXY kinematics in detail. The should help with keeping the x length of the roller assemby as close to the build volume's x length as possible.
The only issue right now is that I didn't expect the assembly to be 300mm instead of 260mm. Part of this is likely to do with the use of a 57mm roller instead of the original 50mm to get 179mm of rolling area instead of 159mm.
It is expected that the roller gear will be 48T, the pulleys 16T and ball bearings 604zz, with a belt of height 1.38mm and non-tooth height of 0.63mm. The pulley flange is 18mm.
I wonder if I can use sensorless homing for this axis or if I'd have to also accommodate a limit switch.
I'm looking at my thickness options for the cast tooling plate and want to balance cost and weight reduction. For a 540x168mm bed, 5mm is £25, 6mm is £30, 8mm is £40 and 10mm is £50.
I went scanning on the internet for some calculator, but I have access to Fusion 360 so I used that instead to get a more accurate simulation. The plate is fixed up to 20mm away from the edges.
Simulating the bed with and without a part on it
The 6mm plate is 1469g and the 5mm is 1224g.
This is the 6mm bed with a 10N (approx 1kg) uniform load on the top. The max deflection is 34um which sounded pretty good so I went to 5mm.
The deflection is now 52.5um. Considering that deflection is over an entire print job, it likely wouldn't affect printing anything.
I wanted to do a more localised force test, but first I wanted to see deflection of an empty bed.
Without any load applied and only gravity acting, the max displacement of the plate is 23um on the 5mm plate. I do wonder if that could cause some first layer complications. 6mm isn't much better at all, reducing deflection by only 3ish microns.
Simulating pressing against the screen
I'm just going to assume a 2kg force is applied to the screen.
5mm isn't looking too good.
There's a displacement of over 100um from no force applied. In practice, the distance between the screen and the part would likely be similar across the length of the bed if compensated in software and would likely not be seen in the part. The force against the screen may also reduce as resin gets pushed out of the way.6mm halves the deflection to about 1 pixel length (40um).
Conclusion
Considering the 250g difference in weight, £5 difference in price and better performance, I'll use the 6mm thickness. Before this simulation, I was thinking that I'd have to use 8, 10 or even 12mm. Considering I don't know all the factors just yet, I think 6mm is the safer decision. It's also closer to 1/4", which is likely a common cast tooling plate thickness.
Hackady, why must you tempt me with a cool and exciting project? I feel like a mini, maybe even wearable printer based on VLM could make some real sense. One of the comments said something about an "Omni-tool".
It sounds like vaguely like a 3D printer and cyberdeck all-in-one. I can see that. A 3d printer, Compute Module 4 and #Tetent Tiny [gd0040] with a screen. To print anything actually useful (eg a screwdriver) , a length of 130cm or so would probably be required. Reading this hackaday article, it seems that dual material would be beneficial for dissolvable supports. Some mini activated carbon filter will also need to make an appearance.
I['m thinking of having the cartridges on the inside part of the arm and the build area on the outer side.
Other than the big questions about the Suspense I still have yet to answer, another 2 questions arise for this forearm wearable.
How to make the cartridges gravity independent.
Light source?
For the second question, my immediate choice would be this one:
Cheapest one on Aliexpress with a good power output for faster printing. I just worry about the overall thickness. Printing fast is usually of higher importance than ergonomics. Maybe a printer that fits inside a fillament spool is a more realistic target. It doesn't have the same apeal though.
Just been doing a few changes here and there, such as changing the model for a 7.6" screen with 495mm X and 163mm Y. I've also modelled in some cartridge placeholders. I'm thinking that moving the lamination cylinder closer and further away from the cartridge is better than my previous idea of moving the cartridge to the cylinder.With the cartridges, I'm assuming 50% of the volume to be resin, with a target of over 1L.To allow for the 495mm X axis over the original 445mm, I've done some cutting and trimming. I also intend for doors to the resin cartridges on the filleted chamfer walls, as highlighted below:
So I'm browsing AliExpress and stumble upon the control board for the Photon M3.
Additionally, the Photon M3 screen was still available for the £54 price, and with more quanitities than 1 or 2.
So I start thinking of what could have been when using the Photon M3 screen. It's still my first choice of screen, partially because it's got the same length as the height of the M3 Max's screen. I dig around for any STL that isn't the #SecSavr Skyrise [gd0092] frame, since I designed that specifically for a CR10 build area. Thus, the largest part I know of is a side section for the #SpecSpotlights [gd0024]. Wait. It... fits?!
Speed Calculations
But surely, FDM printers like the Sublime will still be faster when it comes to draft prints, so there's a need still.
Assuming 3 * 1.5s exposure for a 120um layer thickness and
3 seconds to change resin whilst cleaning the layer:
160 /(3600/7.5 * 0.12) = 2.7778
.78 * 60 = 46.8 minutes
Time: 2h 47m.
*starts slicing with the CR10 0.8mm nozzle + 0.5mm layer height profile*
Ok so I just have to beat 2h 47m and the SecSavr Sublime isn't redundan--
Speeds?
Ok those might be a tad slow, maybe? Ok what if the change takes 6 seconds and not 3?
And I need to be printing at flow rates never before seen in industry for that. I think Vez (VzBot creator) just barely got that with his fancy brass hotend. And then 16k acceleration with motion componenets as heavy as seen in the Sublime might produce some strong vibrations to the rest of the room ( but the motors are low impedance enough and components stiff enough that a quality print might still be possible).
[E1: 18th July] I didn't notice that Cura didn't save the new nozzle size when I changed it in settings.
