New Printer Day: Saturn 2

[ICMF]

Yay!  My Saturn 2 has arrived!  Upgrading from both a Phrozen Mighty (for the bigger size) and OG Photon (for the higher res)  I haven't been able to run it yet, but resin printers are kind of commodity products at this point, so barring any major design flaws, it should be fine.  A few thoughts for those who are interested:

 

Like:

• Power switch is on the front of the machine.  Nice for multi-printer setups and just generally more ergonomic.
• Lid "pre-drilled" with hole for vent fan.  There's a moulded in hole (with screwed on cover) for a ~70mm PC fan to help vent the machine.  Those that want to add extra ventilation won't have to mutilate the lid.
• Vat has max fill indicated.  Simple but handy touch.
• Vat has handles.  Pretty much a requirement at this size.
• Vat has feet, protecting the FEP when you set it on a workbench.  Simple solution with four thicker cap-head screws, but it'll save scratches/damage/punctures.
• Build plate is a positive fit.  Handy if you have to remove and reinstall the plate mid-print for some reason.  
• Plug in charcoal filter.  The printer has a USB socket on top *strictly* for plugging in the included filter.  Easier than using one of the battery operated filters, but DO NOT PLUG IN OTHER USB ACCESSORIES.

 

No-Like

• The vat screws are too long.  They thread about 1/2" into the printer, so they'll be annoying to screw/unscrew.
• The power button is a momentary switch.  It takes about a second for the screen to light up, so there's a bit of "did I press it hard enough or not" uncertainty when you turn it on.
• Tape around screen.  Elegoo have applied a tape/gasket to seal the screen from resin leaks, but the cutout is actually smaller than the build plate, which makes levelling... confusing?  Narrower tape or a thinner build plate (it's larger than the screen) would help.

 

Meh.  (not a fan, but not a big deal)

• The design.  It's a little Transformers, IMO.  I'd be good without the angles.
• USB socket is at the rear left.  I'd much prefer it to be at the front, but it's easy to fix with an extension cable.
• Pouring lip on the vat is in the top left corner (or else max fill line is in the wrong location).  Bit of a nit, but with the fill line visible, I'll have to rotate the vat to pour resin out of the spout; with the spout in my preferred location, the max fill is at the front of the printer, and less visible.
• Ball jointed build plate.  Not a huge fan of Elegoo's levelling system, and with a plate this large (10"), it would really benefit from four point levelling.
• Solid feet.  These are moulded into the chassis.  I'd prefer to have adjustable feet, so I can level the printer itself.
• Plastic chassis.  I honestly don't care - it's sturdy enough for a resin printer - but some people will complain.

[general melchett]

Good little report there ICMF. I've got one on pre-order over here in the UK, looking forward to getting hold of it. Agree with your observations but will add that, while not really a problem, adding wi-fi connectivity would have been a nice feature. Like you, I've never been a fan of Elegoo's ball-joint method of levelling, much preferred the four screw system used on the Phrozen Mighty 4K which always seemed to stay put. Hopefully with the larger build capacity and screen resolution it should come into its own with the larger prints we'll be doing. Certainly worth trying out.

[nick]

How did you get on with this? It's taken me a LONG TIME to get comfortable and good results with this machine, I'm starting to feel the love finally, but it nearly went in the skip a few times along the way.

 

I've got a Jupiter too and I've ended up modifying both of them quite a bit to get the reliability I need.

 

BTW if you do want to fit a fan to your Saturn (For heating in my case) I knocked up a an adaptor plate for it here they aren't standard holes of course:-

 

https://yonda.org.uk/files/Fan Mount.stl

 

 

[ICMF]

Currently?  Not great.  I've run a bunch of normal, supported prints, and the print quality has blown me away.  Like, injection moulded quality with just a bit of anti-aliasing.

 

But...

 

I tried to print a simple replacement turbine part on the build plate last week and the Z-axis dimensions were completely screwed up. https://imgur.com/8lUgfmw (original model in grey; actual print dimensions in red)

 

Then ,I drafted a simple stair-stepped calibration model to figure out where the error lay. https://imgur.com/vo12U7l  Super simple - each 'step' was 1mm high, so even if the build plate level was off, each step should still be 1mm tall.

