Custom Stepper Motor Driver 2.3 - A3982

ThrustMe Wrote:
-------------------------------------------------------
> Duty cycle = U_DECAY / U_CHARGE = 0.322.
>
> Thus with two coils on at 1 A each, the input
> current becomes 2 * 1.0 A * 0.322 = 0.644 A.

I think duty cycle is sort of irrelevant here. What he needed was a psu that can offer the current when the current is at its peak. So peaks matter. Setting current for each coil at 0.7A, half mode current being reduced to ~70% of Imax, thats ~0.49A thus x2 makes ~1A, but this is output. Not input. You cant assume that input = output, so clearly input has to be higher. Either way, my conclusion is the 1A psu wouldnt of been enough. Besides, i wouldnt advice using a psu at 100% of its level, its better to run at at something like 80% or 70% of its max, but thats just me.

ThrustMe Wrote:
-------------------------------------------------------
> U_DECAY (R_DS(on) total = 0.66 Ω): (1 A) * (0.66
> Ω + 2.2 Ω) = 2.86 V.
>
> U_CHARGE (R_DS(on) total = 0.7 Ω): 12 V - (1 A) *
> (0.7 Ω + 0.22 Ω + 2.2 Ω) = 8.88 V.
>
> Duty cycle = U_DECAY / U_CHARGE = 0.322.
>
> Thus with two coils on at 1 A each, the input
> current becomes 2 * 1.0 A * 0.322 = 0.644 A.

If you dont mind, and if its not too much to ask, please explain those stepper "duty cycle" formulas, in words, what do they represent and how do they work? How do they map out on the current waveform? What is the shape of the current waveform? Thanks.

ThrustMe Wrote:

> Thus with two coils on at 1 A each, the input
> current becomes 2 * 1.0 A * 0.322 = 0.644 A.


I dont think i really get the math, but isn't this like and averaged (RMS / whatever)? Yet the power source would need to provide for the instantaneously 1 A right?'

Of course I read the datasheet!

ok I tried with the ceramic filtering caps. It does work better. It still fails some steps - fewer. But the thing is if i try to hold it a bit with my hand - simulating the load - it still fails quite some steps... - and it doesn't matter how fast i have the motor turn.

Edited 1 time(s). Last edit at 02/12/2013 10:09AM by alkalin.

Well i was asking him trying to catch things up, coz i mainly think he used some formulas without understanding. I think copy pasting formulas its a killer of the mind. The definition of statistics is "if you have your feet in a fire and your head in the fridge, then it can be said that in average, you are doing ok" ~well something like that.
If you want a psu for something and you look at amps, doesnt even matter which waveform there is, you need to look at max as in the peak instant values that are drawn at max time. In the big picture its a current waveform, sine, though doesnt matter for this purpose. And rms is just a method used to generally average things that arent average-able by other means and actually the rms trick is much more used in statistics (distribution) than in electronics. At an extremely small time scale i think there is a frequency in changing direction, but not quite a square form as it has parameters like off-time or blanking and decay that can be pre-set or controlled so that shape changes also. So not even that look square to me, nor relevant in terms of duty cycle, so i dont see where the "duty cycle" concept fits anywhere near this. Maybe he was actually trying to say something, and expressed it wrong, coz i really cant understand it.

About missing steps, if you have ceramics on V+ of 470nf and on 5V of 220nf, with electrolitics on both like 220-100uF, i think thats already more than the usual. You can make a short test with more caps, but if it does work like that then it means the psu has a big ripple. Another thing to generally check should be the ground, to have a good ground. Also, if it misses steps only under load, like it doesnt have the power to move, but without load moves just right, then clearly an increase in current setting should help.

If it still misses steps even with no load, the question is what are you using to step it, to generate the step signal. And are you sure that step line waveform is in parameters: e.g. try to make sure that step signal has at least 2usec high pulse width and same for minimum low time. Further on try 3-4usec min high/low time. Also coz that rc filter (r6c6) has a time constant of 1usec by itself. You might even try take that r6 and c6 out and see if it changes something, you can always put them back if you like to.