How does the GoTo work? - A slew correction issue

[Vroobel]

Hi all, 

Do you remember my huge worm gears and GoTo system built on Dobsonian mount and Raspberry Pi?

https://youtu.be/u7_AuXA8eNs

https://youtu.be/2W5UI7m4FnI

It works very well mechanicaly - I can rate it looking on its tracking. Even a slew works well, but the correction is required. A problem appears exactly here. 

I can imagine how the original GoTo works with the Alt-Az mounts, but still know nothing about the correction of the slew. 

For beginning, I assumed that the first part of the slew moves the telescope to exact position of the object at the time of a choice of the object. But the object is moving at this time, so the second part of the slew moves the telescope slightly more. Of course, sometimes the telescope goes backwards then, depending of the direction. 

I'd like to explain my idea of the correction. After the first slew (its part 1 and part 2) I have to move the telescope a little bit towards the object both in Az and Alt axes. Every movement up and down, left and right is added or substracted (depending on the slew direction) to/from the primary numbers of steps and then a kind of factor is calculated, ie. Az=1.009912345 and Alt=0.999234567. That means numbers of next slew will be multiplied by the factors. 

I didn't know if the factor should be applied to the 1st part of the slew or to the 1st and 2nd ones together, so I developed a feature allowing to make a choice during the observation. 

I tried it yesterday and unfortunately it's not working properly. The results are random in my opinion, I can't find any pattern to sort out the problem. I assumed to add the factors to numbers of the steps, but maybe the correction is related with something else? 

I stuck now and have no idea what to do. 

Is someone here knowing the GoTo software and eager to give me a clue? 

Edited August 19, 2019 by Vroobel
Link to YouTube added & title

[Vroobel]

I still try to solve the problem. My system doesn't have encoders as feedback, so I think  stepper motors can loose some steps during striving to bigger speed of the slew.

Any other ideas?

[Vroobel]

Hi, 

It looks like end of my troubles with GoTo. 

The reason of "random"  deviation in the slew wasn't related with software, even it wasn't related with mechanical construction. It was insufficiently leveled base of my Dobsonian mount. I trusted too much earlier the level of my troley and mount without OTA, but this night I got the glare and checked spirit levels of all set and its full load. One of the spirit levels showed that it needs some improvement and 3 turns of one M10 screw (3x 1.5mm = 4.5mm!) was enough to do it. 

Finały I had a pleasure to jump randomly between M31, M13, M92 and M15 and also plenty multiple systems visible from my garden. I started with 38mm eyepiece (33x), then swapped it with 25mm (50x) and every time the objects landed in middle area of its field, but most often almost exactly in middle! I could check it with my laser. So I risked and added a Barlow lens x2 and again, randomly chosen objects could be found in middle of the field even with 100x magnification.

I am proud of that, as the construction is my own at all. 

Another video on YouTube will show my GoTo and its accuracy soon.