Will this laser idea work? On topic

It sounds like it would work, but aren't you off by a factor of 10 on the distance to the target? It looks to me like 1/10000 of a rev at the shaft would move the spot 1" at 1591".

Do you have access to a microscope you could fixture to look at a paper disk with a reference mark attached to the encoder shaft? Or if the guts of the encoder are accessible, you could look directly at the encoder disc. You don't need much magnification to resolve .001".

All that said, unless something is very wrong with the encoder, I doubt you'll find anything other than a very small amount of hysteresis in the detector.

I apologize for not responding further to your earlier requests, but really didn't have much to add other than to stress the need to pay attention to everything that could possibly go wrong when working to tenths, which I know you already know. Which method of driving the encoder did you settle on?

Ned Simmons

Eric,

This should work fine. Note that as the mirror moves through angle A, the reflected beam moves through angle 2A, so that if the target is 159 inches from the mirror and the mirror moves 360 degrees/10,000 =

0.000628 radians, the laser spot moves 2* 0.000628 * 159 inches = 0.2 inches. Should be easy to see.

D> This may seem off topic but it's not really. Measuring is important to

Boy, I need to proofread my posts. There were a lot of errors in the above. Especially the times ten answer. You are right, 132 feet would show 1 inch. Which is what I want. The driving method so far is a belt. I can detect no slipping at all. But I keep getting this .0009 error. I don't know if this is a display error because it's reading the encoder disc in quadrature. I thought about looking at the disc itself. It's 2.000 inches in diameter with 2500 marks on it. which makes the spacing about .0025". This is readable with a magnifier. But I thought that amplifying the error would be easier because I can see the movement at the same time as the display changes. I can very easily see 9 inches movement at 132 feet. And it turns out I have a target just about that far away. A nice clean cement wall. Thanks for reading and responding. Eric

If you've got an o'scope you can look directly at the encoder channels to eliminate any possibility that it's the readout, if that's what you mean by display error. Less convenient, but workable, would be a voltmeter or two. Or even an LED on each channel, as long as you're careful to limit the current based on the encoder's drive capability.

Ned Simmons

Ned, I have a scope. I don't really know how to use it. I need to get a book on oscilloscopes in general. The scope is a Tektronix 465B. The only probe I have is one with a red and black wire terminating in little hook shaped clamps. If this will work can you tell me how? Thanks, Eric

Hey Eric,

I think you "typo-ed" two places. The first mention of the shaft radius should be 0.159154943, rather than .19something; and later on you effectively multiply 0.0001 X 10,000 and get 0.1 instead of a whole number "1".

These didn't seem to affect the idea though, except what does fastening a mirror onto the shaft do for you? Why not just fasten the laser to the shaft, and measure the laser-beam movement at the

1591.54943" point as 1". If you don't have a 132 foot distant surface available, then you might consider using mirrors to multiply/divide that distance to something more practical, or better yet, make the 132 foot point right near where you are going to be working?. As his is a "counting" discovery, you don't need much "wall" space. You can count forward and backward 18 pulses in 3 feet of wall, or along a yard-stick!!

Interesting. Let us know how you make out. I've had very little to do with encoders except use in determining speed and the derived acceleration and deceleration, and they were 1024 X 2 models. Any chance that two encoders on the same shaft would help you? That's how we determined direction.

Take care.

Brian Lawson, Bothwell, Ontario

Greetings Don, I don't understand exactly. Oh, yes I do. I drew a picture and it's all clear. Duh. With the mirror at 45 degrees the included angle is

1. And the mirror only has to turn 45 degrees to make the light shine directly back on itself. Thanks, Eric

The first error that confused me is the radius of the shaft. You said it was .19515", which didn't make sense to me. Working backwards from a 1" circumference, I get .15915, so that was a typo apparently.

As someone else pointed out, if the mirror moves x degrees, the angle between the incident and reflected beams will be 2x degrees. [Sanity check: If you shine a horizontal beam at a mirror tipped at 45 deg, the beam will go straight up, or 90 deg to the incident beam.]

So if your mirror moves 1/10000 of a revolution the change in beam angle will be 360 * .0002 or .072 deg. Tan (.072) = TanX = ~.0012566 (TanX is just a name I made up for this discussion.)

For your measurement, TanX = m/d, where d is the distance to the target and m is the measured movement. So for m = 1",

d = m (1/TanX) = 1 (1/.0012566) = 795.8 inches = 66' 3.77"

I would think that distance should be more than adequate to see what is going on. I think I did that all correctly; feel free to critique anything I messed up.

Seems the whole apparatus will probably be a pretty good seismograph, so stable setup and settling time may be important. Quality of the mirror surface may matter too. I don't have any real experience on that.

As far as I know, most encoders of that resolution are optical. I have used US Digital encoders myself (which I think is what you were planning) and never noticed any backlash. They sometimes have absolute specifications that are less than the number of bits they output. I take that to mean they will detect fine steps, but the number may not be completely accurate relative to the part of 360 deg.

I'd be really surprised if the encoder had any backlash-type uncertainty that was much greater than one count.

-Rex

What encoder are you using, and what is it connected to?

