picclock's modified i3 3d printer attempt

[picclock]

Hi All

So having got to the stage where I can print some stuff I thought I would have a go at designing a new X carriage and printing it. It will incorporate a servo for automated bed levelling/calibration, a longer std Lm8Luu for the upper bearing and a std LM8uu for the lower one. I decided to use Designspark Mechanical because it is the right price (free !!) and simple to use - a definite requirement.

As you can see from the pictures I made the rookie mistake of generating an unsupported edge at the bottom which came out straggly. The part shown is still attached to the raft which will break away easily if I decide to use it.

I have also decided to scrap the underperforming heater bed and replace it with an aluminium plate heated with resistors. See next post.

Thanks for your interest

Best Regards

picclock

[picclock]

Sadly the above part did not conform to the drawing  . There appears to be a fixed extrusion error which makes all the parts 0.14mm larger than they should be in the X and Y direction. This is not a scale error, but a fixed addition. So that whether the part is 10 or 30mm long the length will be increased by 0.14mm. This also makes slots or gaps approx 0.28mm smaller. The part was made with ABS, a 0.4mm nozzle (Std E3D V6) and a 0.2mm layer height. I have tried adjusting the extrusion rate (Cura Flow) to 80% but this just seems to create weaker parts, with minimal effect on size change. The positional accuracy of the nozzle is pretty good (X is amazing < 1 thou deviation over 20mm), see attached pdf. So I think its likely that the slicer is not correctly calculating the extrusion bead thickness. I have checked the filament diameter, and its very good, varying between 1.69 and 1.75 mm. I have also checked that the correct amount of filament is being extruded by the firmware, and its spot on.

Any thoughts or ideas or fixes ( ) much appreciated.

Best Regards

picclock

[PeterE]

picclock,

I am inclined to go along with the answers in Pete49's thread in that the oversize is due to good accuracy 

What I mean is that your extruder nozzle lays the melt string at the center of the line stated in the program but it flows out a bit on either side then making the part ever so slightly wider than was expected. So like already suggested, try to compensate that by decreasing the important measures according to the overflow to give a detail taht is spot on.

You seem to have very consistent measures so accuracy is really not a problem. It is just necessary to compensate for the true width of the melt string to get items on spot.

One of those parameters to compensate for that was not obvious from the beginning.  Oh how much one learns from all these threads.

BR

/Peter

[picclock]

Hi PeterE

I'm pretty sure the slicer is supposed to compensate for the extruded bead width. I tried  the craftware slicer on one of my test pieces printed with 0.4mm nozzle and 0.2mm layer height as before, and although for some reason it failed to complete the print, when I measured the base it was pretty spot on, 19.99x 20.02 ~ so they clearly got that bit right. It sort of confirms my guess that the slicer is causing the issue.

Think I just need to find the right slicer. Am going to try slice3r, kisslicer, repsnapper, and any others that come to mind, feel free to recommend an alternate.

Best Regards

picclock 

[PeterE]

I see. I downloaded a copy of Repetier Host as I have heard good things about it, and if I remeber correctly it included the slice3r. A program I also saw mentioned in a copy of 3D make magazine. Supposed to be quite OK to use.

Then about the mis-settings in your current slicer. Could it be that there is a setting possibility to adjust the extruder centerline so it follows the center of the path rather than the outer edge - which is how I get the impression it does from your description. (Odd description but I hope you can follow my thought.)

/Peter

[Pete49]

picklock just a thought re the glass bed .When the bed on the Da Vinci chipped (cheap glass) I got a replacement borosilicate with the heater attached from a guy in Dallas Tx for about $Au70 including shipping. I am thinking about getting 1 for the CTC kit printer as it will heat the whole plate without needing to worry about it not being uniform heat.
Pete

[picclock]

@ PeterE
I've just done a run through on some of the other slicers. Some are a pain to use or have other snags. Slic3R is OK, but the dimensional error is much larger than Cura ~1.6 times. However the finish is the best I have seen. It also seems to round all the corners, which looks very good, but not suitable for accurate parts. Cura have a new beta version which I may have a go with as it seems to offer lots more controls for tweaking. My current thinking is that the error may be fixable by specifying an oversize nozzle in the slicer - say 0.43mm - which will hopefully alter the final size. I may have another go with craftware to see if I can get it to finish the test piece as it seems by far the most accurate.

