Worst PHD graph in history? Need help :-( [Archive]

codemonkey

11-09-2014, 09:36 PM

Fairly recently I bought a second-hand NEQ6 Pro, and almost simultaneously my first guide cam and an OAG. I've been unable to get guiding to work properly, in fact it just seems to be getting worse if anything.

I was starting to think it was just really bad seeing that was the cause, because with guiding disabled I can see the star move around a fair amount with almost every frame, but the problem exists almost purely in RA, no matter where I seem to point the mount, so I doubt that it's got anything to do with seeing at all.

Has anyone got any suggestions? I've been tweaking this for weeks and I'm spinning my wheels losing my remaining hair.

I've tried the obvious things like balance, checking for cable drag, boosting the exposure (5s in these), reducing the RA guide aggressiveness as far as 50%, upping the minimum move in RA as high as 1px and as low as 0.15px

I've noticed the mount seems to make a kind of "gritty" noise in DEC when slewing, mostly at the start and the finish, but the RA sounds fine...

LewisM

11-09-2014, 10:14 PM

Polar alignment probably way off - seems to be TRYING/fighting, but losing. I had the same issue with an NEQ6 - I would rough align with a compass - and the NEQ6 has enough EM interference that the compass was out 15° or more! There is more ferrous material in an NEQ6 head than I realised.

Also,having oblate stars in the guide image isn't helping you, because PHD and most guide software looks for a centroid position of a circular light mass/pixels. Having weird shapes is NOT helping this at all. The guidescope I assume is a mini-guider, because I also used to get this with the Orion miniguider - just loosen the objective a LITTLE, (releasing the "pinch") and it should improve dramatically. Mine ALWAYS did.

LewisM

11-09-2014, 10:18 PM

Also, try PHD1 instead of 2. I find it works a LOT better with my setup than 2 (which almost always never even tries to calibrate!). PHD1 just locks and stays locked for me (PA down to below minute of error)

codemonkey

11-09-2014, 10:21 PM

I drift aligned for quite a while on both axis before recording that. PHD was recording my polar alignment as being within 1 arcminute, so even though I don't particularly trust it, I doubt it's way off.

I spent about an hour tonight trying to get the stars a more normal shape, but haven't been able to; this was as close to round as I could get it. I'm using the Orion Deluxe OAG with a QHY5L-II, which don't really play well together because if your camera has a 1.25" you have to stick the camera into the OAG, hold it perfectly straight and then tighten a tiny screw with an allen key to lock it all up. PITA.

Shiraz

11-09-2014, 10:21 PM

my guess would be that it's backlash in RA - DEC looks good.

have you tried tightening up the worm clearance as a first step. The transfer gear clearance may need to be tightened up if the worm clearance is OK. easy to check if there is a backlash problem by seeing how much the scope can be rocked in RA by hand - if you can feel it clonking back and forth, then it needs tightening. You could also try deliberately unbalancing the scope in RA so that the worm always sits on one side of the gear, but it is still a good idea to tweak up the backlash.

these mounts make strange noises when starting and stopping a slew. You can make them less noisy with careful adjustment, but it does not seem to make much difference to performance. However, the Dec noise you describe could possibly be due to a slightly over-tightened worm, so it would probably be worth checking that as well.

persevere - these are good mounts, but they require adjustments every few months to keep them in peak condition.

alocky

11-09-2014, 10:24 PM

Actually, the dec graph isn't too bad, and it doesn't look like it's working very hard so your polar alignment is good enough. There's definitely something wrong with the RA though.
What focal length are you imaging/guiding at?
Cheers,
Andrew.

codemonkey

11-09-2014, 10:28 PM

Thanks mate :-)

To be honest I don't know what you mean by "worm clearance." I take it I should basically pull this apart and follow some of the video instructions I've seen on adjusting the gears to remove any play and giving them a good clean?

Thanks Andrew :-)

This is only at 600mm... I've also bought a new scope that I haven't had a chance to use yet, an RC8... I shudder to think what the graph would look like on the RC8!

LewisM

11-09-2014, 10:51 PM

Yes, actually Dec is not too bad. I stand corrected.

