Re: [CAD_CAM_EDM_DRO] Linear Encoders

>
> I started a new thread since we really havent been talking about DRO's in this sub thread anyway.
> I managed to delete the two messages i meant to reply to.
>
> The first was the one that said the HP is a chip with a grid incorporated in its layout.
> If this is true, then do they have the layout we need for 500 lines per inch?

I presume you are talking about the HP optical readers heads?
No, the only have readers for 300 and 360 unless I'm mistaken

> i would guess that they are not willing to give up on the glass scale market.
> Anyone ever break apart an Accurite? The rotary encoders have a metal grid, i seen it.
> if HP or someone else makes a detector that don't need a grid,that is great,
> and i think we know what to do.
> The other point is that if HP doesn't make the right chip that isn't stopping
> us from using the rotary encoder scheme.
>
> Otherwise, I'd like to volunteer whoever it was that emailed UsDigital to ask
> them if they would print us some 500 line per inch and sell it without a head available,
> and we could use some of that in the head.

You don't need to do that.
As I, and a couple of other people have pointed out, these things can easily be
generated by a local typesetter for less than US digital wants for the off-the-shelf
stuff, and the price they want will probably go up by an order of magnitude or two
for a custom job.

Truth be known, they probably print them using the exact same process.

>
> Finally, It seems that you could use the two head outa phase scheme to double the resolution
> of a grid as proposed in the other message i deleted.

Yes, this is simply how quadrature encoding works, If we go to the trouble of
aligning 2 heads at 20 bucks each, why not just build a real encoder
of the right resolution. all they really are is a pair of LEDs and a pair of
Photo-transistors and they can be had for under a buck.

> My thought is that the spacing would have to be correct,

It does, and since the odds of getting it perfect are about nill, I'd lean
in the direction of an adjustable reader. It could be as simple as
cutting a slot in the base of the reader and having an adj screw to
force the 2 surfaces apart.

We are only talking a few thousandths.


+----------------+
| |
| 0 | 0 | 0= LEDs
| | |
| |==========X Screw
+-------|--------+
^Slot

These type of readers used to almost all be adjustable.
IF the HP readers were, we could simply realign them for a new scale.
They aren't

> but that you could use ttl logic to convert the two signals into one.

Then you just lost the advantage of doubled resolution. for no real reason.

You COULD build a multi sensor head and get even higher than 2 X Scale
resolution.

________--------________--------________--------________-- A

______--------________--------________--------________---- B

____--------________--------________--------________------ C

__--------________--------________--------________-------- D
x x x x x x x x x x x x x x x x x x x x x x x x x x x x

A & C are what a regular encoder would put out with 2 sensors.

a 4 sensor head could do all 4 and as each 'x' shows there are
transitions for each "line" in the strip or wheel.
2 to high, and 2 to low.

For a regular 2 sensor head, Aligning them would be a simple
issue of plugging the 2 inputs of a logic analyzer or dual trace
'scope. (That may be the sticking point for some people)
And turning the screw until they move into alignment.

Even IF they were misaligned the error would only be on the 1/2
counts.
They DO need to be out of phase or you won't count at all, but if
you get them to count with whatever decoder scheme you have, it will
be as accurate at the scale resolution.
In other words:
If you have a 500 line/inch scale, 1.000 will be correct, as will
1.002, but 1.001 & 1.003 may be a bit early and late as you move
across the scale.
Simply turning the screw until it starts counting reliably will
get you that far.

Turning it until it just starts to count, then turning it until
it just stops and splitting the difference should get you
pretty close in all reality.

You could use a micrometer and the same method for even better
results.

>
> Finally, i must say i am not totally committed to the "lime jello"
> approach to linear encoders, because my (currently imaginary) budget
> probably would include store bought encoders.
> However i am an engineer that can't say no to a project.
> eric

Marc