@Pete49
Do you have a link for the borosilicate glass/heater site ?. I know that borosilicate expands far less with temperature, which causes less stress on the glass when heating, so hopefully no more cracks. Its thermal conductivity is still fairly poor, so it will not do well at spreading the heat. With 200x better thermal conductivity even a thin layer of Aluminium will be effective. Have read differing things on the internet about borosilicate durability though, with some suggesting abrading the surface to get grip. To  to avoid warping (nylon  ), especially on larger parts, would seem to require an enclosure to prevent localised cooling from drafts. I have thought about using a balsa wood rim extending about 1" above the edge of the printbed to try to contain the hot air (or at least keep out the side drafts) as the bed moves. Problem is that hot air is less dense than cold, so it will always try to escape upwards. Heated enclosure anyone ??  Tricky stuff !!

Good luck with solving your new printer issues.

Best Regards

picclock

[Pete49]

I can't find my invoice with the details but here is his email address jdholbrook33(at)yahoo.com
If that doesn't work here is the link I used to contact him originally
http://www.soliforum.com/topic/9642/da-vinci-borosilicate-glass-with-heater/
Hope it helps
Pete

[picclock]

@ Pete49
Thanks for the info, I'll see if I can track him down.

Slicer Success  (finally)
Using the new Cura 2.1 beta slicer, (it has some bugs). I have found an adjustment for horizontal expansion which allows calibration of the XY sizes. Tried it with -0.15mm and the test piece came out 0.3mm smaller than the original oversize one - so it just needs half to correct the overall size error. Also tried another tricky piece and it worked just fine. Quite a relief after all the slicers I have tried.

Hope this helps

Best Regards

picclock

[picclock]

Hi all

So I have my printer up and running (mostly). I have replaced the original X carriage and extruder, and fitted a bed levelling servo, all very worthwhile. The extruder upgrade allows instant changing of filament and jams are a thing of the past. The bed leveling system is worth persevering with and works really well. Main snag I found was that the cura slicer preamble inserts an G90 absolute positioning code which destroys the bed levelling settings in marlin - so I edited the cura definition files to remove it.
 
But (there always is one  ) the performance is poor by my standards, probably aggravated by the use of ABS, needed for its higher temperature stability. The problem is warping of the part caused by the differential expansion of the plastic. The initial layer is stuck onto the bed by a combination of heat, pressure and a certain amount of magic  in the form of ABS juice, PVA, hairspray, or whatever works for you. Bed adhesion does not seem to be a problem using just dilute PVA, and if you add ABS juice to it you will have a very hard time removing the part, however I digress. Once the initial layer is attached the next layer is added. The temperature of the subsequent layer is determined by temperature of 1st layer, surrounding temperature, and radiation/convection of the same. Subsequent layers therefore experience a different temperature depending of the heat losses of the layer, layer size (more time to cool for larger parts) to which to adhere. If the plastic is extruded from the nozzle at 240C, and the temperature at which it hardens is 100C then the surface temperature to which the subsequent layer attaches is very poorly defined. Once removed from the bed the temperature of the layers will becomes equal and warping occurs, because at the same temperature the layers are different sizes. It always seems that the bottom layer becomes convex which implies that the temperature of the plastic when deposted, bonded and hardend is higher than that of the bed, presumably because the heat maintained in the previous layer is greater than the bed temperature. So the upper layers shrink more and warping occurs.

ABS has a glass temperature of around 100C so one would think that provided all the layers were bonded at that temperature warping should be a thing of the past. I have already taken steps to guard the print bed from drafts but this does not seem to have improved things much, despite having an internal temp of ~ 40C.

