Re: more on sine wave encoder

Hi,

The simplest method would be to use a "flash" A to D converter. This
type of converter is a "tree of comparators" type, 256 of them for an
8 bit version. They convert in 100 ns or less, making them suitable
for video applications and they are cheap since every video capture
board must have one.

The 8-bit output could address a ROM having an arc-tangent scaled
look-up table and the output from the ROM could be quadrature encoded
on 2 data bits.

I have not priced arc-second accurate encoders. Anything accurate to
1 part in 1,300,000 (1 arc-sec) no matter what it is would be
astronomically expensive. A laser interferometer would be a better
bet.

Mariss

--- In CAD_CAM_EDM_DRO@egroups.com, "Elliot Burke" <elliot@h...>
wrote:

> > Date: Sun, 24 Sep 2000 18:23:06 -0000
> > From: "Mariss Freimanis" <geckohall@h...>
> >Subject: Re: sine wave encoder outputs
>
>
> >1) Encoders have a cyclic amplitude modulation; a sine/cosine
> >amplitude of +/- 1 volt may be +/- 1.2 volts when the encoder is
> >turned 180 degrees.
>
> True. The sine wave decoders normalize the output to allow for

this. This

> is why 4 outputs are provided, although it can be shown that only

three are

> necessary.
>
> >2) Any resolution enhancing scheme depends on a 90 degrees
> >relationship between the two channels. This cannot be gauranteed.
>
> The phase relationship has a maximum error which is specified in the
> manufacturers literature.
>
> >3) The waveform more than likely is a "quasi-sinewave". This means

it

> >actual waveform will be somewhere between a sinewave and a triangle
> >wave.
>
> For grating systems not designed to have sine wave output, this is

true.

> The encoder I have has generates quite accurate sine waves.
>
> >4) Even if the encoder is perfect, the circuitry necessary to

extract

> >additional resolution is complex and expensive.
>
> This was my question. Who has a simple circuit? I know a a scheme

with a

> tree of comparators and op amps, but thought maybe there was a

simpler

> solution. The simpler solution should spend very little time doing
> calculations. For example, the arctangent can be calculated with a

look up

> table (PROM and a latch technique), which still isn't very simple.
>
> >Apart from the "non-dithering" advantage, the expense of
> >interpolation is better spent on a higher resolution digital

encoder.

> >If anyone is interested, I can post a circuit I designed that
> >extracts an analog voltage proportional to position between encoder
> >counts.
>
> Have you priced arcsecond encoders?
>
>
>
>
>
> Elliot Burke