DIY Makerbot Replicator - help

[Dom]

Hi,

I'm hoping someone can help on this- I'm about ready to throw in the towel :(

I had the bright idea to build a Replicator (Dual) from scratch, including the frame etc after numerous issues with acquiring the parts and problems with parts turning up DOA i finally managed to get it all together into a semi-working state. As far as i can see everything is as it should be but whenever i start a print in replicatorG-sailfish the extruder/carriage starts hitting the front & left of the machine, the only way I've been able to prevent this is by setting the offset values well out of spec, i think i set the X offset to something like 68 and Y offset to 0 and when starting a print nothing hits or crashes. but the problem is the extruder no longer seems to centre to the HBP and is completely misaligned..... so i don't really know what to do.

Also the "MK8" dual extruder which i picked up seems different in dimensions to the standard one that comes with the official replicator, the spring arms for the filament (if that's what theyr'e called) over extend the extruder carriage by about 10mm either side and if i don't set the X offset to 68 that's the part that hits the left rod and grinding noises follow. i can't find any option to change the carriage size in replicatorG so can only assume there isn't one?.

I was starting to suspect that the issue was with my pulleys and that the default steps per mm maybe didn't apply and perhaps it was moving more that it should, but i checked the diameter off the pulleys i used and they all match up (give or take a few points of a mil) with the pulleys used in the design files.

Anyway, if anyone has any ideas on what I'm doing wrong I'd really appreciate the help, and would definitely owe you a beer ;)

[Ryan Carlyle]

How many belt teeth on your pulleys? It should be 17 for the R1D. You did use 2mm pitch GT2 pulleys, right?

[Dom]

Hi Ryan,

Thanks for your reply, i really appreciate it.

Yeah theyr'e all 17 tooth GT2 pulleys apart from the idler pulley which - is a 15 tooth GT2

[Dan Newman]

>> whenever i start a print in replicatorG-sailfish the extruder/carriage
>> starts hitting the front & left of the machine, the only way I've been able
>> to prevent this is by setting the offset values well out of spec, i think i
>> set the X offset to something like 68 and Y offset to 0 and when starting a
>> print nothing hits or crashes.

Okay, so the bot homes correctly? (Homes X-max and Y-max.)
Could it be that your starting gcode just sends the extruder to far to the
front and to the left for your build volume/build plate? The standard
Rep 1 DUAL starting gcode assumes a true-to-spec Rep 1 Dual build volume
and build plate. And even then, cuts it close to not hitting the front
or the left edge. If your bot is somehow different, then you might
need to change the parking position used by the starting gcode.

>> Also the "MK8" dual extruder which i picked up seems different in
>> dimensions to the standard one that comes with the official replicator, the
>> spring arms for the filament (if that's what theyr'e called) over extend
>> the extruder carriage by about 10mm either side and if i don't set the X
>> offset to 68 that's the part that hits the left rod and grinding noises
>> follow. i can't find any option to change the carriage size in replicatorG
>> so can only assume there isn't one?.

RepG doesn't need to know your carriage size.
RepG can be told the extent of travel for each axis via the axis length
in the machine definition. However, RepG *only* uses that to display
the build volume. It doesn't account for it when slicing or commanding
the bot to move. You just have to not use too large of models (and need
to take that into account if you use the skirt plugin to generate a skirt
around the model).

Again, it sounds like you need different starting gcode with a different
parking position.

Dan

[Dom]

Hi Dan, thanks for your reply.

Yeah everything homes correctly.

When starting a print it homes first, then shifts diagonally towards front left - i assume this is normal behaviour?.. other than the crashing/hitting part :)

The HBP size is the same as the the original replicator i think, i measure my one at 230x150mm. How would i go about verifying/editing the start gcode parking position, is it just trial and error?

If i set the X and Y positions to there minimum positions (physically) and then open the control panel in replicator G and set all the jog values to 0, after homing i get the measurements,
X = 143.701
Y = 76.365

Bearing in mind the X value would be closer to 153 if it weren't for that spring tab sticking out, do these values seem about right?

Cheers for your help Dan

[Dan Newman]

On 10/06/2015 2:00 PM, Dom wrote:
> Hi Dan, thanks for your reply.
>
> Yeah everything homes correctly.
>
> When starting a print it homes first, then shifts diagonally towards front
> left - i assume this is normal behaviour?.. other than the crashing/hitting
> part :)
>
> The HBP size is the same as the the original replicator i think, i measure
> my one at 230x150mm.

There's the size of the HBP and then there's the size of the additional room
for the mechanics to move about.

> How would i go about verifying/editing the start gcode
> parking position, is it just trial and error?

You would measure it. If you have the home offsets set such that (X,Y) = (0,0)
then measure in mm the distance to the left of X=0 and forward of Y=0. Call
that Xp and Yp. Then the parking position is (-Xp, -Yp). [To the rear is +Y
and to the right is +X.]

> If i set the X and Y positions to there minimum positions (physically) and
> then open the control panel in replicator G and set all the jog values to
> 0, after homing i get the measurements,
> X = 143.701
> Y = 76.365
>
> Bearing in mind the X value would be closer to 153 if it weren't for that
> spring tab sticking out, do these values seem about right?

There's actually a script for doing this in RepG,

Files > Scripts > Calibration > Replicator 1, 2, & 2X > Home Offsets Calibration

Here's the contents of that script

(*** This script will guide you through ***)
(*** calibrating the start position on ***)
(*** your Makerbot Replicator ***)
M103 (disable RPM)
M73 P0 (enable build progress)
G21 (Use Millimeters as Units)
G90 (Absolute Positioning)
M18 (Disable stepper motors)
M70 P10 (Move carriage to center of plate)
M01 (Move the extruder carriage until it lies in the dead center of the build plate, then press yes to continue.)
M70 P10 (Check distance from nozzle to plate)
M01 (Check the distance between the extruder nozzles and the build plate. They should be about one business card thickness apart. )
G92 X0 Y0 Z0 A0 B0 (Declare the current position to be (0,0,0,0,0))
G161 Z F1000 (Home Z axis minimum; go until reaching the end stop.)
G162 X Y F4000 (Home X and Y axis maximum; go until reaching the end stop.)
M131 X Y (record the current coordinates to the motherboard)
M70 P10 (Calibration complete)
M73 P100 ( End build progress )
M01 (Your coordinates are now saved! To adjust them, use the 'Machine Onboard Preferences' dialog in the Machine menu to change the Axis Offsets.)

