Why is my PHD graph spiky

[traveller]

Just a quick question, why is my PHD graph spiky?
Should I be adjusting any of the parameters?
Thanks
Bo

[jenchris]

Yes and balance it a bit camera end heavy.
drop ra to 80 and hysteresis to 8.
If it gets worse,you'll know I was wrong.

[codemonkey]

I'd kill for a graph that looks like that at the moment. How are your images looking? If it's not broken don't fix it I'd say.

[Octane]

Yep, reduce your aggressiveness.

In MaxIm DL, the default is 8 for both axes. I was ending up with a spiky graph. I had to reduce the aggressiveness down to 1 before I ended up with a flat line.

H

[lazjen]

I can't tell the vertical scale on that image. How much is it spiking?

[traveller]

Thanks Chris and H,
I will have another play tonight. Can't recall the spike heights, but it wasn't really affecting image quality. I was only able to do about 60 sec in the strong moonlight anyway, so it wasn't a big deal.
Lee, I was very happy with the graph, but things can always be improved :)
Bo

[JohnG]

Hi Bo

I would be downloading the latest full release of PHD, Version 1.14 and using that, on the Graph you will find that you can adjust your Aggressiveness and Hysteresis on the fly, in other words, you can do real time adjustments while the mount is tracking, a real bonus for fine tuning.

If you have a modern mount and electronics, PHD2 is also available and has quite a number of improvements as well.

Cheers

[peter_4059]

Try increasing the exposure to >1sec - you might be measuring fluctuations in seeing.

[Octane]

Oh, yes. I didn't see the exposure time. Anything less than about 2-2.5 seconds, and, you'll chase the seeing! :)

H

[traveller]

Thanks guys,
To do:
Download latest PHD
Increase exposure to 2-3 sec (it was the triffid nebula, so getting low in the west)
Adjust aggressiveness
Cheers!
Bo

[Amaranthus]

Don't worry about exposure time re: chasing the seeing, if you are stacking frames. It will use this info to better estimate the centroid. This does matter however for the guide pulse frequency - for that I use anywhere from 1-8 sec (typically 3-5) with individual exposures of 100-500 ms. Oh, and definitely try out PHD2 (or Metaguide).

[peter_4059]

Barry - how do you set the guide pulse frequency in PHD? I believe PHD will only stack frames if the actual camera exposure is less than the exposure selected on the main PHD screen. Bo has 0.5 sec selected on the main PHD screen so I believe he is making a guide adjustment every half second.

[peter_4059]

Here's where you download PHD2:

http://openphdguiding.org/snapshots.html

[Amaranthus]

Peter, it's a little counter-intuitive, yes. You are correct that I set the guide pulse frequency by changing the "Camera Exposure Duration" pulldown. However, I then set the properties of the ASI120MM to a different exposure, and so achieve the desired true exposure and frame stacking.

So the 'exposure' for PHD2 is also the actual frequency of guide pulses, whereas the 'exposure' for the ASI120MM is the WDM exposure setting of the camera. So if I set the camera exposure to 500ms, and the PHD2 exposure to 5 sec, then when guiding, PHD2 will report a stacking of 10 frames, as expected.

This 10-frame stack is useful, as I noted, for getting a better 'averaged' estimate of the centroid, compared to a single long WDM exposure, and also avoids guidestar bloat etc. Once can control the FWHM of the guidestar nicely by playing with the WDM exposure time (and gain, if necessary), without mucking about with the actual guide pulse.

I think this only works for video stream type guiders like the ASI120MM, the QHY-5LII and the new Orion Starshoot AG. I don't think it works for CCD-based guide cams, like Lodestar & ST-i.

There is another way to control the guide pulse frequency, and that is to set the Time Lapse (ms) in the Advanced settings.

[Octane]

Barry, that makes perfect sense for video-type cameras.

Not sure what guide camera Bo is using, but, if it's anything non-video, then, I think it's prudent to guide at 2-2.5 seconds at a minimum. :)

H

[peter_4059]

Barry,

I thought PHD effectively sums the 10 individual exposures rather than averaging them. How do you quantify the stack of 10 is better than a single longer exposure in terms of estimating the centroid? I believe the main consideration in terms of centroid estimation was to ensure you are not trying to guide on a saturated star hence the use of the star profile graph.

Cheers,

Peter

[traveller]

How can you tell its camera end heavy? Just courious.
Bo

[traveller]

It's an Orion starshoot autoguider/QHY5 (original)
Currently guiding at 2.5 sec graph is smooth :)
Bo

[Amaranthus]

Peter, the reason I think it is an average rather than a sum is because if I change the WDM exposure length, the size/saturation of the star changes accordingly, i.e. shorter camera exposures yield a lower FWHM and visually smaller stars. I wouldn't have expected that if it were a summed stack. But I could be wrong...

I was also, I admit, thinking of Metaguide, which very explicitly seeks to characterize the centroid via averaging, rather than modelling based on an assumed star shape/light spread distribution function.

[peter_4059]

Barry,

See attached explanation from the PHD2 developers....

https://groups.google.com/forum/#!topic/open-phd-guiding/WWY_tdKCpw8