Yet another DSC encoder project. A Bresser-Messier Dob254 one...

[cwis]

In short:

I have a Bresser 10inch dob and want pushto via Star Safari  et al. I have a garden with restrictive views and star hopping is a little frustrating. I had a look at available solutions available and decided it would be tidiest and most satisfying to roll my own.

I started off experimenting with magnetic 12 bit rotary encoders (here's one mounted to the azimuth axis on a little 3D printed mount).  The magnet is mounted to the pivot bolt below and the bolt held stationary so it does not move relative to the dob stand - so turning the mount rotates the magnet relative to the sensor.

 

Results were very encouraging so I splashed out on some 14 bit sensors on their own I2C development boards (with matched magnets).

After some time working out the i2c commands etc to pull meaningful data from them on my test rig (Raspberry Pi Zero using Python):

I moved to "production" with an ESP32 development kit  - I could solder plugs on directly for neat, easy, robust connections, I could use two hardware i2c buses so I didn't have to change the default configuration of the sensors to use different addresses or have any extraneous wiring connecting the two sensors together or use flaky software buses...

(I used pins set digitally high to power the sensor boards - power draw was fine doing this, but I did use the actual proper ground pins). 

(BTW I hope you like the carpet in the spare room. Genuine '60s -  It's not one for looking at with a hangover!)

After an issue with a dry joint on a clock pin was diagnosed using the worlds cutest oscilloscope: That's I2C right there. You can even see the weak pullup resistors on the ESP32 making the return from ground back to 3.3V slower than ideal.... Mind you my cables are a bit long - I'll trim them when it's on the scope.

Next- print some mounts that are a snap fit for the sensor boards, print something that snap fits in and provides a pin to centre the mount on the dob base:

The taper is a bit messy - printing PETG on a printer set up for ABS so there's no part cooling fan. I should have printed 4 copies of it so each layer had a chance to cool. Nevermind.

Ask a grownup with a lathe to make lower profile replacement screws for one of the Alt bearings out of aluminium - One with a recess for the magnet:

I pick those up tonight... More soon!

 

Edited June 30, 2021 by cwis
Horrific grammar then missing pics

[faulksy]

for some reason i had no pictures except the last one ?

nexus system is fairly cheap and it is ready to play with

[cwis]

3 hours ago, faulksy said:

for some reason i had no pictures except the last one ?

nexus system is fairly cheap and it is ready to play with

Oh yeah that is broken! I'll have another go... Cheers!

[cwis]

4 minutes ago, cwis said:

Oh yeah that is broken! I'll have another go... Cheers!

Right - that's better, pics show in incognito mode too.

I had a look at the Nexus along with many other systems and although it's a great system (fantastic, even!) it's far more than I need. 

I don't need a star catalogue - Sky Safari has one.  I don't need an OLED display or GPS, my phone has those. etc. Nexus is designed to be stand alone - where as I will always be using mine in conjunction with a phone, table or laptop. I'll power it off a USB pack so don't even need an on-board power source.

My base functionality is: 14 bit encoders with noise suppression, linearity,  averaging etc etc for probably 0.1 degree accuracy (maybe 0.05 if I mount the magnets well) , and bluetooth capability to connect to an application written by grownups.

 

 

[Scooot]

18 minutes ago, cwis said:

Right - that's better, pics show in incognito mode too.

I had a look at the Nexus along with many other systems and although it's a great system (fantastic, even!) it's far more than I need. 

I don't need a star catalogue - Sky Safari has one.  I don't need an OLED display or GPS, my phone has those. etc. Nexus is designed to be stand alone - where as I will always be using mine in conjunction with a phone, table or laptop. I'll power it off a USB pack so don't even need an on-board power source.

My base functionality is: 14 bit encoders with noise suppression, linearity,  averaging etc etc for probably 0.1 degree accuracy (maybe 0.05 if I mount the magnets well) , and bluetooth capability to connect to an application written by grownups.

 

 

This is the Nexus to go for as @faulksy mentioned.

https://www.astrodevices.com/shop/index.php?product/page/8/Nexus-II

Its a stand alone box designed to be used with Sky Safari etc and a phone or tablet. It doesn’t have its own Sky catalogue or OLED display. All you need are the encoders and mounting kit. I have an older model but it’s been superb.

Edited June 30, 2021 by Scooot

[cwis]

9 hours ago, Scooot said:

This is the Nexus to go for as @faulksy mentioned.

https://www.astrodevices.com/shop/index.php?product/page/8/Nexus-II

Its a stand alone box designed to be used with Sky Safari etc and a phone or tablet. It doesn’t have its own Sky catalogue or OLED display. All you need are the encoders and mounting kit. I have an older model but it’s been superb.