[Dan Newman]

[Error/omitted text]

>> How would i go about verifying/editing the start gcode
>> parking position, is it just trial and error?
>
> You would measure it. If you have the home offsets set such that (X,Y) = (0,0)

... If you have the home offsets set such that the CENTER OF THE BUILD PLATE is
(X,Y) = (0,0) then ...

> then measure in mm the distance to the left of X=0 and forward of Y=0. Call
> that Xp and Yp. Then the parking position is (-Xp, -Yp). [To the rear is +Y
> and to the right is +X.]

Dan

[Dom]

So i ran that script Dan and checking onboard preferences it gave me the offsets X = 68.643 and Y = 37.221 i assumed i centre between the nozzles and not one or the other?

If i follow you correctly, if i set it to it's centre point, then measure the remaining X distance on the left side & the remaining Y distance forward until they both hit there "minimum" limit (physically hitting the rod on the left and physically hitting the pulleys on the front), then that will give me the required measurement for the start G code? is that correct or am i completely off on that?

Thanks again Dan

[Dan Newman]

On 10/06/2015 2:42 PM, Dom wrote:
> So i ran that script Dan and checking onboard preferences it gave me the
> offsets X = 68.643 and Y = 37.221 i assumed i centre between the nozzles
> and not one or the other?

You should center the right nozzle.

>
> If i follow you correctly, if i set it to it's centre point, then measure
> the remaining X distance on the left side & the remaining Y distance
> forward until they both hit there "minimum" limit (physically hitting the
> rod on the left and physically hitting the pulleys on the front),

*without* physically hitting anything. If you hit something, then you lose
steps. Don't do that. You should keep about 5mm clear of hitting anything.

Dan

[Dom]

OK i re-did the calibration and i get the centre home offsets in onboard prefs as X =  76.769 & Y = 37.721 .

Measuring the remaining space on the X Axis left - i had -65.5 and Y forward i had -37.4, I deducted 5mm like you said off each to give me X = -60.5 & Y= -32.4

In the file "Dual_Head_start.gcode" i changed these settings to "G1 X-60.5 Y-32.4 Z50 F3300.0 (move to waiting position)" & "G0 X-60.5 Y-32.4 (Position Nozzle)"
In replicatorG i generated the gcode based on the edited gcode file and start a print with the right nozzle selected, i used the whistle example as a test. for the most part everything seems to work well, the carriage moved to where it was supposed to and didn't hit anything, the print started and it looks as though the nozzle is off centre to the right by at least 40mm, maybe a bit more. I stopped the print and loaded up the snake example, i left the placement on the bed and everything else to default, generated the gcode etc, when the print starts the carriage does hit the front of the unit this time, although it's not as severe and doesn't last too long. Is there something else I'm missing?

Thanks again

[Dan Newman]

On 11/06/2015 4:32 AM, Dom wrote:
> OK i re-did the calibration and i get the centre home offsets in onboard
> prefs as X = 76.769 & Y = 37.721 .
>
> Measuring the remaining space on the X Axis left - i had -65.5 and Y
> forward i had -37.4, I deducted 5mm like you said off each to give me X =
> -60.5 & Y= -32.4
>
> In the file "Dual_Head_start.gcode" i changed these settings to "G1 X-60.5
> Y-32.4 Z50 F3300.0 (move to waiting position)" & "G0 X-60.5 Y-32.4
> (Position Nozzle)"
> In replicatorG i generated the gcode based on the edited gcode file and
> start a print with the right nozzle selected, i used the whistle example as
> a test. for the most part everything seems to work well, the carriage moved
> to where it was supposed to and didn't hit anything, the print started and
> it looks as though the nozzle is off centre to the right by at least 40mm,

Which nozzle? Left or Right? If it is the right, then

-something went wrong when you set the home offsets, or
- the wrong steps/mm are being used when you generate x3g, or
- some mechanical oddity.

When you slice for and print with the right extruder (tool 0), then
the gcode position (0,0) is exactly where you centered the right extruder
when you calibrated the home positions. When you slice, your model is
typically centered at (0,0) unless, of course, you didn't center your
model.

> maybe a bit more. I stopped the print and loaded up the snake example, i
> left the placement on the bed and everything else to default, generated the
> gcode etc, when the print starts the carriage does hit the front of the
> unit this time, although it's not as severe and doesn't last too long. Is
> there something else I'm missing?

Look at the generated gcode and see if it has your starting gcode. If it
does, then it sounds like you have some sort of issue with your system whereby
the carriage moves too far some times. One possible case would be if the
system is loosing steps occassionally. If that's the case, then when you
determined the home offsets, it may have physically lost steps and measured
too long of a distance. I.e., it had to step 100 steps to physically move
90 steps. Then sometimes when it doesn't lose steps it moves to far forward
and to the left.... Or maybe not. But if each time you have the same
starting gcode and sometimes it hits the front/left and sometimes it doesn't,
then you clearly have some form of non-deterministic behavior. Most likely
issue would be lose of steps along one or both axes. Could definitely happen
if you aren't using the correct stepper motors and consequently are not
supplying them with the currents they need.

Dan

[Dom]

Which nozzle?  Left or Right?  If it is the right, then

This was after calibrating the right nozzle in the centre of the build plate, and printing with the right nozzle.

 When you slice, your model is
typically centered at (0,0) unless, of course, you didn't center your
model.

