Jump to content


Photo

Periodic RA spikes in PHD with Atlas

  • Please log in to reply
8 replies to this topic

#1 Jon Rista

Jon Rista

    Mercury-Atlas

  • -----
  • Posts: 2836
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 29 July 2014 - 05:07 PM

I've been working on tuning the performance of my Atlas lately. I hypertuned it a couple weeks back, but only recently had a couple clear nights to put it through it's paces. I am still tweaking gear meshing and bearing preload on both axes (things still feel a bit too tight, trying to find the right tension on everything).

During the tests of my mount's tracking after I make adjustments, I noticed that I'm getting periodic spikes in RA. Here is a screenshot:

Posted Image

Does anyone know what causes this? If it wasn't for these spikes, my RA tracking would be about 2" peak to peak, maybe even a bit less, which I would be pretty happy about. Would really like to figure out how to tune this last issue out of my mount and get some good, reliable results.

Thanks!

#2 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5637
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 29 July 2014 - 08:35 PM

What's the period of the spikes?

Might be the spur transfer gears.

That said 0.53" rms is a very good figure. You probably can't do any better, so stop obsessing already :-)

#3 Jon Rista

Jon Rista

    Mercury-Atlas

  • -----
  • Posts: 2836
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 29 July 2014 - 10:36 PM

Eh, I'm an obsesser. :p I see spikes that reach well above the average peak height between them...and that makes me want to eliminate them. :D

I am also honestly not really sure what's "good" or not...your saying 0.53" RMS is good? I've been looking at the peak error. IIRC, seeing this night was around 2.8-3.1, so having a peak error of 3.9 would mean the mounts tracking error was making my stars worse.

Anyway, period wise...I think, just doing a rough visual measurement in PHDLab, the closest spacing seems to be about three minutes apart, the longest spacing 12 minutes apart, the average spacing about 5-6 minutes apart. So the base period seem to be around three minutes. I need to get PECPrep working. It usually gives me more info, but all I'm getting is this:

Posted Image

#4 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5637
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 29 July 2014 - 11:38 PM

It's probably noise. Spur gear errors would be very regular.

0.53" RMS guiding is a fairly good figure although Frank may disagree. To be honest I usually get around 0.75" to 1" (!!!) and that's on a Mach1 with < 1" of PE. But then you have to keep in mind that guiding errors are not only due to the mount, but also the seeing (and how good your polar alignment is).

I found somewhere on the net that an AP mount will not respond to guide pulses shorter than 30ms. At 15.04"/second, that's 0.45". This means 0.45" is the smallest distance an AP mount can move. That's 0.90" p-p. Which of course would put a lower limit on how tight your guiding can be (and how low your RMS error can be).

I'd expect an Atlas to be significantly rougher, the gearbox alone will guarantee that. Would it be 3.86" peak as your PHDLab result shows? maybe... most motor gearboxes have around 4" of periodic error themselves (AP has less, because they cut their own reduction gears) - this is quite aside from the PE of the main worm drive itself.

At the end of the day what is your stellar FWHM? note that with your gear, you can't measure FWHM properly (you need to be significantly oversampled, which you're not).

#5 Jon Rista

Jon Rista

    Mercury-Atlas

  • -----
  • Posts: 2836
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 30 July 2014 - 12:06 AM

Ah, interesting about the minimum guide pulse. I did notice that EQMOD, when using ASCOM PulseGuide mode, defaults to a minimum 20ms guide pulse. I've never messed with that, but I figure that's probably something similar. I can't imagine my Atlas doing better than your Mach 1, so I guess 0.53" is pretty darn good then! :D I guess the only other thing that is really going to improve star quality would be a better scope, and maybe a sensor with smaller pixels. (I've noticed that, since moving to the QHY5L-II for my guide cam, that the star field image that PHD shows has taken on a whole new life...it renders a very three dimensional star field, especially when you have larger stars, such as the primary stars in constellations, in the field of view...I can only imagine what a larger APS-C sized sensor with similar sensitivity and pixels that small would be like for real imaging...would be a thing from heaven!)

If the gearbox is a prime source of roughness, then I'm going to belt mod. I've got a bunch of cloudy days ahead, so the only real astro things I'm doing right now are tuning my mount and building an arduino-powered, thermally regulated cold box for my DSLRs, (and, of course, processing and reprocessing old data). ;) Might as well add belt modding to the mix.

Regarding my star FWHM, I've always gone by what BYEOS tells me in it's focusing mode. It ranges from around 4.5" on a bad seeing day to maybe 2.3" on a very good day. I probably have somewhere between 3.0-3.5" most days. This would be with stars that are roughly "medium sized" as far as what's visible in BYEOS' focus mode...so they are probably pretty large, bright stars. But, that would be measured with my current equipment...as you say, I'm definitely not oversampling.

