PHD2 graph - Comments please

[malc-c]

[removed word]... I thought it was too good to be true.....But it worked

I must have missed that when I created the new profile - but then I ran the wizard, and  would thought you couldn't move on to the next page without entering the info......  

It is perplexing though.... enter the focal length, and I bet the guiding will be all over the place.... leave the focal length out, and run the calibration  and you get amazing results that keeps the mount almost perfectly pointed.... If this is the case then what's the point of specifying the focal lengths.  

 

[michael8554]

PHD2 was probably guiding well after your recent tweeks, but the data on the quality of guiding, the graphs and tabular results,  were wrong.

So you see nice flat lines in your screen grabs, but the reality is different.

The guiding probably won't be any better or worse with the focal length corrected, but you won't see nice flat graphs, you'll see what is actually happening.

In fact it would probably be okay to carry on with your current  Calibration and pixel graphs, no problem, but don't go showing anyone your amazing 0.15 arcsec guiding, cos it's not so !

Michael

[malc-c]

Here's my take...

When it was calibrating it moved the star 20 steps in X direction and back in Y, then 20 steps in A and back in B recording how many pixels or parts thereof it moved.  It then starts guiding based on its algorithms or simply applying corrections based on the star movement.  If the star moved 1 pixel in direction X it corrects sending a pulse it calculated to move the mount 1 pixel in direction Y thus maintaining the stars position, thus guiding.

If the weather holds I might try another run, but this time ensure the focal length is entered.  I'm guessing I could always load in last nights profile again if the results go really wrong ? 

[alacant]

9 hours ago, malc-c said:

I bet the guiding will be all over the place

Hi

It doesn't really matter what the guiding looks like. Look at the -I'm guessing good- images as a result. If you're satisfied, that's all that matters.

It's all too easy to stare at PHD2 all night and you never move out of the 'testing' phase!

From your description of your journey after the break, I'd say you've cracked it.

Cheers.

[malc-c]

Alacant, I think you have a point.  So often we get wrapped up with trying to seek perfection that we never get round to actually doing astronomy.  Regardless of what happened last night's PHD trials, I'm happy with the resulting images.  I still need to perfect my processing skills, but that's a topic for a different day, but here is the results of 16 x 400s subs and 16 x 400s darks stacked in DSS.  No flats, bias frames etc....  I've resized the image, but took a crop around the galaxy from the full frame.... stars look round to me

I might try one more profile this evening just to see what the results are from curiosity... I'm guessing that as Michael has commented, with the backlash reduced and the results from the guide assistant the scope's balance and CG along with low backlash and decent star profile values it should mean the traces should be good.  All I know is last night resulted in one of the sharpest images of M81 I've ever had.

 

[alacant]

1 hour ago, malc-c said:

I'm happy with the resulting images

Nice images. 

Now would be a good time to move on. Leave everything as it is and start imaging [1].

Cheers

[1] Correction: I see that you already have:)

[malc-c]

For some reason this didn't get uploaded  the other day

 

One final image, NGC 2309.  Again, just a stack of 18 x 240s subs and 18 matching darks - I'm happy with that, and no doubt with processing would be even better

 

[malc-c]

OK, well as its clear tonight I thought I would run one last series of tests, and this time the focal length of 181mm was correctly entered.

I'll upload the log file later as I'm running a series of subs on NGC2903 and will stack them later to see if the stars are any worse or better.  So to start with, having set the scope to 00:00:04 DEC and 08:50:55 RA I did a run of around 20-30 minutes.  I then ran the guide assist four times (restarting the tool each time) and accepted and applied the suggested improvements after each time.  I've attached the screen captures of the final three - The final result was a backlash value of 804.

I then pointed the mount at NGC2903 and began guiding.  Comparing the screen grab to the one shown at the same settings uploaded on Wednesday the trace is slightly but noticeably improved, and a lot flatter.  But the target graph is more scattered, presumably as with the focal length inputted this affects the spacing ?

I'll upload the log file tomorrow so you guys can comment on how good or bad the mount / scope is performing

[michael8554]

Calibrate at zero Dec but run Guide Assistant ON YOUR TARGET. 

Now that you've finished testing, don't forget to set the Graph Y Axis to arcsecs instead of pixels.

RA 0.69arcsecs and Dec 0.59arcsecs look good and are similar, so should yield round stars.

Michael

[malc-c]

Thanks for the tip.

Log file attached....  best trace was the last....

I'm guessing that for an HEQ5 / 200P having those RMS values is a good result ?

 

PHD2_GuideLog_2020-04-19_210404.txt

[malc-c]

Well the stars are round  so it must have worked

 

[alacant]

7 minutes ago, malc-c said:

the stars are round 

...and small. Perfect. Let the imaging begin.

Let's hope you've started a trend; phd2 users looking at their images, not their graphs!