 I just opened the model through examples and hit generate G code, i assume the preview showing it  (the whistle, for example) centred on the build platform is carried over to the Gcode automatically?

Look at the generated gcode and see if it has your starting gcode.  If it
does, then it sounds like you have some sort of issue with your system whereby
the carriage moves too far some times.   One possible case would be if the
system is loosing steps occassionally.  If that's the case, then when you
determined the home offsets, it may have physically lost steps and measured
too long of a distance.  I.e., it had to step 100 steps to physically move
90 steps.  Then sometimes when it doesn't lose steps it moves to far forward
and to the left....   Or maybe not.  But if each time you have the same
starting gcode and sometimes it hits the front/left and sometimes it doesn't,
then you clearly have some form of non-deterministic behavior.  Most likely
issue would be lose of steps along one or both axes.  Could definitely happen
if you aren't using the correct stepper motors and consequently are not
supplying them with the currents they need.

I just checked the Gcode tab and everything matches the start settings i changed for the waiting position & nozzle position.

I've ran through the whistle example again to see if anythings different, it's hard to tell but it looks almost like it printing the part bigger than it should be..maybe twice as large as it should be, as opposed to it being off centre... I'm not feeding any filament through it yet, as I've still got to adjust the nozzles etc still - so this is just how it looks to the naked eye. checking the scale it says it's currently set to "1" which i guess is correct so i don't know whats up there. the carriage doesn't hit anything as before, so that's good.

I've also done the snake example again to see if it consistently hits the front and isn't just random. - Here's what happens, it starts by moving the carriage to the front left (with 5mm clearance between the carriage and front/left side) like normal, then when the print starts the carriage briefly hits the front of the unit, makes a terrible grinding noise for about 4 seconds, then the carriage continues on it's path, while it moves towards the rear of the machine it triggers the endstop multiple times, then the process repeats as before, as it continues on it's path.

I'll check the belt tension, the only belt that looks even slightly slack is the X Axis belt, I'll tighten it up and see if it makes and difference.

I forgot to mention this, when i was ordering the parts, i couldn't find the 83 tooth closed loop belt anywhere in the UK or EU, a couple of places in America were selling - but they didn't ship to the UK, the only thing i could find that was close to a match was an 82 tooth belt, 2mm pitch. do you think because it's one tooth shy - that that is affecting steps? i assumed it wouldn't matter because i thought the steps were calculated on pitch and pulley diameter... but chances are I'm wrong, also i remember seeing somewhere that the clone machines tended to use different tooth counts on their closed loop belts.

Edit*
Just checking over the stepper motors i got, i was quite conservative looking at the specs, and as far as i can see they should be a good match when compared to the ones used in the official replicator, although it's unclear if the official ones (moons something or other) current rating are per phase or not? the ones i got are 0.4A per phase, for a total of 0.8A. they should be OK right?

Cheers Dan

[Dom]

So it definitely looks like it's a scaling issue, i very crudely held a pen against the right side of the carriage while it was printing the whistle, and it shows a very sketchy looking whistle about twice the size of the model shown in replicatorG, I've attached a couple of images.

It does look like it's actually printing in the centre of the bed though (if you take away my pen position - offset), but obviously just in the wrong scale, which is why it seems to crash against the front of the unit when printing the larger model of the snake.

Any ideas why it's doing this?

[Dan Newman]

> Any ideas why it's doing this?

Perhaps your stepper drivers are set to 1/4 microstepping and not 1/16th?

Dan

[Dom]

Good point, they could well be.

How would i go about checking, and changing it if it turns out they are?

Thanks for your help this far Dan, i'd still be banging my head against a wall otherwise :)

[Dan Newman]

On 11/06/2015 1:03 PM, Dom wrote:
> Good point, they could well be.
>
> How would i go about checking, and changing it if it turns out they are?

What type of stepper drivers are they? Depends upon the type. Some are fixed
such as the botsteps shipped with most mightyboard clones. Good news is they
are fixed at 1/16th (or should be). However, if you have something other than
botsteps, then usually there's pins you jumper to select the amount of microstepping.

And I know that some folks have bought a mightyboard clone from Geetech and gotten
TI 8825-based stepper drivers with them. Anyone's guess if Geetech hard wired them
to 1/16th stepper or did who knows what.

The other way you can end up with a model twice as large as it should be would
be if you accidentally had your machine type (Machine > Machine Type (driver)) set
to a Thing-o-Matic in RepG. It has the same 17 tooth pulleys but uses 1/8th stepping
rather than 1/16th. (P.S. Did I write 1/4 stepping in my prior message? Meant to
write 1/8th stepping. Rep 1 should be set to 1/16th but if for some reason you
had 1/8th then things would be 2x as large in the XY plane.)

> Thanks for your help this far Dan, i'd still be banging my head against a
> wall otherwise :)

Not a problem.

[Dom]

Your right on the money there, the board and stepper drivers are manufactured by Geetech, as far as i remember the drivers are the botstep version though.... i'll have a look at them later and see if i can eye the traces to make sure they're in 1/16 config -  assuming the replicator design files schematics illustrate this.

The stepper motors I'm using - in the datasheet they state the "Step Angle = 1.8º " is this the same as 1/8th stepping?  i do hope i haven't made a schoolboy error with these motors ;)

The machines definitely set to the replicator 1 dual in replicatorG, i guess i could set the gcode profile to the replicator 2 model to double check - assuming the step info is stored in the gcode?

[Dan Newman]

On 11/06/2015 2:00 PM, Dom wrote:
> Your right on the money there, the board and stepper drivers are
> manufactured by Geetech, as far as i remember the drivers are the botstep
> version though.... i'll have a look at them later and see if i can eye the
> traces to make sure they're in 1/16 config - assuming the replicator
> design files schematics illustrate this.
>
> The stepper motors I'm using - in the datasheet they state the "Step Angle

> = 1.8º* "* is this the same as 1/8th stepping?

