EQ6 Belt Mod (Belting Online)

[MarkS]

I carried out the mod today.  Just waiting for clear skies so I can record a PE chart and check the guiding performance.

Here's the kit of parts:


The kit preserves the 12:47 gear ratio so the handset will continue to work as normal.

The pulley grub screws are 1.5mm hex which is different from the existing EQ6 grubs (2mm).

The only tricky part was that the toothed pulley on the Dec worm axis just touched the bottom of the housing as it rotated, so I had to grind down the height of the flange by 0.5mm by mounting on a drill and rubbing it on P60 wet'n'dry paper as it spun:


When putting the motor in place, a bent piece of wire holds the belt in the right place so it goes over the new motor pinion:


Forget all those scary online discussions where folk have to drill out the pulleys to a bigger diameter, make modifcations to the worm housings to give clearance and cut longer lengths on the motor mounting plate slots - none of that is required for the kit I used:
http://www.beltingonline.com/heq6-belt-mod-drive-kit-12220

Since I have two EQ6 mounts with almost identical guiding performance, it will be easy to test the effect of this kit in terms of guiding. once we get some clear sky.

To be continued ...

[Rocket Pooch]

This will be interesting to see, I would expect some of the gearbox harmonics to drop off.

[MarkS]

I forgot to mention how quiet the mount is now, when it slews 

[Rocket Pooch]

Ha!  we compaired EQ6 noises in France and mine was quite quiet, wonder how yours compairs.

[JohnP]

Mark really look forward to seeing results. May go for it myself if positive.

How long did it take to install...

John.

[MarkS]

How long did it take to install...

Around 5 hours in total, I reckon. That doesn't include the 45 minutes I spent hunting round the house and garage for a 1.5mm hex key.  I eventually found one on a set of bicycle hexes. And it doesn't include the 30 minutes spent working out the best way to reduce the height of the flange

The fiddly bit is getting the belt around the motor pulley - if hooking it with a piece of wire failed I would have drilled a 10mm hole in the side of casing next to the pulley so I could manipulate the belt directly through hole.  Luckily I didn't need to resort to that!

[Mac]

Hmmm,

Looks interesting, as mine is in bits on the work bench at the moment (long story, Rusty bearings!!)
I might just wait and see before i rebuild it.

Mac.

[MarkS]

It might be clear tonight.  I can feel a test run coming up!
If it fails to impress, it might be time to order the EQ6 transfer gear set from Aeroquest. 

Maybe, in any case, I should mod one EQ6 with a belt kit and the other with the Aeroquest kit.

[The Thing]

Or cut your losses and start saving for a Mesu...

[MarkS]

Or cut your losses and start saving for a Mesu...

or two ...

[RobertM]

I've read that the aeroquest parts need lapping and lack the precision to make them worthwhile.

Make sure you do your homework first !

Robert

[Mike]

Check out Gierlinger.

[Rocket Pooch]

I think before you replace anything else get it looked at again, I'm not sure after replacing the bets the gears will be an issue because they won't be there, it should only be the RA Bearings and meshing now.

[MarkS]

I now have the results of the BeltingOnline EQ6 belt mod and it's quite a mixed set of results.

Firstly here are the graphs of PE before and after the mod:




The above graphs cover a few worm cycles of 478.69 sec

There is a slight overall improvement in peak to peak PE but it is not dramatic.  The main difference is in the amplitude of the superimposed 122.2 sec cycle (the time of one rotation of the motor).

This is a lot clearer in the graphs below which show the PE due to components running at motor rotation frequency:




Note that both graphs show the 12 peaks which is caused either by the 12 teeth on the motor pinion gear (or is 12 tooth belt replacement) or of something else running at this frequency (maybe gears within the motor body, teeth on the drivebelt?)

