You should all listen to Dr TAO interview on the vision encoder where he explains the difference between resolution, precision and accuracy.
1 Like
He offers another nugget of advice which is to use a filament with low shrinkage, perhaps one with CF in it, because he says the shape of the part may affect how well the shrinkage compensation works.
If youāre expecting to visually see a difference after using the plate, I think youāll be disappointed. However, if youāre printing something with gears that mesh, maybe youāll notice a difference in how well they mesh. Likewise for parts that fit together with tight tolerance.
1 Like
Iām glad I read about the hard-to-spot film here. I actively looked for one but was unable to see it so ran a calibration with it still there - which completed just fine. Recalling this thread I then looked again and again until I saw a faint bubble in a corner. Sure enough there was a film.
Interestingly enough, a new calibration after removing it ended up within 5-12 Āµm from the previous result.
Apart from that story I had a maximum improvement of 1054 Āµmā¦ wtaf!? Over a full millimeter?! Iāve had the printer for like 10 days, maybe I should print some serious tests instead of filament clips.
1 Like
tighten your belts i guess
Yeah, im thinking the h2d actually needed this plateā¦For some reason. 1mm is nuts. Imagine if they shipped it to someone who couldnt afford the plate with the printer. With 1mm, id expect the plate to come in the box. At first, I thought the plate took the precision to ābeyond normalā, but the more I see people post their compensation numbers, the more im thinking that without the plate, the h2d is sloppy.
One of my printers was off by 550ish and the other by 450ish. Imagine if I was stuck with that variance.
It makes a difference if you print parts that need tight clearance.
Iāve found that after using the vision encoder plate, my H2D will reliably, repeatably print articulating functional parts with clearance gaps down to ~0.05mm on straight parts, and 0.07mm on circular parts enclosed all around the circumference.
Before using the vision encoder plate, there were inconsistent results with gaps <0.15mm (some positions were good, some were unfreeably āstuckā).
4 Likes
I wonder whether Aurora Tech Channelās clearance test could detect the improvement? Or if not that one, what sort of clearance test could?
Thatās the tolerances Iām used to on my p1s. For a 6x3mm magnet I have to make the hole 6.15x3.1mm. I have been doing the same prints on the H2D and Iām finding the same holes are a little looser and I need glue to hold the magnet in.
I think itās calibrated once at the factory so itās more of a maintenance item. My printer said it was max 0.65mm off but there wasnāt any difference in the printed parts before/after. So new printers are probably ok (?)
1 Like
Looks like Iām buying one when I restock then. Iām working on something that needs exactly that kind of precision. Iām going nuts with about a 0.15 variance print to print.
For reference hereās the vision encoder calibration results of my H2D after a few hours of break-in.
I havenāt printed any closely-mating parts to do any comparative testing, but the real value I see in this is ensuring that mating parts printed on different H2Dās fit reliably.
Edit: on second thought, if mating parts are printed together on the same plate, the orientation of the parts on the plate matters, so this could improve the fit of parts printed on the same H2D.
1 Like
According to my testing (which was very extreme), this is true and the vision encoder calibration makes a difference.
One of the (slight, but made a difference in my tests) problems before calibration was exactly this - differences in clearances/tolerances based on position of parts on the build plate.
I canāt emphasize enough how impressively a āgoodā H2D (one that had proper QC and also didnāt get damaged in shipping) that is well calibrated with the vision encoder is - I can print function, articulating, close-clearance test prototypes on the H2D that couldnāt even be considered printing except on a well-calibrated good quality resin printer.
1 Like
Those screenshots of the vision encoder are nice and all but they donāt mean much without real world testing before and after parts. Make some blocks with fittings, with holes, and measure them with calipers. Then print the same set of parts again with the vision encoder results on.
The one thing I can say my vision encoder results were nearly identical across 8+ tests. After I changed out my tool head and x axis those numbers jumped up 150 consistently on every test since. So it was a huge benefit after changing out the x-axis. Something changed when I replace the x-axis, and the vision encoder corrected for it. All my results after the replacement are consistent with the new numbers. So the machine was able to recognize the difference, and the calibrations are consistent.
1 Like
If anyone has a moment to giggle, Iām really curious to see what happens when you:
- Take that vision encoder sheet to kinkos (or a mediocre quality copy shop near you) and have them copy it on one of their large format printers.
- Rerun calibration on your H2D with the copy
Like, I kind of want it to give you a malformed benchy. That would be hilarious. But I have a suspicionā¦
1 Like