No, that's the arc subtended by a single, full step of the motor. So, to
see the number of full (aka, natural) steps the motor can do, divide 360
by 1.8 for 200 steps. And that's what you want. Then with 1/16th stepping
you get the desired 3200 microsteps per full rotation of the motor.

> The machines definitely set to the replicator 1 dual in replicatorG, i
> guess i could set the gcode profile to the replicator 2 model to double
> check - assuming the step info is stored in the gcode

step info -- steps per mm -- are in the machine definition (.xml file) for
the machine type you select with Machine > Machine Type (driver). You can
then pull down Machine > Machine Info and see it. The gcode is in mm.
When you generate the x3g, the units of millimeters in the gcode are
then translated to steps via the steps/mm value in the machine definition.

Dan

[Dom]

all looks greek to me ;)  the machines XML says "motor_steps="3200" "towards the bottom, which is in-line with what you said, but it looks like it's related to the extruder motors rather than the X/Y motors. maybe i just can't see it,

[Dan Newman]

Output shows the correct steps/mm for the X and Y axes. Look for "stepspermm=".

So, if you're getting a model which is 2x too large in X and Y, then the stepper
drivers are incorrectly wired for 1/8th microstepping. Should be 1/16th.

Was the whistle also too tall along the Z axis? I would expect everything to
be off by 2x; otherwise, the print may fail with too much or too little plastic
being extruded. Too much in the case where X, Y, and extruder is 2x but Z is not
and so too much plastic builds up. Too little in the case where just X and Y
are off by 2x but Z and the extruder are not and thus too little plastic is
output.

Dan

[Dom]

Thanks for checking, at least the settings are correct.

Was the whistle also too tall along the Z axis?

I don't know yet, I've just been doing dry runs without the filament for testing,i  thought it best to do that for the time being, as I'm likely going to have to make some adjustments to the nozzles etc, and will be easier if they aren't caked in plastic. I'll do another dry run tomorrow and physically measure the the distance with a ruler between the nozzles and bed towards the end of the print and that should give a rough estimate of the whistles height. i suspect your right though, because if i cancel a print via the units  onscreen LCD menu the Z axis will move the bed all the way down until it hits the base of the replicators frame (much like the X & Y Axes when the hit front & left) in these situations i just turn the unit off....quickly :)

Found the product page where i bought the board and stepper drivers, and they're listed as "A4988 Stepper Motor Driver" I'm pretty sure they had "botstep" printed on them... I'll double check tomorrow. In the images on product page here it looks as though the steppers have trimpots on them, I'm 95% sure the ones i got don't have trimmers... so I've got no idea whats going on there. Worst case scenario, if i can't rewire to 1/16th for some reason are there other models/brands that are 1/16th and drop in compatible?

[Dan Newman]

> Found the product page where i bought the board and stepper drivers, and
> they're listed as "A4988 Stepper Motor Driver" I'm pretty sure they had
> "botstep" printed on them... I'll double check tomorrow. In the images on
> product page here

> <http://www.aliexpress.com/item/Mightyboard-kits-inculding-A4988-stepper-motor-driver-heatsink-LCD-display-ect-for-makerbot-dorp-shipping/32235733465.html>it

> looks as though the steppers have trimpots on them,

They probably just used photos of knock-off pololu A4988 stepper drivers.
The botsteps use A4982s but do NOT have trimpots like the Pololu ones.

If yours have A4988s then they are not direct clones of botsteps since
botsteps use A4982s instead. If yours have A4988s, then you'll need to
download the data sheet for the A4988 and ensure that they have wired
MS1, MS2, and MS3 correctly to achieve 1/16th stepping. And they may
have done something wrong like presenting MS3 on the header going to
the motherboard instead of VREF. A botstep connects VREF to the mightyboard.
A pololu-style driver using an A4988 does not present VREF and instead
presents MS3. It's anyone's guess what Geetech did. After all, they
don't provide a true Mightyboard clone in the first place. (They literally
followed the "published" schematic for the MightyBoard rev E. However, the
actual, released MightyBoard rev E is different from that schematic.)

> I'm 95% sure the ones i
> got don't have trimmers... so I've got no idea whats going on there. Worst
> case scenario, if i can't rewire to 1/16th for some reason are there other
> models/brands that are 1/16th and drop in compatible?

You'd buy clone botsteps from WanHao or FlashForge if you cannot rewire what
you have. For a A4982 the MS1 and MS2 pins should be tied to +5V to achieve
1/16th stepping. The data sheet and botstep schematics are downloads of
Thing 16059,

http://www.thingiverse.com/thing:16059

Dan

[Jetguy]

Cut the MS 3 pin, solder bridge on top of the driver MS2 to to MS3 to set 1/16th mode.

Stupid error by Geeetech and again, why we do not recommend this board.

[Dom]

Hi Jetguy,

Do you mean cut the MS3 trace or just physically removing the MS3 header pin?

Cheers

[Jetguy]

Remove the MS3 header pin.

Caveat, this is based on the understanding you have drivers with current potentiometers on them right??? Even if that's because they are on the underside of the stepper drivers (obviously, check before clipping the pin).

The reason is as Dan pointed out, on Pololu standard A4988s used on Reprap boards, MS3 is on the pin header and current is set via a potentiometer on the driver board.
on REAL botstep drivers designed for the mightyboard MS3 is not the same purpose. Instead it's used to set current instead of a pot on the driver.
Because the voltage on the vref pin on the mightyboard lines up with MS3 and the typical voltage on vref is about 1.6V max this is considered logic low AKA the same as ground.

On an A4988 driver, MS1, MS2, and MS all need to be at at least 3.3V logic high to be in 16th stepping mode. Otherwise, if MS3 is connected to a Vref pin at 1.6V or lower, it's sensed as logic low kicking the driver into 1/8th stepping.
Given your problems, this explains everything.

Literally, every axis is going 2X further than commanded (1/8th VS 1/16th microstepping modes).