Following the steep peaks and troughs of these 12 peaks is difficult for the guiding - far more difficult than following the PE of the worm gear.  They have the infamous 10.2 second period. Since the guiding finds the second graph much easier to follow, I have seen a factor of 2 improvement in guiding accuracy.  The unmodified EQ6 was guiding at around 1.1 arcsec RMS error and the modified EQ6 sitting next to it was simultaneously guding on the same star at 0.5 arcsec RMS error.  This was on a night of pretty bad seeing.

This is the best guiding I've ever seen on my EQ6!  Note that it was carrying a payload consisting of a Celestron C11 plus guidescope/guidecam plus DSLR camera which is around 17-18kg in total.  Plus 4 x 5kg counterweights.


Regards,

Mark

[Rocket Pooch]

Mark,

The belts will not improve the PE much but the frequency issues, your half way there.  The next step is get the correct end bearings, measuring the centres of the worm and case and shim if needed then mesh the gears correctly under load not on the bench.

Just for reference at 1.4 meters OAG I was getting .3 pixel for good seeing and .79 for bad seeing with jumps to 1.4 pixel.  Whats the Pixel guiding accuracy of you setup post mod with focal length of the guide scope?

Also what are you trying to achive, I was imaging at 1x1 with my QSI at 1.4 meters with an OAG at .79 and getting bang on round stars, well within a nats balls anyways.  The reason for asking is our skies are rarely better than 2 arc seconds seeing anyway, so if your trying to get less than that you will be lucky to get it.

I can help you with setting up the mount properly if you like, but you will have to come to the dark side, i.e. West Sussex.

Chris

[MarkS]

Hi Chris,

Last night with a 600mm fl guidescope I was getting 0.24 pixels RMS error on the QHY5 guidecam (post mod).

Multiplying this up to a 1.4m fl equivalent that would have given me 0.56 pixels RMS.   I was purposely guiding on a star close to 0 Declination and the seeing was not at all good.  If instead, I was guiding on a star higher in the sky, I would (of course) have got a smaller figure in terms of RMS pixels.

My goal was to reduce the amount of star bloat when imaging at longer focal lengths - so if I could reduce the RMS guiding error to well below the RMS seeing then I'd be happy.  I reckon I've just about achieved that now but I need to try an imaging run to check.

I understand the shimming issue but what do you mean by the "proper" end bearings?  I assume you mean the worm bearings?  What is "proper"?

Mark

[Rocket Pooch]

Hi,

Yes worm bearings, something like the link below the temperature resistant ones not the £3 ones which look the same but go all wobbly when the temperature drops which causes an odd noise when slewing.

http://simplybearings.co.uk/shop/p7313/SKF+6082RSHC3GJN+Sealed+High+Temp+Deep+Groove+Ball+Bearing+8x22x7mm/product_info.html

Chris

[MarkS]

By more careful analysis I've identifed another frequency component running at more or less exactly 8x the tooth frequency i.e. a period of 1.27 seconds (=10.18/8).  The only explanation I can think of is that there are step down gears within the body of the motor itself.

This isn't a sampling artifact because I'm sampling PE at 10 frames/sec - so a period of 1.27 seconds is easily identifiable.  It has a peak to peak of 0.5 arcsec.

Does anyone know if this is the case - i.e. is it likely there at least one additional gear within the motor body?

In any case, it has a very interesting implication for guiding - for optimal guiding the guide camera should be run with an integration period of 1.27 seconds - to integrate exactly over this period to prevent "false" guiding corrections being issued.

If it's not caused by an internal gear then maybe it is an artifact caused by the microstepping - each motor step is subdivided into 64 microsteps.

[Mike]

Does anyone know if this is the case - i.e. is it likely there at least one additional gear within the motor body?

Highly unlikely Mark. They look like standard cheap 1.8 degree stepper motors to me. It's more likely a micro-stepping issue as the motor isn't turning smoothly but jumping along each tiny step.

[The Thing]

Does anyone know if this is the case - i.e. is it likely there at least one additional gear within the motor body?

