encoders- and troubleshooting

[Kev Williams]

So after bragging up the GCC the other day after cleaning the #1 mirror in the bottom of the machine, I engraved a bunch of white/black Romark since then, was nice to be able to back down the power substantially--

Yesterday my new V-Tech 'see-who's-at-the-door' phones & video doorbell came yesterday. After hooking it up, I engraved this small sign to point out the new doorbell.
be1.jpg
As you'll notice, it looks HORRIBLE!

So I did another test, repeated a bit of engraving that's been turning out very nice the past 2 days--
be5.jpg
--now, this lettering is pretty small, but it's HORRIBLE!

Not sure exactly what's going on, but the last time something similar (<keyword) happened, cleaning the encoder wheel on the X servo did the trick.
So, I pulled the motor, and sure enough, the wheel's pretty gucky...
be2.jpg

Now, this is the first machine with encoders I've ever owned. I understand the hall-effect thing, but I didn't understand how this thing tripped an HE switch. I mean, the edge is transparent-
be6.jpg
-I would expect everything BUT the edge would be transparent!

However, it wasn't until I checked these pics after I took them, because I couldn't see this with just my old eyes-- and now I get it!
That's a lot of hash marks! or teeth, or whatever they're called!
be3.jpg

And it's also clear why these things need cleaning, pretty soon the guck can fill up the gaps between the teeth!
be4.jpg

Are these machines cool, or what?!

Ok, so I clean it up, put it back together, and Voila! Uh, wait a minute--- NOTHING CHANGED!!??
My next test looked exactly like the first test...

AGGHHH... Ok, so maybe the Y motor encoder needs cleaning... Do you know what a pain it is to get THAT one off? (I'll bet Gary does!)...
And once I got it out, and got it apart, the encoder wheel looked like brand new. Not a speck of dust on it. But I cleaned it anyway, and re-centered it within the HE trigger, got it back on and--
--same bad engraving...

And then it dawned on me-- which is where we come to the "-and troubleshooting" part of this story...

Yesterday I removed the #3 and #4 mirrors, and the lens to clean them...
Check the lens---- tight.
Check the mirror over the lens---- tight.
Check the #3 mirror---- and it's anything BUT tight! Just kinda sittin' there rattling around...
I do remember tightening it, but obviously I didn't put much muscle into it!

Snugged up the mirror and----MUCH better--
loose mirror version on top, tight mirror version on the bottom...
be7.jpg

--all the difference in the world!

So I spent 2 hours tearing out servo motors and cleaning their encoder strips, when the problem was nothing more than a loose mirror...
Not that the encoders didn't need it, and now they'll be good for another year!

But I SHOULD HAVE checked out all the EASY stuff first!

[David Somers]

Dont feel bad Kev,

I am one of those people who can calmly listen to someone describe a problem and then listen to a someone else say the problem is likely in the hanafran which will need to be removed and dissasembled and cleaned and Oh boy what a job that will be. Another person will strongly suggest the whoozi assembly is gone and describe an insanely complicated process to go through to deal with it. And i will suggest using a kleenex to wipe off the whatsit and see if that helps. People will look at me oddly and then finally ask, what? Why would you do that first? And I will say because it could be involved and it is the simplest thing to check. Often they will do it and voila! That took care of the problem.

However....when the problem is mine to deal with personally my pea brain is drawn inexorably to BOTH the hanafran and the whoozi assembly before I turn to wiping the whatsit.

Brains (or lack thereof) are wonderful things aren't they?

Glad you got this figured out without getting too much deeper into the pit Kev!! <grin>

Dave

[Bert Kemp]

Hey Kev ya know the best thing about readin that whole post? Was you bangin your head against the wall

[Gary Hair]

YOu only spent a few hours... how about spending MONTHS with every software configuration possible only to discover the coupler connecting the stepper to the chuck was loose the whole time!
Oh, and I'm glad I can say that I've never left a mirror loose... That's a lie, but I'm glad I can say it

[Bruce Volden]

Kev

Those "hash marks on the round "thingy" are contrast readers, the encoder reads the light/dark areas and can keep track of where it's at in regards to positions.
Glad you found the culprit though!

Bruce

[Kev Williams]

'contrast readers' - optical trigger rather than magnetic (hall-effect), now I know what to call it!