Check RA spur gear tolerances. Replace the factory grease. Remove any detritus that has stuck in/on the grease. Backlash in Dec was always a major issue for me, until I cleaned, polished, re-meshed the tolerances.

cometcatcher

11-09-2014, 10:57 PM

Your guide star looks horrible. I'm not surprised PHD is having trouble guiding. Are you trying to guide on a waaay off axis star? I haven't used my OAG for years but star quality wasn't it's strong point.

Shiraz

11-09-2014, 10:58 PM

No, you certainly don't need to take it to bits - only do that as a last resort. There are some hex head bolts that lock the worm carrier to the main body. You undo them very slightly so that the worm carrier can slide on the body and then use the small opposing screws to wind the carrier in or out to adjust the clearance. Then tighten the main bolts back up. takes about a minute when you get the hang of it. It is a normal part of owning an EQ6 - they are even supplied with the key wrenches to do it.

This shows you how and explains what the process does (you should not need to do the end float adjustment on the worm bearings though)
http://www.astro-baby.com/EQ6%20rebuild%20guide/EQ6%20worm%20alignment.htm

others have commented on the guide star image quality. I have found that the centroid calculation in phd is quite OK with images like yours - it seems to be doing well in DEC in your case and I doubt that there will be any problem in RA. However, it is worth choosing an unsaturated star, rather than one with obvious saturation like the one in your example.

Peter.M

11-09-2014, 10:58 PM

Honestly, I assume you are using an OAG due to the star shapes. I would take an image and see how it comes out, if star shapes are less than ideal then look into it. My graphs are regularly like this (in arc seconds) and I have no issues with my images.

Peter.M

11-09-2014, 10:59 PM

It dosent matter what shape it is to guide on

Amaranthus

11-09-2014, 11:15 PM

Lee mentioned that he was using an OAG. As others noted, star shape is not critical - the centroid is based on an assumed not characterized star shape.

I'd strongly advise you try, as a first step, to increase the Westerly bias on the mount. This can really help in keeping the gears engaged. It should NOT be perfectly balanced!

PHD and PHD2 use identical guide algorithms.

DJT

11-09-2014, 11:39 PM

This is what you could get using an RC8 and OAG so dont panic. I grabbed this tonight. I use CCDSOFT, but the settings are 50% aggressiveness , 0.01 min move and 1.0 max move using an OAG and the SBIG RGH, 5 second guiding exposures with a 5 second delay.

Www.ccdware.com has a good guiding calculator which works pretty much the same as what PHD2 will do.

I know nothing about the neq6, can't help there, but my setup is well within the capacity of your mount so you just need to tweak.

jjjnettie

11-09-2014, 11:49 PM

We should get together one evening Lee. Do some trouble shooting. Not that I'm all that technically minded, I've had a little experience with it.

atalas

12-09-2014, 06:15 AM

Hi Lee

Mate don't go pulling the mount apart!!its a very we'll tuned mount.
I never ever had a occasion where it failed to guide well with guide scope.
The PHD graph is showing it to be too aggressive and slight balance.
Balance isn't equal in all areas of the sky.
The noise is normal for all EQ6.

codemonkey

12-09-2014, 07:40 AM

Ok, thanks for the tip Lewis :-)

I get all kinds of crazy shapes with the OAG, from slits, to triangles to crescents, depending on how it's adjusted, but the one thing that I never get is a normal looking star. I'm pretty confident that this isn't related to the star shape as I saw the same kind of movement when trying to guide with the Atik 314L on axis.

Awesome, thanks mate.

I've tried various exposures to try and reduce the effect of seeing, lending me to blown stars (worse than seen here) and well exposed ones, but the problem persists regardless.

I agree, the fact that DEC is fine suggests to me that the shape has nothing to do with this.

Thanks Peter, that's what I believed.

I took some exposures in the beginning and ended up with square stars, thus the quest to optimise the graph, but you're right... I could have a horrible graph and good results.

Thanks mate :-) I've biased the balance in every possible way, perfect, east heavy, west heavy... it no doubt had an effect but nothing resulted in me having a graph that looked like it would give usable results.

That sounds great Jeanette!