[Dom]

Yeah i just checked and i was mistaken originally, turns out the driver board i got from Geetech does indeed have the trimmer on them.

The image Here on the reprap wiki shows the unit i seem to have,

If that all looks in order? I'll cut the MS3 pin and bridge it to MS2.

Thanks Jetguy.

[Dom]

Jetguy, gotta say thanks for your help, followed your instructions on modifying the stepper drivers and now everything is operating as it should.

Cheers,

Dom.

[Dan Newman]

On 12/06/2015 7:36 AM, Dom wrote:
> Jetguy, gotta say thanks for your help, followed your instructions on
> modifying the stepper drivers and now everything is operating as it should.

BTW,

You may want to check the VREF voltage set by those trim pots. If they are not correct,
you may be sending too much current (or too little) to your stepper motors. For A4988s,
the current limit Cl is

Cl = VREF * 2.5A / 1V

Now, if your stepper motors match the specs of those used by MBI, then you want
around 0.72 A for the X, Y, A, and B stepper motors. (MBI shot for 0.72A with
a digipot setting of 118.) For the Z stepper motor, around 0.25A. (MBI went for
0.24 A with a digipot setting of 40.)

Note that MBI used, IIRC, +/- 20% digipots and so, there's, well, 20% latitude in
what you set I guess. Again, if I recall correctly....

To set these stepper drivers with no external VREF, you would supply them with +5V
on the logic Vin and ground on the logic ground. (That's two adjacent pins at one
corner of the daughter card.) Then adjust the trimpot and check the voltage between
the VREF checkpoint/pad and logic ground.

(I'm stating "logic Vin" since there's also inputs for the motor supply current/voltage
input. The logic Vin is the +5V input for the chip's logic circuit.)

Dan

[Dom]

The X & Y motors I've got are rated at 4.5V 0.8A, do you know if the ones in the official replicator are the same, and they just set 0.72A to err on the side of caution? or is it just to take into account the +/- 20% on the digipots

My Z motor is a little different (JK42HS40-1004-AL) it's rated at 3.9V 1A. (BTW if any ones sees this and considers it for a replacement in their replicator > the rod on it was a little too long, maybe about 3mm, i had to drill a hole in the top of the wooden frame to give it slightly more clearance)

I'll have to look into the extruder motors, i don't know the the model off hand as they're just what came with the extruder, can't imagine they're far off.

Dan would it be OK if i just remove the stepper drivers from the unit and use an arduino (or similar) to provide the 5V to tweak the VREF?

Thanks again.

[Ryan Carlyle]

If you supply full rated current, the motors can get up to about 90C in an enclosed printer, and any plastic brackets you're using are at risk of failing. Extra much so if the digipot tolerances are on the high end and you're over-driving the motors. It makes sense to under-drive the motors a little to reduce the heat generation and temp rise.

[Dom]

Right, cheers Ryan.

So if i do the calculation Dan posted for my X & Y motors and then deduct about 10% from the end result, that should put me at a safer tolerance level?

As for my Z motor, do you think i should apply a similar current to what the official replicators use of "0.25A" or apply a current closer to that of my Z motors rating of 1A ?

[Dan Newman]

The MBI Z stepper motor has a much lower max current rating. Something around
0.4A perhaps? Don't hold me to that. But point being, it's a much lower max
current motor that MBI uses for the Z axis.

Dan

[Dom]

Ok, i'll try 0.4A for the Z motor then.... it can't hurt regardless.

[Ryan Carlyle]

Easy enough to test it and find out. Start low, and dial it up if you run into skipping problems. Or measure the running temp after an hour and dial the current down if it gets too hot (eg over 75C surface temp). 

If your motor is rated to 1A, there's probably no reason to run it below 0.7A or so, that'll put it at half its rated temp rise which should be very safe. Under-driving motors by a large amount makes them unreliable and lose precision. 

[Dom]

Oh i wasn't aware of that, i'll start the Z motor off at 0.7A then and see how it goes. Thanks Ryan

[Dom]

So i set the values on the X & Y stepper drivers to 0.32V to give me a current limit of 0.8A, getting quite allot of missed steps during a print. they were originally set to 0.955V so it's a good job you said to check.

Dan when you say the digipots have +/- 20% do you mean that in the same way that a physical resistor might have +/- 20% accuracy? or that when adjusting the value between 0 to 118 in the settings, it provides 20% range. Hope that makes sense.

Edit*
I'll just add though, that when i'm measuring the temp on the X/Y motors (50% of the way through the whistle example) they aren't even going above 28 deg C . the skipped steps could well have something to do with my belt tension though, so i'll double check all that once i've got the current limit dialed in

[Dan Newman]

On 13/06/2015 10:34 AM, Dom wrote:
> So i set the values on the X & Y stepper drivers to 0.32V to give me a
> current limit of 0.8A, getting quite allot of missed steps during a print.
> they were originally set to 0.955V so it's a good job you said to check.

Oh my. That would have made your stepper drivers nice and toasty. Keep
your cuppa warm.

On the Core-XY bots some of us build, we use X & Y stepper motors rated
for 1.5 A and we tend to set the A4988's current limit to about 1.2A.
That works well for us. You have different motors and mechanics so
obviously your needs will be different.

> Dan when you say the digipots have +/- 20% do you mean that in the same way
> that a physical resistor might have +/- 20% accuracy?

Correct. I cannot speak to the ones on the Geetech boards, but the ones
I've seen used on other mightyboards were the cheapest ones with the greatest
errors (lowest tolerance).

> or that when
> adjusting the value between 0 to 118 in the settings, it provides 20%
> range.

No.

Keep in mind that you aren't using the digipots -- they don't do you any
good since they do not electrically connect to your stepper drivers.

[Dom]

Right, thanks for clearing that up Dan,

Given that I'm missing steps, and my temperature is pretty low on the X & Y - do you think it will be alright if i set them closer to the 0.9A/1A area?

