So i just finished my @nop_head mendel90, the kit is good, but i think it needs some updating. There are a number of things i plan on changing, but i wanted some input as almost all of you have more experience then me.
I am planning on changing:
1: z home at bottom (homing directs the printer into a position that if the switch fails will break the printer, homing at top has a number of other problems, but that one is the one i don’t like the most)
2: using the nut traps and interlocks similar to the printrbot in the cnc cut panels. (cutting these panels and using printed brackets is extra steps that i think are unnecessary, by using the panels we have already added the cutter, why not run with it)
3: safety fall off at the top (currently if the z axis hits the limit without the switch working the printer will harm its self, by getting rid of the top home, you can make it safely un-thread the z axis when it hits the limits )
4: Switch to 1.75mm filament (direct drive or bowden, i am not sure yet)
5: fix the fact that the wades tension nut hits the screw box when not tightened enough.
6: switch electronics to something that doesn’t have screw blocks, taking advantage of the ribbon cables at both ends would be a plus ( i hate screw terminals with a passion)
Let me know what you think, i do like the over all design, i think it needs some updating though.
Doesn’t the Mendel90 use MDF/particle board? If so, the nut traps that PrintrBot uses in its plywood parts would just break the particle board wood apart.
Currently the kit i am using is using dibond (aluminium clad with polyethylene core) but there is no reason why it couldn’t use a suitable material for these nuts, it would decrease the number of printed parts by 12 and screws by up to 24 probably more like 15 or so, but still it would increase throughput and build speed (maybe even stability because of the interlock)
@Ashley_Webster by moving the z-stop to the bottom you have a little knob that you turn instead of having to recompile and upload marlin for every fine adjustment.
If it is off by .1mm, turn the screw 1/4 turn, if it is off by 10 mm, give the screw a few more turns until it is close. Homing takes less time, to set the Z level perfectly, takes less time then it took me to get the rough home.
1.75 mm has lower back pressure, you don’t need the giant gear reduction system in the wades, also i have allot of 1.75mm filament around from my other printer. I think i am going to do a bowden setup because i want to eventually do multiple extruders, and there is no way you are getting 2 wades on this frame.
The frame would be the same, it would just take less time to build.
I have finished the openSCAD mod to make the top un-thread, now i just need to add the holes for the z-stop tuning peg and i will be set. (then i have to print it and use a normal version of marlin to run the machine)
@Ashley_Webster In the normal convention, the limit switch would be down by the bed (i will put it on the left z motor bracket) and a screw comes down and pokes the switch. If this fails then the head will crash, but because of the way the tower is designed there will be no damage. in the current build if the home switch fails then you are sending it to a crash that could harm things. (it did actually happen to me, one of the wires was not properly inserted into the screw terminal, i had a quite nasty crash, nothing broke as i didn’t have the top completely tightened down yet)
if you’re recompiling and uploading firmware to change your z-height, you’re doing it wrong. Get it close in firmware, and fine tune from the slicer.
screw terminals may be kind of a pain, but they last, and are more reliable than any other type of connection in this type of high vibration environment.
I’m not familiar with the printrbot nut cutouts, but it seems more trouble than its worth.
@Anthony_White meh, I would rather turn a nut then deal with this in software.
I would think d-sub would work, the screw type terminals and I have a history of hatred, yes they do work, I am not as against them as I am against homing at the top.
Basically it is slot and peg joints with screws sprinkled in to hold it together. If you do it right it makes things much easier to assemble. The pegs hold themselves together, the screw locks it down. I need to find the design link for those.
I once would have agreed with you on tuning - but you can’t turn a nut to create an exacty .1 or .05mm difference, which you can in software. After doing a print and measuring the skirt with calipers, I can adjust to get exactly the thickness it should be with a change in the slicer z-offset.
As a mechE I understand what you’re talking about with the slot and peg joints, but I think the current method is quicker to assembly and provides equal if not greater rigidity, albeit with a few more parts.
If you home at the bottom, why can you still adjust the offset?
I think you meant to ask why can’t you still adjust the offset in the slicer - and of course you can but the position of the nut is subject to change - from vibration, careless bumps, or whatever.