 

I tested the model on my Anycubic M3 Max, and the results were as expected - each step was +/- 0.02mm of the expected amount, which is well within the expected margin of error of my digital calipers.

 

When I ran the test on my Saturn 2, the results were terrible. My first 1mm was 1.57mm thick (which is WAY thicker than the actual tramming of the build plate), but the top of the part was only 9.7mm thick, meaning I had actually *lost* 0.87mm over the 10mm. https://imgur.com/Aqpxm4Y

 

No idea where the problem lies, or how to fix it.  It feels like it has to be something mechanical - it can't be a slicer setting. since it's the motion of the Z axis that's the issue.  I guess my next step is to print a taller calibration piece, to see if the error eventually zeroes out after 20 or 30mm.    And wait until Elegoo get back from vacation to hopefully help out.

[nick]

Since my last post, I have got both printers dialled-in now. It all comes down to bed levelling IMHO.
 

It is super-critical on these bigger machines. 

the fastest and easiest way to do this is to slice something, anything, a cube say. Set the layer height to 0.01mm. Then drain the Vat of all resin, loosen the build plate clamps and start to print your test job with the empty vat in place.  Pause it as soon as the UV lamp comes on for the first layer, then slightly push the plate down and tighten up. 
 

I did a lot of research and experimentation about what Z=0 setting does and doesn’t do, plus on top of that, the printer, in its wisdom,  does all kinds of ‘helpful’ stuff to the first layer itself. Very few people measure or care about Z axis accuracy from the bed up as most things are rafted/supported and it does sort itself out after the first few layers. 

 

try you calibration print again, I got mine absolutely dead on with this method. There is Z scaling as a Gcode option, but you need to get it linear fist, yours isn’t, I suspect because your z=0 is out. 
 

BTW, if this is out, in my experience failures can occur 150+ later. I was getting utterly inexplicable failures at these heights and it all went away when the height was right. I can’t fully explain this, but there is without doubt, some kind of ripple effect if the first layer is out, that can manifest itself as really strange failures many layers later. I’ve got no science here, just my own observations, but I am absolutely convinced after chewing through gallons of resin to get these things reliable. I wish I had a better explanation, I suspect it’s something to do with pressure build up with a large plate and a big pool of resin playing out but I don’t know enough about the physics or fluid dynamics to offer a proper explanation. 
 

levelling this way is a mile away from the traditional piece of paper method in terms of where the plate ends up. 
 

Have fun, it took me 3 months to do this, but it’s a distant memory and they work like Swiss watches now.

 

Apologies to any normal readers that stumble upon this, for inserting a large block of resin printer gibberish in the middle of Britmodeller.

 

Nick

[ICMF]

Minor update, but I seem to be on the way to fixing things.  Unfortunately, the fix kind of sucks.

 

After reading about the Z axis flexing here https://blog.honzamrazek.cz/2022/10/elegoo-saturn-2-review-is-pixel-size-everything-in-depth-look-disassembly/ I decided to add a ridiculous amount of light off delay.  I went with 90 seconds for the base layers and 60 seconds for the normal layers.  The print time was 7x longer than 'normal', but I wanted to see if I was on the right track, and it was better to err on the side of caution.

 

The results were markedly better.

 

My first mm was still off (which I expected) but the subsequent steps were much closer to the expected dimensions, and broadly linear.  They're still consistently UNDER 1mm each, so I'm losing ~0.15mm/10mm and expect I'll have to play with the firmware to tweak the e-steps, but at least things are moving in the right direction.  Now it's a matter of dialing in the light off delay to figure out the shortest time I can get away with while still maintaining accuracy.

 

As for bed level...  well it is as far as it can be.  My bed calibration tests showed that my bed is racked, so 'level' is a matter of compromise.  https://imgur.com/3ZqoMZI  But now that I know there's so much flex in the Z axis, and that it throws print dimensions so far out of whack, I'll have to re-level it with a large light off delay for a more accurate calibration.