MB Faceting Machine Build

[MakkyBrown]

I'm going to be the guinea pig so to speak  .  New encoder has arrived, will be awhile till I fit this as the machine build is on hold till I finish other bits and pieces.The big question will it be too accurate and will step change flick still be usable as a stop. This was a bit expensive and twice the resolution of my current encoder. Has 0.009 degree steps, really hope it works and larger size doesn't cause issues.

[RoughCreations]

The big question will it be too accurate and will step change flick still be usable as a stop. This was a bit expensive and twice the resolution of my current encoder. Has 0.009 degree steps, really hope it works and larger size doesn't cause issues.

I have been looking at the Arduino encoder  programming code from Tom Herbst's book. You should be able to store the calculated angle in the first position of an array during each loop, moving the value from the previous loop into the next available array position and so on and then calculate the 3, 5, 7 (or more) moving average for each loop. For example if you found a 7 loop moving average was the most suitable, you would declare a float array in your code with 7 elements, and use that. The moving average value could then be used as a smoothed display value, or be used to display a moving bar. I reckon you will have lots of flexibility, and even be able to interpolate artificial angle values between steps using a bit of logic.
Hope that makes sense,
RC

[MakkyBrown]

Perhaps I could have an adjustable smoothing amount for when laps are very uneven. Preference being no smoothing Lots of possibilities.

MB

[RoughCreations]

This was a bit expensive and twice the resolution of my current encoder. Has 0.009 degree steps, really hope it works and larger size doesn't cause issues.

What's the diameter of the new encoder?

[MakkyBrown]

Case is 55mm OD plus a bit extra for the reader/plug part.
EDIT: Just thinking if there is clearance it might screw straight onto the mount I made for your facetron.

[FlashGP]

RC,

I think you would also need to check the sample rate, and keep it at a reasonable number, say 14 per second for a 7 step smoothing, a high sample rate would still cause a smoothed sample to jump around.

I recon 2 sets of samples per second would see the numbers move fast enough.  Perhaps 1 set per second would be enough.

[RoughCreations]

RC,

I think you would also need to check the sample rate, and keep it at a reasonable number, say 14 per second for a 7 step smoothing, a high sample rate would still cause a smoothed sample to jump around.

I recon 2 sets of samples per second would see the numbers move fast enough.  Perhaps 1 set per second would be enough.

Yes, that makes sense. The sample rate could even work on its own for smoothing, depending on the lap.

[FlashGP]

Hi RC, sort of, but a rolling average of a series of sample rated would take some or a lot of the lap jitter out.  At sub hundredth of a degree accuracy, even the flattest lap will induce jitter. 

It may be better to use the extra resolution to give a more accurate 1/100th of a degree than try to have 3 decimal places

[MakkyBrown]

You need the jitter flash, but you also need to be able to read the display. The stop is when the display changes to the target number with no jitter. The issue is going to occur when lap runout is greater than a single step on the encoder. Smoothing would most likely only be needed over the time period of 1 rpm.

MB

[FlashGP]

I was thinking of smoothing over 1 or 4Hz (0.25 to 1 second) at most to take out lap and vibration jitter. 

[MakkyBrown]

I wouldn't smooth with a 5000cpr encoder Flash, cut to the point the flicker stops.
If I need to smooth the 10000cpr encoder I'll smooth each rpm. I have a 500cpr encoder on my drive motor I may try and split the signal on to get an accurate rpm measurement.

MB

[Faceting Frank]

I take it your using encoders for both indexing and angle settings? Bit over my head, everything I have operates with a screwdriver or spanner.

[MakkyBrown]

That is the plan Frank. Jury is still out on how well the index stepper/encoder will work. Height measurement will just be steps like on a cnc machine and calibrated by homing.
MB

[Faceting Frank]

Just did a bit of maths on my cheater, which operates on a 5mm spring loaded thread, the thread pitch is 0.8mm. My index wheel is 157mm circumference x 96 index = 1.63mm per index. That means one full turn of the cheater = 0.5 of an index. I made the part of the turning screw that the scale is on 25mm circumference with 1mm markings. So, the scale on the cheater divides one index by 50. I keep a record of any cheater settings on facets if needed. On wide facets, say 8mm plus, the 1mm increment on the cheater scale makes a big difference on meet points. So 96x50=4800 increments I need for wide facets.
Hope that makes sense.

[MakkyBrown]

Thanks Frank for your feedback, atm I'm planning 3200 steps/increments, 6400steps maybe possible but both will require testing. So I am sort of in the ballpark I think going on your 4800.

MB