Any prints from the Beta team?

[User_Defined]

Those updated prints came out wonderful, and this is already well within 90% of usability!

My biggest question now from the latest update is this:

The layers seem to be small, but I can still clearly see layers/artifacts in the prints. Do you know what the reason for these artifacts is? Could it be laser accuracy? I know the peachy is not the Form1+, and I don`t expect it to be as good as such an expensive printer, but I am very curious about what the reasons are for the difference in print quality.

For example, remember the column prints? They looked almost like glass. What is preventing the current prints from printing like that?


If it is simply laser accuracy, this can be resolved with a linear magnetic angular feedback IC. It would take your input voltages and then tie them directly to an angular output! Many of these sensors have analog outputs, so you could theoretically just make an analog error amplifier with your existing coil driver amplifier. These chips are under $2 and have only 3 pins! I mentioned all of this earlier, and let me mention once more that this could elliminate the need for magnetic dampening. Just make sure to get a medium precision IC.

If it is not laser accuracy, maybe the forum members can help brainstorm and throw ideas on the potential causes? I remember Jstraw (hehe) saying his laser spot was shaking randomly, but this was resolved?

-UD

[rylangrayston]

Those updated prints came out wonderful, and this is already well within 90% of usability!

My biggest question now from the latest update is this:

The layers seem to be small, but I can still clearly see layers/artifacts in the prints. Do you know what the reason for these artifacts is? Could it be laser accuracy? I know the peachy is not the Form1+, and I don`t expect it to be as good as such an expensive printer, but I am very curious about what the reasons are for the difference in print quality.

For example, remember the column prints? They looked almost like glass. What is preventing the current prints from printing like that?

-UD

This is exactly what were working on today at peachy. There are many theory’s we have, We know that the peachy can do glassy clear prints with g code now because we have finally seen it do that!
Ill be talking about it more in an update soon.

We found out that one of the settings we used in cura has a bug that causes the laser to blink off for a very short time in between nearly every g-code! Thats what was causing the 21 spots on the owls back.
once we fixed that we got an owl with about 500 layers of glassy clear printing at a time, then every 500 layers there was an ugly spiral or ring up the print.

My theory is that the spiral is caused by the laser on off, we spiral the spot that the laser turns on and off as we go up the print.
others here think that its not a spiral, its a ring, and that its the furnace or the fridge turning on periodically.
We will know soon!

Im confident we can get much closer to what those columns look like with g-code! Time will tell.

I Really like what your suggesting with the Haul effect sensors.. last time we looked into this I think we found that the sensors themselves had to much Hysteresis.... can you recommend a chip that you think would work?
Basically what your suggesting is an analog PD or PID system with magnetism as the source of the feed back right?
That could work reallllllly well!

[Feign]

They've already mentioned that the Peachy so far is high precision, but low accuracy. I wouldn't mind seeing the exact same g-code run on two identically built and calibrated Peachy printers, to see just how different they are.

An interesting thing just occurred to me: Could it be possible that the neodymium magnets they are using don't have exactly the same amount of magnetic force? If so, then the equation for voltage in to angle out might vary from printer to printer. Calibrating at just one size might be insufficient if the strength of the magnets is an assumed number. Also, is the output to the coils really a smooth curve as it increases? Even tiny fluctuations could account for the apparent roughness of the final print. Most likely the root of this goes all the way back to the sound card (which makes it a problem that has been known for quite a while time now).

[User_Defined]

sorry for the late reply, off topic: I just replaced my windows XP with LUBUNTU, it is a crazy learning curve but totally worth it!

When I suggested this I was doing a quick search on digikey:

Product Index > Sensors, Transducers > Magnetic Sensors - Hall Effect, Digital Switch, Linear, Compass (ICs)


I use digikey for small orders, yes it is more expensive but they have amazing stock and the website is very organized. Once I find my parts, if they are too $$ you can cross reference on www.findchips.com for cheaper competitors.

[edit] Wrong link here, but it was a digikey link to a magnetic Hall sensor. I lost the link but since the following sensor is better, you can find it if you are curious!

- Hysterisis error is from temperature cycling (not magnetic hysterisis)
- symmetry error only applies if you are flipping the magnetic field polarity during operation.
-ratiometric error is also irrelevant since your supply voltage is fixed.
- Linear sensitivity seems like it will be the only relevant error, and it is only ±1.5%

1.5% is still too high in my opinion though... but this is a $2.31 chip.


Diodes Inc sensor looks better:
http://www.diodes.com/datasheets/AH49E.pdf

the linearity of span is 0.7% which is better than the allegro. I think that this if the chip I would try! In this rare case I think the through-hole version is better since you can easily manipulate its orientation to fit the field on the sense magnet. (maybe eventually make a circuit board for it. Be sure to have it on the opposite side of the mirror driving coil!

