Telescope project gets to first major milestone

[rfdesigner]

Hi

I've been keeping a bit of a log of progress on my telescope project on my blog... however tonight I've managed to get first all round functionality.. so I'm a bit chuffed and wanted to share here alongside some of the other DIY projects.

What's the project?

A computer controlled telescope mount, I have the hardware (a fullerscope MkIV mount) what was missing was an up to date Goto control system that could do anything else besides should I want it to...

What's it made from?

Stepper motors, direct driving the worms. Dedicated microstepping motor driver chips driving the motors, a PIC driving the stepper chips, another PIC dealing with handset instructions and coping with an LCD display and a keypad. I've left room for a USB interface so it can talk with a PC later on.

So what's just happened?

Tonight for the first time I've got the entire system working as a whole.. :D:D:D:D:D:D:D

Turn on, it starts tracking as that's the default mode, select slew mode, select slew rate, then slew the mount, both axis at the same time if desired. Time for a flip of the GEM is 90 seconds, I might work on that to make it a bit quicker.... when I've got nothing better to do.

I'd had various bits going before, even managed to take a couple of astro pics, but this is the first time I'd got the lot working together, so I'm just a bit chuffed.

What's next?

Autoguider port.. actually very easy the way the rest of the system is built. I'd give myself something around a week of half an hour here or there to get that working, then it's onto tidying up the hand controller.. proper PCB/case etc which will take a little longer.

Outstanding Issues?

Mechanics. I need a new worm cut for the Dec axis, the old one is ok for now but it's plain steel and some of the teeth rusted a little, the RA used stainless so it's fine. I cleaned away the rust but it means it's no longer quite as complete as it should be. Careful alignment means I'm happy for now that it isn't eating the worm wheel, but it does need replacing. I'm also going to need some metal cut to fix the steppers on better. The shafts are secure, but the bodies of the motors are held in place with wood.. not a long term solution. Anyone got any ideas on where I could get this done?

Anyway.. next weekend if all goes well and the clouds keep away I hope to be swearing at my autoguider as I try to achieve lock for the first time.

Derek

[Tom How]

Blimey - impressive! That is no mean feat. Pictures please?

When I made my mount, I dismissed the idea of doing the motors and controller myself. Too much like hard work. I guess having a spare set of Meade DS motors + autostar contributed to this decision.

Does the GOTO part work yet - how do you tell it where to go to?

How about a picture of this damaged worm - if it is only slightly damaged, I wouldn't worry too much on the DEC. Getting a replacement could be tricky.

[rfdesigner]

Tom

Glad you like it...

Pictures to follow... (no video clips as I realised I don't have a video cam of any sort at home)

RE Goto

Whilst the handset doesn't yet have a Goto function the base unit is built with this in mind, so tracking is done by adjusting 'time' thus slewing a celestial sphere in memory, the mount then plays catchup via error correction. To get Goto I'll need to allow the hand controller to change the reference point (easy, already proved it can slew to a setpoint). To make it work well I need to do some trig to correct for misalignment, atmospheric refraction etc, this means upgrading the base PIC to a dsPIC, I've run some sims on one and all the necessary maths will only cost about 20% of the computing power.

RE the DEC worm. I am more worried about further degredation of the worm, going stainless would sort this... not this week though.

There's a whole heap of other stuff I'd like to add, like 'endpoints' (to stop it running the scope into the mount), 'viewable objects' (goto warning before trying to slew to an object below the horizon / behind next door). But that will all take time, and I want a robust system before I put bells and whistles on it, but putting bells and whistles on won't take too long due to the basic design.

Derek

[inksmithy]

Derek, that sounds fantastic - its an awesome feeling to get to a milestone like that, congratulations.

I'll be watching with interest!

Alan

[Tom How]

yeah, end points are vey handy. I ended up writing a extra little program that works with my autostar to automatically park the scope when the object being imaged had reached a "pier crashing" position. (I'd long since gone to bed).

Actually this reminds me... I need to look at that again in view of my new mount... it can track a heck of a lot further beyond the meridian.

Another things worth thinking about at this stage, if you intend to do astrophotography, is putting some kind of switch/indexer to note the position of the worm wheel so you can put a sub routine in for permanent periodic error correction.

[wackyscot]

Crikey, well done that man. I thought I was doing well making some anti vibration feet for my tripod lol.

David

[rfdesigner]

Tom

Thanks.. yes I plan to use PEC.

Actually I've already proved I can do PEC, but I need to write the bit where it learns what the PEC is. (I've worked out how to do this entirely internally to the mount, but I need to autoguide first) A shaft encoder could be handy, but my intention was to never disengage the worm or clutches, So I need to add a 'park' feature so it can turn off into a known state then shaft encoders may not be needed. I may still add a shaft encoder of some sort though.

anyway, autoguide first.

Derek

[Gina]

Great stuff!! Well done

[rfdesigner]

It's taken a week.. and it the driver has an autoguider port. Not a huge step forward in terms of complexity, but it's a point I've been driving towards for quite a while.

so PHD guiding installed and..

oh look it's getting cloudy despite all the promises of a clear night, I must be getting close, the weather knows!

