Printable View
Hi,
now that the first printers are shipping out, I'd have the question if anyone figured out how much overhang can be printed with the peachyprinter?
It's not really urgent (didn't order yet, and I guess if I ordered now, I'd have to wait for another year ;)), but I'm curious.
Thanks a lot,
Lukas
Unless I've got this the wrong way round. The peachy prints on the surface of a rising reservoir of water. so the print is supported and always enclosed by the water. I would imagine this would let you print almost out at right angles with out any problems.
And makes it way more versatile than standard resin machines that pull the model from the surface of the resin.
You can pretty much print anything you like with no supports. Can't do that on a formlabs printer :-)
Yeah I'm right :-)
<br>https://www.youtube.com/watch?v=V6T81zxlHgY
Thanks for the great question tinloaf and
the great answer from curious aardvark
Ill just add this... as its a great clip of the longest over hang print we ever did. the hole print is one big over hang! talked about here at 4:20
https://www.youtube.com/watch?v=M8V1pu8hoU8#t=279
Just curious, can it do an overhang that drops? I'm thinking of something like a small case r where the arm droops, so when the printer is printing, the first bit will not be attached to the it, until a few layers are made. If the printer isn't jostled can the two separate parts join up without a rough surface?
Hey iplayfast,
I'm pretty sure, that won't be possible.
As the resin floats on the water, a created Island will also float and not stay at the same height with rising water.
Also although the printer doesn't exceed any force on the resin, wind, surface tension (or some jedi) could make the Island drift away.
Fortunately this also enables the PeachyPrinter to use really small support structures as there is virtually no weight to be supported.
Considering overhang, You probably can print full 90°, thats something to test for one of the guys that already have a printer!
I hope, this answers your questions.
Greetings,
quertz.
Isn't the cured resin slightly more dense than salt water? if so any floating island would eventually sink when they get massive enough to overcome the surface tension of the water
You probably can't print a "perfect" 90 degree overhang simply because of the mechanics of the Peachy printer - the water level is rising continuously (the printer has no control over that) and it's going to take time to print a large, flat plane. As a result, whichever bits get printed last will be higher. For accuracy, I suspect the goal will be to print a lot of layers, with minimal time spent on each one - so that the height change during each layer is kept as small as possible.
Only to clearify this, you can't do it with the default dripper system. If you change the dripper into a controlled system, like a pump or something, than the peachy should be able to print a perfect 90° overhang.
Umm you sure about that.
I understood the even on the original model the dripper was controlled.
Either that or the laser has to cover the entire surface of the print vessel for every single layer. Which would set a specific print time per layer that would allow any part of the surface to be printed.
Otherwise it would be impossible to print different things on each layer - the very definition of 3d printing.
Either way you can print the entire surface per layer.
And the resin must be less dense than the water otherwise the whole thing just would not work.
Hey guys,
I'm pretty sure, Peachy's default dripper system is uncontrolled, at least not by electronics.
Using extreme numbers:
If the water takes 20min to rise 10cm (I guess, that's a fast, but realistic print), you get .5cm per minute and during those 10sec per Layer (see videos) the water will rise about 1mm per Layer.
The Platform being maybe 5cm wide, this gives you 88.5°.
Seems pretty close to 90° to me (As I said before, I think, these are really extreme numbers).
Making the dripper slower (I guess, .2cm per minute is a realistic value) you can get angles as close to 90° as you want.
Greetings,
quertz
I watched the assembly video a while back, and to me it looked like some heated rubber thing was wrapped around the dripping tube in the dripper. I figured that the printer can heat that thing to slow the dripping...
Can anyone with a printer confirm this?
Ohh napking math time :) ...Quote:
Originally Posted by quertz
here is how I would do it.
we have 100 layers per mm so each layer we move up 0.01 mm, and standard .7mm laser spot size ... lets say we need to over lap the spot by 75 percent each layer to make sure the layers connect( this is very untested)
so every 4 layers we move .7 mm over and .04 mm up and Im feeling lazy so ill go here http://www.cleavebooks.co.uk/scol/calrtri.htm
I get 86.7 deg
wow our napkin math answers are very close quertz!
here is a pick of a rook printed with the standard v1 kit printer:
Attachment 8611
The rook has stairs that have rather flat tops:
Attachment 8612
With the flat tops lit just right from behind, and a macro lens we can see what appear to be layers!
