Posted 25 September 2012 - 06:38 PM
Basically it's not a matter of choosing the right encoder, but rather trawling ebay for the cheapest possible analog encoder and then working back from that..
Posted 26 September 2012 - 01:19 PM
encoder, but it must be expensive I fear!
The Heidenhahn was quit expensive compared to others,
so right now I'm checking cheaper alternatives.
The interpolation scheme will depend on the strategy
to follow, we'll see.
Typo, wasn't 10.000 but 5.000 cycles
Posted 26 September 2012 - 06:45 PM
Posted 27 September 2012 - 12:08 PM
which is very apealing to me, eventhough they ask 220€
plus shipment ( another 60€ !! ).
I'm looking at some external hardware to hookup between
the arduino and the encoder to speed things up.
With a modest 864x interpolation I'll have 8.640.000
pulses ( quadratured ) which should be enough I believe.
Some interesting Flash-ADC chip does the conversion in
only 1.18usec! ( 8 bit only ), so using 2 of them
for sin and cos.
I'm still thinking which way apeals most to me to setup
the whole thing.
Posted 27 September 2012 - 01:51 PM
the key issue is that any form of interpolation will be inaccurate. So the lower the interpolation, the better.
with a 2048-ppr and 8-bit interpolation (256X) you should get 524288 tics per rev, or 2.47" per tic, or 6 tics per second. I think that should be enough...
I am planning to run my 5000-ppr encoder at 200X only = 1 million tics per rev, 1.296" per tic.
Posted 27 September 2012 - 02:54 PM
Posted 28 September 2012 - 02:24 PM
Thinking it over again, with a 5000 cycle encoder:
Sidereal day has aprox. 86164.091 seconds.
That means 1 turn of the encoder in a sidereal day gives
us 0.058028814 cycles per second.
Doing 259x interpolation on 1 channel would give us
aprox. 15.0294628 pulses or ticks, which equates quit
nicely to the sidereal rate of the mounts.
Accumulated error in 1min and 1 second would be
aprox. 0.7716" which can be compensated for.
360º = 1296000 seconds
5000*259 = 1295000 pulses So we will miss 1000 pulses
each 360º with this scheme.
Looks like it isn't too hard to get there.
Posted 28 September 2012 - 08:41 PM
Since there are 1296000" per sidereal day and 86164 seconds, that is 15.04" per second.
To get 10 tics per second (which means we can update the RA speed 5x / second - much better than any autoguider!) we would need a resolution of 1.5" per tic. Which translates to 861K tics. The 5000ppr analog encoder, with 4X digital quadrature and 50X analog interpolation, would be 1 million tics. That would be sufficient.
So even a 2048ppr analog encoder would do.
Posted 08 October 2012 - 02:10 PM
A-9523-6051 RGS20-S scale 510mm long
Price $ 197.00/ea
9531-0342 – epoxy - stock
Price $ 5.00
A-9523-4015 – end clamps – stock
Price $ 20.00
A-9531-0239 – scale applicator – stock
Price $ 58.00
Posted 08 October 2012 - 06:47 PM