Highly unlikely Mark. They look like standard cheap 1.8 degree stepper motors to me. It's more likely a micro-stepping issue as the motor isn't turning smoothly but jumping along each tiny step.

There was a guy on the eqmod forum asking about replacing the stepper with a more accurate one (2x steps per rev) and how that would affect EQMOD - could it be accommodated. http://groups.yahoo.com/neo/groups/EQMOD/conversations/topics/37492

[Mike]

It's certainly possible to replace them with better resolution motors, you'd just need to modify the EQMOD settings accordingly and ensure they work with the existing circuitry. Though if the circuit is doing standard 1/8th micro-stepping it should work with any motor with the same amount of coils.

[Rocket Pooch]

This isn't a sampling artifact because I'm sampling PE at 10 frames/sec - so a period of 1.27 seconds is easily identifiable.  It has a peak to peak of 0.5 arcsec.

You don't want to be doing that sampling rate, guiding subs should be 1-2 seconds to average out seeing, or you will try to over correct for seeing.

[MarkS]

The 10 frames/sec is done by PERecorder for PE analysis - I wouldn't use it for guiding

It's because PERecorder only uses a webcam.

[Rocket Pooch]

Same thing the stars will be subject to seeing.

[MarkS]

Same thing the stars will be subject to seeing.

That's right - the seeing caused a lot of jitter in my PE charts but averaging adjacent samples removes most of this.

The reason for using PERecorder is that it indexes PE against EQASCOM motor position count.  This will allow me to unleash Part II of my fiendish plan - to produce a PEC file accurate enough to remove most of the 10.2 second period.

[Mike]

This will allow me to unleash Part II of my fiendish plan.....

Mwuuhhhaaaahhaaaaaa!!!

[JohnP]

shame that mad professor doesn't look like Mark....

[RobertM]

It doesn't now 

[RobertM]

Mark,

Have you got any further with this ?

Robert

[MarkS]

Yes and no.  Unfortunately I've been a bit busy recently for one reason or another.

I did produce a PEC file for the correction and the first night I tried it my guiding was down to .08 pixels at 300mm i.e. around 0.3 arcsecs.  The next time I tried it the PEC file made no difference and there was a lot of jitter - guiding was .16 pixels at 300mm.  I'm not yet sure of the cause of the variability.  In any case, I'm planning on replacing the worm bearings next - probably the weekend after uDSC.

I think the belt has bedded down a bit now - the PE graph below shows much less amplitude of the 122sec period of the motor pinion than straight after the conversion:


Here's part of the PEC file I used:


It contains the 479 sec worm, the 122 sec pinion and the 10.2 sec tooth frequency.  The whole PEC file contains 12 worm cycles because the pattern repeats every 12 cycles now the 36 tooth transfer gear no longer exists.  I had to do some jiggery pokery to allow EQMOD to accept this non-standard file.

The worm cycle is not sinusoidal (on my mount) but has quite a definite hump shape.  On that basic hump shape the 122 second and 10.2 second periods are superimposed.

So much to do and too little time to do it!

Mark

[RobertM]

Lots of good work there Mark.  It's a bit difficult to see just how peaky or smooth the curve is for guiding purposes - any chance of showing an expanded single worm cycle ?

I agree with you about the belt teeth, if they have right angled edges rather than beveled or rounded then they are likely to catch on the cog teeth.  If so then that should ease over time or you could just run the motors for a few hours to bed them in.

Robert

[MarkS]

Hi Robert,

Here's an expanded single cycle of the worm:


The data comes from PERecorder and webcam at 10 frames/sec (each frame 1/25sec exposure).  For this plot I have shown a moving average over 1 second.

Here's the raw data without any kind of averaging:




It is impossible to know how much of the spikyness is caused by the seeing (which wasn't good that night) or by the mechanics of the mount.  That is always the problem when analysing PE data.