I’ve “crashed” my z axis upwards several times when setting up a different print controller, and while it was scary no damage ever occurred - I was able to cut the power before the top brackets were stressed to the point of fracture.
Just a limit switch’s top and bottom it solves all problems but if you auto level do you need a bottom limit switch?
I think the ultimate solution to the “home up or home down?” question is “Home up, probe down”.
Your probe finds level, but will neccessarily have some Z-offset from the nozzle. For which using the current software tweak of offset is probably best.
Having to upload firmware to change Z-Home offset on a nophead mendel90 annoys me, but it is more reliable and less variable than tweaking screws. I built about 7 mendel90s with bottom homing, but am now converted to homing up.
It would be nice if marlin actually implemented the Z-offset GCODE functionally.
@Sanjay_Mortimer I think probing bed is the best overal solution (using the bed auto level hack, http://forums.reprap.org/read.php?151,246132) to me it makes the most sense. I can see some cases where homing at the top would be nice, but still i don’t like it, maybe it is because i have only done things homing down and it make more sense to me, but we will see how it goes. It still seems like a solution looking for a problem.
@Camerin_hahn With respect, I think that your assessment of the Mendel90 is flawed due to it comparing features with other RepRap designs. What you perceive as things in need of an update are actually features of the design.
Top-Z homing offers more control over the zero point than manual adjustment of a lower limit switch (a failed lower limit circuit could allow the head to hit the build platform).
The use of flat-edged Dibond permits the builder to achieve a perfectly square structure without the need to put additional holes in the baseplate (and deal with additional tolerances on material width as a consequence).
The bipolar X-Y coordinate system with zero at bed centre is another great innovation which can only be truly appreciated through using the machine.
Open your mind to the differences of the Mendel90 and put it to some use before you embark on needless modifications. You might be pleasantly surprised by what you find.
@Neil_Darlow1 I just because they are design features, doesn’t make them right. There is some benefit to these design choices, however there are detraction as well.
Top-Z homing up:
pros: allows you to resume failed prints where homing down would cause damage be impossible.
less likely to “drift”
cons:
if the switch were to fail for any reason you will break you Z mounts.
any mistakes in calibrating the distance will cause a crash in to the bed, this is far more likely then homing down
most of your print is further away from the z=0 point (any inaccuracies in the z drive will encountered twice)
Time to home the machine is much longer (at least 2 times as long)
Moot:
accuracy (you can do the same “fix” for home up as home down just change the location of Z=0 in software to be …05 or -.05)
The flat edge allows the design to me made with a simple saw and a drill with a template, this is not a bad design, but you could reduce the number of printed parts and there by speed up manufacturing if you use peg and slot fitting. the actual manufacturing is not using these simple tool, why not optimize design of the product to fit manufacturing?
0,0 is the center. This is better for many reasons (simplifies things for filling the bed and such, but there is on major problem, the host software doesn’t like it right now. If it gets changed in the host software i would immediately switch to this method.
@Camerin_hahn peg slot is more likely to involve inaccurate location of parts as theres more chance of errors. Thats simple engineering fact. As for z home its should be where the print starts as thats standard manufacturing procedure. But its up to the individual to choose their own way and both ways have pros and cons.
@Nigel_Dickinson In this case the important point is that the walls Z-axis is 90 degrees orthogonal to the xy plane If the support piece is at 89 degrees it will not affect print quality. the important point is that the z plane and and the support sit flush and tight. by being flush (and the cut square) the parts will make a 90 degree angle and prevent wobble. these joints do have problems if not properly cut (ie things not cut, then you need to file these parts to get them to fit)
@Camerin_hahn Both Pronterface and RepetierHost are OK with 0,0 at bed centre. Also Skeinforge and Slic3r have no problems with it. Do you see a pattern here? The well-known open-source hosts and slicers handle this situation well. Are the problems with hosts tied to a particular commercial printer?
I think we need to address the 0,0 as a center in your host software - other than a hassle with Cura, what package are you having problems with? It’s pretty easy to setup in repetier host and skeinforge
@Camerin_hahn really and you were using a Simple before yet you didnt redesign that. You forgeting that the diabond is laser cut, the printed parts are more likely to be imperfect more than the diabond. So I think you need to reappraise where errors will come from as the
Machine should be square to its self.