Optical encoders and geared mountings: possible info overload?

[mental4astro]

Hi all,

Here I'm exposing my ignorance of electronics with the following question:

Gears I know can be used to increase or decrease the number of revolutions an element is caused to rotate. Now, as an optical encoder punches out information at a particular rate to a cpu, say one running DSCs, is it possible to create a gearing ratio that can overload the cpu?

As an example, say the optical encoder I'm using has a resolution of 2048, and is connected to azimuth of a dob mount, and the effective geared ratio is 8:1, 8 encoder revolutions per one rocker box revolution, would I be oveloading the cpu of the DSC? If not, then this could give me an effective encoder resolution of over 16,000.

Yours in ignorance,

Mental.

[gary]

Hi Alex,

Argo Navis is designed to be portable and therefore battery powered. Optical
encoders, however, are power hungry devices. Therefore Argo Navis "samples" the
encoders by literally powering them on and off thousands a times a second. A
caveat of sampling is that if the encoders are rotated too quickly, one can
exceed their "sampling rate". In engineering, there is a concept known as the
"Nyquist rate (http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_th eorem)". It dictates that one has to sample at least twice the rate the encoders
are producing "ticks" at when turning. Argo Navis has a second CPU just devoted
to sampling the encoders. Therefore it can sample at a much higher rate than
older DSC's. However, ultimately it too has a maximum sampling rate, dictated by
both the processor speed and also by the electrical switching characteristics of
the encoders themselves.

Argo Navis has a SETUP ENC TIMING menu whereby one can establish the encoder
timing, thus trading off power for a higher sampling rate. The maximum sampling
rate is around 16,000 samples per second. Hence the maximum slew rate should not
exceed 8000 times a second for the Nyquist sampling theorem not to be violated. Working
backwards, one can then determine by how much you can gear the encoders based on
what your expected maximum slew rate is likely to be. This maximum sampling rate
is achieved by setting TOFF=0 in SETUP ENC TIMING. However, this value has the
highest power consumption.

Argo Navis can determine when the sampling rate is exceeded and will display
this as an encoder error.

The factory default values for SETUP ENC TIMING of TON=17 TOFF=17 give a good
compromise between sampling rate and power consumption and provide a minimum of
about 7000 samples per second.

The following calculation is based on a rate of 7000 samples per second.

For a 10,000 step encoder geared 1:1, you can slew at approx. ((7000/2)/10000) *
360 = 126 degrees per second. The 7000/2 is owing to the Nyquist rate. Therefore
you can also work backwards and if you know in advance what your maximum slew
rate will be you can then determine the max. number of encoder steps for one 360
degree revolution and hence your gearing ratio of your pulleys.

As another example, take a 2048 step encoder geared 8:1 to provide 16,384 steps effective.
(7000/2)/16384) * 360 = 76.9 degrees per second.

Please keep in mind that these types of slew rates will only work with Argo Navis.
All the legacy DSC's have much slower sampling rates and some just silently drop steps.

Timing pulleys and belts generally provide a better solution that spur gears for encoder
applications because unlike spur gears, timing pulleys and belts are mathematically designed
to have close to zero backlash.

Armed with the above information you can select pulleys with gearing
ratios that meet your needs with your Argo Navis.

Best Regards

Gary Kopff
Managing Director
Wildcard Innovations Pty. Ltd.
20 Kilmory Place, Mount Kuring-Gai
NSW. 2080. Australia
Phone +61-2-9457-9049
Phone +61-2-9457-9593
sales@wildcard-innovations.com.au
http://www.wildcard-innovations.com.au

[mental4astro]

Cripes, Gary, thanks.

The reason for my asking is a mate has his azimuth bearing made out of a lazy susan. I've thought of fitting an encoder's shaft with a little wheel, and set the wheel to run parallel to the tangent of the pivot point. Depending on the distance out from the centre of the pivot, the effective number of encoder revolutions can be quite large.

No backlash with this direct drive method either.

[gary]

Hi Alex,

Rubber friction rollers have the advantage of zero backlash. Many professional
telescopes such as the Keck Observatory 10m and the Lick Observatory Shane
3m employ incremental encoders that are mounted off-axis and coupled via
a friction roller.

However, even on professional telescopes, great care has to be taken to avoid
slippage. Any slippage will result in a pointing degradation. On amateur telescopes,
even a well engineered roller solution can prove challenging. Even very tiny
angular misalignments can result in significant slippage.

For the types of telescopes you are talking about, I recommend to avoid friction rollers.
One can spend countless hours trying to get them to work reliably and for any
scope that is portable, the challenges of maintaining alignment of a roller is made
even more difficult whenever the scope is transported.

Rather than a friction roller, I recommend you consider a timing belt and pulley.
For example, take a look at this image on our web site -
http://www.wildcard-innovations.com.au/images/hector4.jpg
This is a 20" Dob and you will notice a black strip of MXL timing belt has been
glued to the inside of the Alt trunnion. An encoder with a small aluminium
timing pulley fitted on its shaft is spring loaded so that the pulley pushes down
on the belt. The shaft spins many times as the mount rotates around the Alt axis.
The encoder employed had a very low resolution but the effective step count is in the
order of 56,000 steps.

This particular timing belt arrangement never slips and this solution has proven very reliable.
However, determining the effective step count on the Alt axis using this
arrangement can be challenging, made all the more difficult because the Alt axis cannot
be rotated through a full 360 degrees.

On the Az axis, rather than exploit the diameter of the lazy suzan bearing, it
is generally advisable to run a timing pulley and belt off the main Az pivot bolt instead.
When a roller runs on top of the lazy suzan like a needle around one revolution of an old vinyl record,
it is extremely difficult to avoid slippage. Even when one tries to run a wheel on the outside edge
of the lazy suzan, one has to ensure that the lazy suzan is perfectly circular and again
great care has to be taken to avoid slippage.


Best Regards

Gary Kopff
Managing Director
Wildcard Innovations Pty. Ltd.
20 Kilmory Place, Mount Kuring-Gai
NSW. 2080. Australia
Phone +61-2-9457-9049
Phone +61-2-9457-9593
sales@wildcard-innovations.com.au
http://www.wildcard-innovations.com.au

[mental4astro]

There is no azimuth pivot bolt with this lazy susan. That's why I thought of the 'frictional roller'. And I thought I was the first to come up with that idea! :rolleyes: :lol:.

[gary]

Hi Alex,

Alas, there is no shortage of prior art. And alas there is also no shortage of stories
even in the professional telescope community of engineers battling with alignments
of friction rollers for encoders. :lol:

Would you be able to get a picture of your friend's scope? I could have a quick look.

As advised in the P.M. I sent, I will follow-up for you on a recommendation for the
encoder installation on your own scope.

Best Regards

Gary Kopff
Managing Director
Wildcard Innovations Pty. Ltd.
20 Kilmory Place, Mount Kuring-Gai
NSW. 2080. Australia
Phone +61-2-9457-9049
Phone +61-2-9457-9593
sales@wildcard-innovations.com.au
http://www.wildcard-innovations.com.au