Even with more realistic flow rates of 120mm/s and 9K acceleration isn't that much of a time penalty:
It's still quite a shame that print speeds and accelerations more-or-less cut in half are only a 20ish minute difference apart. Seems that most of the gains are in acceleration anyway.
Doesn't look like the 0.1mm layers would even need the 5 axis ironing step like for a 0.5mm FDM print.
Looking at the 8.9" screen again
8.9^2 / 6.6^2 = 1.8184
108 * 1.8 = 194.4 W source to match 108W COB
Now, writing this log, I realise that the screen sizes aren't square so the area difference is less than this. Anyway, I went off to look for that kind of power since the one at the store was only 120W.
Longgggg story short (like 5+ hours), anything with a power over this one cost a significant amount more, with some over £400.
The 7.6" 4K screen
At 4:53 am, I asked this:
The seller said that the 6.6" COB was the best option for what they had. I can confirm, since one of the options was this beast:
Some time later, the listings were up and the total cost was only £8.60 more than the 6.6" setup.Truly innovative stuff. Also, the seller sent me an image of the COB height:Looks to be under 45mm.
kelvinA
Whilst filling out the BOM spreadsheet to get a guage on pricing, I thought I'd just quickly sketch the reachable area of the SCARA. With a shoulder rotation of 180 degrees, the whole bed would be reachable, though there would need to be a fibre break somewhere in the middle if adding fibre to exceptionally long parts
I think this solution is better than the original idea of the moving gantry as it stays within the outer footprint of the Suspense at all times, allowing it to still be enclosed with doors while printing with continuous fibres. The concept Suspense design also uses a shorter bed than what I'm expecting, so it's likely that the 90 degree reachable area grows closer to matching or exceeding the 320mm of the Onyx Pro.
Another option could be something resembling a SCARA, which may allow fitting the attachment inside the printer's footprint. Prints that use fibre are unlikely to need the full 32cm Z height (and neither Markforged or Ainsoprint have a print height over 21cm anyway). Full Y height would be most beneficial and X would probably be fine considering that it's fairly long already. 
Everything about the printer except the application, collection and washing of resin is more-or-less standard stuff. The exclusions are things that could only really be fully tested when inside a system (imagine trying to find a good FDM 3D printer bed material by using a 3D pen). With only 8 cartridges, 1-bit colour is the only thing I can think of presently that I won't be able to try out.
2x300 for the rollers, 4x400 for the Z, 2x500 for the X axis and 4x600 for each set side of cartridges. The cartridges also get 4 additional MGN9C carriages. The roller linear rail length is the only one I've confirmed in CAD for now.
As you can visually see, the design of the rail mount has been changed. Aluminium channel wasn't going to fit in the space available and so I've used 10mm x 10 x 1.5 aluminium angle pieces which is 2.3mm away from the roller surface. The LEDs are 120l/m 2835's. ceil() has been used to make sure that the strip is longer than the X axis and that it is a length that is cuttable (every 25mm I'd assume).
It feels like a while since I've actually done some grind modelling. It's likely because there are so many moving parts to gd0036 that I never really dove deep into a computation since things could change, so most designing has been pseudo conceptual. Modelling this reminds me of the days spent modelling the 
It took a few hours yesterday, but I thought of this belt path to be able to move the roller across the Y axis and spin the roller whilst having the motors mounted elsewhere. I was able to think this up because, a few days ago, I got to print on an Ender6 and get to stare at the CoreXY kinematics in detail. The should help with keeping the x length of the roller assemby as close to the build volume's x length as possible. 
It is expected that the roller gear will be 48T, the pulleys 16T and ball bearings 604zz, with a belt of height 1.38mm and non-tooth height of 0.63mm. The pulley flange is 18mm.
The deflection is now 52.5um. Considering that deflection is over an entire print job, it likely wouldn't affect printing anything.
Without any load applied and only gravity acting, the max displacement of the plate is 23um on the 5mm plate. I do wonder if that could cause some first layer complications. 6mm isn't much better at all, reducing deflection by only 3ish microns.
Simulating pressing against the screen
There's a displacement of over 100um from no force applied. In practice, the distance between the screen and the part would likely be similar across the length of the bed if compensated in software and would likely not be seen in the part. The force against the screen may also reduce as resin gets pushed out of the way.
6mm halves the deflection to about 1 pixel length (40um). 
Cheapest one on Aliexpress with a good power output for faster printing. I just worry about the overall thickness. Printing fast is usually of higher importance than ergonomics. Maybe a printer that fits inside a fillament spool is a more realistic target. It doesn't have the same apeal though.
Just been doing a few changes here and there, such as changing the model for a 7.6" screen with 495mm X and 163mm Y. I've also modelled in some cartridge placeholders. I'm thinking that moving the lamination cylinder closer and further away from the cartridge is better than my previous idea of moving the cartridge to the cylinder.
With the cartridges, I'm assuming 50% of the volume to be resin, with a target of over 1L.
To allow for the 495mm X axis over the original 445mm, I've done some cutting and trimming. I also intend for doors to the resin cartridges on the filleted chamfer walls, as highlighted below:
Additionally, the Photon M3 screen was still available for the £54 price, and with more quanitities than 1 or 2.
Wait. It... fits?!
Phew. Wait. The height is 131mm not 160mm.
It's still quite a shame that print speeds and accelerations more-or-less cut in half are only a 20ish minute difference apart. 
Seems that most of the gains are in acceleration anyway.
Longgggg story short (like 5+ hours), anything with a power over this one cost a significant amount more, with some over £400.
Some time later, the listings were up and the total cost was only £8.60 more than the 6.6" setup.
Truly innovative stuff. Also, the seller sent me an image of the COB height:
Looks to be under 45mm.