Drip Governor - A most valuble first hack

[jjmouris]

Pete, nice work!!! Keep it up!!!

With a large enough reservoir using your method we could really get some accurate height management going with potentially the option to "dip the print" by very small amounts and great return accuracy. Something currently not possible with just a drip.

[Pete]

Thanks for the kind words, Rylan/Anuvin/jjmouris.

jjmouris, working on a few quick things from my initial testing
-I can easily control a micro to do discrete steps since the pump is only a handful of Hz (it's actually a really cool epicyclic gearbox, I'll do some macro photos and post them)
-approx 0.12ml per rotor arm

A 150mm square reservoir would give ~0.005mm Z height per count. I've seen a lot of printers using 0.1 or 0.05mm Z height resolution so to achieve the 0.05 you'd need 10 counts which is around 2/3rds of a second to pump in that water. If you had a print liquid that maybe needs 1mm over that and then back out again (don't know if this is reasonable) then the total time would be about 30 seconds in between each layer, so it's not all perfect. Just to put this in perspective printing a 25mm high object would run the pump without any printing happening for 4 hours! All this said, I'm probably going to be happy leaving my peachy printing stuff overnight and if it resolves a host of issues then I'm good with it!

[jjmouris]

Pete, without having any real experience with peachy printing. I would imagine that 1mm above the printing level is way more then needed to break the surface tension when printing at 0.05mm accuracy. I guess it depends on the resin but my gut feeling says that twice the accuracy should be enough (0.1mm), thus reducing your time estimate by a factor of 10.

For me however it is not about speed but final accuracy. I can see how surface tension would stop the resin from flowing properly even if left for hours if the change in Z level is not enough. Breaking the surface tension and then lowering the fluid level down to the next layer height sounds like a more accurate method to me.

I guess this is where much more trial and error is to come in the near future. Your work will no doubt prove essential in making this possible in the first place!

With regards to controlling the pump, maybe we need to look into stepper motors and absolute digital control of the stepper motor rather then sticking with the rather old hat brushed motor combined with timed runs. You can use a cheap of the shelf brushless RC motor and simply make it step round by controlling the current on each of the 3 phases.

J

[Pete]

Pete, without having any real experience with peachy printing. I would imagine that 1mm above the printing level is way more then needed to break the surface tension when printing at 0.05mm accuracy. I guess it depends on the resin but my gut feeling says that twice the accuracy should be enough (0.1mm), thus reducing your time estimate by a factor of 10.

I've got no experience either, yet! I really hope you're right but I fear might not be, the meniscus on water can be easily seen and these are much thicker liquids, although it's been a long time since I've tried doing physics of this nature so hopefully my feelings are wildly inaccurate.

For me however it is not about speed but final accuracy. I can see how surface tension would stop the resin from flowing properly even if left for hours if the change in Z level is not enough. Breaking the surface tension and then lowering the fluid level down to the next layer height sounds like a more accurate method to me.

Ditto...although I have seen a thread somewhere talking about floating another liquid level on top to effectively lower the surface tension which may prove a good addition or just better on it's own. I'll also be experimenting with ultrasonics to try and resolve this issue.

I guess this is where much more trial and error is to come in the near future. Your work will no doubt prove essential in making this possible in the first place!

trial and error......thank you, you god! When I was leaving school they were trying to change this phrase to 'trial and improvement', some ofsted nonsense about positive motivation......it's all about the error!
I hope my future work will yield much more than this ;-)

[jstrack2]

Here are a few pictures in CAD of an idea for the drip governor I had after doing some online searching:

valveA.jpg valveB.jpg valveC.jpg
The first picture is how it would look assembled, whereas the second and third picture I moved the outer casing aside to show what is going on internally. The blue thing is a rubber ball cut in half. The dark gray thing attached to it is iron (coated so it won't just rust). Normally the rubber ball stops the flow of water (as shown in picture 2). When the solenoid is powered it raises it and allows water to pass(as shown in picture 3). Because the ball is rubber (and spherical) it will form a good seal. The whole setup is vertical, so it will fall from gravity when not powered, thus closing the valve.

