Mesu 200 Encoder Upgrade

The Mesu 200 mount is a great mount and for the price I don’t think you can get better value for money anywhere else. It’s only slightly more expensive than an EQ8 but has a far superior load capacity and is built more sturdily. It is much cheaper than something like a 10 micron which has a way lower load capacity and for a matching load capacity you have to go to something like a Paramount which is again way more expensive. The great thing about the Mesu is that it uses the Sitech Servo II control system which can be adapted for use with any telescope and is packed full of features which allow you to adapt the system to your needs and specific setup. One of the things that I have always found with the Mesu being a friction drive system is that even though it tracks beautifully and has no backlash and small periodic error I have always thought there was room for improvement and the goto accuracy even though the manufacturer tells you it is not important as you can always just plate solve for better pointing needed improvement and these factors are down to what can most likely be attributed to slip.
Now I am not saying there is any kind of huge slip in the friction drive because there is not but by default with this design it is prone to slip especially as you get larger payloads on the mount due to the momentum the slewing equipment has and the fact that balancing accurately is much more difficult on a friction drive due to the lack of a clutch to release the drive. This slip can also be a good thing as should your mount slew to a position the OTA or camera make contact with the mount or pier then the drive will actually slip and the mismatch between encoder positions will cause the mount to go into error and stop driving which can prevent damage to your valuable equipment. The trouble for me is that this is a 2 encoder system and the higher resolution part of the system is the part that is subject to mismatch of position due to this slip (the motor encoder) and the encoder that is on the axis is not of a very high resolution so the motor encoder thinks it is in the right spot but is out due to minor slip and the axis (scope) encoder is not accurate enough to tell it that it is slightly out of position. The numbers for my reasoning are as follows:

  • The axis encoder has a resolution of 10000 ticks per revolution or 0.036 degrees per tick or 2.16 arcminutes per tick 0r 129.6 arcseconds per tick
  • The motor encoder has a resolution of 4000 ticks per revolution but goes through a 2000:1 reduction ratio to give an effective resolution of 8,000,000 ticks per revolution or 0.000045 degrees per tick or 0.0027 arcminutes per tick or 0.162 arcseconds per tick.
  • On a smaller rig I might have an 8″ RC with an 8300 sensor which gives me a FOV of 28.6×38 arcminute and a image scale of 0.68 arcsecond per pixel
  • On a larger setup i might have a 12″ RC with an 11000 sensor which gives me a FOV of 33.8×50.8 arcminutes and an image scale of 0.76 arcsecond per pixel.

Now you may think I am just being a bit picky with this as you will say “surely you get the target on the sensor every time when you do a goto and you can plate solve from there” and I could do that and wait while it goes through iteration after iteration of plate solve until it is close enough to the target coordinates. But I live in Scotland where the weather is terrible and time for data gathering is very limited and I do not want to spend lots of time targeting every time I go to a new target or if as is normally the case when I go back to a target on another evening I want it to be pointing bang on within a couple of pixels to where it was the previous night every time.

General pointing aside the next big thing to be impacted by slip is the calibration map. Having a Mesu mount means I use the Sitech software and part of that software is Point XP which is a bundled astrometry program that you can use to set up your calibration map and model your mount. The recomendation for the calibration is anywhere from 16 to 50 individual calibration stars and I like to use a model with no less than 32 points. Now each calibration point requires a slew to the object where it will them take a short image and plate solve it to get the exact coordinates and Point XP allows me to fully automate this. What I think happens however is that each time it slews there is a tiny amount of slip  and due to the large difference in resolution between the two encoders there is a poorly corrected feedback as to the actual position which makes the model more inaccurate. What a good mount model will do is help to compensate for minor imperfections in polar alignment and leveling of a mount and give you far better goto accuracy and better tracking as well before having to correct through autoguiding but if that model does not have accurate and repeatable positions due to any slip that might occur at any time then your pointing and tracking will be adversely affected.