How to use it:

Ive never done this before, but I would start by looking how a magnet could be connected to the mirror axis, and then look how the sensor would need to be placed to have that field move in such a way that the sense would see it.

You might need to do a bit of reading and look at some diagrams on how to use these sensors.

After the system is set up with glue or tape, you would then run the thing to see what kind of an output you get! Then feed that to an amplifier to get your proportional feedback signal, and send it back to your main amplifier.

From there youll have to use the Zieglar Nichols method to figure out the PID values.

http://en.wikipedia.org/wiki/Ziegler...Nichols_method


I know it may seem like quite a bit of work, but I have a feeling it would be worth it if you are having significant accuracy problems. At the very very worst case, you develop it for the pro version and then open source the circuit so I can copy it and add it to my peachy

good luck and dont be shy to shoot me a line if you guys get stuck.


-UD

[rylangrayston]

sorry for the late reply, off topic: I just replaced my windows XP with LUBUNTU, it is a crazy learning curve but totally worth it!

When I suggested this I was doing a quick search on digikey:

Product Index > Sensors, Transducers > Magnetic Sensors - Hall Effect, Digital Switch, Linear, Compass (ICs)


I use digikey for small orders, yes it is more expensive but they have amazing stock and the website is very organized. Once I find my parts, if they are too $$ you can cross reference on www.findchips.com for cheaper competitors.

file:///home/frank/Downloads/A1324-5-6-Datasheet.pdf
- Hysterisis error is from temperature cycling (not magnetic hysterisis)
- symmetry error only applies if you are flipping the magnetic field polarity during operation.
-ratiometric error is also irrelevant since your supply voltage is fixed.
- Linear sensitivity seems like it will be the only relevant error, and it is only ±1.5%

1.5% is still too high in my opinion though... but this is a $2.31 chip.


Diodes Inc sensor looks better:
http://www.diodes.com/datasheets/AH49E.pdf

the linearity of span is 0.7% which is better than the allegro. I think that this if the chip I would try! In this rare case I think the through-hole version is better since you can easily manipulate its orientation to fit the field on the sense magnet. (maybe eventually make a circuit board for it. Be sure to have it on the opposite side of the mirror driving coil!

How to use it:

Ive never done this before, but I would start by looking how a magnet could be connected to the mirror axis, and then look how the sensor would need to be placed to have that field move in such a way that the sense would see it.

You might need to do a bit of reading and look at some diagrams on how to use these sensors.

After the system is set up with glue or tape, you would then run the thing to see what kind of an output you get! Then feed that to an amplifier to get your proportional feedback signal, and send it back to your main amplifier.

From there youll have to use the Zieglar Nichols method to figure out the PID values.

http://en.wikipedia.org/wiki/Ziegler...Nichols_method


I know it may seem like quite a bit of work, but I have a feeling it would be worth it if you are having significant accuracy problems. At the very very worst case, you develop it for the pro version and then open source the circuit so I can copy it and add it to my peachy

good luck and dont be shy to shoot me a line if you guys get stuck.


-UD

Ok very good advice. We just may order that chip and give it a shot, I was wrong, historeses was not the isue last time we looked into it, it was how to actually position the chip in the field, but I have more ideas for that now ... I dont think its too much work( its definitely worth it!) , Scott has already made an analog PD system that acts on capacitance feed back for the pro, its on a bread board at our maker space right now... we may have to add I to it, but other than that we sould be able to calulate, change some values and just use the same circuit.


Thanks again this is a very good lead.

[User_Defined]

Im truely glad to be of help!

I just saw your ted talk and I am thrilled that so many people on this forum, as well as beta testers are helping to propel this project to its amazing potential.

It would be really cool to have some closed loop mirror system in place on the standard peachy, then really all that would be left for extreme accuracy would be proper Z sensing!

As I said those latest prints are already well worth the base model value, and definitely usable IMO. I really cant wait to see what the final fine-tuned version will yield. If you do end up going closed loop with those little chips, I expect great things, and I expect peachy to have the amazing problem of keeping up with orders

keep it up, and thanks for keeping us forum members "in the loop"!


-UD

[jstrack2]

Here are some birds that I printed and then my sisters and I painted for Christmas.

[jstrack2]

Here is another picture of them hanging:



Here are three other birds:



Finally here are all of the birds prior to being painted:




This is different resin that is a lot thicker. I am having some trouble with it because of this, but I think I will soon be able to deal with that.

[jstrack2]

Also here is a T rex that I printed:

It isn't yet perfect, but I think that prints are getting better!

[MagicDan]

Can you post a picture with something in it for scale? (Coin, paperclip, etc.?) It looks awesome! (T-Rex scale)