Mount and Reservoir Discussion

[rylangrayston]

Nice informative reply Rylan. But as always I have more questions.

Do you plan (I'm sure you do) allow serial input to the peachy board? I ask as this would allow an arduino (are in my case as I had a bad habbit of blowing them an atmega chip) to bridge the gap between a PC (via wifi) or an SD card or even just tell the peachy that you added XXml to the build area of salt water.

along these lines here is what we were thinking ( but have not done yet)
use the extra IO pins to create ic2 and or SPI communication ports, so that an arduino can send and receive any message to the printer via the peachy circuit and so that
the peachy circuit could posibly control things like LCD screen, button panel etc.

What we do have working now is the peachy software sending commands thru a separate serial port, this is how we do our time lapses of prints an arduino is instructed to open the shutter while the laser drawing 1 layer. We also have send codes to the arduino to turn the peristaltic pump on and off, or whatever else we want.

[rylangrayston]

Yes, thanks for the great response. I'm really excited about the printer.

Are there any other I/O pins for output on the peachy board? It sounds like there is a drip sensor pin on the board ; wondering if there's another pin for output besides USB, or if the expectation for this hardware version is that the Peachy software would signal other (separate) devices (like a drip assembly) through other I/O mechanisms on the computer.

It seems like long term, the printer (or its successor) should have the capability to signal the zlevel rise directly rather than relying only on external passive decoupled drip + detection. Since the travel path of the laser for each layer is variable, to get optimal print speed it would benefit from being controlled by signal rather than depending on an external sensed rate. But I can certainly see where it makes sense to focus engineering on a whole list of other areas, like laser speed and accuracy, and variable laser focus and power output that need to be tackled first. It might be a bit silly to build a super accurate, super fast Z level control mechanism to support larger build areas if the laser speed / power needed requires minutes to cure each layer. =) Still, accurate, controlled z-level on smaller volume build areas is still useful and important.

>We have yet to Write a feature where you get N calibration points. ( Right now you just get 8.)
Can you comment on this further? 8 points for which calibration? for Z or for XY?

I picked up a used 300cc fuel injector on Ebay for really cheap. The math suggests that for the ~2'x2' container, @~40psi the injector should be held open about 80ms per pulse to get 0.33ML, the amount needed to raise 1 micron in this container. (roughly) Assuming that salt water flows similarly to gasoline, and the used injector is in more or less good condition, and the thing is actually accurate. That's a lot of assumptions. Will need to experiment and see if it's in the ballpark.

I also picked up a peristaltic pump. It's a bit hard to tell from the specs what the pulse size will be. Anyway, I can play with both ideas. What size pump are you playing with above?

I'm not going to attempt to run anything bigger than 2'x2' for now. That size is still manageable to build UV cage around for safety (not to mention it will be a while yet before you guys are ready to ship). And anyway, even if the injector can do the z-level rise at ~12.5 pulses per second as the napkin math suggests, the laser speed has a long way to go before it can cure that many layers per second (if it's even possible to go that fast).

What is the layer height you are using on your test prints? With surface tension and other effects, how would you say the printer is performing with respect to z-layer height?

There are about 8 extra io pins( i dont have the details memorized because we havent started using them yet)

a more direct Z height will definitively get implemented something like a g code to the micro that says goto this height travleing at this speed.

more on calibration..
We are working on a 3dmentional matrix of points that can be used to interpenetrate out all the non-linearities for all the places the laser can be in the build volume... but dam thats a wordy way to say it, and a year ago I wouldn't have understood what I just wrote.. so ill try to put it into an analogy too. This analogy by the way is actually how i prefer to think of it:

say you had a clear cube of Jellow and inside this cube of jelow was a bright red capital letter T.
To get the Red T in the clear block of jellow you cast the T first, then placed flopy red T in a cube and then poured clear jellow around it.
Great just one problem, The Upper Right tip of the T is sagging down, but the upper left side is bent towards you and also bent up.
What can you do to correct this.
what if you had put iron balls in the upper right and upper left parts of the T. Then you could use magnets pull on the balls just enugh to unwarp the
Saggy T. The balls would squish and warp the jellwo around them, Such that the T could become less saggy.
Now what if you had 100 magic balls that you could move magicaly in the jellow, then you could really straighten out that saggy T to perfection.

In this analogy above the magic balls are our calibration points. We get to magically move the balls because we have a mathematical representation ( a 3d matrix ) of the jello cube / build volume. The atoms of jellow are akin to every resolution point in our build volume, just like the position of each atom in affected by the balls near to it, we interpolate the position of every resolution point so that our calibration points affect the resolution points near them.

