Focuser Backlash compensation!

Hello all,
I kind of have a question and maybe a feature request if appropriate or maybe a little understanding instead with ref to a focusers backlash compensation.

It seems strange to me that you can only set backlash compensation to either ‘In’ or ‘Out’ but not both.

Recently I used a dial gauge to measure the backlash on my Moonlite with ref to working out step sizes etc. What dawned on me while doing this is that the measured backlash is the same in either direction, what I mean is, if my focuser tube is traveling outward and keeps going outwards the backlash is on the inward side and vice versa.

This got me thinking all because of the clouds LOLOL

When the tube it travelling outwards, If I were to suddenly change direction and go inward by 50 steps then my focuser has to first move the relevant amount of steps inwards taking up the backlash and then the tube will move the remaining amount, if your backlash step size was lets say 20 steps then the inward movement of the tube will only move 30 steps which is 20 steps out from what it thinks it’s at. Thereafter though, for every step inwards a move of 50 steps will move the tube 50 steps cos the backlash has been taken up.

Once again, if you then change direction to outward again and move 50 steps then the change in direction again needs to take up the 20 step backlash leaving you with a tube move of again…Only 30 steps.

I can see this kind of thing will have major effects on auto focusing. SGP when running it’s autofocusing routine begins by moving the tube outwards to the predetermined position based on half of data points * step size, if backlash comp was set to say 100 steps then it will move 100 further steps out and back again to take up the backlash for the inward direction, then SGP does it’s data points going inwards each time, past originally set focus point and ends inwards of the focus point and calculates best focus position. At this point SGP them moves the focuser out (change in direction) to send the tube to the position it calculated, but because it changed direction from inwards to outwards, if the tube was told to move out by 300 steps to hit that golden point then it will miss that point by 20 steps because 20 of the 300 steps were swallowed up by the backlash. Now 20 steps off might be fine for some but catastrophic for others based on what their CFZ is, it might end out of their CFZ.

If im talking complete rubbish then someone please correct me now !

So with the above in mind and no one makes me look like a clown, let me explain what I think should happen:

Backlash compensation options should contain a third option called ‘In & Out’ or ‘Both’, SGP will keep a note of which direction the tube is traveling every time there is a tube move. If the focuser changes direction and the move is 50 steps, SGP will in response to the direction change add the figure set in the backlash comp steps size box to the 50 steps move.

Example:
Backlash comp is set to 20 steps

If travel is inwards 50 steps at a time and the focuser tube changes direction then the signal sent to the motor will be 50 + 20 = 70 steps. This way, on direction change the 20 steps takes up the backlash and the 50 steps actually moves the tube by the required 50 steps, providing the next command is still outward by X amount then the backlash figure is not applied, again, if direction changes to inwards then SGP will again add the 20 steps compensation to the figure sent to the motor.

Surely a method like this gives a more ‘Absolute’ position to every move of the tube. Even it you cant pin your backlash step quantity perfectly, or there is a little bit of variation in the gears for example, you should at least get the tolerance to within a couple of steps from perfect, at least then every direction change move would only vary by a couple of steps from absolute position which has got to be better than 20 steps, When SGP does it’s final ‘Go to ‘X’ focus position’ it would also be only a couple of steps out instead of 20 steps.

So, again, unless I’m corrected I’d like to propose the following:

  1. Add a new option to Focus backlash compensation dropdown box called ‘In & Out’ or ‘Both’

  2. Have SGP monitor focuser tube direction and identify that a direction change is about to take place and to add the ‘step value’ in the compensation box to the figure sent to the focuser motor but only when the direction changes.

Well, im going to shut up now

Many Thanks
Paul

1 Like

Please change the topic category and place it in the Feature Requests category. Features outside this category (like this one) will not be considered for implementation (it is too time consuming to try and figure out which posts are requests).

Thx.

When the tube it travelling outwards, If I were to suddenly change direction and go inward by 50 steps then my focuser has to first move the relevant amount of steps inwards taking up the backlash and then the tube will move the remaining amount, if your backlash step size was lets say 20 steps then the inward movement of the tube will only move 30 steps which is 20 steps out from what it thinks it’s at. Thereafter though, for every step inwards a move of 50 steps will move the tube 50 steps cos the backlash has been taken up.

Once again, if you then change direction to outward again and move 50 steps then the change in direction again needs to take up the 20 step backlash leaving you with a tube move of again…Only 30 steps.