Cheers

[malc-c]

18 minutes ago, alacant said:

...and small. Perfect. Let the imaging begin.

Let's hope you've started a trend; phd2 users looking at their images, not their graphs!

Cheers

Yeah, I think you're right, 

It's been fun learning what all the bells and whistles in PHD2 do... but I think in my journey documenting the process of getting hung up on seeking the holy grail of a perfectly flat line and tight clustering on the target graph raises several points.  One of those  that I found interesting was that I did a comparison of a single sub taken when I inadvertently left out details of the guidescopes focal length with one I took of the same target last night having created a new profile with that information correctly entered.  At full frame, looking at the same groupe of star there was absolutely no difference in shape or size of the stars between the images.  Now maybe I have been lucky and managed to hit that sweet spot on the balance and CofG  and my PA an PEC are good enough that this omission had no impact.  Or it could highlight an "issue" with PHD2 in that the numbers are all meaningless ?  - Comparing the two target graphs they were so different, but there was no difference in the resulting image

Anyway, onwards to the next learning stage, how to process and get the most out of the data I'm capturing 😉

[malc-c]

Here's the comparason 

[RayD]

50 minutes ago, alacant said:

...and small. Perfect. Let the imaging begin.

Let's hope you've started a trend; phd2 users looking at their images, not their graphs!

Cheers

Well said, Sir!

[michael8554]

45 minutes ago, alacant said:

Let's hope you've started a trend; phd2 users looking at their images, not their graphs!

You couldn't be more wrong. 

Those results were the result of deep dives into the graphs and guidelogs.

13 minutes ago, malc-c said:

Or it could highlight an "issue" with PHD2 in that the numbers are all meaningless ? 

I predicted that despite not entering the focal length, PHD2 did a  good Cal with some sort of preset, and I suggested you could continue with that Cal if you wished, of course the graphs would be meaningless. 

Entering the focal length resulted in the same Cal but with meaningful graph and guidelog results. 

Michael 

[Freddie]

During calibration, PHD always moves a set number of pixels (can’t remember the number) and then uses that calibration to determine how to guide. It only ever uses the focal length to calculate the arc sec figure, so you could completely miss out entering a focal length or put in a completely wrong by a factor of ten figure and it would still calibrate and guide in exactly the same way as when a focal length is entered. With the wrong FL the graphs would not be accurate in terms of arc sec but they would still be accurate in terms of pixels.

[michael8554]

You may be right Freddie, but I thought the Cal step employed was calculated by the number of steps selected, and the focal length of the guidescope. 

Michael 

[Freddie]

I don’t know if EQMOD makes a difference and allows different settings etc as I don’t use it but straight PHD always moves 25 pixels during calibration and the step size (in ms) that you select will change the number of steps it takes to move those 25 pixels. So in terms of calibration, guiding and the graph showing pixels, it makes no difference if you add a FL, put an inaccurate FL in or just don’t bother it will still calibrate and guide the same with or without FL.

[michael8554]

Thanks Freddie

I have still much to learn about PHD2.

Michael

[malc-c]

Freddie, so my basic analogy for when I omitted to include the focal length was close?  I'm guessing that the reason the software asks for the focal length and camera details is so it can calculate the movement in arcseconds and thus be meaningful when compared to other peoples results, or results on the same scope when different guide scope or camera is used. 

[malc-c]

Just a follow up... I've been able to do three or four imaging sessions since and now get into a workflow where having calibrated I now use the guide assistant when on target.  Currently imaging M106 and after an hours worth of guiding this is the result.

Stars are nice and round, and I'm happy with those RMS error results

[michael8554]

Well, the thread is titled "PHD2 graph - Comments please", so:

A great result, round stars and sub 1 arcsec guiding.

Those two red messages on the bottom line of the window ?

"SAT" and "Dark". 

Without a Dark or Bad Pixel Map you run the risk of guiding on a hot pixel. That would show SAT and have a perfect star profile like yours, but I don't think you were guiding on a pixel 🙂

Changing the y axis scale y:+/-8" to +/-4" and ticking the "Corrections" box would let us see what's really going on, instead of red and blue lines glued to the X axis.

Michael

[alacant]

9 hours ago, malc-c said:

Stars are nice and round

That's great news. Remember to look at the images, not the rms values!

Cheers

[RayD]

If you do look at the graph and want it to be meaningful, I would change the viewing scale to one where you can see spikes and other issues that the graph can highlight for further investigation in the logs, rather than this scale which doesn't really show much other than the much desired flat lines.

A typical scale of x: 100 and y: +/- 4" will give you a graph that doesn't necessarily look as pretty and flat as you may want it to, but it will tell you an awful lot more about what is going on and when.

Glad you got positive results as guiding can be the cause of much frustration.