seeing individual layers is very rare... sometimes we see artifacts that look like layers but are some other effect like break over. In most situations our layers are to small for the human eye to see:
Attachment 8613
PS that last pic is my favourite peachy ever :)
The dripper is only "controlled" by a manual valve - the printer itself can't change the drip rate. As Synchron has said, if you replace the dripper with a pump then you can print a perfect 90 degree overhang, or maybe even more (in that you can run the pump backwards and effectively re-print earlier layers, as long as the parts you need to access aren't shadowed by parts already printed).Quote:
Originally Posted by curious aardvark
To be able to create absolutely any part (with overhangs to 90 degrees, but no more than that) with the whole surface being touched at every layer, you would indeed need to set the drip rate to a very low level so that the laser would have time to cover the whole print area. This is not practical. Instead, you make an estimate at how much of the area the laser will actually need to cover (probably something like 1%, since you'd normally just print a shell rather than a solid object) and set the dripper to allow that much time.
Because the layer height is so small, the laser can probably cure a couple of layers at a time - so if it takes three layers to get to every part of the job then it'll probably still be fine (although possibly not as smooth).
The challenge is when you do want a very large amount of stuff done on one layer, like a big flat surface. If you want to print fast, you need active water level control so it can slow down for this part and then speed up for others. With the dripper, you'll just have to turn down the speed until the result is acceptable.
So overall print speed can be optimized if we augment our Peachy with a micropump of some kind, and for some prints this can be a dramatic increase? Has anyone started trying to use a motorised pump yet?
I'm just new to the Peachy (learned of it from a friend), but AFAICT negative overhangs can also be done in theory, as long as you tether them with a (very thin) tether from the base (or a suitable part of the object)... As long as they're not completely free floating from the point they start at, it should work...
I'd love to see an option for a controlled pump in the peachy, and some kind of measurement device so it can keep track of the factual level, and not just the estimated level. Supersonic echo-location (distance measurement by high frequency sound wave) has become pretty accurate, and could be part of the setup in this.
It would allow the Peachy to control the rate of rise as well as reasonably accurately (not in single layers, but close enough) measure the actual levels. (If only to be able to assess the pump/drip rates as it prints...)
quick question re the accuracy of the measuring, is it able to repeatedly measure to the 0.001mm level or is it a LOT coarser than this?as with the drop meathod it is possible depending upon the container dimensions to get near to the 0.00001mm resolution
No, it is much coarser than that IIRC, which is why I mentioned using it as a way to maintain a continuous active calibration (And possibly use it for initial calibration too). It may not be able to measure 0.0001m, but it can average out that 50.000 drips is still 10cm (for instance)...Quote:
Originally Posted by harpo99999
Standard laser would be more accurate (up to 1mm approx, IIRC), but there are 2 issues with that: 1) possible unintended resin hardening (if there is a UV component) and 2) the resin may be too translucent for it to reflect reliably.
And of course, the more precise the measurement instrument, the more expensive it is... :(
I don't have all the answers, obviously, or I'd have made the printer myself! ;)
Just trying to think how we could improve reliability without breaking the bank!
(As a matter of fact I have plenty questions myself, and will start asking once I have read through the topic lists, so I know what has or has not been answered yet)
Hmmm... Interesting read I found when I started googling: http://www.olympus-ims.com/en/applic...ckness-gaging/
that was an interesting read regarding the thickness gaugeing but the article implied that it was mostly for testing solids, not delta distance in atmosphere to a liquid that would be usefull for the peachy
True, but it does seem a reliable measurement can be gotten that way. Possibly by submerging the sensor in the fluid, we could get better results (measuring delta is more important than the absolute level here. We can give it a one-time offset to compensate for the measurement difference too, so the absolute is corrected)
the main catch as I see it is the interfaces between differing materials ie salt water to resin to air and each having differing speed of propigation of ultrasonic sound, and assuming you are placing the sensor in the salt water to get the 'z' height then there is the corrosion issue for the sensor, and if my understanding is correct the higher frequencies are used for the higher accuracies/smaller distances
Why not stick two long copper probes close together and held vertically in the water and then measure the resistance? It should drop as an inverse function of fluid height as the resistivity of the salt water should be constant if it doesn't heat up too much from the resin hardening (which is exothermic) but the fluid wire crossection would be a linear function of height.
Why not use the laser and a floating mirror on the resin. Messure the time, the laserlight needs to a sensor.