I can see this kind of thing will have major effects on auto focusing. SGP when running it’s autofocusing routine begins by moving the tube outwards to the predetermined position based on half of data points * step size, if backlash comp was set to say 100 steps then it will move 100 further steps out and back again to take up the backlash for the inward direction, then SGP does it’s data points going inwards each time, past originally set focus point and ends inwards of the focus point and calculates best focus position. At this point SGP them moves the focuser out (change in direction) to send the tube to the position it calculated, but because it changed direction from inwards to outwards, if the tube was told to move out by 300 steps to hit that golden point then it will miss that point by 20 steps because 20 of the 300 steps were swallowed up by the backlash. Now 20 steps off might be fine for some but catastrophic for others based on what their CFZ is, it might end out of their CFZ.

I considered this myself and wondered if this was the reason I kept getting the focus warning!
Having said that, I considered that the SGPro team know what they are doing and ive not seen a 3 way backlash option anywhere else but certainly worth the question.

Ray

My apologies Ken, completely missed that one…Moved

I don’t see why you need backlash in both directions. You only need backlash when you change direction. Normally you work against the weight of a camera against gravity. If you always end up moving in the same direction, QED, there should be no effective backlash.

Buzz,
Your right buzz, you do only need backlash comp “when you change direction”, and yes, if you always move in the same direction then the backlash is taken up ‘on the first move to that direction’

 Maybe gravity will play a part with some focusers I'm not sure but my Crayford moonlite

CR2 is solid as a rock whether the camera were to be below the focuser or above the focuser (in relation to up & down gravity), it just doesn’t move a millimetre, it takes quite a pull or push to actually make the tube itself move inside the focuser body manually even with the weight of the camera kit connected, with this being the case, I can grip the loosened clutch assembly which connects the motors output spindle to the focus knob’s spindle and there is a fine but definite wiggle there on the motor side, this is the backlash within the gears inside the focus motor housing not at the focus tube end and this backlash is relevant whenever the focuser changes direction, not just in or out but both directions.

I worked out my backlash to be approx. 5 steps. Using the dial gauge and moving 1 step at a time, On a direction change I have to make 5 ‘1 step moves’ before the gauge registers a tube move, exactly the same when doing the opposite direction, I will say that now & again the tube made it’s first move on the 4th click and now & again on the 6th one but 97% of the time it was 5 steps for the focus tube to begin moving…That’s Backlash in the motor housing ! Tried this with the focus tube and camera vertically upwards(downward gravity force) and horizontal with the focuser & camera to the side (no actual gravity force in play)…Same result so there is no gravity here.

Mounts solve the problem of backlash in the gears by unbalancing the mount by moving the weights in or out thus ensuring that all the gear teeth lean tight against each other but this only works in one direction east of the meridian, if the mount does a flip then the mount would move so the gears go tight against the other side of the teeth. This small move from one side of the teeth to the other side would end up with the scope itself being slightly out of position which is why after a flip SGP gives you the opportunity to pause and move your weights in the other direction so the teeth engage against the same side as before the flip.

The mount backlash compensation method wont work on focusers though

Paul

There is an implicit assumption that you are always finishing focus in one direction - and moves the other way will involve backlash compensation. So moving one direction will always be well behaved - because you know you have removed backlash already. And moving “backwards” the other way will require overshooting - then coming back.

So I don’t see a need for this feature. Gravity gives you a reason to choose one direction over another as the direction of final focus - but even if it was symmetrical, you would choose one way as the one good way - and the other way requires backlash unwinding.

This is all concrete during automatic focusing when you make all steps in the same direction. You would not want to unwind backlash at each step - and you would not want to take the focus curve the “wrong” way.

There is already a feature request to let the focus curve go from low to high numbers - or high to low - depending on the compensation direction. I think that feature really is needed and there is already a request for it. But I don’t see a need for this request to unwind either way - because you only need to unwind backlash in one direction no matter what.

Frank

Frank is spot on - the mention of gravity was to favor one compensation direction over another. The default direction in SGP goes from higher numbers to lower, which is often the ‘conventional’ direction opposing gravity. I’m guessing that some focus motors work around the other way.

Thanks - yes - my focus motor went the wrong way on my sct and I had to rewire it to reverse the direction. So it can definitely happen that the focus curve goes the wrong way. But that feature request has already been made and I hope it’s still in the system.

Frank

Sorry guy’s, been away and couldn’t get on here…

freestar,
I hope the feature request to which you refer gets implemented at some point, it’s obvious it is a valid request for ‘some’ as you have discovered yourself, question is, if it does get implemented, will you then rewire you motor again ?

LOLOL

“There is an implicit assumption that you are always finishing focus in one direction - and moves the other way will involve backlash compensation”

Interesting statement - ‘implicit assumption’ - to mean: Suggested though not expressed, Accepted without proof !