#6 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5637
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 30 July 2014 - 12:29 AM

I use the FWHM estimate in DSS. But someone (Frank I think) said that if your FWHM is only a 2X to 3X multiple of your pixel scale, then the FWHM will be inaccurate.

Back when I used to rent on iTelescopes, I got to use an FSQ106, ST8300C (one-shot color) on a Paramount ME. And the FWHM was... around 4".

Now that's some pretty high end, auto-focused gear, and a measly 4" FWHM? I can get that on my balcony with my old AP600 mount.

It's because the pixel scale is only about 1.8" / pixel. To get a good measure of FWHM, the pixel scale should probably be in the 0.5" range at least.

By all means do the belt mod since it's a relatively cheap mod. The Avalon mounts are all belt drives, so it must be a benefit (the AZ-EQ6 also uses a belt drive instead of spur transfer gears).

You do have to consider that the Atlas uses 200ppr steppers, and even with the most optimistic of micro-stepping, a stepper can't really hold position in-between poles (it will tend to detent to the nearest physical pole). A 200ppr stepper with a 5:1 reduction on the 180-tooth Atlas main worm gear is 7.2" per pole step (!!!) of course the Synscan controller micro-steps the stepper to get finer resolution but arbitrarily small micro-step increments don't work in the real world. Best case you can hold position to 1/4 step (even if you do 1/16 or 1/32 micro-step) so that's almost 2" of stepper motor "grittiness" (7.2" / 4) that a belt won't get rid of...

#7 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5637
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 30 July 2014 - 12:34 AM

.. you're probably thinking, what with all the objections I've put up, how does Paramount and AP (and the rest) do it.

Well.. many use belt drives (the Paramounts do). And they use servos with very high gear reduction, so there's less motor "grittiness" because of the high reduction.

Then your second problem is, the reduction gear can have substantial error (the infamous CGEM 8/3 comes from the 50:1 reduction gearhead). The solution is to use expensive Maxon or Dunkermotoren motors with very finely-made gearboxes. These motors are $300+ each, which is why they don't show up in the CGEM, Atlas, etc...

#8 Jon Rista

Jon Rista

    Mercury-Atlas

  • -----
  • Posts: 2836
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 30 July 2014 - 10:06 AM

Thanks for the details, Orly. Interesting stuff. I'm planning on moving up to a larger/longer scope soon, and I wanted to tune my tracking performance to I could get good results with it. It sounds, though, as though I'm already doing quite well. I'll still do the belt mod, because I think it will be fun to do, to eek the last bit of performance that I can out of the mount, and particularly to get rid of that terrible grinding noise that occurs every time I slew.

I will say that Frank seemed to be dead-on about dialing in a very close polar alignment. According to PHDLab with some of my logs, when doing a drift check, my polar alignment is around 0.3-0.5', or 18-30 arcseconds off the pole. That does seem to have helped immensely. I did a few 15 minute long test exposures, and as far as I could tell, the only thing that was causing any star elongation was drift due to differential flexure. I'm redoing the way I attach my scope to the mount, doing away with the side-by-side, using some saddle rings from ScopeStuff, and an extra vixen dovetail across the top for the guidescope. I'm hoping that tightening things up right around the center of the mount, rather than having everything hanging off the sides, will help as well.

#9 Jon Rista

Jon Rista

    Mercury-Atlas

  • -----
  • Posts: 2836
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 03 August 2014 - 11:11 PM

I think I nailed it! :D Kept tweaking, and kept working my polar alignment with PHD Drift until it was reporting 0.06' (3.6") off the NCP. I programmed PEC, and got an EXCELLENT curve. I also ordered a new dovetail from ADM, one with nice long slots on either end, picked up a 6" and a 4" set of cradle-type scope rings from ScopeStuff, and very tightly mounted my lens to the dovetail. Switched back to the Orion 50mm Mini guidescope, QHY5L-II guide cam, mounted that to the top of the forward 6" scope ring, and tightened everything down very securely. That seems to have eliminated my flexure issues. Did a nice long imaging session of Bubble/Lobster nebula region, it ran all night without a hitch, I managed to get perfectly round stars at 480 second exposures, using my 600mm lens and a 1.4x teleconverter (840mm effective). I hadn't wanted to try longer focal length exposures until now, as I had so many problems before. This is a screenshot of a 1200 second period (400 samples) in PHD 2.3 from last night:

 

Ut2KcTI.jpg

 

RA RMS is down to 0.22! Dec RMS is 0.20, and total RMS is 0.29. My peak to peak performance is about 1", based on the +0.5 to +1.5 point sampling in the Target area of PHD. Here is my PEC curve from EQMO:

 

kZQNYNc.jpg

 

Looks like an Atlas really can perform extremely well, when it's used right. ;) Thanks for recommending (and, basically, insisting) I get the Atlas instead of the CGEM, Orly! Best advice I think I've taken so far on these forums! :D








Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics