https://www.researchgate.net/publication/284330089_Knife_and_Roll_Coating

Well that explains why BCN3D is using the roller approach. Not sure how viable that would be for my design. I'm less concerned about the viscosity and streaking limitations of the knife as I am with the complexity and cost disadvantages of the roller. 

Leadscrews to replace ballscrews

About an hour ago, I found out that the 13.6" Prometheus MSLA uses a good old fashioned T8, 8mm-lead leadscrew with a 0.9mm stepper motor. The prints look absolutely fine. Considering that they're cheaper, quieter and have a higher linear stroke : length ratio, I've decided on switching out both sets of ballscrews for T12 leadscrews. I'm using 8mm-lead since that would mean each whole step is 0.04mm, which is the same size as a pixel and I'm only interested in printing in cubic voxels. 

T12 was chosen out of concern for the cartridge axes. They're used on the Z axis too to keep the parts consistent (like the linear rails) and because I believe that seeing 12mm diameter leadscrews on the Z axis would make the printer look stronger and more premium.

The savings are quite significant, more significant than chasing a couple £ on which Monster8 version to go with. Speaking of which, I just found out that Zonestar made an 8 stepper motherboard. It's smaller than the regulars at 117x95, but unfortunately doesn't seem to come with UART support and costs more than the BTT Octopus.

Activated carbon filter efficiency

While I'm on the subject of price, it doesn't seem that my hopes of filtration are going to pan out all that well:

Even with some thick filters, it doesn't seem that the efficiencies are ideal. Thus the log 2 logs back is likely the better solution. I don't think people in their appartment (like me) would object to venting a tube outside for 10 minutes; it's having to keep a ventilation tube for the hour or day long prints that I have an issue with. The data from the video below suggests that the seal doesn't need to be airtight to keep polutants in the print chamber.

Additionally, cooling down the print chamber may both reduce VOCs and increase viscocity. I can't imagine it'll be power efficient though. Cooling down the resin from within the cartridges sounds more efficient while reducing the chance of condensation.

Expected pricing target

Since the current BOM is down £110 and the filter system is no longer happening in the forseable future, I think I can target £1099 for the first prototype, which includes the arm attachment and extra screen. Oh and I haven't mentioned this yet, but after some thought, I'm using the same Nema 17 motor for everything. This also saved a bit on the BOM.

I just put together the BOM for the belt system as an alternative to the ballscrew and in terms of cost saved... there's no cost saved.

It also seems that not a lot of holding torque is actually needed for a 1605 ballscrew:

Even less for a 1204:

Which means that the 14:1 reducer motors can be removed too in favour of the £4.80 motors used in the rest of the printer.

Then there's the risk that if the top pulley assemby breaks, the resin cartridges are going to fall down like underengineered lifts in action movies.

Overall, it seems that belts would end up being more expensive than ballscrews.

[16:32] Confirmed:

Computing a mental simulation, I was arriving at the conclusion that a slower filter air flow would lead to a higher vapour concentration which would make filtration easier. I've just read a comment on this reddit post.

Therefore, it may make more sense to make the Suspense airtight (perhaps with a very slight negative pressure) and then fully clean the air as a final step before the user is notified to open the door. Lower evaporation means that the filters should last longer. 

Considering how VLM can (hopefully) print support material and lower Z heights result in faster prints, it makes more sense to priorities less frequent topups of the resin cartridges than a couple of extra centimetres of Z height. As long as I can theoretically print an entire half of #Teti [gd0022] (256mm) or tile cover for the #T^2 Tiles [gd0095], (260mm ish) I'm fine.

To get a rough estimate, I just threw the model with a cutout for where the peristatic pumps will be into Cura and sliced 4mm walls and 999 walls, and then take away the latter from the former to get the material contained inside. For a 48mm high cartridge, I get about 1700g and for 37mm it's ~1300g. That disparity increases when opting for a double height cartridge. 

Since there's a few other walls that need to be inside the cartridge, I'm expecting the 37mm cartridge to be around 1000 - 1100g and the 48mm one being 1500g. The double stacked for 48mm would be something closer to 3200g.

My main concern now is that they'd be something like 2.3 * 5 + 0.5 = 12kg of cantilevered mass. It's cantilevered since the sides are the ony way to install the cartridges so I need to be able to slide them in and out. I'd have to see if the quad 9H carriages can take it and look into a very thick belt with the pulley mounted onto the 42.4mm tube.

I found this article: https://www.fabbaloo.com/news/question-of-the-week-polyjet-patents

Turns out the polyjet patent expired all the way back in 2019, but as I and the author can deduce, there's no prosumer budget printers around. The cheapest I know of is Mimaki's newest offering at £35K. 

Jettability really puts a limitation on available materials and quite a lot of support is needed since no overhangs are allowed. VLM would still need support material since any and all islands will stay on the PET film and get whisked back to the cartridge, but it could be more like Cura's tree supports (or even Lego's) due to the tolerance to overhangs. Additionally, expensive additives to the resin like thinners will not be needed. This 2016 paper on low-cost polyjet printing acknowledges that a cheap printer with expensive materials defeats the purpose of a cheap printer. 

Speaking of polyjet...

I just remembered that I was sent a sample part from the J55. I asked about it days after release and I wasn't actually expecting to get anything, but it did eventually come through. I was expecting some show piece model, not what looks to be the 3D equivalent of 2D printer test page:

Unless transparent CMY colours are loaded into the printer, I belive coloured transparencies (see the cyan and yellow checkerboard pattern) are where the most visual difference is going to be when comparing this test piece to the VLM-like process of the Suspense (I'm thinking of calling it SLSL). It's probably going to look cloudy.

The knobs turn, and smoothly. There's no dicsernable backlash, so the tolerance is likely under 0.2mm clearance. Overall, this is a good test print to have now, since I can design a very similar test and tune the printer to get in the ballpark of tolerance and colour reproduction.

So last night I was listing the motors needed and all the stepper drivers needed. I came to 16, with 2 using a Y adapter and 6 needing to be connected to relays to switch with 6 other motors. I'm assuming that 1 line of each coil needs to switch, meaning that I'd need a 12 channel relay. They only seem to come in powers of 2 and 16 is cheaper than 8 + 4.

They seem pretty sizable though.

BTT EXP MOT

Anyway the slippery slope started when I remembered that the "scraper" axes can't be shared as the PET film will move the unpowered one when the "tension" motors tensioned it. Now thinking about it, the scraper roller would've just moved to its axis limit and all would've been fine again, but last night it meant bringing in the BTT EX MOT.

There's 2 main reason's I dislike the motor extension board. The first one is that it comes with some unreasonably short joiner cables and it's only after seeing user posted images that I understood why. The second thing is that BTT has their EXP1 and 2 inverted, compared to the FysetC Spider and MKS Monster8 that have the ports in a similar location. I could buy (or probably ask for) longer cable but that's more expense on a £6.50 board, more wires and more PCB footprint. 

Anyway, now only 3 motor pairs used relays, and that 8 channel board is still pretty sizable, and I was like "Man, I don't really want to be cutting and trimming and risking a timing error, causing fried stepper drivers just for 3 motors. Add another BTT MOT.". 

FysetC F6 v1.4

Since I'm using Klipper and the Manta M8P comes with 2 USB ports instead of the 1 I thought it had (I had to look closely at one of the listing images that wasn't a birds-eye view), I was thinking that I could replace the 2 MOTs for a £19 F6. I'd spend a couple quid on what is looking to be a £1-1.2K printer just for quality of life. 

I was thinking that I could save a stepper motor by having the fibre extruder be a racheting system, but then I remembered I needed a pump for the (clear) UV resin that the fibre would be placed on.

I was wondering if I should just go for 2 Monster8's instead of a Manta8+Monster8+F6 merger but all was well...

...Until the string stepper attacked.

I totally forgot about that.

Motherboard8

The next maybe hour or so was me deciding if I should go with the Monster8 V2 for £31 or the rediscovered Monster8 V1 for £27. AliExpress search isn't the best, so I all the Monster8 V1's I found were around £30 and I didn't even know that there was a V2 and that I was looking at it until I saw this image:

I will be honest. If price was equal and I had the choice between the Spider V1.X and Monster8 V1, I'd go spider just because it looks so much nicer and has a more compact footprint. However, the Spider from any non-shady seeming seller was £40, even more than the BTT Oct. Speaking of which, I didn't choose the Oct because it required a second input for motor power and I didn't want to pay more for a board and then have to strip and crimp more wire.

The Monster8 V2 is certainly more apealing to my eyes.

As you can see, all the capacitors are aligned and all the headers are neatly on the bottom of the board, as well as the USBC port on the same side as the M8P's USB ports. For the aesthetic, I plan to place the boards in the following orientation:

Having the ports in the V2 location helps with wire management. The MonstV2 also has easily ressetable fuses, unlike the V1 and M8P, so I could put the UV lamp on one of them if needed.

Especially since the board is designed by FysetC, I never understood why it didn't come with the classic blue heatsink in the first place. The potentiometer is covered when used with the long silver heatsink. 

Conclusion

I'm currently at 23 motors if I can get away without a stepper motor for dispensing the fibre. If I can't that'll be motor 24. I do wonder if there would be some mysterious motor 25 down the line, but I don't think s--

remembers that the PnP also needs a solderpaste dispenser

oh. Yeah I should see to it that I avoid the need for a fibre dispensing stepper then.

For some reason, I don't feel like physically modelling the arm attachment even though I've mentally modelled it a few times already. Thus, I decided to start thinking about the actual resin cartridges. 

Volume

Part of the reason was to decide if I should reduce the thickness of the cartridges to gain another 50mm of Z height or prioritise more resin volume. The latter is divided over 5 cartridges so it'd probably gain more Z than lose in resin volume. As long as I can get the amount of resin that would correspond to 1kg of filament printing, I should be ok. I can always make a double high cartridge for the 2 most used materials if not printing in full 1-bit + alpha channel colour.

Shape

The cartridges are still intended to be relatively flat, house shaped containers that maximise the available area inside the outer case. Considering Matterhackers' resin bottles are 90mm in diameter, I'd likely still only be able to fit 5 different materials in the space provided. It'll just be harder to access them.

Motors

The resin is unlikely to perfectly laminate just by gravity alone, so a pump motor would be needed. There might be a solution that only requires 1 motor, but I don't have it. I believe that there would need to be one to move reclaimed resin into a buffer and another one to move it out of the buffer to be applied to the PET film.

Pump

Until proven not to work, I'm going to aim for a mostly 3D printed peristatic pump.

The cartridges are actually intended to be the permanent storage container of resin. While it means that I can't use a cool, see-through tube like a custom PC water loop (which would've helped with resin identification without labels or fancy LED's like Me In The Past envisioned right at the start of the project), it does mean that I intend to print new cartridges when there aren't any empty ones instead of pouring the resin back into the bottle and cleaning the cartridge out. The pump is also intended to be a seal so that resin doesn't escape when stored. I know that this sounds like the consumable ABS build plates I heard Stratasys uses for their older ABS printer, but the containers are printed and so are unlikely to last forever and the main motivation of this project is to make resin printers safer to use. 

Where possible, I'd like to avoid screws, steel dowels and ball bearings in the design.

Due to the cartridge having a wall thickness, I think that the optimum solution to maximise resin volume is to have both pumps stacked ontop of each other. This could spur another appearance of the double stacked motor.

Applicator

When I started the project, I knew that applying a material to a surface must've been a problem that has been solved decades ago. I've only now found a lead on Google.

Moments before seeing this, I was actually wondering if I could use the scraper used to reclaim the resin could also be used to set the height of the resin coating from the other side. The thought came from SLS recoaters. 

I haven't looked into it much, but the "resevoir" approach might require a greater amount of pump and manufacturing precision to perform and more likely to partially clog (as in, a void in the lamination). The blade allows me to apply a lamination without a roller as a moving part, and with lamination height set to 0 (or negative), I might be able to make a seal to scrape the material into the collection area. 

Oh but I won't be able to suck back the resin still in the application zone due to the same seal. Would it be possible to skip the scrape reclaim alltogether and collect all the resin back in the application zone? No... I think it'd have to: 

1. scrape the resin from the PET film
2. move the scraper away from the film a minute amount so that there isn't a seal and the application pump draws back in what resin it can
3. apply any resin still in the application zone and scrape it off.

2 and 3 would only be needed for changing to a different resin. Also, when I'm talking about movement, it's actually the application roller that's moving towards and away from the scraper and not the other way around.

There would probably be something like the string used in the cleaning roller on the cartridge to catch any resin that makes it though all the main seals.

Lastly, the tubes for the reclaim section are likely to be on the far edges since it's more likely that nothing printed there and so would see the highest amounts of reclaimed resin.

This all started when I realised that "KGT Factory Store" didn't have a listing for a 550mm linear rail (they had 450 and 650 though) and didn't want to bother them to add it. Instead, I looked around AliExpress and found "KGT manufactor Store" which had almost identical prices, but they were giving out a 5% discount for the order quantity I needed. This, as well as the existant 550mm listing, reduced a £204 order to £180. Then I saw that I was 2pcs away from a 10% discount, and so increased my 9H carriages from 4 to 6, saving £1 in the process. It could be possible to save another £3+ by going with 50mm shorter rail on some axes, having the carriage extend over the edge by 10mm, but I don't know if that would work so I don't want to risk it. I've got a better chance of reducing the length needed for the resin cartridges anyway; going from 600mm to 550mm would save £8.

I then decided that I was spending too much (£61) on the cartridge ballscrews and, after reading this forum thread on lead screws, I'm switching to belts and a 19:1 gear reduction Nema 17. I was also able to find 5pcs Nema17 motors for £24.09 (down from £28 from the same seller on ebay).

Back on AliExpress, I was going through my basket and noticed that the 7.6" screen was missing. Turns out the listing got adjusted and the screen is now only £35.20! That's even cheaper than the 6.6" 4K mono at £39.50. Now it's gone from a good deal to just plain BetterTM.