It seems all of the main Astro-Imaging software’s also use the IN or OUT method for backlash compensation so SGP is not alone in this arena, its obviously the ‘implicit assumption’ is that that’s how it’s done and that’s how it is and that’s it.

I’m merely here to challenge that ‘Assumption’ with real world proof based on my observations !

Gravity - Lets look at this in a little more detail - You have a refractor pointed at your zenith, the focuser end is pointing down towards the ground with the camera attached and an autofocus motor is attached. One could ‘assume’ that gravity is acting on the camera and focuser tube and one would be right but there are two possible mechanical situations here I can think of which make a world of difference to the use of backlash compensation with the said scope setup:

  1. Loose focuser mechanism - If this were the case, the weight of the camera would try to pull the tube out(or downwards in this case), this would in turn twist the focusing knob spindle which would in turn twist the gears inside the focus motors housing until all the looseness or backlash was taken up and the tube would stop outward movement all in a fraction of a second - In this situation you need no backlash compensation at all because GRAVITY is always keeping the backlash at bay due to the weight of the camera on the loose mechanisms, you can go in or out to your hearts content and wherever you move your focuser, it will be in the exact position you intend it to be…Result = NO backlash compensation required due to the helpful effects of gravity.

  2. Stiff or well adjusted Focuser mechanism (This option is more likely valid today in 99% of focusers) - Same situation with our refractor pointing at the zenith but something is different here. The weight of the camera is still pulling on the tube due to gravity but the tube wont pull out as the mechanisms are nice & tight. As a result, the focusers spindle doesn’t get twisted which in turn doesn’t twist the gears in the motor housing and thus, the looseness in the motor housing is not taken up - Backlash is now present in spite of the ‘acting Gravity assumption’ Our tube can only move position by a solid defined rotation of the focusers spindle as the gravity effect has been over ridden, but, there is backlash present in the motor housing due to loose gears.

Further to point No 2:
If you now move the focuser OUT by 1000 steps, your tube wont initially move, first the backlash has to be taken up until all the gears in the motor housing have engaged tight against each other (you just lost, lets say, 20 steps taking up the backlash), once the 1000 step move is complete, and if you were to use a dial gauge to provide ‘PROOF’ you would see you have only moved the tube 980 steps not the intended 1000 steps. Likewise, if you now move the focuser ‘IN’ by 500 steps, again the tube is solid, before it can move inwards you have to take up the backlash in the motor gears first, so on the change in direction you again loose 20 steps taking up the slack in the gears and you only end up moving the tube only 480 steps inwards.

The other situation to consider here in relation to Gravity is a Newtonian reflector, the camera here sticks out of the side of the tube so is mainly horizontal to the ground( if your focuser is pointing towards the center of your mount), as the camera will never be pointing down to the ground like on our zenith pointing refractor gravity will have practically zero affect on this kind of setup.

On to SGP and all the other software:

I see the point and I believe I understand what it’s doing when applying backlash comp when doing autofocus, as long as SGP’s backlash step figure is more than your actual backlash size in steps (and you have the direction set to IN) then the move out and then back in WILL take up you backlash for the inward journey to get your data points. But when SGP does final calculations and decides how many steps to go ‘OUT’ again to hit best focus it’s final position will be off by the size of your motor housings backlash which could mean the difference between focus or slightly out of focus. This is not SGP’s fault but the fault of the ‘implicit assumption’ itself.

From my observations with the dial gauge - Backlash is not directional, it cant only apply in one direction unless you have a very loose focusing system assisted by gravitational effects like the one highlighted in point 1 in which case you wouldn’t need to apply the feature anyway in any form. Backlash can not ‘accurately’ be dealt with by using an ‘implicit assumption’ just because that’s the way it is and just make sure the figure you use is more than your actual backlash size. Continuous improvement, challenging older concepts to see if better ways can be found to deal with old problems and there’s certainly lots of people on these forums and many other forums who suffer the same problems when it comes to motor driven focusers but no one seems to challenge things.

I would urge anyone to use a dial gauge to test out these claims because it COULD pave the way for a new way to deal with focuser backlash compensation which MAY help lots of users and further improve an already outstanding piece of software.

Again, these are just my own opinions, doesn’t make them right or wrong and no malice intended

Regards to all
Paul

I don’t think that is right. Looking at the focuser position values, the backlash is applied on the final move too. It goes past and comes back again.

No I would not need to rewire the motor if the requested change is made - because the request is to allow SGP to work properly with any focusing system.