I set the Z Motor to 0.68A in the end and that seems to have worked well, so much quieter (on all the motors), i can't speak for the extruders steppers yet, i just set them to 0.68 as well but am yet to put filament through them.

Edit* correct me if i'm wrong Dan but the calculation you provided is dependent on the fixed value current sense resistor on the stepper driver boards?

[Dan Newman]

On 13/06/2015 11:04 AM, Dom wrote:
> Right, thanks for clearing that up Dan,
>
> Given that I'm missing steps, and my temperature is pretty low on the X & Y
> - do you think it will be alright if i set them closer to the 0.9A/1A area?

1. I'm not a stepper motor expert.
2. I have no idea what the specs are for the motors you have.

If you know their specs, or at least part numbers you may want to post that info
here so that folks much more versed in this issue can chime in.

Dan

[Dan Newman]

On 13/06/2015 11:21 AM, Dan Newman wrote:
> On 13/06/2015 11:04 AM, Dom wrote:
>> Right, thanks for clearing that up Dan,
>>
>> Given that I'm missing steps, and my temperature is pretty low on the X & Y
>> - do you think it will be alright if i set them closer to the 0.9A/1A area?
>
> 1. I'm not a stepper motor expert.
> 2. I have no idea what the specs are for the motors you have.

Two key specs here will be (1) max current, and (2) winding resistance.

Dan

[Dom]

No worries Dan,

The motors I'm using for the X & Y are 17HS1011-20B

Unfortunately i can't find any info on the motors for the extruders, but i expect they will be in the same ballpark as the X & Y

[Jetguy]

Guys,the problem here is the varying specs of current sensing resistors companies are installing on the Pololu knockoff drivers.

The correct formula is 8 times the resistor value on the board times Vref (the wiper of the pot) = current in A.

You MUST look at the drivers and identify or measure the 2 current sensing resistors. Because folks are using different resistor values, this modifies the vref voltage VS current output curve.
What I'm saying is, we used to thing 0.4V =1A on most Pololu drivers. The knockoffs are using alternate resistors and might be 0.8V or more to reach the same 1A output current. The only way to know is examine or measure the current sensing resistors and plug that into the formula.

[Jetguy]

part 2, "Unfortunately i can't find any info on the motors for the extruders, but i expect they will be in the same ballpark as the X & Y"

No, many of the motors coming from China are NOT the normal 0.84A rated NEMA 17.

The best way to test and KNOW is measure single coil resistance.
MakerBot motors are about 6 ohms
Many knockoff motors are 3 ohms or less
2 Ohm motors are on the verge of being unusable in an extruder. Those motors at 2 Ohms are 2A rated current in order to reach rated torque. They are built with low inductance and thus low static resistance for speed (the coil can be switched very fast).
The problem is an extruder load is a slow and very fine movement. Such a low inductance stepper cannot reach torque without nearly smoking the common A4988 driver maxed out and even heatsinked.
I personally would not attempt to run any motor for an extruder at less than 4 Ohms coil resistance.

[Dom]

Ah, that explains it.

I'll check the current sense resistors and see what they are, cheers Jetguy

Yeah the model number is actually printed on the back of the extruder motors (42HS3415-19B Brand is NN) but hunting around online, i couldn't find it. Thing is, the seller listed different motors in the specs for the "MK8" extruder but upon receiving the motors they were the 42HS3415 models.

I do have another two motors spare, which are the same as as my X/Y motors, they're rated at 5.57ohms per phase. Do you think they'll be OK to use?

Thanks Jetguy

[Dom]

You were spot on Jetguy, just measured one of the coils on the extruders stepper motors (42HS3415-19B) and it measured dead on 2 ohm (after Rel my measurment probes) so i guess i won't be using them then.... that seller really screwed me over, first i discover the thermocouple was infact a thermistor so i had to change that, then the nozzes feed screw was bent so i had to replace that, and now the motors aren't suitable, so all i'm really left with is (before my replacements) is the nozzle bracket and fans, which i'll no doubt replace at some point as theyr'e very loud, haha... only in china :)

[Dom]

You MUST look at the drivers and identify or measure the 2 current sensing resistors. Because folks are using different resistor values, this modifies the vref voltage VS current output curve.

I just checked and they're 0.200Ω each, would i sum them together for the calculation (0.400 Ω) or just plug a single resistor 0.200Ω into the calculation?

Cheers,

Dom.

[Jetguy]

Single resistor since it's one resistor per phase.

https://www.pololu.com/file/download/a4988_DMOS_microstepping_driver_with_translator.pdf?file_id=0J450

Current setting in Amps= Vref divided by (8 times resistor value)

0.2 Ohms times 8 is 1.6 It takes 1.6V Vref to make 1A.

Now, at very low resistor values, not all meters are accurate. Did you measure or get a number or marking to determine the resistor value??
Most come with either 0.05 Ohms or 0.1 Ohms. ) .2 Ohms is NOT an ideal value. You waste heat in the resistor at even moderate currents.

[Dom]

Thanks for confirming that Jetguy.

Did you measure or get a number or marking to determine the resistor value??

 

I checked the numbers on the resistor, each one has "R200" printed on them, i verified with my meter also.

.2 Ohms is NOT an ideal value. You waste heat in the resistor at even moderate currents.

I guess i could get some 0.05 Ohms later on down the line and switch them out, do you happen to know the (physical) size? in the meantime i think I'll set the current limit to 0.68A for my X & Y (motors rated @ 0.8A) do you think with the 0.2 resistors currently in place I'll be OK temperature wise, for now?

Thanks Jetguy

[Ryan Carlyle]

Just turn on the motors without doing a print (go into jog mode or whatever) and check the temp every so often while you do something else. If the motors stabilize under 70C, you can dial up the current. If they get over 70C, dial down the current. In the end, the goal is to get the most torque you can without overheating anything, so this is the simplest way to tune without trying to do math that may or may not be accurate after considering component tolerances. 