Currently we only have done the work to have 1 ball at each corner of the cube shaped build volume, so we only have 8 balls and there only in the corners.
This means currently we can only remove linear artifacts, for example we can correct for things like the entire T leaning, but we cant yet unwarp a specific part of the T. Some day we will do work to allow you to decide to have as many calibration points as you like, and when that day comes your printer could be much more accurate.

hope that analogy made thing better not worse

Glad to here of the progress on the injector pump idea already Toasterboy!
lets see whats next.., pump size.. We have been making our own pumps so now its time for my napkin math
Today I was printing in a 6 in diamter cylinder, and there are 206 pump rotations per mm so ....
6* 2.54 = 15.24 cm diameter
15.24/ 2 = 7.62 cm radius
7.62 * 7.62 = 58.06
58.06 * 3.14 = 182.32mL
182.23/2000 pumps per cm = .091 mL per pump rotation

The pump uses between 100 and 200 mA depending on speed which is controlled via votages between 3 and 20 volts.

as for how its doing in the z resolution. I think there is a recent post where I explained this in some detail... ill see if i can find it...
EDIT: found the posts with details about z resolution, there on the 3rd page of update 51 thread.. here is a link
http://3dprintboard.com/showthread.p...Printer!/page3


Great questions everyone, Im happy to answer them now while the community is still small, someday there may be thousands of people joining and posting, and Ill be glad to have all the help I can get from those of you who are here now learning about the peachy printer in detail before anyone else

[Toasterboy]

Ok, Cool.
You might be able to play with the Mesh Deform Cage in Blender to rapidly zero in on the required deformation matrix. i.e. model a concentric onion cube (with smaller cubes inside) like shape, print, measure, and play with the deformation in Blender .. seems like it wouldn't take too many prints and a caliper to get the deformation seed matrix close to correct. Probably help the larger you can print the test object.

Cool.. so the sample prints are about 10 micron Z or so.

Your results with the peristaltic pump sound promising. =)
Has anyone done any tests with vibrating the build area to overcome surface tension/meniscus?

I got the injector in the mail today, but I'm waiting for the pump still, so it will be a bit (and a trip to the hardware store for other bits) before I can play with it further.

[jontelling]

I got the injector in the mail today, but I'm waiting for the pump still, so it will be a bit (and a trip to the hardware store for other bits) before I can play with it further.

Are you a beta tester or are you just doing a ad-hoc test of water levels before the V1.0 arrives? Please post pics etc, will be nice to see.

You mentioned the meniscus, ive only thought about it a little, but was there a mention that it was an issue or are you thinking once you go to a super high Z resolution? heh you could have a rod moving up and down in the build fluid, as long as it doesn't fully emerge there shouldn't be any ripple. hehe i can see it now, a big pully and a rod attached.

[rylangrayston]

Ok, Cool.

Has anyone done any tests with vibrating the build area to overcome surface tension/meniscus?

I got the injector in the mail today, but I'm waiting for the pump still, so it will be a bit (and a trip to the hardware store for other bits) before I can play with it further.

Its alwase great to here of people taking actual actions like ordering parts for unique setups, this kind of thing really excites me! I second jontelling, post pics!

as for vibration, very early in the project I spent hours tapping the printer in different spots watching how vibration actualy helped in many situations ( such as when your using a higher powerd laser) I bough a bunch of cellphone vibrators, sealed them in hot glue and diped them into the resin while printing, but i didnt spend enugh time experimenting to really say anything other than there seems to be potential in this idea. My next step on this one would be to get big speaker to set the bottom tank on, and then make the speaker do a range of feq and wave forums over the course of a print to see what works. It been funny because usually I here people talking about how vibration will make the printer fail. I Think controlled vibration could improve things, and if the same vibration that improves surface tension also causes the mirros to shake, its prety easy to vibrate in between the drawing of each layer. Oh ultasonic mixers are another thing to try ( if set to a low power, so they dont actually mix the resin and water)

[Toasterboy]

=)

I'm not a beta tester. Would be fun though.. wish I had my hands on a Peachy. =) The dripper assembly is decoupled from the actual printer, so it's quite feasible to experiment with the dripper without a printer.

I was thinking meniscus would be more of a problem the faster you go (i.e. more powerful laser) and the thinner the layers are, and the thinner the laser beam is. If the printer ever gets "accelerated" to the point that it can do multiple layers per second for example.

