Encoders for 800mm project

[Michele Scotti]

Hi, it’s about time to spec the encoders for the telescope.

The current set-up is as follows - I have a SiTech II and I deem the axis encoders essential for tracking capability and long exposure.  There will be an off-axis guide however I’d like to get the most out of unguided operation - and I have a couple of doubts to clear....

Right from the design phase, I wanted some easy way to attach a linear scale to an accurate surface rotating with the axis. The pac-man is an example of that – once it’s ground it can accommodate a linear strip.

 

Currently my first (*) route is to use linear incremental encoders and sourcing them as used parts from scrapped CNC machinery. Linear scales can be purchased as strips and glued, taped or somehow fixed to the trunnion.

To start with: Resolution.
The trunnions provide a ground SS strip with an approx. diameter of 1200mm. This provides a ratio of (360deg x 3600arcsec) / (1200mm x Pi) è 343.8 arcsec/mm or 0.34arcsec/um.
Linear scales with 1um resolution are pretty “standard”.
QUESTION #1: is 0.34arcsec/um or better ca. 3 ‘ticks’ per arcsec enough to track?

Accuracy:
Just like Sun Divisional Error, I’m not that concerned if adopting CNC equipment – they are usually pretty decent and unlike a CNC machine, when it comes to tracking for several minutes the portion of the scale is relatively small.
QUESTION #2: or should I be concerned??

If anyone here has some advice that’s much appreciated – I have a background on encoders but I lack insight and practicality.

 

QUESTION #3: also, can I directly connect the TTL (0-5V) A and B phases directly into the SiTech?

Btw, I’ve ordered a Renishaw readhead – it was just too cheap not to buy it. Looks like scales that go with that are not that inexpensive. I’ll post the image as soon as I received – so we can see if I just wasted some (more) money…

(*) I have a couple of concepts to develop a similar or better system suitable for DIY but in our club we are currently lacking somebody with Electronics experience to develop a functioning prototype.

 

Cheers, Michele

[markse68]

Does the renishaw read head have  a b  quadrature out? The ones I know are analog sin/cos and need separate interpolation.

Mark

[Michele Scotti]

HIgot this one - according to the data sheet it's a 1um resolution with digital output. for phase A and B in quadrature. But I don't think there's a "quadrature" signal. However the SiTech II has separate inputs for A and B so I suppose it performs the quadrature?

[markse68]

6 hours ago, Michele Scotti said:

However the SiTech II has separate inputs for A and B so I suppose it performs the quadrature?

Yes that’s how it works- the device reading the A and B signals does the quadrature bit to work out direction and multiply lines per revolution by 4 effectively to give you quadrature CPR.

Mark

[DaveS]

The ones on my DDM85 are Renishaw (I think) absolute encoders running at 0.02" per "tick".

[Michele Scotti]

5 hours ago, DaveS said:

The ones on my DDM85 are Renishaw (I think) absolute encoders running at 0.02" per "tick".

That's the same as Planewave i.e. 50 ticks per arcsec anyway I just talked to the chap that developed the tracking system and he says that I should aim at 10 ticks/arcsec 

[Michele Scotti]

Just for info, I've started a thread on a CNC forum - great chaps there too- and  getting a lot of suggestions.

All in all it looks like a tape scale + encoder readheadwith a 0.1um is the way to go.

The challenge is to find the hardware in the used market to contain the expense and stay true to the spirit of the project.

Read-heads seems fairly attainable on ebay ( like less than 100$/£ for a RGH22/RGH24Series if you're lucky and patient).

If I could only find linear scales that are inexpensive then bingo! but I seem not to find any of them – the optical ones with 20/40microns gratings, at least. Still puzzled about how to salvage them from a CNC machine.

Also, what is not entirely clear is if and optical encoder is better than a magneic one. The both are available to 0.1um resolution

[markse68]

1 hour ago, Michele Scotti said:

Also, what is not entirely clear is if and optical encoder is better than a magneic one. The both are available to 0.1um resolution

In your application i’d have thought magnetic or capacitive or inductive (Posic do linear scale versions) would be better than optical purely from the point of view that it’ll be difficult to contain and shield the read head/grating from dirt and humidity which the other options would be more immune to.

Mark

Edited October 23, 2020 by markse68

[Michele Scotti]

12 hours ago, markse68 said:

In your application i’d have thought magnetic or capacitive or inductive (Posic do linear scale versions) would be better than optical purely from the point of view that it’ll be difficult to contain and shield the read head/grating from dirt and humidity which the other options would be more immune to.

Mark

From what I got so far the resolution resides mainly in the readhead  - i.e the digitalization of the sinusoidal information -. whereas the accuracy/repeatibility is in the grating of the scale -regardless if it's magnetic or optical. 

Apparently magnetic scales are """"fairly""" affordable - the optical ones (with 20/40um grating- are nowwhere to be seen...that's weird because the matching readheads are plentiful on Ebay...just like the one I bought.

[markse68]

11 hours ago, Michele Scotti said:

the optical ones (with 20/40um grating- are nowwhere to be seen...that's weird because the matching readheads are plentiful on Ebay...just like the one I bought.

The Renishaw encoders i know rely on the grating being bonded to the moving part with epoxy- much easier to unscrew a read head and the grating would likely be ruined trying to remove it- could explain your experience.

I wouldn’t worry too much about accuracy of the different technologies as i’d imagine you’ll exceed their accuracy just with the inevitable sag you’ll get going from vertical to horizontal with your ota?

Mark

Edited October 24, 2020 by markse68

[Michele Scotti]

2 hours ago, markse68 said:

The Renishaw encoders i know rely on the grating being bonded to the moving part with epoxy- much easier to unscrew a read head and the grating would likely be ruined trying to remove it- could explain your experience.

I wouldn’t worry too much about accuracy of the different technologies as i’d imagine you’ll exceed their accuracy just with the inevitable sag you’ll get going from vertical to horizontal with your ota?

I agree about the grating - i guess they are glues inside the Alu casing -although I'd be ok to try and debonnd a scale. The only source of an optical grating in a reel is from China and it's £400 a meter....that won't do for me. I have the impression that magnetic scales are cheaper.

Sag is inevitable but I have a couple of considerations:

1) the design is somehow taking that into consideration - see post:

2) sag in a properly designed and executed mount -not saying it's my case!- is predictable and proporsional i.e. it could be mapped out

3) the sag for given an amount of tracking time seems -at first glance- manageable. IT should be less than 1arcsec each Altitude degree. Aiming at 10min integration time I'm not that worried about - we'll see iwhen in operation if all of this ijust theory....

[Michele Scotti]

Brief update: had a chat with the sales manager at Renishaw - he confirmed the feasibility to use their magnetic encoding system - 0.24um resolution.

Pro: it should be cheaper than the optical one 

Cons: magnetic has some hysteresis when changing direction. The amount is less than 2um which is less than an arcsec in the 800mm project set-up. On top of that - possibly to simply convince myself!- i'd say that hysteresis would affect mainly the pointing accuracy rather than tracking provided an exposure that doesn’t change the direction of the axis.