Brian, The mirror is about 1/2 inch on a side and .09 thick. The laser is 1.5 inches dia and 8 inches long. With wires coming out. I feel it may have more influence on shaft movement if mounted directly to the shaft than the mirror will. Direction is determined by only one encoder. It has two readheads out of phase with each other. So A leads B turning one direction and B leads A in the other. Thanks, Eric

A US Digital H6S-2500-I. And I realize I made a bunch of errors in my first post. I am using the ED3 display from US Digital too. I need to eliminate the encoder as a source of error before I continue. I don't think it's the problem but so far it seems like the culprit. Eric

I was just thinking... Why not make a test setup to check the encoder outside of your intended system?

Say you want to move the shaft 1/10000 rev with .01 inch applied. The circumference required will be .01 * 10000 = 100". The radius for that circle is about 15.9".

Why not attach a light, but firm, arm to your encoder, 16" long. To remove as much load as possible, put the encoder so its shaft is horizontal with the 16" arm hanging straight down. Clamp a micrometer so it touches the rod near the end and put a light spring (rubber band?) to pull the lever against the micrometer barrel. Now, moving the micrometer .01 should move the encoder 1/10000 rev.

Obviously, you could shorten the lever by some factor and still get decent measurements.

Just one other approach, if you feel so inclined.

-Rex

You don't really need probes for this, Eric, though you can use the one you've got on one channel. Plug it into CH1 on the scope and hook it to channel A of the encoder with the encoder connected to the depowered readout. (I'm assuming this is not a differential encoder - it seems unlikely - you'd have wires called A and A', and B and B', or the like if it is.) Connect the ground clip to the readout chassis or signal ground.

All you need for encoder channel B is to make a connection to the center pin of the scope's CH2 BNC connector. If you don't have a mating connector, you can stick a pin in the center of the connector and use a jumper, just be careful not to damage that nice old scope.

Set the toggle under the VOLTS/DIV knob to GND on both channels.

Push in both CH1 & CH2 pushbuttons on the left of the screen, all the rest in that group should remain out.

Set the TRIG MODE to AUTO.

Set the TIME/DIV KNOB to around 1ms. Make sure the coaxial knobs are locked together.

Now fiddle with the POSITION controls way over on the left to get the two traces separated on the screen. If you don't get the traces, try fussing with the SOURCE and LEVEL on the lower right. (I'm winging this by looking at a picture of a 465B that Google found for me.)

Set the toggle under the VOLTS/DIV knob to DC on both channels.

Set the VOLTS/DIV knob to 2V on both channels. If the probe is a 10:1, set that channel to .2V

Turn the readout power on. You should the encoder transitions when you turn the shaft. Play with the trigger settings if you want to synch the sweep to the signal with continuous rotation.

That should get you started. Don't be afraid to play with the knobs, you can't hurt anything. Hopefully someone else will chime in if I've missed anything obvious.

Ned Simmons

Thanks Ned. I'll let you know how it turns out. Eric

What I'm trying to do is move the encoder in the system and see what happens. Since I have the laser, mirror, and some clay the setup will be easy. I don't want the measuring to influence the encoder. So if the carriage moves .0008 and the display doesn't I can't tell where the problem is. If the carraige moves .0008 and the encoder shaft turns the requisite amount, and the display doesn't change then it's the encoder. But if the carriage moves .0008 and the encoder shaft doesn't move, and the display doesn't change, then the error is in my setup and I can look for it. I don't think it's the encoder, but it seems to me that the mechanical setup is backlash free. So until I eliminate something it won't be solved. And the encoder check is the easiest so far. Eric Eric

H6S-2500-I? That's 2500 CPR (Counts Per Revolution). For a shaft with 1" circumference, that gives a resolution of .0004". I'm thinking you've got an uncertainty in the +- 1 count range, which is pretty close to what you are seeing.

Had to go look to refresh my memory. As I recall the best A2 encoders, which I have used, are spec'd to 12 bits output although they give 16 bits. 12 bits is 4096, or 2.4 tenths in your application.

I think you need a smaller shaft.

This should work. You can shorten the beam path by folding it multiple times - ie, multiple bounces with mirrors.

Be sure to use first surface mirrors or the extra reflections will drive you nuts.

Basically you just want to calibrate your encoder.

Jim

Actually, since the encoder is being read in quatrature the resoultion is 10,000 pulses per rev. When moving the carriage one direction it counts right along with a 1/10000" .500" travel indicator. When the direction is reversed the carriage must move .0009" before the display shows a change. After the .0009" movement is satisfied everything works OK. The problem occurs no matter where in the .500" of the indicator I use. It also happens if the indicator is moved anywhere in the 8" travel of the carriage. If the problem still shows with the laser measure then I'll try it with 1X counting instead of 4X. But that should not make any difference. I think. Eric

Hey again Eric,

I must be missing something about using the mirror. Maybe I should just shut up and see what you do, or the results anyway, but that's not me!!

My lack of understanding is in/with the deflection of this mirror. As I understand it, you are going to

1 - fasten a mirror to the shaft, 2 - shine a laser beam on the mirror, 3 - note the point this reflected beam contacts a wall 132 feet away, 4 - make a "mark" on the wall at that point 5 - rotate the encoder one pulse (by turning or stepping the shaft to create a single "output" pulse of the encoder) 6 - repeat step 5 for ten or more "pulses", and then reverse the . step direction for the same number, and see if there is a . difference between the start point and the end point.

To me, that should work, but if you are going to measure the distance the beam "moves" from pulse to pulse, then the mirror will cause a "doubling" deflection type error with each pulse. That was the reason I suggested fastening the laser direct to the shaft.

Am I wrong about your intent, or the error?

Take care. Watch the laser in the eyes.

Brian Lawson.