Also this can be made super easily by using a thin male pvc pipe adapter and pvc pipe. These two PVC parts will be everything that is beige in the CAD drawings except the cao at the bottom and top to connect the tubes to. The caps can just be a thin sheet of plastic cut in a circle with a hole drilled in it. PVC pipes can be as thin as about 1 centimeter outside diameter. So it could be small. Also if it needed to be cleaned (besides just turning it upside down and running water through it, it could also just have the tube pulled off and then be cleaned. Of course no parts here even cost close a dollar. And assembly should be very easy. With a transistor (and diode to protect against a back emf) all that would be needed would be a simple signal to turn on the coil.

male pvc.jpgpvc.jpg
By the way here are the male PVC pipe adapter and PVC pipe I am talking about.

Happy Father's Day!

[jjmouris]

Pete, any progress with your pump?

[Pete]

Hey jjmouris,
Haven't played with pumps recently, there's a few more fundamental issues to sort out before addressing pump improvement i.e. software efficiency and hysteresis in the damping system, both of which are improving at a great rate but still are not there yet. I am building the control and feedback mechanisms for the pump into my USB controller for the peachy so I expect when the bigger issues are sorted that my addition of these will be relatively quick and I'll be sure to post the results.

[Chayat]

Anyone else read the title of this thread in the voice of Keanu Reeves from Bill and Ted?

[jstrack2]

Here is a little video showing what I was talking about in the CAD files I posted last week. I put food coloring in so the water is easier to see:

I used a neodymium magnet I had laying around instead of iron since it had a nice shape, performed well and won't corrode. Therefore Here is a video of it going up and down:

It only uses about 50 mA at 5V. I used 36 gauge copper magnet wire btw, but a little higher gauge may be better. Also a smaller PVC adapter would be better, but the one I used was just what they had at a local hardware store. It has a diameter of about 27 mm. A tube can be glued to the top and bottom of it, or to have things sealed you can use a little PVC cap. That way to clean things all that is needed to be done is just pull the little PVC cap off and it can be everything can be super easily cleaned then. Since it uses so little of current a transistor is not needed to power the solenoid (even if powered by a microcontroller), although a diode should probably still be used to protect against backwards voltage. Here is a picture of the rubber ball glued to the magnet and a picture of the PVC with the solenoid glued on:

halfball.jpgPVCsolenoid.jpg

The total cost of everything I used is only a little over one dollar, and especially if bought in any volume could be less than a dollar. It should be very reliable too. Nothing should corrode (I think the rubber used in bouncy balls won't, but worst case things can be coated). If the solenoid was only powered for a small fraction of a second (instead of me slowly flicking a switch) then only a tiny amount of water would flow through. So the software could totally control exactly when and how much water flows through. I think that this will be cheaper, more reliable and more precise than using a pump. Let me know if anyone has any questions or would like me to show something else (like actually having the tubes attached for example).

[BrockMcKean]

It only uses about 50 mA at 5V.

That looks really simple and could be very effective. However, I think you should modify it to work around the current modulation scheme the peachy uses for mirror positioning. This would need to use FM, because AM is already used using it or PM would just be too complicated and costly in my opinion. Also, using FM directly should translate into an up and down motion in the magnet, so it should be able to easily control the actuation directly. Problem is, the line level of most audio output is very low, so I think there's really no way of getting around an op amp or some other kind of amplification, which also means you need a DC bias, and as far as I know there's basiaclly no way to get around this without using requiring separate USB power, batteries, or a transformer. A transformer could be used to directly power the bias from the audio output, but then you're at least doubling the cost and probably quite a bit more ( think) to achieve the Voltage and current simultaneously. So, in my opinion USB power would be the way to go here since it's already connected to audio and practically every source of audio is going to have some sort of USB connection. I'm not sure what kind of power mico and mini USB provide, though. So, powering this entirely from a device that doesn't have a standard USB connection might be a little bit tricky.

As for the magnet itself, it would probably help to have some sort of support system inside the pipe as well to provide some surface the magnet can press against when it is up so that the viscosity, pressure, and flow rate of the resin does not affect the orientation of the magnet very much either. Any ideas there?