[pixpop]

I'm attempting to use a chinese glass linear encoder as feedback for a linear motor.

eBay item 170462020380 is an example.

My question is, how do I know if I have it mounted correctly? The info from the vendor is of course sketchy at best.

It seems very sensitive to alignment.. if it's not aligned just right, it binds.

No matter how I align it though, it seems there's something touching the glass scale inside. I say this because if I put my ear very close and move the positioner back and forth manually, I can hear a sound as if something is rubbing on a surface with tiny regular ridges.. the pitch of the sound increases with velocity.

It's not a grinding or scraping sound. Imagine rubbing your fingernail across the surface of an LP record.. it's a bit like that. It's very quiet, and I have to listen very carefully to hear it.

Is it likely they have something touching the scale inside? I would have thought not, but maybe there's a wiper or something to keep the scale clean? (I can't see inside the scale to verify).

I've been shimming and adjusting, and have got it to move freely. It feels right, and looks right, but this sound bothers me.

Are these scales very fragile, easy to damage?

[I'lldietrying]

If it's like the the standard DRO setup that measures axis travel, the rubber wipers that are supposed to keep the crud from getting inside to the scale may be dragging on the body of the reading head. I mounted a new scale/reading head on one of our lathes a couple of weeks ago that felt a little funky. It doesn't bind or anything, just feels like the wipers are tight.
I didn't have any problems making the linear error compensation. Just wrung together 300mm of gage blocks, indicated them parallel to the spindle since it was for the "Z" axis and referenced the length with a 1µ indicator. Entered the correction factor in as ppm and so far it has repeated just fine and everyone has been satisfied.
The scales I'm familiar with have plastic clips to hold the reading head and protect the scale during transport. They should be removed after installation of the scale, but I've never noticed them interfering with anything. I've seen scales in use that never had the clips removed.

[Jason3]

Fwiw, my Acu-rite scales have rollers that guide the actual read head along the scale. Its quite possible they could make the noise you're describing. They would handle quite a bit of misalignment without damage, though of course the accuracy would suffer unless a correction was applied. It wasn't hard to align them better than 0.01 mm though, so I did.

Regards,

Jason

[I'lldietrying]

Fwiw, my Acu-rite scales have rollers that guide the actual read head along the scale. Its quite possible they could make the noise you're describing. They would handle quite a bit of misalignment without damage, though of course the accuracy would suffer unless a correction was applied. It wasn't hard to align them better than 0.01 mm though, so I did.

Regards,

Jason

Lol, am I misreading your post or am I that unlucky? Come to think of it, let's leave my luck out of the equation because sometimes it isn't too good. Are you saying you haven't had to apply a linear error compensation factor when replacing the scale and reading head?
I've mounted replacement scales quite a few times, aligned the scale/reading head assembly with the axis being measured/ensured that there wasn't any binding caused by the mounting surface/fasteners, compared the actual distance traveled(based on gageblocks or some type of a calibrated standard/known value ) to the value displayed on the readout and I almost always have to apply a linear error correction (entered into the readout itself in ppm). Generally I try to get a manual mill or lathe to consistently match the DRO within 5µ unless the machine tool isn't capable of it.

[Jason3]

So far I haven't had to apply any correction, no. Maybe I need to be more fussy What would be a typical amount of error correction you'd need? My feeling was that the error I was seeing with the last set I did would likely be cancelled out when winter rolled around and the workshop cooled down - and possibly was accounted for by poor technique, or stiction, or any other kind of error that was as likely as the scale needing compensation. I felt I'd be as likely to make it worse as better, basically!

[I'lldietrying]

So far I haven't had to apply any correction, no. Maybe I need to be more fussy What would be a typical amount of error correction you'd need? My feeling was that the error I was seeing with the last set I did would likely be cancelled out when winter rolled around and the workshop cooled down - and possibly was accounted for by poor technique, or stiction, or any other kind of error that was as likely as the scale needing compensation. I felt I'd be as likely to make it worse as better, basically!

If you have your Acu-Rite DRO manual handy, look for a section called linear error compensation or something close to that. It has a formula that is used to calculate the value. It is something along the lines of displayed measurement-actual measurement / displayed measurement. I may have that formula backwards, but it's in your Acu-rite DRO manual.
The displayed measurement is the value that the readout says you traveled. The actual value is the value of your gage block stack or reference standard.Sometimes the value will be a negative value and sometimes it will be a positive number depending on your displayed and actual values.
The largest value I ever recall using was a -286, but remember this value is in ppm (parts per million).
Make sure to look up the formula in your Acu-rite manual, because I may have it reversed.

[I'lldietrying]

..and I am known for being a tad fussy. I've just always felt that if a machine was a clunker, then a DRO wasn't really very helpful and if it's a good machine tool then I want to give it every chance to make my life easier.
The lathe I replaced the scale on was a Hardinge toolroom lathe (a pretty accurate machine tool) and the last mill I did was a manual Makino mill..but it's a manual mill that uses a ballscrew for movement. So again, I want to take full advantage of the ballscrew by tweaking the scale/DRO as close as is reasonable.

[I'lldietrying]

.... I felt I'd be as likely to make it worse as better, basically!