Hey Louie, thanks for posting mate :-)

The last part of the graph has minimum move set to 1px, aggressiveness to 50% and hysteresis to 10, so I think I can safely rule out the aggressiveness of the software settings as being a factor.

The maximum pulse was also reduced to 400 in RA and 120 in DEC (RA was down at 120 at one point a few nights ago but I bumped it back up because PHD seemed to be struggling to get it back after it went out in one direction).

I also tried shifting the balance at various points after slewing to that location but before recording this chart that you see (there were other charts, different but equally bad).

This was somewhat low on the eastern horizon, and I think it did seem worse than when I was pointed close to the meridian, which might suggest that balance is a factor in it though...?

Logieberra

12-09-2014, 08:37 AM

Do you have a cheapy guide scope handy? OAG'ing is a lovely thing, but I'd start simple. Master a separate guide scope first, play with the guiding settings and in time go back to OAG.

traveller

12-09-2014, 09:21 AM

Hi Lee,
I recommend you do drift alignment using PHD first (I used my guider with my ED80). It's so much easier to do and the feedback can be very quick.
http://njstargazer.org/PolarAlignment.asp
Take your time and be patient, you will get there.
Bo

pluto

12-09-2014, 10:32 AM

Or even better use PHD2's built in drift alignment helper tool.

Octane

12-09-2014, 02:43 PM

Lee,

What happens when you reduce your aggressiveness down to 10%?

If you're within an arcminute of the pole (for the region of the sky in which you drift aligned), you might find that 50% at 600mm might be too much. This has certainly been the case for my FSQ/STL/G-11 combination. I've had to reduce my aggressiveness down to 1-1.5 to obtain a flat graph in MaxIm DL.

H

Peter Ward

12-09-2014, 03:38 PM

An AP1600 will fix that graph :D

codemonkey

12-09-2014, 05:28 PM

Good call. Wish I did... I was agonising over guide scope vs OAG for a while but decided on OAG since it was where I'll end up anyway as I'll be shooting at 1600mm very soon (hopefully). I imagine with a guide scope the issue in RA would be much less pronounced in the images thus probably enabling effective guiding. Still kind of a bandaid though.

Thanks mate, I did drift align with PHD2 and apparently had it within 1 arcminute on both axes.

Yep, that's what I used.

Thanks H. My understanding of the aggressiveness setting is that it's a percentage of how much guiding should be applied relative to how much the software thinks it should apply.

The software should have established how much it should move to correct for any given error during the calibration process, so the focal length shouldn't make a difference if calibration was successful.

Based on that I would expect that having to reduce the aggressiveness so... aggressively indicates that the software has gotten it really wrong.

Speaking of calibration though, I have noticed during calibration that it completes E-W in about 15 steps, says that it's clearing backlash twice, and then completes N-S in about 7 or 8 steps which seems interesting.

lol I wish.

SamD

12-09-2014, 05:35 PM

Yes, your PHD graph is awful ! I've been "getting to know" my second hand NEQ6 over the last few months too, I feel the pain !

I reckon your polar alignment is fine, the 0.43'' RMS is typical for seeing.

RA guiding looks like the mount is not getting enough corrections, or not responding to the guide commands properly. I see about +-15'' drift in RA when not guiding, so the guiding commands do seem to be getting through eventually, since you're getting about +-3''.

In fact, on your graph there's a couple of minute periods where RA guiding is OK (probably correspond to periods when the unguided plot reaches plateaux on the the main +-15'' sine wave).

Don't suppose you have a small Max RA duration in the PHD settings limiting the corrections being sent ? If you haven't already tried, you could also set the mount auto-guide speed up to 1X, and up the aggressiveness.

I've found that 1s intervals give me best results in PHD with an ED80 and RC8 (ST4 guiding and guidescope). I don't think moving up to 5s will help control RA guiding.

