print parts are smaller than designed

Hi guys and thanks in advance for any help with this,

So I made my reprap from scratch (as close as possible to a Prusa i3). Overall I made it and I'm printing. I started with ABS and I've somehow managed to overcome everything - prints are sticking to the bed, I have quite a nice flow of material coming out of the nozzle etc.

After some failed attempts I managed to print a part taken from thingiverse -it is a z axis endstop because I also need it to modify the printer - I now have a fixed z endstop.

The thing is that my print is clearly smaller than the model. After printing I get a 43x20x3.7. The model was a 55x30x8 )

So I clearly have an issue there.

Also, what I've noticed is that after the first 3-4 layers, the nozzle is somehow sinking in the plastic. So, instead of printing OVER the other layers, is somehow pushing in the already existing, last layer.

Any kind of idea of where to start checking would be appreciated.

Thanks!

First, check your steps/mm in the Marlin firmware. A good starting point is multiplying whatever number is there right now for each axis by the ratio of the model dimension by the actual dimension. So for the axis that should have been 30mm but came out 20mm, multiply the steps/mm value by 30/20 = 1.5. Same for the other value. Then print a test object (a small thin-walled box is a good idea so as not to waste plastic) and measure it. Repeat until your dimensions match perfectly with the object.
If problem persists, check that your motors have enough current by adjusting the pots on the stepper drivers (make sure printer is off or use a special anti-static screwdriver).

As far as the nozzle sinking in the plastic, several things to troubleshoot:
1. Check that your Z is actually moving, that is the motors are moving on layer change.
2. Calibrate your Z steps/mm. This is done the same way I described above for X and Y. Move the Z axis 100mm. Check how much it really moved. Set your new Z steps/mm as the old value * 100mm / value axis moved.
3. Calibrate your extruder steps/mm. This time extrude filament in 100mm increments and check how much really extrudes. Easiest way to do this without wasting plastic is to remove your nozzle, disable cold extrusion prevention using the command M302, feed about 200mm of filament down into the extruder, then tension it and retract 100mm of filament, marking how much was really retracted. Then multiply your current E steps/mm by 100mm / actual value of filament moved.
4. After you print a thin-walled box fine, switch to a small solid cube. Calibrate your extrusion multiplier by printing it until the top of your cube is smooth (no gaps) and not concave (not protruding past the perimeters).

Edited 1 time(s). Last edit at 01/31/2017 04:13AM by anvoice.

It looks as though all your steps/mm are too small. Use the M503 command to find out the current settings. For the 43mm direction multiply the steps/mm by 55/43. For the 20mm direction multiply by 30/20. For the Z direction multiply by 8/3.7. For an initial test use the M92 command to change them, If the new value you got for the X direction was xfigure then the format is M92 Xxfigure. Repeat for the Y and Z directions. These values may reset at the end of the print.

If you can edit your EEPROM directly then use this method to set the new figures. Do another M503 and see if they are as you set them.

Edited 1 time(s). Last edit at 01/31/2017 04:27AM by Supermec.

Thanks a lot guys for the help. I imagined I had the steps pretty wrong but I wasn't sure about how I should caibrate. I will check with formulas given and come back.

Once again, many thanks!

Please don't use the suggested "Calibration" method!!!!!

The above method works well for the e steps and thats about it. Use this calculator to calculate your steps per mm

[www.prusaprinters.org]

If your print is still off, look at belt tension or maybe speed settings or stepper current.

I could go on a rant about why the above method is just a bandage solution, but Tom does a good job in one of his videos

[www.youtube.com]

Thanks for the info.

The issue is that I've already used that calculator so something must be wrong.

On the z axis I am using simple M5 threaded rods and the only thing I have to check is if the exact step of the thread is 0.8mm as the calculator has in it's the reprap calculator database (I'll go to the Diy I bought and check).

X and Y are normal - GT2 belt with 20 teeth on the pulleys. I use NEMA17 with 0.9 degrees and 1/16 drivers (with the jumpers in).