It occurred to me that it may be possible to solve two problems with one solution. Instead of using a heated bed which adds to the Y axis movement mass, why not heat the whole interior of the machine to 100C using circulated air to even out the temp differential. Could use heat bed controls with fan for circulation. This would eliminate the need for an underglass aluminium heat spreader as the whole interior would be at one elevated temperature. To prevent heat damaging the extruder, belt, motor etc a plastic oven bag or sliding fibreglass/srbp heat seal could be used, attached just above the print head as a possible solution, minimising heat loss.

Still at the thinking stage on this so if anyone has tried this or has any further thoughts on this solution please contribute.

Best Regards

picclock. 

[efrench]

Are you removing the part before it has cooled to room temperature?  Try using a larger brim and/or adding pads at each corner to help it stay attached to the bed.

For ABS, I start the first layer at 240 degrees for the hotend and 115 for the bed, then drop these to 230 and 100 for subsequent layers.  My bed is 8mm cast aluminum with a 2mm glass mirror on top, so the actual bed temperature will be somewhat less.   I'm using ABS juice and if the print shows any signs of lifting, I'll give that spot a little squirt of the juice being careful to just get it on the brim.

I think it's more important to eliminate drafts than it is to heat the entire chamber. I'm using this to enclose the printer.  I printed a few brackets to attach it to the aluminum extrusion. 

[picclock]

Hi efrench

This is not a problem about bed adhesion, though I found that tricky at first. I'm using 100 or 110 for the bed and 260 for the filament - which may be a bit high but it generally seems to work well. I usually print with a brim and I don't get corner lifting, or if I do its very minor. Some of the best results seem to come when using a raft which effectively throws away the first three layers. The trouble with using a raft is that the finish is poor on the detached face.

Using PVA and a light smear of abs juice attaches prints almost too well, but that is not an issue. I always wait for prints to cool but once released they start to warp. The upper individual strands are under tension, the lower under compression and the only thing resisting this change is the stiffness of the material. I did a test with a heat gun warming a part and the warp increased from about 3mm to 7mm over 100mm.

Clearly this is an issue that needs solving. I understand that Ultimaker uses a fan blowing temperature controlled air across the top layer during printing as a way to improve this issue. Having thought about the heated chamber idea a bit more it occurred to me that I could do something similar and attach it to the print head. By attaching a conical hot air duct (made using soldered tin can) to the nozzle and heating the air with a couple of 40W cartridge heaters with a heat exchanger, I could use the waste air from the head cooling fan to cool the extruded bead in a controlled fashion, hopefully eliminating (or at least minimising) the problem. It should be possible to control the air temperature and set it just below the glass temperature of the plastic being used. Ramps and marlin allow for a second extruder so the temp control is already there with the addition of a thermistor. Not sure if that makes much sense but it is very much an idea in progress 

Best Regards

picclock
 

   

[efrench]

Have you tried normalizing the print with the heat gun before removing it from the bed? It might work.  I haven't had any warping problems when the print stays attached to the bed, so haven't tried this.

I had a jamming issue which turned out to be lousy diameter control on some cheap filament.  Before I discovered the source of the problem, I increased the temp to 260. While this lessened the jamming problem, it didn't cure it.  After putting on better filament, I printed a temperature calibration object (just a tall single wall rectangle). I edited the gcode to change the temperature 5 degrees every 10mm.  After inspecting the print I lowered the temperature to  240/230.

[picclock]

hi efrench

The reason for 260C temp was to try to improve layer bonding. On tall objects there are visible gaps in part of the layers which start typically at 2-3 cms and occur every few cms after. I suspect the layer bonding and object warping are all part of the same problem. Thanks for the idea of heat treating the part before removal. It may well be a simple, viable solution and I will give it a try. I am not overly keen on it as it may introduce other dimensional errors but its certainly worth a shot. I suspect the trick is just to supply the right temperature for the right amount of time .. .   Also thought I might experiment with a heated chamber using my heat gun as a 1 off to try it - probably about 100C should do it.

