Hello, the 400 step motors are only the ones I happened to have - there's nothing special about them. 200 step motors will work fine too, but you'll just change a couple of settings during initial configuration.
The main spec to pay attention to when sourcing motors is the current - less 0.6A. The voltage doesn't really matter, but lower is better, for heat dissipation, if nothing else.

sn

In principle that's true, and that's the real reason I went for 400 step motors back in the day. In practice I struggle to see a difference, there's so much play in the rest of the system that the extra resolution ends up a little wasted.

For vector work, I'm sure it's irrelevant, but for pixel shading, with very fine pens, and at a high density (dark pixel), you could notice a difference. So yes, the maximum density of a pixel an inch wide would be higher with a 400 step motor.

Working it out, a sprocket moves 95mm in one turn, and has 200 steps. The interleaved stepping style that gets used on the polargraph gives us half-steps, so we've really got 400 steps. That's 95/400=0.23mm is the smallest detail that can be resolved. There is a lot more than 0.23mm play in this collision of string, gravity and friction, even if you do find a pen that really lays down a line that is that fine.

That said, where you will notice it first is in cases where the maximum possible resolution cannot be evenly divided amongst the number of lines to actually render. Imagine your maximum possible lines in the pixel is 100. You can imagine exactly what a pixel drawn with 100 lines would look like - a nice, even shade, one motor step between each line. One with 50 lines would look ok too, even, smooth, two motor steps between each line. But one with 99 lines would look bad - mathematically it would need to have lines with 1.01 motor steps between them, but that's not possible, so we end up with 99 lines with one step between them, and then a final two-step space, which ends up looking even more startling.

So a 400 step motor (effectively 800) would help there, not in terms of maximum density, but just to give more headroom really. I did do a bunch of tests of this kind of thing, ages ago, but I can't find the pics now.

In the PolargraphSD, the solution to this is microstepping. All of the addressing is done using plain full motor steps (200 or 400 steps are plenty enough to do the big moves), but then the hardware works internally with 8x microstepping, and because the pixel shading patterns are generated internally, it has an effective resolution of 3200 steps per revolution (or 1600 for 200 step motors).

sn

That's a good point. The first time I bought stepper motors I did it based on the voltage, because I thought lower voltage would be easier to drive, and got these 1.2v ones. Forgetting that low voltage usually means high current, so these things were rated to suck up 2A, and performed for a short time, moving weakly and erratically, before killing the motorshield and power supply. Oops.

Second time around I paid attention to the current rating only (Brian Schmalz's pages about the Easy Driver are a really useful plain-speaking resource) and had much better luck. In fact, contrary to my advice that lower voltage is better, if the current rating is fixed, higher voltage is actually better. So if you had two motors, both rated for 0.6A, one with a voltage of 6v and the other with 1v, the 6v would be better.

sn

I've just added a second choice - http://www.adafruit.com/products/324 which would do too.

sn