So clearly on the calculator I get 160;160; 4000.

To be sincere I tried in my first attempts 16 tooth pulleys as I couldn't really remember what I chose when I bought but it clearly moved the axis to much with those steps/mm. It hit the limits eventhough the travel I've inserted in Marlin was a clear measurment I made.

I will check the tension as you say also, though I really don't see an issue there. I don't see it skipping steps and really, my printed piece is very ok but the dimensions are clearly off by some %. It's like it scaled down the whole thing. It didn't mess the layers as I believe it would if the tension was too low on the belts. All the layers are on top of each other.

Allover the web I've seen steps/mm in Marlin EVEN like "164.68" so why would people have it like that if the calculator gives such clear answers?

With the Z axis the best way is to tell the axis to move at least 100mm and the measure how far it actually moves and adjust the steps to compensate for error.
On my M5 threaded rods i ended up with 4030 steps/mm to get an accurate distance

Belt tension is one of those tricky ones, Just because it's not skipping steps doesn't mean that its too lose.

Also look at the stepper current, and also your speeds. Turn down the acceleration and federate and see if you are getting the same dimension error. I think in marlin the acceleration is set to 3000, try 1500

As for why you see different steps/mm all over the web, well its simple. The early repraps had plastic parts and used builders rod as the frame not very ridged and hard to get square. so as sort of a hack the step/mm "calibration" method came about as a simple solution. Now most printers use an aluminium frame and are a lot nicer to assemble, as well as using proper lead screws for the Z, this hack or compensation is no longer needed. The thing is most people still follow the early methods blindly and dont understand why its a bad idea.

I'd like to see how that calculator solves a 50% size difference on one axis, or see a loose belt accounts for that much.

As far as the method being a bandage solution, it's only one if the steps/mm are already perfect. My guess is they're way off because as I said, loose belt doesn't account for that much inaccuracy.

If you have a 50% size difference on one axis thats a sure sign of a bigger issue.

And your right, it might not just be the belt, the bearings could be binding, the drivers under power, kinematics issue, there is a whole list of things to check.

I use to do the print a cube and change the steps/mm on my original prusa mendel, and for years was constantly re calibrating. Till I went back to the calculated steps then started crossing off other issue. Now my printer has printed everything I throw at it with in 1mm for the past few years.

Even the reprap wiki has changes its calibration pages and suggests going back to the original steps.

[reprap.org]

Im only here to help, at the end of the day its up to you how you setup your machine. Ive been been building Repraps since 2009 and always came here to search for help but never posted, thought it was time to give back to the community.

Edit

Sorry just re reading, if you have NEMA17 with 0.9 degrees and 1/16 drivers and M5 rod your z steps should be 8000 not 4000, that would explain the 50% in the Z

Edited 1 time(s). Last edit at 01/31/2017 07:00PM by scottybfg.

Yeah, I already figures that one out but the issue is that my steppers can't move the rod if set at 8000.

Also, that doesn't solve the other two axis.

Right now I was thinking about something else. I might have DRV drivers and I've inserted all 3 jumpers. That would mean that all over the axis I have used measurements in calculator for 1/16 but my drivers are set for 1/32.

Could that be it?

At 8000 steps/mm the default Z homeing speed of 4*60 is far too fast for your motors, this is why they wont move.
Try 1*60 or possibly 2*60

tell please where in marlin I can find that homing speed? Configuration.h?

In config.h its under "MOVEMENT SETTINGS"

#define HOMING_FEEDRATE

and yes if you have 1/32 micro stepping and used 1/16 in your calculation that wold affect the movement, if you post a pic of the driver someone here should be able to identify it for you

Ok guys,

Thanks for the help it seems I've made it work.

Just for the record, in Z yes I needed to have 8000steps/mm and yes the speed of homing had to be reduced. And so did the feedrate of Z + speed when changing direction.

Works llike a charm now.

Solved X and Y with 200steps/m as the calculator said.

PS. I had 1/16 drivers after all.

Thanks everybody for the support and understanding for a newbie.