Hello,
my do-it-yourself-mount suffers from a relativly great backlash on DEC. Even more, it changes. I guess the worm wheel is not 100% concentric to the axis. So normal backlash compensation doesnt work. This makes some trouble for PHD2 pulse guiding.
My idea is to solve the issue with renishaw high res encoders on DEC-scope axis (I have already encoders in cascade mode on RA-axis working well).
So if PHD2 sends a pulse to DEC, sitech automatically fist compensates the individual backlash (move the motor until encoder reports a move of the scope axis) and then executes the pulse.
Will it work this way? Before I spent a lot of money....

Thanks to the specialists!
Ralf

Hi Ralf,

The high resolution encoders that will help Dec and correct backlash can not be interpolated encoders - they won't work to see and correct backlash.  Interpolating encoders only work on RA because it is constantly moving, tracking at a nearly constant rate that allows interpolation to work.  A 1 million tick encoder has 1.3 arc-sec resolution, so it would never get closer than the last 1.3 arc-sec.  Only the Servo II can operate with mount encoders of this number tick/rev and higher - the Servo I can't read that many tics.

Don W

On Sat, Apr 4, 2020 at 10:11 AM, Don W wrote:

The high resolution encoders that will help Dec and correct backlash can not be interpolated encoders - they won't work to see and correct backlash.  Interpolating encoders only work on RA because it is constantly moving, tracking at a nearly constant rate that allows interpolation to work.  A 1 million tick encoder has 1.3 arc-sec resolution, so it would never get closer than the last 1.3 arc-sec.  Only the Servo II can operate with mount encoders of this number tick/rev and higher - the Servo I can't read that many tics.

Hi Don,
the intended ring encoder has 15744 lines x800 = 12,6 M tick/rev. (RESM20 100mm dia + VIONIC 25nm). I have the Servo II. But no chance working due to interpolation?

On Sat, Apr 4, 2020 at 11:21 AM, @radinkel wrote:

RESM20 100mm dia + VIONIC 25nm

Hi Ralf,

Well, 15744 lines gives a resolution of 82.3 arc-sec.  If your backlash is less than 82 arc-sec I don't know how interpolation could work on DEC.

Don W

Hi Don,

Next question: You recommend Renishaw 26 bit absolute encoders in Cascade Mode. This will recognize and compensate backlash? It sounds so:

"Cascade Mode:

You can ONLY use this mode if you have High Resolution Encoders on the particular axis.

If you do have accurate high resolution encoders on your axis, then you will have a nearly perfect telescope! We like the Renishaw 26 bit absolute encoders, or the Renishaw minimum of 13,000,000 tick encoders (about 10 ticks per arc second).

The cascade mode works by positioning the motors to a location where there is zero error on the scope encoders."

But these encoders do interpolate to.
Absolute encoders use different protocol for absolut position information. Renishaw provide: BiSS, DRIVE-CLiQ, FANUC, Mitsubishi, Panasonic, Yaskawa. Does Sitech really deal with one of these? I thought Sitech uses only the A/B digital input?

Hi Ralf,

There are two main types of encoders used with SiTech controllers.  They are A-B (quadrature) encoders common for the Servo I and Servo II controllers.  These are NOT sign wave, they ARE square waves.  The second type encoders are Absolute encoders, with 65 million unique tics per rev, these require a BISS box to interpret these individual tics.  The BISS box is a ~ $500 box.  I don't know if that can work with Servo II, I know it is being used with the SiTech Brushless Controller.  The Renishaw Absolute encoders are about a $5000 (or so) plus the machining to install these complex and precise encoders.

Interpolation of A-B encoders is adding square wave tics in between the Optic tics in the encoder.  Interpolation works IF the encoder is moving at a constant speed of rotation.  For instance the Gurley 320K and 500K encoders have 5,000 tic optical encoders.  The 320K adds 63 bits between each optic tic (5,000 + 5000 * 63 = 320,000 tics).  The 500K Gurley adds 99 interpolated tics between each optic tic.

So what happens for Dec??  When tracking, DEC doesn't track.  It isn't supposed to move.  Due to many factors when guiding, the Dec may move slightly in either direction, so Backlash becomes very important and prominent.  This motion is probably well withing the "space" between optic tics, so How can the interpolation tell how many tics to add???.  This is exactly why I said an interpolation encoder is no good on Dec.

Don W

Hi Ralf,

OK, I am learning something too.  The output of the A-B type encoders are all square wave.  Internally, I don't know, but they could be sine waves.  I would like to know the model number of the encoder that has 125.66 million tics/rev.  What does it cost?