I'm missing a few parts still. Attaching a pic of what I have so far.

The question is really just whether the fuel injector will accurately/consistently dispense in a range appropriate for controlling z-level in a build vat (of variable size). Since that's similar to what injectors are designed to do in an engine, chances are good that it will.

The injector seems to fire just fine at 12volt 3 amps. (some of them need 6 amps) I don't have specs on it so I'm not sure if it needs more. Don't have a scope. 3 amps might not be enough to get the injector open as fast as it can go.

Initially I'll just program the arduino to dispense at a fixed rate and dispense into a measuring container with stopwatch. If that works out to be in the ballpark over a good range, I'll probably add a display and buttons to make it adjustable without uploading new code to the arduino.

I'm thinking the peristaltic pump makes more sense for the Peachy at the speed it's currently operating at. Something like this injector setup probably won't be needed unless the laser is beefed up and the Peachy can handle moving the laser much faster so it can cure multiple layers in a second.
arduino_injector_bench_small.jpg

[rylangrayston]

Intresting that you brought up layers per second

so way back in this update we were doing some very small prints. they were esp small in the xy
at a square collum that is 2mm x 2mm for example has a 2* 4 sides so a circumference of 8 mm
100 mm/s /8mm = 12.5 layers per second.
so even the basic printer when given small jobs can do many layers per second.

you can see lots of examples of this fast printing here:


PS Thanks for posting that pic!

[Toasterboy]

Fantastic! Yeah, I thought you had mentioned that early on. Do you think that prints #76-78 might have gone better with a buzzer breaking the surface tension and an injector or pump providing really stable Z raise in synchronization with the laser? The implication is that with a 1 or 2 watt laser (or focusing the 20mw one) the printer could go very fast indeed if the surface tension effects can be solved.

[Toasterboy]

Ok, so I programmed the arduino to fire the injector 20 times a second and hold it open 80 ms each pulse. Wired up the pump, and connected the hydraulics with surgical tubing (and zip ties, after the hoses popped off the first time) =) Using regular water for now.

Latex tubing expands in such a way that it provides approximately constant pressure, apparently. (people use variations of it to power homemade cps supersoakers).

#1 Wow, the clicking from the relay I bought is annoying... it's a lot louder than the injector, which is quiet.

#2 It basically works.... fired it up and the injector pulses water at what appears to be the programmed rate. Dispensing volume and pressure unknown, but it's pretty quick... maybe approaching the rating of the injector, which is 300cc per minute.

#3 Need to solve pressure regulation before it makes sense to do any measuring... the pump I bought is adding pressure to the system faster than it leaves the injector, and that's with an undersized power supply. I might have gotten a pump that's too beefy (500 ml/ minute and 40psi). Clearly need to get some kind of pressure sensor, and shut off the pump when pressure is above target level, and probably control the current too. The latex tubing can only handle so much volume before it expands (and that's destructive, it can only handle that a few times).

#4 I didn't have the right arduino pin cables, so I jury rigged with bits of wire and alligator clips. It works, but if water leaks it's going to short out and potentially electrocute me. Need to get the right bits and bobs and re-arrange the parts so as to minimize mixing water and electricity.

#5 The jets from the injector come out with a good bit of force. Looks like some of the nozzles on the used injector I got are clogged... will need to get some brake cleaner and try to salvage the clogged ones.

But yeah, it basically works, and it's pretty clear that it will be able to adjust dispensing rate to match a given container. (above some minimum drop size, which depends on the pressure, and physical characteristics of the injector).

Yay, fun!

[Toasterboy]

Here's a video of the proof of concept injector setup:


It is configured to open the injector for 80ms, and close it for 50ms. That works out to ~7.69 pulses per second (not 20). After the video I did another test and measured it dispensing ~100ml in ~91 seconds. I don't think that it's operating at full capacity yet though.

That works out to ~0.14ml per pulse. (which is about 1/2 micron per pulse z level rise in a ~2'x2' container) Also, the dispense rate would be ~4000ml/hour with this setup, which is about twice as fast as the siphon dripper. Haven't played with varying the open/close durations yet.

I suppose the next steps are to research and select a pressure sensor, and figure out a better tubing solution, so it will take a while before I'll be able to proceed. I'm also pretty sure that the latex tubing isn't sufficient to run at the right pressure.... suspect it's well below 40psi currently and will perform better at higher pressure. I used latex initially because it fails gracefully, and pops like a balloon rather than exploding on overpressure.

It seems pretty clear that this concept can provide some pretty fine control over volume dispensing and timing though.