Z axis calibration idea

[Anuvin]

Isn't that pretty much how it works now? The software lets you put in how many drips it takes to fill the container, and then the software counts the drips, and estimates the Z. What would the caliper do, in your suggestion? Are you suggesting that the software compare the Z based on caliper data to the estimated z based on drip count?

[jstrack2]

Yes the software would compare the two values, and with each drip comes up with a new estimate on the position. Without the caliper (or something comparable, although I really like the caliper idea!), just counting drips obviously leads to significant inaccuracies. Only using a caliper can limit the resolution. By using the Kalman filtering method you get the caliper accuracy and drip resolution. And this level of detail of position will be known with each drop. The Kalman filter is pretty great. It produces the statistically optimal estimate of the system state. For example it is used to determine the state of charge of a battery, and it is used with accelerators and gyroscopes in a phone to give accurate orientation readings. It would be sooo awesome for the Peachy Printer to be super accurate and have extraordinary resolution in the z-axis! Plus I think this would be pretty robust too.

http://en.wikipedia.org/wiki/Kalman_filter

[mike_biddell]

I like the Kalman filter Idea. I once use such a filter for a navigational invention.....very effective. We could start with a basic drip calibration (an expectation about what a drip is worth in Z terms) and every 0.1 mm, redatum that calibration. So in other words, the picaxe would receive the drip pulse (not the Peachy) and the Picaxe would put out the rationalised pulses on the basis of the accurate drip value.

[jstrack2]

Yeah I think this will work well. One detail (although this matters relatively little since it is a small fraction of 0.1 mm) is that I think that the calculation should be redone each drip. If it is calculated every 0.1 mm and otherwise each drip just adds a fixed amount then there will be a correcting jump every 0.1 mm, rather than a more smooth fix every drop. Again though this doesn't really matter much. With decent initial drip calibration I doubt this effect will be noticeable. Getting this right will make the printer slightly more robust.

Anyway I am real interested in hearing updates about how the caliper idea works for you. I was pretty surprised how cheap they are.

[Anuvin]

Ah, I understand now, thanks for explaining. That does make sense. Calipers are generally inexpensive, so this could really wind up awesome and cheap.

Call me an optimist, but I think that the baseline Peachy setup will end up being adequate for most users. I have been getting to get very consistent numbers with a cheap burette tip. http://labsuppliesusa.com/burette-tip-3e3620-each/ I only mention this for anyone reading the thread who is concerned that the provided kit won't be adequate.

I can't wait to see what comes out of this thread. Mike, be sure to keep us updated.

[jstrack2]

Yeah that's a good idea. If the drop size is consistent and the caliper & drip Kalman filter system is implemented then even if the drip rate varies the z position will still be known with extreme resolution and accuracy.

[jstrack2]

Of course as long as the drip size doesn't vary too crazily the Kalman filter will also clean that up nicely.

[BumpingSpheda]

in a sense, the argumant is moot,
as get better with our calibrations and others develop ways of fine tuning the laser and mirrors we would be able to get finer and finer resolutions possibly even the the point of nano. the whole purpose of this thread is to try and make the Peachy the most reliable 3D printer on the market even if everything we are talking about is to be done after market. to be able for the printer to accurately know what its current Z height is simply means you get that much more precise of a final product with less fussing around.

3D printing technology is constantly advancing, and there is no law that states that it has to advance in a certain order, if we can measure more accurately than our printer can print it just gives us an excuse to make the printer more accurate to match.

Badass.

What about something like a potentiometric transducer linked up to a tone generator? You have your float device, it's in between and closes the gap between two vertical traces, one of which is Copper (or some material with very low resistance), the other a different material/form factor with a known/calibrated resistance per unit of distance. The measured voltage determines the height of the float which is converted to a frequency that the tone generator can push to the mic-in.

[jstrack2]

I think that a potentiometric transducer would work as well. It might be easier than the caliper I think. I would prefer the caliper though all things being equal since I think it would be more precise and more importantly much more accurate. A weakness of the caliper's approach is it takes a relatively long time to make measurements. But this is irrelevant for the Peachy Printer, since the liquid level shouldn't rise to the top in a second or two! haha

As for sending the signal to the computer I think maybe the voltage could be just directly sent to the computer via the headphone jack just like the drip does. If using the caliper a microcontroller would probably be needed, such as a PIC. Some microcontrollers can be bought for only a dollar or two though so I don't think this rules calipers out. It's nice that there are multiple options!

[jstrack2]

Also adding a microcontroller should be pretty easy since the Peachy already has a circuit.