003 :: Stepstick power NOISE !
Today i am tinkering with some power supply noise measurements on the Vmot pin (using different condenser combinations). In this example, i am using a +22VDC PWM supply, powering a NEMA23 stepper that is drawing about 0.5A approx. The following are A4983 stepsticks (followed by a series of A4988s later below)
(Fig 1) This is the scope view (using the al cheapo DSO203). A single 0.22uF across Vmot+GND. Ramps 1.4 board modded with additional 2x100uF, 1x 0.22uF and 1 huge 2200uF to create a rather large cap reservoir. as such we can still see some 100mV worth of noise going around the Vmot pin.
(Fig 2) +100uF (+0.22uF) is tied across the Vmot+GND, we see some nice smoothing.
(Fig 4) Stepper swapped, NEMA17 sanyo. drawing approx 1.2A. Vmot is now @ +14VDC.
0.22uF alone. almost 200mV of noise
(Fig 5) 100uF additional. notice the ringing is gone. noise reduced to about 110mV
(Fig 6) +2200uF + 0.22uF. noise goes just under 100mV
(Fig 7) +2200uF +100uF +0.22uF --> quite alot of noise taken off, almost 75mV of noise.
(Fig 8) STEPSTICK swapped A4983 --> A4988
Now while trying out this cheap china A4988. I am not sure if this is a PCB design flaw but the noise is really amazing.
(Fig 8/9) these 2 are rough captures. Vref approx 0.26v (noise is really mayhem !). on the A4983s,@ 5ms timebase, there is no such large sine wave to be seen. the spikes are so intensely erratic that it is quite tough to get a stable picture @ 5us timebase.
(Fig 10) Vref adjusted to 0.3v (notice at anytime there is a really nice spike running 100mV++ riding the wave)
(Fig 11) Vref adjusted to 0.4v (noise going over 300mV). my guess here is that the R-sense is not really 0.05 ohms, and it is giving the chipset hell about what control current to give, at 0.4v --> it is telling the chip to give it all out (supposedly 2A++). However, based on some stepstick guides, @ 0.4v, the chip is only giving 1A--> which is impossible because the chip is frying itselves off to the moon overdriving and overheating a poor little NEMA17!!! Sometime later on in my tinkering, I will swap the Rsense on this stick to approx 0.33ohms and see if this same noise pattern persists. I have also ordered 2 other A4988 from other suppliers to see what are their differences with this perculiar piece. Based on manual torque feel, the 0.26v Vref gives somewhat a full torque sensation from the steppers. Meaning the Rsense on this strange A4988 to be approx 0.02ohms (DUH??? did someone designed this stepstick to output 4A???)
(Fig 12) Vref adjusted to 0.14v. (all of above are done with 1x 0.22uF across Vmot)
(Fig 13) Vref @ 0.26v additional 2200uF + 0.22uF (noise just above 100mV). at this rate, the noise accounts for over 1/3 of the sense voltage, this definately severely depreciates the chips ability to control the u-step output capabilities.
this piece of A4988 is really interesting (from reprapdiscount). it is really very hard to control any stepper on it, and the sink temperature is classified as ---> beef well done! for newbies who do not know the exact science, this will present many hours of tinkering and pondering about the source of strange problems (like steppers going up in smoke? having a nice burn on the fingers?) ...
so the conclusion of this run is :
1) to feedback to the RAMPS blokes to take note of vmot noise. possibly fatten supply tracks (triple it?)
2) to mod in actual, all stepsticks vmot pin to incorporate multiple condensers to try and achieve a noise value of below 75mV (hopefully)
3) what is the purpose of clearing up all the noise? so that the A49xx chipset can achieve a better current sense and achieve more accurate u-step control. (and therefore this test run is a success)
4) (pictures not shown) when mechanical resistance is applied to stepper shaft, there is an increase in Vmot noise, cant say for sure how much but it is roughly 5-10%. I hope folks take this into consideration in future mods.
care has been taken above to avoid noise from any USB channel (therefore USB is not plugged into any devices including the DSO)
1x cheap china A4983 stepstick was harmed in the making of this blog. *USD$8 poofed*
cheers yall !