Issues with calibration and timed out warnings PHD 2.611dev6

[Simon Thompson]

When calibrating I'm getting a warning about dec backlash and also getting a 'timed out' error towards the end of the nights guiding using PHD2 v2.611dev6 whereas before upgrading to dev6 no issues at all. 

I get more or less perfect PA in NINA using 3PPA and confirmed my PA in SharpCap whereas when I run guiding assistant in PHD2 my error is over 10 arc mins which is strange as normally PHD tends to agree with my NINA PA. Towards the end of the nights session every dither tends to chuck up a warning in NINA  'timed out warning waiting for guider to settle' whereas the majority of the nights guiding I had no errors.  Log file using PHD2 v2.611dev6.

Log files using PHD2 dev6  https://openphdguiding.org/logs/dl/PHD2_logs_nRjr.zip

I downgraded to the release version of PHD2 2.611 and the results of lasts nights guiding appeared to be great with zero 'timed out warning' Warnings and no issue with dec backlash.

Log files using the release version of PHD2 https://openphdguiding.org/logs/dl/PHD2_logs_85PN.zip

Using a CEM70EC with an Esprit 120 and OAG

[Bruce Waddington]

Hi Simon, thanks for moving this to a new conversation thread.  I'm afraid I don't see what you see.  In the dev6 log, you had 6 settling failures out of a total of 137 dithers, so a failure rate of 4.4%.  Not a big number by any means, almost certainly didn't cause you any particular problems.  In the 2.6.11 log, you had 5 settling failures out of a total of 126 dithers, a failure rate of 3.9%. Not zero. Hopefully we can agree this isn't a significant difference and we can move on to explaining why you get them at all.  Remember, a timeout result simply means that you didn't allow enough time for the mount to do what it needed to do in order to complete the dither and settle down.  That's pretty much a function of the mount mechanics, the size of the dithers you're doing, and your settling parameters - actually very little to do with PHD2 which is just measuring what happened and telling you about it.

I think the underlying source of most of your problems is an unusually large drift rate in RA.  Here is the RA drift seen with no guiding active on each side of the pier (RA in red):

West side of pier, RA drift is 7.8 arc-sec/min, drifting west

East side of pier, RA drift is 8.35 arc-sec/min, drifting east

So something isn't quite right here, the RA drift on mounts is rarely larger than the Dec drift.  One possibility is that you have some sort of "sag" in the equipment that's mounted to the rear plate of the OTA.  That can happen with long, unsupported extension tubes or even bolts that aren't quite tight enough.  This isn't a lot of error and PHD2 does a generally good job of staying on top of it.  But it means there is a nearly constant RA offset of the guide star from the intended position:

You can see that the RA track is positioned about 0.5 arc-sec away from the lock-point.  This doesn't cause any problems for your images because the star isn't bouncing around.  But the problem does affect your calibrations:

1.  Your calibrations are rather poor.  In fact, the one you used all night on the dev6 release triggered an alert message that you ignored.  That's already a red flag that your guiding results could be impaired.

2.   The calibration problem was mostly caused by the drift in RA which caused an orthogonality error of over 13 degrees.  That means PHD2 can't determine the RA and Dec components of the guide star movement as accurately as we want.  This is exacerbated by your use of relative long 3.5 sec exposures which increases the total amount of drift that will be experienced during calibration.

Nearly all of the dither timeouts occur with large dithers, often over 13 arc-sec in RA.  Coupled with the long exposure times, this again lets the RA drift slow down the settling.  Further, your mount usually over-corrects in RA when it gets one of the large guide commands.  Here's just one example:

At point 1 we see the large dither of 9.8 pixels in RA to the east caused simply by changing the lock point.  At point 2, we see two consecutive RA guide commands trying to move the guide star to the new lock position.  The first one apparently accomplished nothing at all, but the second one caused an over-shoot.  Now there are two large excursions that have to settle out, and that takes more time than you have allowed.  I can't tell whether this is because of a load imbalance or excessive gear mesh in the RA drive system - but it is quite unusual I would say.

Hope this clears things up for you and you can stop worrying about the dev6 release.  Just fyi, none of the code that is involved in any of this has changed in several years.  I think your choices here are:

1. Find and reduce the amount of RA drift

2. Use smaller dither amounts

3. Relax the settling constraints

You can refresh your understanding of the settling parameters here:

https://openphdguiding.org/man-dev/Supplemental_Info.htm#Dithering_Ops

Regards,

Bruce

[Simon Thompson]

Hi Bruce

Thank you for your detailed explanation. I was getting concerned because the previous versions of PHD2 and NINA didn't show any errors with settling after dithering whereas dev6 did for some reason. In regard to dithering I just left it at default within NINA -  I will now reduce the dithering amounts. I did notice the large amounts of drift when I ran the guiding assistant but as my stars appeared to be fine I wasn't too concerned. I'll strip the focuser off the OTA (Essato 3inch) and break down the optical train and try to get to the bottom of the sag you mentioned, I didn't notice anything unusual when setting up my gear (portable) but I'll remove the components and see if there's anything out of the ordinary. 

Many thanks

Simon