For all these reasons I decided I wanted to upgrade the encoders fitted to my mounts scope axis there are two main options here which are to either go with a medium resolution incremental encoder or a high resolution absolute encoder. At first I had been quite keen to go the route of an absolute encoder as the precision these can provide is incredible and once the mount has had a single star calibration stored it will always remember where it is pointing even if you switch it off and move the mount (it won’t help you if you set it up wrong though as there is no fix for human error) but these are hugely expensive and you could probably expect to spend in the region of £2000 per axis on that option depending on how high of a resolution you wanted and they are notoriously tricky to install properly and in addition to this also require an additional hardware box to allow the sensors to talk to the mount controller. The other option and one that is well supported within the mount control software is for a medium resolution Gurley incremental encoder and there is even a study done on their use and documentation available online which tells you the exact models that are compatible (there is an interpolation error which must be corrected for and this is an option already designed into the software for 2 of the encoder models) so I decided to go with the 320,000 resolution option as there are no benefits really from going to the 500,000 resolution according to the study and unless there have been some changes to the software since that was written there were actually better and more stable results from the 320,000 encoder.

At first I did go back to the mount manufacturer and enquire about having them perform the upgrade for me as I knew of others who had had the work done already but the response I got was pretty disappointing as they seemed to be very reluctant about it, give me very limited information and generally tried to talk me out of the idea yet they did offer it as an option on a new mount and suggested if I was worried about resolution maybe I should upgrade to their latest mount which has absolute encoders but is a monster of a mount and starts at a price tag of €16,000. I kindly declined the offer of a new mount and they pretty much showed no interest in doing the modification and although they had encoders in stock told me if I was going to do the modification myself they would not sell me any of the encoders. Thankfully I managed to find a UK distributor for the encoders who was actually quite familiar with the application in astronomy and was happy to order a couple for me but they did have a lead time of 5-6 weeks (they are made to order in the US) and still a little pricey at £525 a piece but much less than an absolute encoder so I bit the bullet and ordered them with the intention of doing the installation myself.

The original encoder is a fairly simple affair with a small and lightweight encoder fixed to the rotating axis by means of the stub shaft of the encoder going into a hole in the main axis shaft and being secured by a small grub screw. The encoder itself is then secured by means of a simple metallic tab with a small sprung fork that goes over a pin to stop it rotating and ensure there is no backlash. The one shown here is for the declination axis which is more difficult to get to but easier to take a picture of while it is still attached once you have the axis dismounted which is doem simply by unscrewing 4 allen head bolts.

The original brackets were not going to work as they were designed to be fixed by a nut to threaded collar over the shaft of the encoder which the new encoders did not have. What I did was get some 1.5mm stainless steel sheet and mark out a template that would allow me to use the three M4 mounting holes on the encoder to affix the plate while having a central hole for the shaft to come through and also to have two finger sticking out which could be formed into a sprung fork mechanism that would grip around the existing pins and hold the encoder in place. After the initial drilling of the holes I formed the S shape into the fork to give it clearance from the encoder casing and then pit it in the lathe to cut the circular shape to fit the casing.

After mounting the newly manufactured plate to the encoder I then had to twist one of the forks to the right position to enable it to act like a sprung fork and grip the pin when I mounted it. The one thing I had not counted on was the lack of space between the encoder housing and the mount on the RA axis and I had to knock in the fork slightly back from its right angles position to allow for this and also had to do it on the dec axis as the smooth machined length of the pin was shorter than it was on the RA axis (as I had made the parts while away at work a couple of thousand miles away where I had the machinery for doing the fabrication I was stuck with what I had).

This is how the new encoder looked finally mounted to the axis.

For anyone wishing to do this modification, here you can find links to various resources I have mentioned above:

Data sheet for the compatible encoders

(suitable compatible part numbers can be found in the below)

Dan Grays explanation of Tick Management and Interpolation error correction

Sitech telescope control system website