Quadrature decoder

[GregS]

Been searching all morning for one, and did find one by LSI/ LS7184.

I have been looking at a drawing of one from the mid 70's, drawn up by
a coworker who is long gone.
using 74121;'s and 7400.

I got a sample request, but I think i can do this myself quickly. I got stepper motors
and controllers requiring step and direction input.
Good thing I have this diagram because I can't find others.
Must be some other stuff around ??

I don't need indexing or counters.
I also see optical encoders that have the outputs I need for the controllers.
But I need something fast.

greg

 

[GregS]

http://groups.google.dk/group/sci.electronics.design/msg/8b9bba503375d924?h=
l=3Dda&dmode=3Dsource

I was snooping around Google Groups. Thanks.

I'll look over the diagram. Here is what i have been looking at......
http://zekfrivolous.com/misc/quadrature.pdf

greg

 

[GregS]

Work statement too vague

What _exactly_ are you looking for? A _detector_ that produces a
pulse train _and_ direction?

...Jim Thompson

Right, pulse and direction.

greg

 

[GregS]

Work statement too vague

What _exactly_ are you looking for? A _detector_ that produces a
pulse train _and_ direction?

I also found this. All I would need is a separate toggle
switch to choose direction, but not so cool......

If you only care about speed and not direction, one edge on one channel m=

ay be sufficient.



greg

 

[GregS]

Its been a long project. i tried to get an existing controller to work
with my steppers. For a Z axis motion. I thought i needed more power
so I used two audio amps to act as buffers from the stepper
output. That did not help much. i didn't figure out why the motors
did not like the drive output. So now I am wiring a separate Z axis
system, which is a lot simpler than using the existing system.
The movement is by knobs, and the speed will not
be over 500 Hz. I also did not want to drive two steppers
in sync, but thats what I was given. I could have also
ordered an opto knob with step and direction outputs but its
a time thing right now.

greg

 

[GregS]

Like this...

http://www.fpga4fun.com/QuadratureDecoder.html ??

Missed the diagrams, I was looking at earlier today.

I've seen self-clocking versions.... found this...

http://archive.electronicdesign.com/files/29/21422/fig_02.gif

But I've seen it done with a dual-D, but it escapes me right now :-0

...Jim Thompson

My first thought, simple, and perhaps even better, was
to use just a toggle switch. I thought about using a VCO
that would speed up the longer you hold the toggle
up or down. You go up, and you go down. Not as precise though
as in focusing a microscope.

greg

 

[Jamie]

Been searching all morning for one, and did find one by LSI/ LS7184.

I have been looking at a drawing of one from the mid 70's, drawn up by
a coworker who is long gone.
using 74121;'s and 7400.

I got a sample request, but I think i can do this myself quickly. I got stepper motors
and controllers requiring step and direction input.
Good thing I have this diagram because I can't find others.
Must be some other stuff around ??

I don't need indexing or counters.
I also see optical encoders that have the outputs I need for the controllers.
But I need something fast.

greg

That can be done with a 7474 (dual Data flip flp).

I'm not going to get into details but, you do need to
invert the CLK signal for each FF in the package..
Basically:

Channel A drives FF #2's Data and Clr pin while it
also drives the inverted circuit for the CLK of FF #1

Channel B drives FF #1's data and clr pin while it
also drives the inverted circuit for the CLK for FF #2

Net results are:

FF #1 outputs CW pulses while FF #2 outputs CCW pulses..

Using a basic NPN common emitter type inverting circuit for the
CLK with a pull up R, should work just fine.. I guess one could also
use a pre biased transistor..

P.S.
Most encoders use a open collector scheme. You'll need to supply a
pull-up or use a PNP inverting circuit which would remove the need of the
pull up of the encoder line.. You would still need a pull down R on the
inputs of the CLK I guess..

 

[Jamie]

for most things that is a terrible way of doing it, a slight
mechanical mechanical vibration
or electrical noise that gets into the channel you use for pulses,
will looks just like
it constantly moving in one direction

do it the right way and it will just toggle back and fourth

-Lasse

Really, I've never seen that. I use a dual D-ff, the other Channel holds
the Data and CLr pin to one state. Even if lets say channel A is
chattering, channel B is in the middle of the phase supplying the DATA
pin for the state which isn't moving.

Oh well, maybe we have different circuits..

 

[[email protected]]

If one phase clocks a D-flop which has as "D" the other phase, you get
a direction signal.

You only get the direction once per "cycle". The four flip-flop (two two-bit
shift registers) and XOR solution updates every quarter cycle and is more
stable around edges.

 

[GregS]

If you are planning to use the quad decoder to feed into the stepper
motor controller
why not just beef up the output drive of your quadrature knob and
drive the motor directly?
regards,
al

Well, I liked the idea of having current control, and I allready
had one channel wired up. I DO have a driver chip that will do what you said.
I am going to consider it for simplicity.
Darn stepper is only 3.3 ohoms.

greg

 

[Ian Bell]

http://groups.google.dk/group/sci.electronics.design/msg/8b9bba503375d924?hl=da&dmode=source

-Lasse

The problem with all clocked or sampled quadrature decoded such as those
referenced above is transition jitter. I know this from personal
experience where such a detector on a mechanical system was subject to
vibrations that could occasional cause a rapid set of transition
forward&backwards some of which would be missed by a clocked system.
This leads to false counts.

Cheers

ian