In any case, here's the same data with my complicated PEC removed from it:



Now it might be a bit simpler to separate the seeing related spikes from the mechanical related spikes.

Finally, here is the above graph again but with a 1 second moving average.  This should remove most of the seeing noise leaving just the mechanical.



Mark

[RobertM]

Considering that's removed most of the short term turbulence it isn't bad at all.  I think you really are into bearing swarf territory now.  Not only the worm but also shaft bearings could cause a lot of noise.  I also wonder how far you can go with the discrete movement of stepper motors, they will also cause tiny vibrations through the mount depending on load.

I assume that you'll only be guiding in RA because of precise polar alignment so aren't concerned with the dec axis ?

Robert

[MarkS]

I assume that you'll only be guiding in RA because of precise polar alignment so aren't concerned with the dec axis ?

That's a very interesting point.  Once I got my alignment bang on, I found there was very little difference whether or not the Dec axis was guided.  I still think Dec guiding is worthwhile, if only to counteract any residual slight drift.  Just need to work out the optimum Dec settings to prevent it trying to guide out random noise.

[The Thing]

Everything I read about guiding says you want PA to be off slightly in so that guiding corrects the DEC drift in a consistent direction avoiding any back and forth bounce movement between worm groves and gear teeth. There is probably an optimum PA error and direction of misalignment for each object position though as drift will be different. And I suppose it depends on how precisely unbalanced each axis of the mount is as well. I had to add a finder to the side of my OTA the other night to unbalance the DEC axis a bit more to stop it bouncing as shown on the guide graph.

[MarkS]

Thinking about it, of course you need to guide in Dec.
Otherwise how will dithering work?

[The Thing]

You can do RA only dithering in PHD2. It's on the Global tab in the Brain along with a dither scaling factor option.

[The Thing]

In your humble opinions, is it worth doing a belt mod to my HEQ5? It's already been rebuilt with new SKF bearings everywhere.

[RobertM]

If your guiding is good then is there much point ?  Don't forget that good guiding is also very dependent on the conditions so it may only when there's little turbulence that you might notice.

The new SW mount is belt driven but I don't know how much of a sales ploy that is.

Robert

[MarkS]

In your humble opinions, is it worth doing a belt mod to my HEQ5? It's already been rebuilt with new SKF bearings everywhere.

You are imaging at a long focal length so your guiding needs to be as accurate as possible.  Every little helps.  For me, the belt mod made a significant difference because it removes the transfer gear with its associated periodic error component.  It means I can now usually guide at 0.5arcsec RMS, though it depends on seeing conditions.

It's worth preparing a Periodic Error chart like Ivor has been doing and that will uncover the mechanical source of periodic errors and give some indication of likely improvement.  If the transfer gear is not contributing to the error then there is no point.  On the EQ5 the belt replacment is a really simple job so there's not much to lose.

For my 2nd EQ6 mount I already have on order another belt mod kit with a 12:48 ratio instead of the usual EQ6 12:47.  It's a special order and quite expensive because the 48 tooth pulley must be specially made.  It means I can no longer use the handset (it will be EQMOD only) but it does mean I'll be able to properly use PEC (since all gears will be turning at integer multiples of the worm speed).  I'm expecting to get guiding accuracy well below 0.5 arcsec RMS using PEC which means the seeing will be the main limiting factor for my long focal length imaging.

Mark

[The Thing]

Mark, how have your belt mods worked out? Was the 48 tooth pulley a worthwhile move? I know that my HEQ5 has been transformed by a belt mod.

[RobertM]

Thanks for the reminder, I must do this when I've finished my other projects.

[RobertM]

Just done mine, it sounds like it should have done in the first place now

Proof of the pudding though comes at Kelling.

[MarkS]

Just done mine, it sounds like it should have done in the first place now

Proof of the pudding though comes at Kelling.

A lot quieter and more refined?

[RobertM]

If say so !  Only the steppers are audible now and they are quite quiet.