A Post I Sent To The Stark Labs Group on PHD Settings

[Neil Heacock]

I thought some folks here may find this helpful. This is the response to a person's request for help with his PHD Brain settings because his graph was all crazy and his stars misshapen after making some changes and not being clear on what he was changing.


On May 20, 2011, at 1:56 AM, Neil Heacock wrote:

1) RA aggressiveness at 120 means that you are going to apply 120% of the correction to the RA movement. This is probably too much. Anywhere from 80 to 100 should likely be what you want. But not over 100.

2) RA Hysteresis is applying a portion of the previous correction(s) to the current correction assuming that there is some error due to lag or seeing or some other such thing and the current correction is exactly.. umm.. correct. The higher the number, the more of the previous correction is applied to the current correction.

3) Min Motion (pixels) - this is the minimal amount that the star is allowed to move *without* sending a correction. If that number were .25, then the star would be allowed to "float around" a quarter pixel without PHD sending corrections to the mount. This is partly to counteract seeing and partly to counteract normal gear error. A setting of 0.05 means that at a 20th of a pixel fire off a correction. This is sure to introduce oscillation and loads of over correction. I'd raise this to nothing lower then .15 or you are sure to send a correction for every guide frame - which is *not* what you want if you can help it.

4) Max [RA/Dec] Duration is how large of a *maximum correction* PHD is allowed to make (in milliseconds). 360/350 (respectively) are probably adequate unless things are considerably bad like massive periodic error or significant polar mis-alignment. You can raise it to, say, 800, but If PHD needs to send an 800ms pulse (for example) then you've most likely got serious polar alignment problems that you're asking the guider to accommodate for that are probably unreasonable. I'd recommend rather then increasing that number too high (500 should be more then enough for any mount), learn how to better polar align. There are lots of tutorials and options online. I've personally been using PHD to aid in drift aligning and it works great.

5) Calibration step - this is the length of a pulse in milliseconds that PHD will send to your mount during the calibration process. If 500ms move the star a pixel or so during calibration, then leave it there. But if it barely moves and you need to make like 50 steps in the various directions and calibration is taking 3 or 4 minutes, then it's way too low. To know where to put that number, play with it until you see the star moving about a pixel for each step and PHD is only taking about 15 to 20 steps in any direction and fully calibrates in about a 90 seconds or so. For me, my 50mm finder/guider is set to 2500 and my 400mm ZS66 guider is set to 1000.

6) Star Mass tolerance.. I haven't figured this one out yet... but in Craig's documentation he says if we set it to 1.0 then it will basically turn it off (as of 1.12.2 I think). I have mine set to 1.0 because until I understand what it does, I want it to not affect things. I have no idea if your 0.3 is the right number for the typical guide stars you get. My recommendation, it sounds like you have enough on your hands to deal with... turn it off by setting it to 1.0.

7) Regarding your guide exposures of 10 seconds in one link and 1 second in the other link, it's important to understand that too short of guide subs will possibly correct too frequently causing the problem of "chasing the seeing" (which isn't too bad if you lower your aggressiveness to 75-80% or so), while too long of guide subs - coupled with bad polar alignment - will cause lots of Dec corrections and never RA corrections because PHD can only send one correction at a time (unless you're pulse guiding) and if every 10 seconds it needs to correct for Dec because of bad polar alignment, then it will never correct for RA. So, my recommendation is to find a sub exposure that is somewhere between error and seeing. Probably 1.5, 2, 2.5 or maybe 3, but based on what you are showing us, I'd be surprised if 3 is good for you. Try 2 second exposures along with the other things I've mentioned here and see if you can get things to settle down.

I hope this helps you and possibly some other folks who read this.

Cheers,

-Neil
___________________________________________________
Visit My Astrophotography Gallery
http://gallery.me.com/nheacock

Visit My Online Astronomy Journal
http://www.myastronomyjournal.com/journal.php?id=nheacock

Join the DSLR Astro Image Processing Yahoo Group
http://tech.groups.yahoo.com/group/dslr_astro_image_processing

[Greg Marshall]

Here's some more tips on PHD settings:

1. DEC guide mode (Off/Auto/North/south): Normally you would leave this set at "Auto". If your polar alignment is excellent you might prefer "Off" to avoid spurious corrections. If DEC is drifting you can note which way it's drifting (easy in "Off" mode) and enable corrections only in the opposite direction.
2. DEC Algorithm (Lowpass filter/Resist switch): Although the default is "Resist", it is probably slightly better to use the lowpass filter. What this does is to effectively calculate your DEC drift rate and apply correction for it in very steady increments, greatly reducing the effects of seeing, etc.
3. DEC slope weight (N): Craig doesn't explain this well and I certainly have no insight, but suggest taking him at his word and leave it at "5".
4. Search Region (N pixels): The default of 15 pixels (i.e., a 30 x 30 area) is generally quite adequate. Larger areas take more time to process and if it gets that far off you've got a serious problem that is probably not going to recover anyway.
5. Noise reduction (none/2x2/3x3): Normally set at "none", but if you're working with a dim guide star and can't increase the exposure time, you might want to try one of the other settings. What it does is to trade off resolution for sensitivity (like binning, but done after the capture rather than in the camera). This can be especially useful for cameras that have very small pixels. Such cameras may provide more resolution than is useful, but suffer from low sensitivity and noise that degrades the accuracy with which PHD can calculate the centroid of the guide star.
5. Time lapse (N milliseconds): A delay after a guide movement before the next guide camera exposure. In some cases the mount movement may not be complete or may not have settled when the "pulse" is done. Setting some delay here will provide time for the mount to settle so that the next exposure is done entirely at the new position.

Another good debate on guiding is how critical it is to focus the guide scope. If you have a sufficiently bright star for the sensitivity of your camera, focus is not critical. In fact, in a camera with very large pixels it may be difficult for PHD to find the centroid of a tightly focused star because it doesn't cover enough pixels to get useful relative values. However, the much more common and likely situation is that the guide star is not as bright as PHD would like. In that case, getting better focused will increase the brightness, albeit over fewer pixels.

[Greg Marshall]

That's a good summary, Neil, thanks. I would add to (5) that you can trade off between calibration step size and the mount's setting for guide rate. If you have to go really high in cal step size, maybe you should increase the guide rate - it would certainly make calibration go faster. But after you've done calibration don't change the guide rate. That's the same as changing aggressiveness. Guide rate is usually best kept at 0.5X (aka 1.5X or 50%), but might go as low as 0.25X and as high as 1X. You should NEVER go above 1X, even if your mount allows it.

I agree that it's a good idea to minimize DEC corrections because they will delay RA corrections that are more urgent. But you seem to be suggesting that PHD will always do DEC corrections first. Is that right?

I dislike the term "pulse guiding" because as a description it better suits the OTHER method, the ST-4 guide port. I have sometimes referred to it as ASCOM driver guiding, but that's not really good either, since it doesn't always involve an ASCOM driver. It's an annoying little bit of AP trivia - like the many different ways of specifying guide rate.

[Steve Coates]

Nice contribution Neil. I'm sure this will clear up a lot of confusion in using PHD.