Google+ post by Matt Joyce on 2013-01-02 00:48:36 UTC

Another example of people not learning from lessons of the past year.
This is not how you set your steps per mm value.

Ok, that’s interesting. Can you explain why and/or point me in the direction of the correct method?

Steps per mm are a direct result of your stepper’s resolution (usually 200steps/revolution, times the microstepping resolution, which usually is set to 16x), the pulley’s number of teeth and the belt pitch. With those numbers you can precisely calculate many steps you need to move your extruder or bed a certain distance - that’s your steps per mm value.
Any other irregularities regarding the final dimensions if a print are a result of other settings being off - most likely the flow rate, which (at least in Slic3r) is set by the filament diameter and an extrusion multiplier. Basically, you calibrate you extruder (as of Richrap’s tutorial), measure the filament diameter and then use the “single wall thickness” Slic3r puts at the beginning of every .gcode file it generates and compare it against the real width of the skirt to set the extrusion multiplier.

If you set steps per mm with the method you linked to above, while it might give you accurate dimensions for the exact case you calibrated for, the dimensions on anything else will be off, especially on larger and smaller prints. And of course, the printer is still not calibrated for the right flow rate, so the prints you’re getting could possibly still be way too weak (when you’re extruding too little) or look ugly (extruding too much).

I have used the richrap tutorial for calibrating my extruder, but how do you calibrate the x/y/z motion if you’re not using the steps/mm to correct your motion? Is there another tutorial that covers that somewhere?

@Laura_Walton When you have your steps/mm set to the right value and calibrated the flow rate, you should get pretty precise parts. Any inaccuracies that remain are not something that you can calibrate away, but are the limits of what the FFF process can achieve on that specific machine. To improve from there, you’ll need to improve the machine itself - using better belts, pulleys, linear bearings, a smaller nozzle, more precise filament and so on.

Hopefully @Matt_Joyce doesn’t mind me hijacking his post here a bit. So @Thomas_Sanladerer I have a printrbot. And the printrbot community is pretty hot on this calibrating your steps per mm method. Last night I went home and spent hours printing calibration objects.

I recalibrated my extruder, extrude 30mm… measure, set steps per mm. I realized I hadn’t done so since my large extruder gear broke, and I replaced it. So there was some significant change.

The first thing I noticed was that having used the method recommended in the printrbot manual I had set my x and y steps per mm to be different. Less than 1 step per mm different but different. So I picked one of the values and set it to match the other and reprinted and my nickel wouldn’t fit. I changed them both to the other value, and it fit. So clearly and obviously in retrospect they should be the same.

I then printed a 20mm hollow cube. It is coming up like 20.2mm in each direction.

The wall thickness as specified in the gcode is .71mm, layer height .2mm, with a .5mm nozzle. I am getting .73mm thick walls.

Is this all close enough? should I be messing with my extruder calibration again to try and get .71mm thick walls?

I don’t have a US nickel, but I am interested in knowing how good the quality can be (should be) from a Printrbot (LC).

@Laura_Thomas you can set your steps/mm to the calculated value and adjust the extrusion multipler in Slic3r (71/73 ~ 0.97), but that’s about it.

+Matt Joyce, you could just use openscad or something else to make a device to fit your local currency. A US nickel is 21.21mm in diameter. But presumably the size of your local currency is documented somewhere.

When it fit correctly it was snug and wouldn’t pop out when set down/dropped small heights, but would push out with gentle pressure.

@Matt_Joyce You can also just look up the diameter of some non reeded (aka smooth edged coin on your local currency) and then make an openscad object that has the same dimensions.

I made a test object that was a 30mm x 30mm x 4mm cube with a 21.21mm hole in it. 21.21mm being the diameter of a US nickel. When correctly calibrated it fit very snugly, popped in, and out. When held up to the light after a few dozen itterations of demoing the fit I could see a bit of light leaking around the coin. It might have leaked when it started too.

Ooh quick search of thingiverse shows me this which might have your local currency included.