[Dom]

Ah alright, cheers Ryan.

It's a bit of a mission to set the pots on the stepper drivers because i can't do it while they're in the machine so I'll set them to 0.68A based on Jetguys calculation first, and then measure the temp and see how they go.

So current flows to the motors while in jog mode, stationary?

Cheers.

[Ryan Carlyle]

Yep, steppers get max current when stationary. They just have to be enabled to lock them in place. They get enabled when you jog or use "enable steppers" via the LCD menu, or start a print, or whatever. 

[Dan Newman]

As I was helping someone else get ready to wire up some Pololu A4988s for
use with a MightyBoard, I snapped some pictures and added some details
to a Flickr album,

https://www.flickr.com/photos/d-newman/sets/72157654553957262

Dan

[Dan Newman]

P.S. While the pictures themselves have a little text apiece, if you navigate to
each individual picture, there's more detail in the description section.

[Dom]

Your way looks a lot neater than the way i dd it, i snipped the pin with side cutters then poked whatever was left protruding up with a dental pick - while heating with the iron :D

Also i should add, the test pad on my drivers were hard to spot, they applied the solder mask to it haha, so i just used the trimmer instead.

[adam paul]

http://www.openbuilds.com/threads/triple-c-bot.1255/page-2
I have some good pics here also.

[Dom]

I set the current limit to 0.72A on my X/Y stepper drivers but I'm still having issues with missing/skipped steps on the X axis, at least that's what i think it is.

I'll try putting them at my motors rating of 0.8A and see if that makes any odds, but I'll be surprised if does.

I'm just using the whistle example again to diagnose, it always happens 2 seconds before hitting 5% using default settings etc, it's repeatable, i tried disabling acceleration but it still happens at the same point. checked belt tension and all that good stuff.

Does anyone know the model number for the X/Y motors used in the official replicators? so i can check the datasheet specifications for torque and compare with my own.


*Edit
So i just removed the extruder assembly from the X carriage, thinking losing the weight might make the problem disappear, but no cigars, same issue. no idea whats going on. I'll try swapping my X & Y stepper drivers around next in case they're the culprit.

[Ryan Carlyle]

Stock motors are Moon's 17HD4063-xxN, where xx appears to be a customer-specific code or batch code (Makerbot uses 03, FlashForge 05, some others have been seen in the wild). But no one has ever found an actual spec sheet as far as I'm aware. Holding torque is something resembling 2 kg-cm. 

Is an overhang on your print curling up, by chance? That can cause repeatable lost steps.

Also, try reslicing the same file rotated 45 degrees, that will often fix random slice-related issues. 

[Dom]

Thanks Ryan,

Just tried rotating the model, and no change unfortunately, still at 5% does this juddering/vibration thing on the X axis.

In the datasheet for the motors (17HS1011-20B) I'm using, it says "Holding torque = 280 mN .m" is that millinewtons a metre?

Btw i swapped the X & Y stepper drivers around earlier to rule them out, no change. Increased the X axis current limit to the motors full rating of 0.8A, still no change, i unplugged the X axis endstop in case it was causing interference, still no change

Running out of ideas :)

Is an overhang on your print curling up, by chance? That can cause repeatable lost steps.

I still haven't got filament running through it yet, just doing dry runs until i can sort out these issues.

Cheers Ryan,

Dom.

[Ryan Carlyle]

Yes, 280mN-m is 28 N-cm or 2.8 kg-cm. Should be adequate. Note that's how much torque they develop at full current. 

What slicer are you using? Is this a perimeter with smooth curves where the issue happens? Sometimes, 8bit machines will judder when fed a series of very short segments in a curve because the motion planner has trouble with that kind of geometry. The segments are all nearly linear because they're so small, and thus the firmware can't identify corners that require deceleration, so it tends to launch around the curve too fast. S3D produces a lot of curves like this because it doesn't enforce a minimum segment length and will output a ton of tiny segments. Slowing down the perimeter speed or decimating the model mesh to produce lower segment count through the curve will usually fix that issue. 

[Dom]

Thanks Ryan, good to know my motors likely aren't the issue then.

What slicer are you using?

I've just been using ReplicatorG-Sailfish, hitting "Generate GCode/Replicator Slicing Defaults" i guess that's Skeinforge?

Is this a perimeter with smooth curves where the issue happens?

This is hard to tell without filament, but i think so, for the first 4% of the print the movements look very compound, and then that changes when it hits 5%, of course it depends on the model, this is just with the whistle example. If i make and print a  model, such as a cube and one a sphere, that should make it pretty obvious during a print, right?

[Dom]

I just made a couple of simple models in solidworks, a sphere and a rectangle (images attached) the issue with the X axis didn't occur when printing either of these shapes, perhaps the shapes aren't complex enough to re-produce the issue.

But when printing any of the examples skull, whistle etc the X axis judders/skips

[Mark Walker]

Seems like there is judder and then there is skip.  If skip, you'll get a gradual distortion of the path in the direction of the lost movement. If judder, that is, shaking of the carriage, you'll get ringing in your print.  Turning "off" acceleration turns off the computations in the planner that tries to compensate for backlash when changing direction.  If I understand what you mean by judder, turning off acceleration should make it worse, not better.

You might want to watch (or perhaps listen to) another rep printer print your model to compare notes.  My FlashForge makes a fair amount of music while printing things like hexagonal infill, but seems to get the plastic where it is supposed to be.  You might be getting more backlash than a factory r1d and that vibration is backlash during a bunch of turns.

You might want to try slowing down your print speed dramatically and compare that.

[Dom]

Hi Mark,

Your right, when turning off acceleration the juddering or skipping (whatever it is) seems to be audibly louder. I think i already tried bringing X/Y axis speed down and didn't notice much of a difference.... if at all, but I'll give it another go. As I'm only doing dry runs atm i have the extruder/fans, HBP etc turned off, so perhaps I'm just drawing attention to something which ordinarily would be masked by other sounds ;)

I think i should start actually printing with actual filament next instead of avoiding it ;) and see if i get something which actually resembles the model. It might be as you said, a lot of noise, but with half decent prints.