The scope's still got the target star in it's cross hairs so I let PHD guiding loose and see what happens.

a bit of calibration, I can hear the motors respond with their gentle singing, the star moves back an forward up and down, then..

Guiding!

it's actually done it!..

Of course I couldn't actually use the setup to take any photos, the clouds saw to that... and there's quite a lot that needs tidying but with a bit of luck I might actually be taking photos with this rig soon.

Must get some pics to add to this thread... maybe in the morning.

Derek

[yeti monster]

Thia does sound like am impressive project. Looking forward to the piccies...

[mag10]

I read something like that and I'm amazed.

Not only do you program everything, but then you build the hardware that makes everything go.

Out of my class?

I'm not even in the school-house!

Congratulations!

[Essex Stargazer]

Derek

Boy o Boy your electronics skills sound very good

Most of what you said went straight over my head

But I do understand the chips bit , yum

[Gina]

Nice progress - well done

[rfdesigner]

I must confess.. I do electronics for a living (RF in my name stands for Radio Frequency)

The problem with RF is everything needs to be 'just so' and most other enginners just don't 'get' RF so if you need a PCB made or software written it's often much quicker to do it yourself, or at least work extremely closely with the appropriate specialist. So I've been forced to learn most electronics disciplins over the last couple decades... that process of just getting on with it yourself is an excellent grounding for this project.

Anyway.. now it looks like this prototype is functional, it's going to be time very soon to migrate to a proper PCB (it's all on Veroboard at the moment) and some nice boxes. Thankfully I can get PCBs made for quite a low price these days so making it reliable and looking nice won't be too expensive.

What I am beginning to wonder is whether to make it good enough to sell.. but that won't happen this week!

Derek

[rfdesigner]

As promised: Pictures

Various pictures of the scope slewing round. A pic of the main board.. which is a PIC16F690 and two Allegro microstepping stepper motor driver chips, plus various other bits and bobs to keep power supplies solid, or add functionality, like the autoguider port which is a real bodge mechanically, but totally Ok electronically. The 'hand controller' can be seen sitting on the carpet on the end of a length of ribbon cable, it's just some more veroboard with a keypad and LCD attached.

Power supply images showing normal voltage and current demand. (I have a DC-DC design I can steal from another home project that can deliver 40V at an amp from a 12V supply)

For testing this outside the mainboard is put in a sandwedge box and the lid stuck down and attached to the pier with duck tape... the hand controller is all low impedance stuff and can cope with a suprising amount of dew on it! (not reccomended )

So as everyone can see... prototype works.. now I need to work on a couple of PCBs and some boxes.

Sorry if it doesn't look much, but 95% of the interest in this one happens in embedded software, which now runs to about 3000 lines of code.

Oh.. and I've now got self programming PEC built in, but untested (which means I think it works and am about to get a rude suprise when I test it )

EDIT: just realised, I can add the guiding trace.

Derek

[ismangil]

This is awesome.

I am just starting to tinker with Arduino to build a hand controller for an EQ5 clone mount, still figuring out best way to turn the motors! Single, double, microstep, speed....

Might ask for some advice later...

[rfdesigner]

Might ask for some advice later...

feel free, the worst I can say is "dunno"

Derek

[rfdesigner]

Last night I took the opportunity to test the next mod: PEC.

Well to be blunt, it worked, but showed up an error I suspected still lay at the heart of the project which I yet to root out.

Firstly I managed to get Phd guiding to get the guiding quite good (rms error of 0.4 or less at 1.1 arc second / pixel), osc value drifting between 0.35 and 0.55... I only wish I had the off axis guider right now, but it's still stuck on my wish list.

I turned on the 'learnPEC' function and all seemed farily well except I noticed that the corrections had a lot of 'west' in them.

So I turned it to 'usePEC' and turned off the guiding, then took some 240 second images (240 seconds = 1 worm rotation period)

Now the alignment was deliberatly off a little so I would end up with a north-south star trail, I could then measure the wigglyness to determine the resultant error.

Anyway. here's the with PEC (left) and without PEC (right).

It is clearly obvious that the 'with PEC' has a step, caused by the learning algorithum pushing the PEC values in an ever increasing progression. When the learn PEC algorithum was stopped, the result was a step in the curve. What is clear is that the PEC does smooth things out a lot, and that I must get the tracking rate bang on.

back to the spreadsheet.

Derek

[inksmithy]

What kind of learning function are you using there Derek?

Alan

[rfdesigner]

I use an array of values associated with the worm angle. Currently due to memory limitations I only have 50 values.

As the autoguiding process continues every time a correction to RA is made and the learnPEC is active, the approprate value in the PEC array is incremented or decremented in line with the autoguiding command.

This is then added to the appriopriate stepper motor positions, so that the PEC values gradually hone into the best positions for the minimum overall error.

Of course it all falls down if the normal tracking rate is to fast or slow.

Derek