Yeah it's a lovely bit of kit and more in line with my stated requirements but I still think it's overkill for my application.  It's basically a piece of scientific apparatus!

My goal is something with a far lower level of precision.  I'm not going to be seeking out dim hard to find tiny objects from my light polluted back garden so don't require that level of accuracy. I'm not actually sure WHAT level of accuracy I'm going to get, and that's one of the things I'm finding interesting about this little project.

As a newbie I could buy another eyepiece with the difference in cost between my implementation and one made by a skilled grownup - and that's without buying encoders that suit the Nexus - I've not even looked how much they cost. Mine were 30 quid.... If I was willing to do the I2C implementation myself they would have been less than a tenner. I'll pay 20 quid to not solder surface mount stuff...

I've got my kit talking to Sky Safari now via Bluetooth so the last bit is mounting on the 'scope. I actually think this will be the hardest bit!

[cwis]

Bit more progress!

Magnet holding screw and lower profile screw fitted. Had to hacksaw the original lower hand screw off - due to a poor fit it had galled being soft aluminium and I couldn't get it out.... It had bottomed out in the thread/on the shoulder of the bolt and wasn't clamping the bearing - it was resting on the bolt and cutting in to it. I've put washers behind the remaining original bolts - might be worth looking at if you have one of these.

I had to print spacers for the above screws cos I gave the machinist the wrong numbers. We'll call these version 0.1 and get some more made - luckily I married his sister so he owes me....

Wrap tape round the mounting guide until it's a snug fit in the azimuth bush, put it in place and mark the mounting holes for the azimuth sensor and reassemble the base:

Immobilise the lower bolt:

Temporarily fit the sensor for testing.

Version 0.2 of the sensor mount will have slotted screw holes to make alignment  easier.

Next - The Altitude sensor. I'm going to route a mount out of ply and clamp it to the base for version 1.  I think I'll be mounting the microprocessor to this too.

I HATE routers so I'll be girding my loins and counting my digits first.... It should occur this weekend.

 

[cwis]

More progress....

I chickened out of using the router over the weekend because:

a). It terrifies me.  I'll be honest here - angle grinders, table saws, routers, chainsaws and power planers all belong in the category of "try and avoid". I avoided.

b). I had no flat plywood.  Plenty of warped ply. Hey ho - I suppose it's a few years old.

I improvised with some fibreboard cladding panels - they are flat and inert - I'll need to paint them before the dew touches them though.... Managed to reuse one of the original bearing thumbscrews and printed a spider to act as a clamp in the existing hole in the dob base with a printed 1/4 inch 20 tpi thread:

 

The version 0.1 Alt magnet holder ended up far more recessed than I wanted - hey ho. Let's see if it works. Tape the sensor in place, plug the esp32 into a power pack and fire up SkySafari:

It worked! Initially LOTS of noise in the sensors (the view was wobbling about in SkySafari) but I removed the spider clamp as shown above and pushed the Alt sensor so it was basically touching the Alt magnet and all the Alt noise vanished. All of it!

Pushing the telescope worked as expected - no lag, and the sensors seem correctly calibrated - a 90 degree turn of the base showed a 90 degree movement in SkySafari. Chuffed! 

So next steps:

1).   Reprint azimuth sensor holder to remove 6mm of gap between magnet and sensor. Hopefully that removes the noise from that axis too.

2).   Request version 0.2 of Alt magnet holder from handsome, intelligent, generous brother-in-law (who has a lathe and 1/2 inch aluminium bar stock) to remove the gap in the alt sensor setup:

 

3).   Reprint spider for slightly more lateral adjustment  (thread printed OK though - chuffed about that!).

4).   Find out why it takes a couple of minutes to connect via Bluetooth.

5).  Print up a holder for the ESP32 and a power pack - remember to mark on it what plugs into what. I might buy another one without the headers pre-soldered on to make it tidier. 

I have to be honest (touching wood etc etc) this has gone a lot better than I thought  - so far...

 

 

 

[cwis]

Still waiting for my Alt magnet holder so I've been tidying up the code and optimising the noise filter.

I also fixed an issue with a pin on the sensor boards floating high and causing errors on the I2c buses that stopped me sampling as quickly as I would have liked with the first iteration of the code - with that sorted the noise in the signal has dropped right down using simple averaging - the reading now typically oscillates between two minor bits when the axis is at rest (as expected really) so I'm looking at weighting algorithms to stop the telescope view in SkySafari from wiggling 0.02 degrees back and forth...

I could drop a bit of resolution (only use the 13 most significant bits in the signal) and remove the issue altogether but I'd rather find a better solution!

