Page 13 of 14

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 12:52 pm
by JeremyBP
I did end up finding good springs. I had some that came out of an old royal portable typewriter. They're about 4mm ID, and 16mm long.

I was looking through a lot of the test prints I've done when I noticed an odd pattern. On larger objects, the wobble was smaller, almost gone. It was much more defined in smaller objects. Also, objects printed at a higher speed had a more pronounced wobble. Hm. So I started looking for a cause that would be time-based. I started where the vast majority of printer problems stem from -- the extruder. And what I found was that I have +/- 5c swings in temperature, oscillating regularly. Tonight I'm going to properly tune my PID controls, and see if that helps at all. I do know that prints are very thermally sensitive, especially at fine resolution and high speed. So this might just be the key.

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 4:45 pm
by RSWeaverAz
I've been following Jeremy's posts as we all could have these problems as well and he's taken some of the risk out of the equation.
So I become aware of that there is a PID controller for both the hotend and the heated bed.
So i ask myself whats a PID controller that I have to tune, and I turn to the Wiki: http://en.wikipedia.org/wiki/PID_controller

Under section 6 Limitations of PID control, Linearity it discusses the issue:
"For example, in temperature control, a common use case is active heating (via a heating element) but passive cooling
(heating off, but no cooling), so overshoot can only be corrected slowly – it cannot be forced downward.
In this case the PID should be tuned to be overdamped, to prevent or reduce overshoot,
though this reduces performance (it increases settling time)."

This exactly describes the hotend used in the extrusion process, it suffers from no active cooling!
So I'm wondering if there's room to add a small Peltier Heat Pump (usually mounted to a heat sink)
to the hotend ... or would a FAN blowing over the extruder hotend do the trick?
Jeremy, Do you have a fan installed that can cool the hotend, I've seen this in several peoples setups?

Also as someone pointed out in another http://buildlog.net thread PID control could be taxing the controller that interprets the gcode.
Jeremy ,Is your PID controller a separate processor?

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 4:53 pm
by JeremyBP
I do not have a cooling fan.
The PID control is on the same processor, but that means nothing. It's a minimal process, and doesn't slow anything down at all.
Going back to the fan, it really isn't needed. Tuning is required regardless, and if I tune it properly I can make the hot end lock on to +/- 0.1c. It gets easier now that auto-tuning is more common.

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 5:03 pm
by mike
Do you have any extruder temperature PID tuning tips?

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 5:07 pm
by RSWeaverAz
Thanks, so its YAA, yet another adjustment or parameter setting(s).

Do you know if anyone tried using peltier heat pumps on a hotend?
This site has some with a center hole: http://www.ferrotec.com/products/thermal/modules/centerHole/
my thoughts are it probably would be to slow,
but peltier heat pumps can do both cooling and heating by changing the polarity.

P.S. I suppose pushing room temperature filament into the heating chamber could be considered a "cooling" effect.

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 5:16 pm
by crispyfry
A heat pump is going to add yet more mass to the moving hotend, even if the heat exchanger is located elsewhere. In my experience I've seen no need for a way to dump heat out of the hotend faster than free-air convection does.

Re: Jeremy's ORD Bot Build

PostPosted: Mon Mar 26, 2012 5:22 pm
by JeremyBP
I think the best thing for pid tuning is the new marlin firmware. It has a new command, M303, which auto-tunes the PID. I'm going to try it tonight at HACDC and see how it works.

Re: Jeremy's ORD Bot Build

PostPosted: Tue Mar 27, 2012 2:45 am
by JeremyBP
Well, I'm blown away. I set up at hacdc tonight, plugged in the printer, and sent an M303 command. Five minutes later, my PID had tuned itself. It now comes up to temperature and locks on. It's amazing.
In terms of quality, initially things look bright. Now that my extruder temperature is under control, I need to change my temperature settings. That said, an initial print of Colbert's head looked good, and showed no wobble. When I print next at hive76, I'll try something more linear and boxy to really see how the results are.

Re: Jeremy's ORD Bot Build

PostPosted: Tue Mar 27, 2012 3:58 am
by mxk
Really? No (noticeable) fluctuation? I'm gonna have to try that on my new RAMPs board, my Mendel with bang-bang temperature control bounces two or three degrees around the target. I finally installed the new Slic3r and Pronterface tonight but I've been hesitant to upgrade Marlin beyond the older beta I have because it just works. Sounds like you've identified my justification for making the leap.

Re: Jeremy's ORD Bot Build

PostPosted: Wed Mar 28, 2012 1:40 am
by Enraged
I just wanted to show the quality of parts, ORDbot vs Reprap.

Disclaimer: Now obviously, there are setup differences and the files aren't exactly the same, but you get the drift. The black parts are printed on Jeremys ORDbot, the natural ones are printed on a reprap by an ebay seller. Jeremy did these for me before he further tweaked his machine, so I bet his results are even better now.

First, the big gears.

Photo 2012-03-27 6 13 43 PM.jpg


Photo 2012-03-27 6 13 53 PM.jpg


Photo 2012-03-27 6 18 23 PM.jpg


As you can see, the black gears are nice and smooth, with the actual teeth very well defined. The black parts teeth are also "square" to the face, while the natural parts are shifted slightly.

Next, the small gears.

Photo 2012-03-27 6 15 00 PM.jpg


Photo 2012-03-27 6 15 17 PM.jpg


The black part has well defined teeth, and you may notice it actually sits flat when placed upside down. The natural part is shifted a bit.

Next, the main extruder body.

Photo 2012-03-27 6 17 14 PM.jpg


Photo 2012-03-27 6 17 29 PM.jpg


Notice how clean the edges are on the black parts. I just need to drill the holes for the motor and the pinch bolt, and it's ready to go. It's hard to tell, but the natural part bolt holes are basically unusuable without serious work, while the black parts allowed the bolts to drop straight in.

Finally, the bearing holder.

Photo 2012-03-27 6 14 08 PM.jpg


It's hard to tell from the picture, but the filament guide on the natural part is shifted, rendering it useless. Also, I could not fit the bearing in, while it snapped right into place on the black part.

I think the ebay seller has some more tweaking he has to do, as most of the parts were shifted slightly, so that could be due to a heated bed, warping, etc, I'm not entirely sure.

Overall, I am very impressed with the black parts that Jeremy printed for me. I just wanted to say thanks again, and I can't wait to get my Hadron so I can start printing stuff like this!