Please help improve my guiding (esp on the DEC axis)

[Henry Kwok]

Hello

I have a self-hypertuned EQ6 running EQMOD and PHD2.

I image with a 6"RC at native FL of 1350 with an Atik 314 imaging camera and a 200mm FL guidescope on a QHY5L-ii guide cam.

The best guiding I get is 1" RMS.

At the pixel scale I image at it is borderline - about 50% of the subs are good, 50% of the subs are discarded.

While I am happy with the subs when guiding is good (just under 1 arc sec RMS), I am hoping for more consistent results.

In particularly the DEC axis seems to be giving me trouble.

I don't know whether it is the polar alignment, the mount tuning or whether it is the EQMOD/PHD setting or a combination of these.

I have uploaded an image when it is good (but thats after discarding 50% of the subs), a guide log of the corresponding night, a DEC / RA drift graph with guiding off and the guide graphs of the RA and DEC axis.

Your help is much appreciated.

Henry

[Henry Kwok]

mod: sorry i did not read the group guidelines properly. if this is the wrong place to post this please delete my post. sorry.

[Andy Galasso]

Hi Henry,

This is a good place to post. We'll try to help as best we can :)

Looking at your dec axis guiding, we see the classic sawtooth pattern that we all struggle with at one point or another.

The downward trend is the result of your polar alignment error. You can flatten that out by dialing in you polar alignment.

The upward spikes are the result of the tiny amount of mechanical flexure in the dec drive train that allows stress to build up as the motor applies torque until the static friction is overcome and the mount suddenly reacts and starts to move.  There is usually not much you can do about that other than try guiding in one direction only (North or South). 

Some of the dec problem is probably also due to some amount of backlash. There may not be too much backlash in your case, but the upcoming versions of phd2 will have backlash measurement and compensation you might want to try when available.

I'm not sure what to say about your RA curve.  I'm pretty sure those high frequency oscillations are just seeing. 1-second guide exposures is a bit too short. Your PE looks pretty smooth so you should be able to go up to 2-3 second guide exposures I think, and that should smooth out the high frequency oscillations and allow you to better tune your RA guiding to track your PE (and not chase seeing).  The guiding assistant will also give you an estimate for max exposure duration so it might be interesting to try that and see what it says.

Great start on the M83 image BTW!

Andy

[Henry Kwok]

Hi Andy

Thank you for the reply.

re: the polar alignment. This actually brings up another question re: the PHD2 routine. Is there any reason why RA guiding is enabled during the drift alignment procedure? I found that there is always some interaction between RA and DEC and when RA guiding is on, DEC will be affected and the effect is more prominent when the PA is off by a larger degree. The polar alignment I achieved is already after couple of iterations with the routine already. Even though DEC appears to have flat line during both the azimuth and altitude adjustment, when actually guiding the polar alignment is obviously still off. I am wondering, would it be better if I just disable guiding completely, like how the DEC drift/PEC graph was obtained, and adjust the PA until the red DEC line flat lines? Has any one tried this?

re: backlash vs stiction - are they opposite end of the same problem, or can they co-exist? Looking at my guide log, do you think I should try to loosen the mesh or tighten it some more, or leave it as it is?

I know there must be some kind of backlash because during calibration the south pulses take a while before they move the star and the star never returns to the starting point. I have also notice that when I manually slew the mount to centre a star on the camera sensor, it takes a while for the star to move, though it seems to apply to both RA and DEC axis.

There are also conflicting suggestions re: the EQMOD guide rate setting. The EQMOD programmer suggests that the guide rate should be set to 0.5 x (i.e. 0.5 - 1.5 x sidereal for RA and +/- 0.5 x sidereal for DEC), but the original Stark documentations suggest setting the guide rate to 1 x (i.e. 0-2x sidereal for RA and +/- 1 x sidereal for DEC). Which way shall I go?

Many thanks

Henry

[Bruce Waddington]

Hi Henry.  I'll try to add a few things to what Andy has already said.  The guiding graph shows a "snap-back" behavior where PHD2 is issuing Dec guiding corrections, nothing happens for an extended period of time, and then the mount suddenly moves by a large amount.  We can't really tell from this exactly what is happening, and there may be multiple mechanical events involved.  One possibility is backlash, which can explain some of the "nothing happens" part of the curve.  But the snap-back part, where we sometimes end up with an over-correction, is harder to identify. It could be partly caused by stiction but it could also involve some other mechanical problem where energy is being absorbed by flexure in the assembly that is subsequently released into the gear train.  Unfortunately, these things are rarely simple and there's no reason to think you have only one problem.