Once the change is made, a user can choose the compensation direction to be whatever makes most sense - and then the focus curve and all focus moves in any direction will be accurate because the final move will include the requested backlash compensation in its history of moves. The one thing that should not happen is to move a small amount one way - and then a small amount the other way. You only need backlash compensation in one direction to guarantee that. Removing backlash in both directions is unnecessary and would be a disaster during an autofocus run.

Frank

I know this is an old topic but rather than create a new one for the same item, I would like to add to this one and also recommend adding the backlash compensation option.

While I believe the gravity argument above does have merit, it depends on how smooth the action is on the focuser whether the weight of the cameras (And the angle of the equipment) will pull the focus to it’s outward most point at rest. Not everyone is able to image at or near zenith.

As another hobby, I retrofit manual machines that perform metal work. CNC Mills, Lathes, Plasma tables, etc. They use rotating ball screws that drive heavy tables to move material and no matter how precision you purchase in this hardware, there will be backlash present to some extent. If no backlash compensation is applied in the machining control software based on the measurements of the machine movements, any parts that are created will be undersized because the equipment did not move to the expected location to perform the work. When you are working with dimensions in the .001" range and less, that makes a big difference in a scrapped part or not. The same would be true here in the focusing side.

I use a Rigel System Nstep, I know from using a dial test indicator exactly how many steps my focuser must move to be accurate both in and out but there is no where to enter this into the software for compensation to occur. From my perspective, It doesn’t matter whether moving in or out on the focuser. If adjustments are made and the focuser has backlash steps that aren’t cleared out and compensation applied, then a repeatable location does not exist when running the routine. Due to this, my focus runs always end up with a sharp hook on the right side because of the backlash introduced by the focuser. That said, these are on the factory Explore Scientific focusers on a refractor I use for AP, but it wouldn’t matter if it was Moonlite or Feathertouch, etc. Backlash is backlash and if it can’t be trained out of the system, then the slop introduces errors in the motion and calibration.

Having this setting would allow the system to account for that and not undershoot/overshoot target settings and make focusing less of a hassle. In Mach 3 (The software I use for my machine work) there is a wizard that allows you to easily determine the backlash and something like this would make setup easier for the user getting started. I don’t use the wizard any longer because I use screw mapping to determine wear in a similar way PEC works but initially on my machine. It would help to have a general backlash compensation in the system for users that need it. Starting with just adding that would be a blessing and the wizard later maybe…

We’ve had backlash compensation for almost 2 years now and we added software reversing a couple months back.

Thanks,
Jared

Thanks for responding Jared,

I don’t see any settings for this adjustment in SGpro and have searched the help files and all the menu options related to the Focusing side. Where is this adjusted?

You can find the settings in the “Other” section of the focuser cotrol panel or profile editor.

http://mainsequencesoftware.com/Content/SGPHelp/Focusers.html

Thanks,
Jared

Thanks Jared. That helps a lot.

I really don’t see how I overlooked that but there it is. Thanks.

Hi guys

One thing that I am not quite sure is if the backlash compensation is applied by SGP only when the focuser move direction is changed.
I haven’t done this with 2.6.X but I noticed that the compensation was applied with any focuser movement if the focuser move direction was opposite to the compensation direction while I was looking at the focuser step number.
The BC seemed to be applied even though the focus move direction was not changing.

Am I mistaken or is this how SGP behaves?

Thank you.
Donghun

Buzz, you are absolutely right with this statement. This is very easy for me to see with my RASA11 rig, since I use a backlash setting of 2000. On the initial move to the largest step value, it goes 2000 past that point then comes back down. When it has moved down through the levels and determined the best focus, it moves back up to the best focus + 2000, then back down to best focus.[quote=“cjdavis618, post:13, topic:2573”]
Due to this, my focus runs always end up with a sharp hook on the right side because of the backlash introduced by the focuser.
[/quote]
This is what I was always getting before I started using the backlash of 2000. Now I never have a hook on the right side (or the left either for that matter). Current design of focuser algorithm is working great for me. Actually, getting rid of the hook was the primary method I used to figure out how much backlash I needed. I kept increasing it until the hook went away. That happened around 1000. I set it to 2000 just to be sure. The RASA focusing moves the main mirror so its not as precise as a camera focuser.
My system clearly only needs backlash in one direction.

No, backlash compensation is invoked whenever the direction of the move is the opposite of your compensation direction. So if you have the backlash compensation set to “IN” the every “OUT” move will trigger backlash compensation so that the end of the move is in the “IN” direction. So if you move your focuser “IN” then no backlash compensation is triggered or needed.

Thanks,
Jared

www.mainsequencesoftware.com