[cwis]

First field test! Zip tied all the components to the alt encoder mount and tried it out last night.

 

The current code zeros the encoder positions on boot so setup is basically point the scope North and level. I used Polaris to get North.  One idea is to zero the encoders and then offset the elevation to Polaris so I could point to that on boot, but that ties the setup to my current latitude so maybe not...

I've put all the gubbins on the opposite side to where you sit to keep it all clean but this may have been a mistake...

Power it up and connect using SkySafari.  I did my first alignment on Polaris and Caph and then tested with a few PushTos and some "sky browsing". I need to read the SkySafari instructions again re: alignment but I got there in the end.

PushTo tests:

I thought I had issues with both axis but this was quickly diagnosed as being caused by a wobbly paving slab (dur). Alt looks fine but I have a periodic error in Az that makes it about 2.5 degrees out over 180ish  degrees of travel. (Polaris to Vega for example).

I think it's related to poor centering of the Az encoder over the az magnet, as moving back 180 degrees to (Vega back to Polaris for example) completely removes the error.  

Vega was still in the finder but that's not good enough for DSO.

Keeping the pushto to within 90 odd degrees Az of Polaris allows a sub degree pushto accuracy - Polaris to M81 put M81 nearly in the middle of the 8mm 160x  (0.38 degrees) view. Maybe 0.1 degree accuracy? If I could get that across all the sky I would be very happy! With Alt I'm within the encoder manufacturers 0.5mm positioning error and I'm using their supplied magnet.

Due to the size of the Az bolt etc I don't think I've postioned the encoder with the same degree of accuracy and I'm using a different magnet. I'm also getting more noise on this axis than I should so something is not right.

It should be pretty easy to run some tests during the day so I'll start there.

Edited to add:

One option would be to add some linearization code - I'll work out a test rig.

Some other ToDo:

Mask the Microcontroller.

It's really too bright! I was going to use white ABS for the enclosure but maybe black would be more appropriate. I'm wondering whether I'm using a pin that has the on-board LED attached to it - if I turned that off at least then it would glow red and not blue! Thinking of putting all the electrics in the bottom of the encoder mount so they are masked by the dob mount when in use.

Setup/Teardown Streamlining and Stowing.

I need to make the boss of the spider deeper so you can undo the centre bolt more to get the alt encoder off the bolt without detaching the spider completely. It would make setup and teardown easier as the assembly would stay attached to the dob base. Along the same vein I need to round or shorten the base of the assembly to allow it to pivot round the spider bolt to a stowed position.

Looking at the weather I should be in a position to test the updates when another clear night appears!

Phone/tablet Mount:

Was continually picking up and puttng down the phone. Would it be a good idea to move the encoder mount to the viewing side of the dob and putting a handy shelf on top for the device of choice so you can just glance down?  Food for thought.

Edited August 25, 2021 by cwis

[cwis]

Quick update - I created a linearization jig to see how much the sensors drift using a cheap (very cheap!) NEMA 17 stepper motor I had floating about. It's 200 steps per revolution as you can see here.

What amazed me was how uneven the steps were! They seemed to alternate between 1.7 and 1.9 degrees - they should obviously be 1.8.  You can however see the signal on top of that noise - the sensor drift. It's cyclical, as expected and maxes out at around 0.5 degrees.

I think I have one of those mini geared steppers somewhere - I may replicate this experiment with one just to remove a load of dodgy stepper noise and add (hopefully!) a much smaller amount of gear backlash noise...

 

[faulksy]

just a thought. i bought a few months ago 2 off 40,000 steps per revolution encoders for £100

[cwis]

3 hours ago, faulksy said:

just a thought. i bought a few months ago 2 off 40,000 steps per revolution encoders for £100

That's really not bad! I'm looking at building mounts for mine with bearings to minimise the errors and then calibrate them them in situ to get the accuracy I need.  I'm up for saving a load of pain - do you have a link? Are they optical or magnetic? Absolute or relative?

I keep looking at the Nexus you recommended - very tempting and I'd have something that worked straight away.... But I'll keep battling on for a bit - I'm having "fun".

[faulksy]

9 hours ago, cwis said:

That's really not bad! I'm looking at building mounts for mine with bearings to minimise the errors and then calibrate them them in situ to get the accuracy I need.  I'm up for saving a load of pain - do you have a link? Are they optical or magnetic? Absolute or relative?

I keep looking at the Nexus you recommended - very tempting and I'd have something that worked straight away.... But I'll keep battling on for a bit - I'm having "fun".

it is all about the fun. they are so so accurate. at the beginning of the night you align then 6 hrs later the object is still perfect

https://www.astrodevices.com/shop/index.php?catalog/all/-/name/3