The guide speed question is an interesting one, fraught with ancient history, folklore, and software minutia. <g>  The simplest answer is to run multiple back-to-back guiding tests of > 10 minutes using different guide speeds.  Short of that, the theoretical discussion goes something like this.  In a perfect world, it doesn't matter what guide speed you choose.  If you switch from a guide speed of 1x sidereal to 0.5x sidereal, the guide pulse durations simply double - that's the only difference.  PHD2 certainly doesn't care, it's just math.  But of course we don't live in a perfect world.  If you use a 1x guide speed, you are most likely generating a larger burst of energy into the gear assembly which can help to overcome stiction and will also produce a quicker response to guide star deflections.  Some mount manufacturers, like Astro-Physics, specifically recommend using a 1X guide rate.  But if you are guiding at long focal lengths, this may result in guide corrections that are quite short, on the order of 10's of milliseconds.  And is the guiding software in the driver, mount or camera capable of handling such short guide pulses with any degree of accuracy?  For some mounts, the answer is definitely 'no' - they may literally discard guide pulses less than some threshold amount.  In those cases, you are better off using a lower guide speed setting with the associated larger guide pulse amounts.  Is that confusing enough? <g>  In the time it would take to research the details of your mount's software capabilities - even assuming you can find the information - you could run the back-to-back tests and reach your own conclusion.

Good luck.
Bruce

[Tamas Kajfis]

Hi!

Could You tell us what are the settings in the PHD, I mean min move/ aggression / star mass change and etc.

[Henry Kwok]

I cannot remember exactly, I thought it would be saved in the log file?

But from memory I have the min move at 0.18 or so, aggressiveness at 95, hysteresis at 10, dec guiding on auto, dec min move also at 0.18 or so. Dec mode is resist switch.

[Tamas Kajfis]

Try this
http://stargazerslounge.com/topic/245886-phd2-minimum-movement-calculator/#entry2673711
enter pixel sizes and focal lenghts
TAE pixels

[Andy Galasso]

I'm not so sure I agree with that way of setting min motion.  The write-up at that link says to calcluate min motion based on the ratio of guider scale to imager scale.  I believe a more important consideration for min motion is your local seeing conditions. IMO you should set you min motion to the smallest value that avoids chasing seeing fluctuations.  The Guiding Assistant (Tools => Guiding Assistant) will measure seeing fluctuations and suggest min motion settings based on that.

Andy

--
You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Richard44W]

Hi Henry,

I'm no expert and hopefully someone else will chime in, but trying to guide 1350mm fl with 200mm fl seems a bit of a mismatch? Maybe it does not matter with PHD2?

Richard

[bw_msgboard]

On the face of it, this is not likely to be a problem.  What matters is the comparative image scales between the guider and the main camera.  Henry’s guiding image scale is 3.9 a-s/px.  I don’t know what his main system image scale is, but with a 1350mm focal length, it is unlikely to be much below 1 a-s/px.  This sort of ratio is easily handled with typical guiding software, including PHD2.  This all works because the position of the guide star is computed with a center-of-mass algorithm, so we’re dealing with fractional pixel positions.  In my experience, the need for off-axis-guiding usually arises because of differential flexure problems long before any centroid calculation errors come into play.

 

Cheers,

Bruce

 

---

--

[Terry]

Hi Andy,

Can you please explain what you mean by "You can flatten that out by dialing in you polar alignment".

Thanks, Terry.

[Andy Galasso]

On Jun 27, 2015 11:48 PM, "Terry" <terry.c...@gmail.com> wrote:
>
> Can you please explain what you mean by "You can flatten that out by dialing in you polar alignment".

Sure, what I meant by that was the downward slanting trend in the sawtooth pattern is a symptom of polar alignment error.

​

 With perfect polar alignment there is no drift in declination and the declination trend line is flat (ignoring other effects like refraction, mirror shift etc.) Improving polar alignment would result in a flatter, more horizontal declination line on the graph.  With less declination drift fewer dec guiding corrections are needed.

Andy

[Henry Kwok]

Thanks for all the suggestions. I will try once the weather clears..., say next year?!?!

re: the guidescope - the guide scope pixel scale is 3.87 ax / px and the imager pixel scale is 0.97 ax / px. This is at a ratio of 1:4 - I understand it is ok up to 1:10. Obviously the guiding needs to be good which is obviously my problem. When the rms error is around 1, the stars are perfectly round, but this only happens in 50% of my subs.

I used to use a OAG when I imaged with my short FL refractor but really I found that the process of finding a guide star with the longer FL and slower f ratio of the RC too frustrating. Bright enough stars are just so hard to find and their SNRs are often so low that the guiding is poor. With permanent set up one can use CdC or TheSkyX to plan the camera orientation to acquire a guide star ahead of time, but with a temporary set up like mine it is not really feasible. Finding a guide star becomes a frustrating trial and error process.

I think I am going to wait for the next stretch of clear nights, spend a good night on perfecting the polar alignment and try again, before I open up the mount again to fiddle with the backlashes / gear mesh etc. Often times I envy you guys in the Northern Hemisphere for having it easy with polar alignment.

Cheers 

Henry 

[Scott Astroman]

Hi Henry,

On my azeq6 running eqmod, the guiding in Dec is improved by reducing aggressiveness and max pulse duration significantly. I typically use resist switch with aggressiveness of 30%, max duration of 500ms. This tends to reduce overshoot of corrections in both directions.

I have also reduced min move to 0.15 pix (0.4"). My thinking was that I wanted to try to correct before too much error occurred. In your case 0.18 pix corresponds to 0.7" error before correction is permitted.

My guide scope is 400mm with 2.68 arcsec/pixel. Typical Dec rms is 0.6" and overall 0.9".

Finally my results tend to be erratic if the guide scope is slightly out of focus, so I always check at the beginning.

Hope this gives you a few more things to try.

Regards
Scott