Main > Main Forum
TS-DOT-02 Turn Count Question
BadMouth:
Don seems to have it figured out their button hole spinner: https://www.amazon.com/gp/customer-reviews/RPPSHKXFPFA95/ref=cm_cr_getr_d_rvw_ttl?ie=UTF8&ASIN=B08HNB5933
--- Quote ---There is considerable disagreement about the pulses per rotation of this spinner. An accurate number is needed to properly calibrate sensitivity for each game in MAME. With MAME on the PC playing Tempest, a sensitivity setting of “9” gives the correct 5 lane per rotation sensitivity. Reversing the sensitivity formula, that means pulses per rotation is 72 * 100 / 9 = 800. Others have measured it at 780 ppr. 780 works well with the MAME scaling formula for Arkanoid and any game I have tried. The 4800 mentioned by the seller may be the native rotation of the encoder, but the USB interface provides only about 780 PPR. The PPR value is unaffected by Windows mouse pointer sensitivity settings.
UPDATE: I tested the spinner in Windows 10 using a tool called "MouseTester v1.4". Using the data collection feature, one full rotation of the spinner provides exactly 775 counts, which agrees with others' testing and empirical results in MAME. Use 775 PPR for MAME on a PC with the default Microsoft mouse driver that is installed when the unit is plugged in.
--- End quote ---
Perhaps the same methods could be applied to your model.
bluewave8:
--- Quote from: PL1 on November 30, 2023, 03:16:09 pm ---
For most people, the more rational and reasonable approach to playing the widest variety of games is to approximate.
The key is answering the question, "How close is close enough for me?"
--- End quote ---
I might be rational but not reasonable ;-)
As far as how close, this is certainly a good thing to think about. Close enough for me is probably around 3-5% or less error. I'd like for the average person to not be able to tell the difference in the angular sensitivity of the spinner if they went from my MAME cabinet to the authentic controller.
--- Quote from: BadMouth on December 01, 2023, 08:24:07 am ---
Perhaps the same methods could be applied to your model.
--- End quote ---
Thank you!! Great find. I found and downloaded that software and am trying to get a good reading now. The software from Tsticks that lets you change the CPI from 360, 720, and 1440 DOES change the output in X counts according to the MouseTester software. Maybe the reviewer was referring to the Windows CPI/DPI settings and did not have the Tsticks software, or the software doesn't apply to that Tsticks spinner.
After 23 tries of measuring "one" full rotation with the CPI set on 1440, my average counts reading was 3710 with a max of 4142 and a min of 3216. This is too high of an error rate so I'll have to keep trying. This has been with eyeballing an allen key that is coming out of the setscrew on the side of the spinner knob. I'll have to try to rig up some sort of fixture to get a more repeatable measurement.
BadMouth:
Assuming US Digital makes industry standard encoders and yours uses an industry standard encoder, the most likely cpr in that range would be 4000 or 4096. I have no idea why there aren't options in the 3k range. Must be some reason.
https://www.usdigital.com/products/encoders/incremental/shaft/s16/
EXIT: Eh, 4k is awful high for the min and average you were getting. Hope you can get repeatable results.
bluewave8:
Thanks for that additional info.
I wanted to report back with new findings.
The MouseTester software default action is to have you click and hold the main mouse button while you move the mouse, then release when your measurement is complete. For this push/pull spinner, I had it set up so that pressing down activates the left mouse button. Apparently, when I do this, the spinner sends pulses for its Z direction that somehow get counted or effect the sending of the X direction as well, which is why my previously reported counts were so high and so inconsistent. I changed my test method to use F1 to begin collection and F2 to end data collection.
With that method, the results are much more repeatable. The latest results have a min of 1412, a max of 1468, and an average of 1435. However, the fixturing I'm trying to use for this isn't the best. I'm using two allen keys; I have one allen key extending from the set screw of the spinner knob, and another one coming up perpendicularly from the control surface. The allen key from the set screw will start its rotation by having one of its sides contacting the left side of the vertical allen key, and will end "one" rotation by contacting its opposite side against the right side of the vertical allen key. This would have my maximum measurement off by at least the "width" in pulses of the vertical allen key. The width may be around 100 pulses, which would bring the total of one revolution closer to 1535, but it was very difficult to attempt a measurement of the width of the allen key in pulses (not very repeatable).
**EDIT- just realized that I'm actually off by the diameter of both of the Allen keys. They are very different sizes, the larger vertical one is around 100 pulses, the smaller one is around 20-40
Unessential:
Just based off trial and error (Arkanoid requiring roughly 120-130 degrees to go from one side to the other, and Tempest going 5 sections per full rotation)
I assumed that the CPI = CPR (it's not) I ended up having to multiply the entire thing by 0.6 to get the correct sensitivity when spinning slowly (so, reversing the math, 600 CPR?)
Notice i said while spinning slowly, I noticed that if I spun the spinner fast it would register a shorter distance than if i spun it slow. -- which didn't seem right, because that would mean the numbers people refer to would need to also specify a rough speed to be useful... But I guess it's a good starting point.
Edit: Just wanted to add that I have BSP rather than DOT, but As far as i know they are the same spinner with just a different mounting option plus push pull functionality