Don W

Take a RESM20 ring DA 100mm and a VIONiC head with 2,5nm resolution.
You can also take a RESM20 Ring with DA 550mm which has 86400 optical lines, so you get 691,150,000 tics/rev.
The RESM20 costs (2016) 493,-- EUR and a Head with 100nm res. which gives 3.125 million tics/rev costs 679,-- EUR. I have that combination for RA, working well even in cascade mode.
I asked for a offer.
Ralf

Hello Ralf,

I would highly recommend that you speak with a Renishaw representative before you make a plan or get a Renishaw encoder.  They will help you setup the encoder for what you are trying to do.  I have Renishaw RESM encoders using the TONiC digital system.  413mm (RA) at line count of 64,800 interfaced to 64,800,000 "ticks" and 255mm (DEC) at line count 40,000 interfaced to 40,000,000 "ticks". These are highly sensitive encoders and pick up the slightest vibration even when not moving. I don't need any more ticks because the motors are set for approximately the same amount of ticks per rev.  What good is lots of ticks on the mount encoder if the motor encoder can't match it?  That's my opinion only.
VIONiC encoders do have a digital square wave output and use A & B quadrature.  TONiC systems use both analog (sine wave) and digital outputs (square wave).  Unlike other encoders, the interface output does not include quadrature when calculating Renishaw final output "ticks".   
TONiC systems have a readhead and a separate interface, where the VONiC system has the interface in the readhead, I think. This is why it's important to speak with a rep., so you don't get the wrong stuff and can't return it.

The limit of a Servo II is about 8M to 10M ticks per second, so keep that in mind for slewing when adding everything up.

Best,
Russell R.

Hi Russel,
your encoders are so hugh! No I don't want so many ticks, I said only that it is possible up to....
I thought about 10-15 million ticks/rev. If Servo II can deal with 10 M ticks my mount can do nearly 1 rev/sec. Thats fast enough ;)!
Your encoders on DEC are working smooth? You have a x1000 interpolation?
My RESM20 has 15744 lines (DA 100mm) x1000 are 15.744 mill. ticks
In the sitech help they recommend 13 mill ticks for cascade mode... Ideal!
I have already a VONiC readhead an RA, yes Interface is integrated. And it works without error since 4 years.
But only at x200 interpolation, so I have on RA only 3,15 mill ticks, but cascade mode is working too.
cs
Ralf

Ralf,

Yes, the DEC works very well. Yes, it is interfaced x1000.  I do not use a guide camera.  I think Dan recommended 10M plus ticks for cascade mode in the help files, but I'm not sure of the implications if you don't have 10M ticks using cascade mode. If you use cascade mode, turn off backlash setting.  If you have sever eccentricity in your DEC axis, you may want to consider dual readheads.  I use dual readheads on my RA axis to help make sure in case there is some bearing wander.

While working with my Renishaw rep., he had a RESM ring that had some scratches and said he could not sell it, so he gave it to me (413mm) with my 255mm purchase.  My mount was large enough to handle it, but the machine work was very intense. Needless to say it works very well. 

Nice looking mount!  Well done!
Russ

 

Hi Russ,
thank you. The axis itself is concentric, only the worm gear is eccentric to the DEC axis, so there is slightly different space between worm gear and worm during one revolution. So normal backlash setting doesn't help . Made one little mistake on the lathe...
cs
Ralf

Hey Dan,

What would happen if Ralf got a renishaw on the DEC, but wired the renishaw encoder in place of the motor encoder, so that ServoII would "think" the rehishaw is the motor encoder and would run concentric, not using cascade mode?  I did this with my focuser motor using a renishaw linear encoder.  Works great in that application.

Russ 

Hi Dan and Russ,

that's the million dollar question, I can't answer, because I don't know, how cascade mode exactly works.
"The cascade mode works by positioning the motors to a location where there is zero error on the scope encoders."
That makes no sense to me: the motor are running until both encoders tell the same?
Better would be this scenario: There comes the command "move 1000 tics to the right", so the motor will run until the scope encoder tells: "you reached 1000 tics: motor stop".
But Russ is right, perhaps use normal mode and scope encoders instead motor encoders. Sounds logical. Tried someone else this?
Motor and Scope encoders have nearly the same resolution.
But no doubt: in this resolution scope encoder will jump even if standing, 1 tick in my configuration means tiny 0,1 arcsec or 0,000025 mm on my 100mm dia ringencoder.
But that doesn't depend on the backlash, it will always be the case. I think...i hope.

cs
Ralf

Hi Dan,
saw a interesting video at YouTube from you about increasing pointing accuracy.
- But still don't understand exactly how cascade mode works
- Absolute encoders: are they so much better than relative ones and their price worth?
- Position of Read head: for GEM (I know you HATE GEMS) On DEC-axis towards south? What do you mean?
- Exact time: wired time server: what do you mean? I use an little app on my computer called "Allzeit Atomzeit" that synchronizes my system time on every start to official atomic clock in Braunschweig per millisecond.

By the way: I started in 1990 exactly the same way: stare through a cross-hair ocular for "hours", waiting for developing and then see hooks, waves, zigzag...
Ralf

Hi Ralf,

The question of whether mount encoders can control the mount is answered YES by AstroPhysics with their new Mach 2 mount, which uses only mount absolute encoders (65million ticks/rev).  they completely handle any slack, backlash, and periodic error.  The fact that they use absolute encoders is not the key, it is the high resolution and firm mounting of the encoders that is key.

Note that even with absolute encoders, the mount pointing is controlled by the last OffsetInit (for SiTech) and the last RCal for AP (NOT absolute angles).

So Ralf, try it, it may work great.  Just let us know what you find.

Don W