That's one of the good things about doing this. Make sure to write down the initial value that is currently the correction factor entered in your DRO. If the process doesn't go well and you want to just put everything back to where it was until you get it figured out, then you just enter the original value back into your DRO.

[pixpop]

The scales I'm familiar with have plastic clips to hold the reading head and protect the scale during transport. They should be removed after installation of the scale, but I've never noticed them interfering with anything. I've seen scales in use that never had the clips removed.

Mine had some red plastic shims that were screwed to the aluminum cover, and prevented the read head from moving. I had to remove them before installation, because once installed the screws holding them in were inaccessible. I was able to use them as a gage to set the spacing between the read head and the scale.

[I'lldietrying]

Here's a link to the DRO on the Hardinge. You could DL the manual and find the formula in there.
http://www.acu-rite.com/view/200s-readout-systems.aspx
..and here's a cut and paste from the specs on it.
Error Compensation
Linear and non-linear compensation up to +/-9,999 ppm

It has error compensation up to 9,999 parts per milllion for whichever type of encoder it is set for...linear or rotary.

It's not anything new though. I remember doing this back when Acu-rite III was cutting edge technology..LOL

The main thing is to make sure your standard/reference is parallel to the linear axis you are correcting. If you're correcting "X" on a vertical mill, then indicate your reference standard so that it is running with the "X" axis and hold it in position in some way that doesn't damage the standard. Then after it's in position, wring a gage block to one end of the standard, but leave it sticking above the standard by 5-6mm. Zero your indicator(one that has a fine resolution preferably) on the end without the gage block and zero your readout. Raise the quill just enough to clear the standard. Dial down to the other end of your standard that has the gage block wrung to it (you may want to hold a light pressure against it.) until your indicator is on zero. Compare the length of the standard to the distance on the readout and do your math.

[I'lldietrying]

Mine had some red plastic shims that were screwed to the aluminum cover, and prevented the read head from moving. I had to remove them before installation, because once installed the screws holding them in were inaccessible. I was able to use them as a gage to set the spacing between the read head and the scale.

Sounds like you should be in good shape for clearance then. Some scales have plastic clips on each side of the read head that come off with a little twist after everything is mounted and locked in place.
Since you have it mounted, maybe you could try gently holding the wipers away from the reading head and move it a tad to see if that's where the interference/noise is??? Could the wires from the read head be rubbing on the inside of the housing for the scale/read head?
As far as being fragile, I've always tried to avoid do anything that would scratch them.

[handlewanker]

Hi, regarding the accuracy of a DRO on a mill etc, it depends on what you want to use the length of measurement for.

For example, if most of your work is cut and try around the 160mm mark what do you need accuracy over a distance of say 500mm, unless of course you're going to drill and bore holes as in a jig plate that needs to be pitched at whatever the distance is
required.

I once had to prove the accuracy of an X axis DRO on a Bridgeport mill that supposedly was giving funny results at random.

I set two buttons on the mill table, about 1 metre apart overall, using a 1 metre vernier caliper, previously calibrated in the met department.

Then using a dial indicator to read the first roller face at zero, (DRO at zero too), and moving down to the second roller face, a distance of 1 metre (3 feet approx), and comparing the DRO readings to the dial indicator readings we got an accuracy of better than .04mm allowing for button diameter in the calcs.

We could also have set the button at 100mm intervals and compared the dial readings/DRO readings, but the powers that be were satisfied that the overall reading was sufficient without trying to prove or disprove the accuracy of the DRO scale all along it's entire length.

I've seen people go to extraordinary lengths to achieve so called accuracy when all they need is a reliable means to know at any moment in time exactly where they are, and that can be variable unless you have a temperature controlled environment that is 68 deg F (20 deg C) all the time, and at 68 Degrees F you'd need to wear special clothing just to work there continuously, I know because in our gauge room the airconditioner was set to dehumidify and maintain the temp at 68 deg F, 24/7.

Using just a standard 150mm digital caliper and two buttons set at about 150mm you can move the X axis back and forward a number of times to prove or disprove the ability of the DRO to read the scale and come back to zero each time.

If the DRO reads the scale etchings or magnetic imprints in pulses and misses a pulse or two then you won't achieve zero at the return, and if it misses a couple of pulses/scale erchings/magnetic lines, whatever, a number of times you also won't get back to zero either.

Same goes for the read head if it misses a pulse or two going down the scale, you will be somewhere on the path travelled but less than you aimed for.

So if the DRO is good for a slow count reliability and you fast traverse it from A to B, the reliability goes out of the window when it misses pulses.

It would be different if the read head actually read the position, as in a GPS indicator for navigating, because then you would be indicating the end product position not the accumulated step rate taken, no matter how many pulses/steps taken.

I worked on a Kearns boring mill and they had light verniers that read the exact position of the table X,Y and Z axis's relative to the vernier scales, but that was pre CNC and DRO era.

I recently bought on Ebay two light vernier read heads, (came off'n some old machine), but without the scales, so if I get two rulers that are engraved in 1/10" markings and using the read heads ability to divide the scale markings into 100th increments I can get .001" accuracy, pity it wasn't metric.

Maybe those guys in Singapore who sell the DRO's on Ebay with glass scales can supply me with scales etched in 1/10" increments.
Ian.