006 :: MOAR stepper experiments


Fig 1
Today i will start off with some alternative values to the usual 0.2ohm Rsense on the A4983 to get a sense of how to further improve step accuracy. the alternatives used are 0R33 and 0R68
(Fig 1)
Fig 2

Fig 2 shows 0R68, the current supply permissible with this is only about 0.25A which is not enough to fully utilize the full torque of this NEMA23 stepper. It was then changed over to 0R33. Notice that I have also modded a smaller heatsink using a bent piece of bronze, this little sink is super-glued onto the top of the chip (and yes the little PCB is so full of flux after numerous swaps of test components, real mess there).

The reason for the change in sink is cos i noted that the usual ALU sinks over hang the chip area and sometimes seems to touch other SMD components under it, and there is no way to tell if it offsets and affects the actual contact of the sinking surface of the chip. So this piece of bronze taken from some surplus crafting material is cut to fit the 5mm x 5mm chip nicely (but still looks really crude, anyone knows a really fitting ALU sink for this?)

Vmot 27VDC, Rsense 0R33, R4 = 47k (fixed 1/4 watt trial). Vref adjusted to approx 0.5A drive. In this case, the u-steps appear to have been better stabilized in between steps and is looking very consistent for every step triggered. For those who do not know, this motor is rated@ only 0.38A.

Next up, i drop in a 230k trimmer in place of the fixed R4 SMD. and well surprise surprise, @ a vref of 1.15v (0.44A drive) using a R4 of 235k. the 400step/rev stepper shows a very very stable 1/16 u-stepping. contrary to the notion that higher R4 value will produce audible or noisy stepper sounds, it seems this stepper doesnt have that problem. Maybe this problem is more pronounce on steppers that use higher currents. With that, the test R4 is swapped up to a 440k trimmer and it seems really good too.

With the trimmer in place, I even did a sweep on it. my first time seeing it live reacting to a actively changing R4 value. In certain parts of the sweep, we can really hear some hiss/whines which are quite soft, but audible. It is very interesting to note that, they do not sound like the NEMA17 version which you shall hear.


the trimmer mods became like this (A4988). A small veroboard is added like a mini test bed to mount some trimmers. The 602 SMD pads are a challenge to solder using a normal non-pencil/pointy tip soldering iron and using some normal AWG28 wires, in total, about 700k of variable trim is added to try out a variety of R4 values.


Next, the sanyo 200steps/rev motor. And testing it on a new batch of al cheapo china A4988 drivers. Incidently, these come with 0.18ohm Rsense. In this run, Vmot is adjusted to 21v and we have a Vref of 1.67v (drive of about 1.15A), R4 = 145k. The coils do whine like a loudspeaker. But the u-steps where even better than before.

so in short, when faced with a fixed (or even unknown model) stepper inventory, by careful observation and adjusting of critical supporting component values. It is possible to achieve stable and very nice u-steps (some at the disadvantage of coil noise). The next vid is a negative example, with R4 nearly zero, the step increments becomes unequal (you may have to look at it for some time to see it) even though all 16 steps can be individually distinguished.

And this is an example of the "nintendo" noise created at higher stepping rate (arduino delay = 40). @1.1A drive the little sanyo gets hot really fast, and it is hot to the touch. A quick thermocouple probe shows the Tcase to be around 55degrees C. The Nema23 however does not suffer like little nema17.

(NO stepper drivers were harm in the making of this post/videos)
*edit* check post 044 for more updates :D

Popular Posts