Best Regards

picclock

[Pete49]

picklock just a quick question if I may.  Now I have my kit printer working well  I was looking at the miles of wiring and wondered if I could cut it to a more reasonable size, not the motors they seem to be just right but the other wires such as the sensors and limit switches.. I was hoping to cut and resolder them rather than have them coiled up but that seems to be the norm. Would this affect the components?
Now returning you to the previous program errr thread. 
Pete

[picclock]

Hi Pete49

By all means shorten the leads to suit your layout, there are no technical issues with this. Just be sure you don't make them too short  , allowing enough length for the X and Y axis extremities. Two ways, best is to get a ratchet crimp tool and some 'dupont' crimps for the black receptacles on the ramps board, unclip original (use pin), cut wire to length, crimp new connector, reinsert into receptacle. Easiest way is to use some heatshrink sleeving 2 or 3mm, cut wire in an appropriate place, shorten to length, put heatshrink over one wire to be joined ~ approx 12mm. Strip and tin ends of wire to about 3mm. Hold one wire in vice/clamp etc.,  and then hold the two tinned sections adjacent and parallel. Just touch them with the soldering iron and surface tension will do the rest. Slide heatshrink over the joint and either use a heatgun, lighter, or the hot part of soldering iron under the tubing and it will shrink to size, supporting and insulating the joint. Heatshrink kits can be bought for a few pounds on ebay, and are useful for many things.

Good luck

Best Regards

picclock 

[Pete49]

Thanks for the reply. I have lots of heatshrink to fit the wires so the next project is going ahead.
I made a cardboard surround lined with alfoil to retain the warmth which works well and the power unit and Arduino board are outside of this to avoid heat related complications. I decided on cardboard (the type with 3 layers 2 smooth and an inner of corrugated cardboard compliments of her new TV) due to lightness, ease of access and the fact that our mild winters don't require extra padding. The only thing I felt needed extra work was making a base for it to sit on so I could secure the printer to it to make the print bed less flexible.
Pete

[picclock]

Hi Pete49
The base of any accurate machine is the most important part, as all accuracy depends on its stability. I used a granite kitchentop for mine, granite being commonly used for accurate measuring systems. For this sort of stuff most decent non temperature/humidity affected materials will work OK, and best of all, when you set stuff on it, it accurately stays there.

Good Luck

Best Regards

picclock
 

[Pete49]

Hi Pete49
The base of any accurate machine is the most important part, as all accuracy depends on its stability. I used a granite kitchentop for mine, granite being commonly used for accurate measuring systems. For this sort of stuff most decent non temperature/humidity affected materials will work OK, and best of all, when you set stuff on it, it accurately stays there. picklock

These are my thoughts too and the reason for wanting a base to solidify the machine. At the moment the heater bed section can move about 6mm either way and while it hasn't moved during printing yet I prefer it not to be able to move at all so a good base is my next move. The wiring mess has been brought under control while I wait for my new 2560 board. I haven't found the cause of the reason it went dark but I'm still looking
Pete

[picclock]

Hi Pete49

I had that happen to me early on. Turned out the on board regulator chip had died. I think it might happen when you plug it into USB (3?) whilst its still powered up. If the power to the USB goes the board tries to power it and it overloads the chip and its game over  . I replaced mine with a 7805 (see piccy) and once working then proceeded to blow up a different part of it by accident  . After that I ordered two, one for a spare.

Best Regards

picclock

[awemawson]

Neat leg bending on that regulator 

[Pete49]

Ta I'll check that as I'm sure I did something similar. How do I check it? I can sympathise with you blowing anther place. I managed to kill the Da Vinci by shorting the thermistor by cleaning it with the supplied brass brush while the power was on.  They don't have a fuse as such just a o value resistor which I fixed by removal and a wire across the gap and put a 4A fuse in line. fired it up and the fuse blew now looking at finding that fault.   No rest for the wicked at the moment.
Pete