[Bruce Volden]

I'm certain there is a more scientific name for it but, that is what I've always been told was the "proper"? description.
I fixed stuff for the USPS when automation was coming into age and had to learn all this new age stuff They sent us down to
OK for schooling more times than I thought was possible! Tried to bring us to speed on what was coming down the pike.

Some readers had strips of industrial plastic with the lines spaced for particular machine speeds, some had holes punched
into a disc rotating at a known speed etc..... All the same.


Bruce

[Kev Williams]

So, since servo's don't know what to do without an encoder's 'steps', is it safe to call them 'stepper motors'?

It IS cool that the machine knows where the laser is pointed whether IT moves the laser head, or I move it. Makes it SO much faster to do red-light alignments!

[Glen Monaghan]

It is not "safe" to call a servo motor (with or without its encoder) a stepper motor. A servo motor is totally different construction and operation than a stepper motor. Steppers only rotate in increments (or steps, hence the term stepper) and the coils have to be pulsed in sequence for the motor to work, plus you change rotation direction by changing the sequence of coil pulsing. Servo motors are more traditional motors that will rotate continuously when voltage/current is applied, and you change rotation direction by reversing voltage/current polarity. That said, you could add an encoder to a stepper and create a stepper-based servo, but I don't think you will find any commercially available servo built that way.

[AL Ursich]

This is the optical encoder for the Hobby Grade CarveWright desktop CNC. It homes by bumping into the stops. Dust is a problem...

Until I saw the clear edge I thought it was a silk screened or printed magnetic bumps. The picture of the clear edge solved that mystery.

Great Pictures.

AL

[Glen Monaghan]

That's about the simplest, cheapest thing that you could legitimately call a servo motor. It uses a (rotary) optical encoder, which is that plastic disk with the radial markings, along with an optical interrupter switch (usually at least one led on one side and at least one photo sensor on the other side of the disk) to monitor rotation of whatever shaft it's mounted on (typically the motor's output shaft). When the controller applies a given power level to the motor, the shaft should rotate in the direction dictated by the power's polarity, with a speed dictated by the power level, so the encoder output should pulse accordingly. If there's only one photo sensor, then the encoder merely confirms rotation speed (e.g., the shaft isn't stalled); two photo sensors are needed for that disc to verify rotation direction. The big limitation with that arrangement is this "rotational" servo is that it doesn't actually know where the shaft is rotated at a given moment, any more than a stepper motor knows its current shaft rotation. Like with a stepper, you need to set a "home" or zero position and then count pulses (from the optical encoder) to determine current position. It does have the advantage over steppers though, in that you can tell if the shaft over or under rotated by monitoring the photo sensor pulses.

"Better" versions of this rotary optical encoder have multiple rings of marks with corresponding numbers of optical interrupter switches. The multiple rings of marks are laid out to subdivide the disk. If you have one ring with 1 mark covering half the circumference (and clear for the other half), a second ring with 2 marks each covering a quarter of the circumference and separated by a quarter of the circumference, and so on up to an Nth ring with N marks each covering 1/N of the circumference and separated by 1/N th of the circumference, then the rings create an N bit binary sequence (gray code is the most reliable format but that's an implementation detail) that subdivides every rotation into 2**N parts (so 4 rings would resolve shaft rotation to within 1/16 of a rotation (360/16=22.5 degrees), while 8 rings would resolve to within 1/256 of a rotation (about 1.4 degrees). If you don't ever rotate more than one revolution, that gives you "actual" rotation, even after losing power and/or back driving the shaft. If, though, you allow for multiple revolutions, then you still need some way to set a home/zero position and to count full rotations (the encoder can provide partial rotations at any time).

Epilog uses a linear version of your rotary disk encoder to track the head's x-axis movement. They have a long plastic strip with alternating black and clear lines mounted across the top of the gantry, with an optical interrupter that straddles the tape and moves with the head. I don't know how many tick marks are on the strip but, after homing, the controller always knows the head's x position to within a tick mark or less. But, just like your inexpensive CarveWright encoder, dust can bedevil Epilog's linear strip as well, so standard maintenance and troubleshooting procedures include periodic cleaning of the strip and optical interrupter.