Cheers Mark,

Dom.

[Dom]

Mark it seems you were right, despite the juddering/vibration it did seem to get the plastic where it needed to go.

It took me about 20 attempts to get the print to start laying properly on the bed, I've attached photos, unfortunately at around 70% my PSU blew up.....i just can't catch a break ;) luckily it didn't seem to take the board with it, so I'll crack the power supply open and see what's up, i suspect the unit drew more current than the power supply could supply/handle.... but it really was cheap. i guess the 10A rating is peak only ;)

Does the print quality look acceptable to you guys or is it terrible? the deeper marks on the bottom seem to line up with some poorly applied kapton i put on the bed, i had to patch it up as i accidentally used the machine as a temporary engraver ;)

[Ryan Carlyle]

I don't see any major mechanical issues in the print, that's good. Most obvious thing is needing a little more cooling to get better results on small details like the lanyard eyelet. This is PLA, right?

As for the kapton, the easiest thing is to put some windex on the build plate so you can reposition the kapton as needed, then squeegee out the excess. Then leave it alone for a day or so to let the residual evaporate before heating up the build plate. Otherwise you get lots of bubbles. (Bubbles usually go away with time or can be pricked with a needle, they're just annoying.)

[Dom]

I don't see any major mechanical issues in the print, that's good.

Cheers Ryan, that is good

Most obvious thing is needing a little more cooling to get better results on small details like the lanyard eyelet.

I set the extruder temp to 235c, this seemed to be the only way to get it to adhere to the bed, could that be the issue? i haven't applied any cotton/kapton over the extruder heat blocks yet either - on my to do list :)

This is PLA, right?

it's ABS

I opened up the PSU, the fuse was blown, there were a few things that brought a much needed smile to my face when looking  inside, I've attached a few images if anyones interested,. The two big "Rubicong" caps were actually 280uF in parallel, the caps are hollow, it's just a can on top of some much cheaper? caps.... gotta wonder why they went to the effort when they branded them "Rubicong". The more expensive components look like they've been recycled from previous products - the rectifier, regs etc. There's a resistor where a cap should be, which is bizarre because the vias are on the board 1cm away, but they decided it would be better where the cap was.... there's a footprint with an isolation slot, looks like it was intended for an opto, but they just bridged straight over.  Anyway, i didn't have any through hole 5A fuses so i drilled a couple of holes in the board and fitted a proper fuse holder. bit dodgy but can't be any worse than it already is :D i haven't tested it out yet, but it should work fine again and I'll look into replacing the thing with a higher quality power supply

 

[Ryan Carlyle]

Hmm, ABS doesn't typically have issues with insufficient cooling, although it's possible. Might be something else. Try more diagnostic prints.

[Dom]

Will do, thanks Ryan.

I'll have to wait a few days mind, turns out it wasn't just a blown fuse on the PSU, the reason it blew is because the bridge rectifier shorted, I've ordered a new power supply now anyway - as i was pushing it (current wise) with this one, but i also ordered a few rectifiers as well  to replace the dead one, and that should bring the psu back to life. I'll post a few details on this when they arrive, as I'm sure a lot of the clone printers will have this or a similar flavour supply installed, so it might be useful to someone.

I calculated my total (worst case) current draw with one extuder to be 9.48A, add the other extruder to the mix - and i calculate it at 11.43A, so yeah, surprised the cheap Chinese unit operated for as long as it did :D i should have really done this calculation prior. That is assuming of course everything is on at once, and drawing current.

[Dom]

Was looking into replacing the dodgy regulator on the mightyboard  to be on the safe side - while I'm waiting for my replacement PSU, but came across this image here, on the device AQV257A pins 4 & 5 are bridged together, on my Geetech board they aren't, is this normal? looking at the schematic it's an optocoupler related to the "safety cutoff" pin 5 goes to "EX2-PWR" (extruder 2) and pin 4 goes to +24V. Dan mentioned earlier that some of these features weren't implemented on the final makerbot motherboard, but on the Geetech boards they were, is there any reason why/if i should change this? or should i just leave it as is?

[Ryan Carlyle]

Those pads were connected so the opto could be left off the board to reduce cost. (There's no firmware functionality to control the safety cutoff circuit, and it wasn't a good design to begin with, so it's a waste of parts.) 

If it's working for you, might as well leave it in there. 

[Dom]

Ah OK, i''ll leave it alone then

Cheers Ryan

[Dom]

Got my replacement power supply, all fitted and back in running condition.

Think i might have found one  of the issues when printing, i could hear a rattling noise, turns out it was the Y rod at the rear skimming the Z threaded rod, i did a few tests to see if it was affecting the print - the cube on the left is just a normal print, the cube on the right is me applying slight pressure to the Y rod during the print to prevent it from touching the Z rod, the voids in the first cube seem to have pretty much gone. I thought I'd have to design some elaborate thing to hold the Y rod in place, but as it turns out someone has already done it here, so i set about printing the "pillow" part, but unfortunately in all my prints the hole that houses the bearing has a large amount of play.... close to 1mm, there is a set screw to hold it in place but i don't know if that level of play is normal. *Edit so it turns out it was meant for different radius bearing, i just edited the openscad code, so all should be ok now

On another note, i got the replacement bridge rectifiers for the failed PSU, of course i ordered the wrong pin pitch in my haste to get my mouser order total up (to qualify for free delivery). regardless, i bent the pins and fitted it in, it now works fine. So here's the info for anyone with the same or similar supply, hopefully this is useful to someone. The PSU is branded AD2410-240W The PCB is branded A20112   PCB0350A-V07 the rectifier which failed was a Fairchild KBU8K but chances are you won't have this particular one, as i said earlier, they all look second hand and they likely use whatever is on hand.

[tramalot]

I put a dab of silicone