Re: Looking for scale interprolator circut

The circuit will do what you need if you feed its analog outputs to
an A/D converter. The 2 least significant bits can be decoded into
step and direction.

That is not the problem though.

An unstated assumption being made is the encoder has accuracy beyond
its stated resolution. Usually it doesn't.

Most encoders specify quadrature signals as being "90 degrees, +/-60
degrees". That is the first tip-off that everything possible is
already being extracted from the device.

It's true that all (most?) optical encoders start with an analog
signal. This signal, when specified, is usually called "quasi-
sinewave", meaning it is someplace between a sine wave and a triangle
wave.

The amplitude of this signal can vary over a 2:1 range and there is
a cyclic DC offset to the signal's zero-crossing points. This is
caused by inevitable eccentricity and "wobble" mechanical errors of
the code wheel.

Encoder mfgs compensate for the effects of these errors by peak-
detecting the analog signals for digital output thresholds. This
clever method does not avail for interpolation schemes.

Interpolation requires an accurate analog input; you cannot extract
more information (resolution) from a signal then what's there to
begin with. Most optical encoders have a resolution that matches
accuracy; there is no excess to work with.

Resolvers (a type of rotary transformer) are an excellent example of
a transducer that has far more accuracy than native resolution. They
produce accurate sine-wave outputs of very constant amplitude. They
are used with interpolation circuits to bring resolution up to be at
par with accuracy.

Mariss



--- In CAD_CAM_EDM_DRO@yahoogroups.com, "protman16" <protman16@y...>
wrote:

> Fred wrote:
> > In the absence of any other information see the file analog

output

> > encoder interface.pdf in the Files > CIRCUITS > Pulse Gen and
> Timing
> > Circuits section which was submitted by Marris. This may go
> someway to
> > what you need?
> > Fred
>
>
> The circuit in the files section is not exactly what I am looking
> for. It is a start though. I need the square wave output to be a
> multiple of the sine wave input. Something that can be selectable.

I

> think an A/D converter might work as long as the range is within +/-
> 1 volt. With an A/D converter I should be able to use the any bit
> from LSB to MSB for monitoring purposes. Of course I would need two
> because of the sine-cosine input wave.
>
> Trevor