(sensor)
|
|
/ (semitransparent mirror) - - - - (laser)
||
||
||
_ (floating mirror)
I think a capacitive sense would work best. Resistance is relative to salt concentration, and would require re-calibration every time. I could easily rig up a test with an arduino. you just need to tape a conductive strip to the outside of the container vertically, and submerge a corrosion resistant lead into the water. Gold coated PCB trace would be ideal, and cheap/easily salvageable. You can measure the capacitance which should increase linearly as the water level rises. You can measure it with an arduino using 3 pins, and two resisters, or using a 555 timer, to generate a digital modulated output which could be measured by your drip input on the peachy.
Hardest there would be to maintain the floating mirror's position!Quote:
Originally Posted by Synchron
Now there's a new and novel solution I hadn't thought of! Will it be accurate enough tho?Quote:
Originally Posted by jsondag
That's a good question. One that I think more relates to the parts used, and the resolution of said parts, than the method itself. I can test this maybe in the next day or two, just with some foil tape, and my multimeter, and see if that provides decent resolution. for a micro, the test involves charging the capacitor through a known resistance, measuring the time it takes, draining the cap, then repeating. The resistor value, and build of the "Capacitor" will need to be experimented with a bit.Quote:
Originally Posted by CescoAiel
Capacitance would depend on the dielectric constant which also depends on saltwater concentration so I don't think we've saved much by measuring capacitance instead of resistance.
Laser interferometer with a floating mirror is better but as soon as you add another laser and electronics that can measure time delays that small you significantly increase monetary costs.
The laser is still there, we could use the peachy laser... But you are right, the sensor elektronics have to be added.
At a previous job I built robots that used the Olympus ultrasonic thickness sensor to detect the thickness of metal plates. It needs a layer of water between the probe and the plate to achieve a good reading. I am not sure about corrosion although most of the probe is anodised metal. The end of the probe is also about the size of a dime.Quote:
Originally Posted by CescoAiel
The saltwater is the conductor, not the dielectric. The dielectric is the container wall. It will vary from build to build, but remain constant on a single container.Quote:
Originally Posted by amoose136
I tested this with just a multimeter as a proof of concept. I just dangled one probe in the water, and touched the other to the foil. Seems to work, and level of salinity has no effect whatsoever. I went from a weak salt water, to more than double the salinity, with 0 change on the meter. My meter clearly doesn't have enough resolution for fine detail, I never expected it to. A RC circuit on an arduino should give very high resolution, as you're converting capacitance to charge time, and time is something micros are very good at measuring.
I used HVAC foil tape, in both a thin strip and wider one. I've found that a wider strip gave me greater movement in readings. I also found that the surface tension plays a role, as you move up and the walls get wetted, if you move down there is backlash, because you're measuring how far up the wall is wetted basically. No big deal for measuring only one way.
I'm attaching a graph of readings at half centimeter intervals. (nF on the left, cm on the bottom) and a pic of my test container.
My meter is not high quality or very precise, so the little variation from linearity is most certainly from that.
Attachment 9034Attachment 9035
Wow, I really like this idea!
I think, even for those of us who already got their Peachy or will get a version without this idea (yes, I'm already assuming they will use it) this is a perfect DIY-addition to the Peachy!
They could easily provide a pin or two from the Peachy board and write a software version that uses the new sensor.
I like the idea that you can just make the tape on the outside bigger for more movement.
According to this:
http://www.nessengr.com/techdata/liq...dresistor.html
the saline will probably have a resistance around 10-100 Ohm/cm so if you use like 10kOhms for measuring the capacity you won't see a large change.
Greetings,
quertz
Good point, I didn't think of the resistance in the "leads" running to the "capacitor". The salt water will act as a resistor in the lead, and will be part of the resistance. As you say though it will be negligible, especially if you place the electrode in the water close to the strip.Quote:
Originally Posted by quertz
EDIT:
I'm assuming we are going to be using an amount of salt close to the limit of solubility, and in your link that comes out to about 4.8-10 ohms/cm for sodium chloride. If you put the contact right by the bottom edge of the strip, you're on the order of an ohm or less to the bottom of the strip. Resistance to the portions higher up will go increase, but again negligible. A wider strip also will help this as there is more cross sectional area of saline to travel through to reach the dielectric interface, lowering the resistance. I also think the current peachy will be able to handle this with maybe just a firmware update as you can either tie into an i2c for comms, or emulate a dripper. I think Rylan said there will be a way to do that at some point.
Looks great. I'll have try this myself and see how few components it will take to make this measurement accurately. Does the 1.0 kit have any spare GPIO pins or are they all taken?
Would just take a charge resistor of high resistance, to charge the "capacitor", and a low value resistor to discharge.Quote:
Originally Posted by amoose136
The large resistance needs to be chosen to charge your capacitor slow enough to measure accurately, and fast enough to be responsive enough, and not be too affected by stray emf. This will depend on your build likely. I'd have to experiment to know a good time to shoot for.
200 ohms or something should be good for the discharge resistor.
if there are two free GPIO pins and enough free cycles on the built in micro on the peachy that's all you need. The firmware on the peachy could do that measurement and use it to correct for z-axis drift.
Or you can use a separate micro.
You can communicate with one free pin, or you could even read your dripper with the separate micro, and have it do some logic to insert or skip drips to correct for it's reading. That would require no changes to the peachy.
EDIT:
I lied, you need 3 pins. You'll need an analog pin straight onto the cap between the charge and discharge caps, to set a consistent point to trigger that the cap is charged. Digital pins aren't consistent in their trip point.
D0----~R1~--\
.....................|..............Jar
A0--------------|-----------| |------\
.....................|..........................|
D1----~R2~--/........................ GND
D0 goes high, which charges the Jar capacitor. A0 waits for it to hit a specific voltage, say 67% of supply. When it does, the micro counts how long that took, shuts off D0 (sets high impedence mode), then sets D1 low which discharges the cap. When A0 sees cap hit low (~5%?) D1 turns off (goes to high impedence).]
.... and loop
Hey Nice Work!Quote:
Originally Posted by jsondag
We thought this was a great Idea to and back in 2014 we did a bit of work on testing out the viability ... it certainly looks like a promising way to get feed back on vertical movement. And unlike many Other methods It could be very very Inexpensive. If you can get the readings you need to replace the dripper with an arduino, then I can certainly show you how to connect the arduino to the peachy software, to close the loop.
The its been a long time since we worked on this and its a bit foggy ...so instead of try to explain it in detail im just going to post a big file with pics and video:There is about a gig of pictures and video so here is a goodle drive link:
https://drive.google.com/folderview?...Fk&usp=sharing
basically here is what we did:
1. used a peristaltic pump with a very make shift encoder to move water from one tank to another
2. laminated copper or aluminum foil with a typical lamination sheets found at staples.
3. Placed the laminated strip in the salt water so that the water acted as one lead to the capacitor, the foil was other lead and the lamination was the dielectric.
4 Created an oscillator that changed frequency as the capacitor changed in capacitance. Scott was a big help, he both made the Relaxation Oscillator and brought much wisdom to interpreting the results.
5 Graphed the frequency of the oscillator using python mat Lib library.
7. Thought about what kind of resolutions could be achieved by mapping the frequency to actual height.
Lessons We Learned.
Altho promising there are many things that make this solution harder than it seems at first.
possible problems:
-50 or 60 Hz noise can ruin your measurements each time you cross its harmonic.
-The feed back is not liner .. it is an inverse function of frequency ... I think
- oscillators drift around in frequency due to many factors that you dont want in your measurement ie temperature
- large surface arias help to even out the random way that a meniscus wets the laminated plate.
- The capacitances achieved were very small (in the 1- 10 pF range? I think ) ... therefore to get low frequency required charging the cap thru annoyingly large resistances .. over 10 mega ohms if I remember right.
- meniscus has large hysteresis upon draining quicky..( this is not a problem at the speeds we print tho)
Our goal was to get 10 micron resolution out of this system, at best we got more in the ballpark of 50 micron "resolution" .. not bad considering we only worked on it for about 3 days, but still that 50 microns came with a host of questions about accuracy and precision.
At the time of this work we were still trying to use the micro phone input ... to get the desired resolution we would have had to go below 20 hertz ( a typical cut off for sound inputs). I think using a micro controller to measure the frequency could solve this part of the problem.
I did a rather pour job of documenting this.... the pictures and video are all I have... I dont have micro code or graphing code. although it wasent very hard to create.
In conclusion I think this could be a very valuable way to get vertical feed back, It needs more work So Im glad to see some one taking it on.
Hope this post helps you along the way.
PS, everyone give James Townly a standing ovation for playing merry little lamb first try the resistor array and the oscillator.
Wow, that's a lot of stuff.
I think we should just open a new Thread here or on the Peachy forum.
50 microns seems pretty good to me, at least to compensate for long-term trends (using the dripper for high-res height-delta, the capacitance for low-res absolute height).
quertz
Ya good Idea quertz...
ok here is a new thread for talking about capacitive feedback of resin height:
http://3dprintboard.com/showthread.p...4282#post84282
In this thread, lets go back to talking about maximum over hang.
and remember everyone starting a new thread for a new topic is good for everyone.. it makes things much easier to find later!