You could also try guiding at say -70 to -80 declination to see if your setup can keep up with guiding there (ie when much shorter RA guide commands are needed).

jenchris

12-09-2014, 05:37 PM

I used my oag with an f10 for 18 months.
During that time my graphs looked awful and often my pics came out at least recognisable.
I hated the graphs so much I went over to an f5 500mm sw refractor as a guide scope
my graphs look amazing now but my pics are no better

Amaranthus

12-09-2014, 05:56 PM

Next step, I suggest getting EQMOD running, and then set up the VS-PEC (this is easy to do now). Run this through 5 worm cycles of training. Then, re-test the PHD guiding with the PEC+pulse running. The form of the RA errors are such that it looks a lot like a bad PE, and so VS-PEC might well be your saviour.

By the way, are you SURE that you're using the hysteresis algorithm for RA? You see, your picture doesn't show up the RA aggr/hyst settings next to the brain (probably because the camera/mount is disconnected), but it does leave open the possibility that you accidentally turned on resist switch or lowpass for the RA guiding (this would explain a lot!).

Also, make sure to use the latest version 2.3.1c build

mithrandir

12-09-2014, 06:13 PM

From the look of the RA graph I'd suggest the mount is too well balanced. This causes the worm to lose contact with the gear.
Try making it a fraction "east heavy". I know that's a pain since you have to rebalance after a meridian flip but it could be your answer.

LewisM

12-09-2014, 09:15 PM

I just use a sliding 200 gram stainless steel counterweight - Telescope on the east side, slide weight UP a few inches. Telescope on the west side, slide the weight down a few inches. It's all it needs.

I also make Dec SLIGHTLY camera heavy too.

alpal

13-09-2014, 08:15 AM

I couldn't get my NEQ6 pro mount running properly until I had
the Dec & RA backlash adjusted & I also fitted a better counter weight bar
as per here:

https://www.flickr.com/photos/24719437@N03/6001169049/in/photostream

AndrewJ

13-09-2014, 11:06 AM

Gday Lee

Did you save the log?????
If so, you can also use something like PHDLab to view the runs.
I cant see how big the drive pulses are in RA, but as they swing from positive to negative, it indicates its not just drift.
I agree that maybe the max RA duration is being exceeded,
but the graph scale appears to hide that info
This can be easily checked from the log data
ref a log from a mount that had too low a setting and when the guides cant keep up, you get the bad excursions.

Andrew

mithrandir

13-09-2014, 06:22 PM

Andrew, I'd expect if the max duration was insufficient that once guiding stabilised the corrections would always be in the same direction and the plot would drift off the top or bottom of the graph.

The PHD2 guys might consider having a warning popup if it wants to exceed the duration limit often and in the same direction. I'll ask them.

AndrewJ

13-09-2014, 09:15 PM

Gday Andrew

Dont understand here, why would it drift off the graph????
The plot came from a mount that uses belt drives all the way to the final axis. It has very large but very smooth PE.
When the PE "rate" was small, it guided OK, but when the PE "rate" was changing at its maximum level, it couldn't keep up, hence the star drifted, but was still "locked" onto. ie PHD2 still knew it needed to bring it back.
On the PE "rate" slowing down then reversing, the guiding managed to incrementally catch up and the plot came back to centre and guiding once again worked for a while, then cr@pped out again later.

ie PHD2 can set a max "pulse time". This creates an absolute limit on how much error can be accounted for "timewise" in a given frame.
The scopes guide rate and PHDs aggressiveness, hysteresis and all the you beaut algorithms etc are all irrelevant if at the end of the day, they calculate a pulse time required that is greater than is allowed by the max pulse time.
I fully agree PHD2 should provide a warning if the max pulse time is hit for more than say 2 consecutive guides.

Andrew

codemonkey

14-09-2014, 04:10 PM

Thanks everyone for your help! :-)

I'll try to cover off on all the questions you have asked:

* Definitely using hysteresis algorithm for RA
* Max guide pulse on RA has been set to 400 for that chart. It was at 1000 I think, maybe 1200, but that was causing it to constantly be correcting in alternate directions (and I ended up with square stars) so I dropped it thinking that was too much
* I didn't save a log.. I thought I checked that option but when I went to retrieve the log it wasn't there
* Already using EQMOD, definitely going to look into PEC... how long does that take? I understand you should do it each time you change anything (including balance, I assume), so it seems like it should be more or less part of the daily set up

I tried shifting the RA axis by hand today... if there was a backlash problem should I be able to feel anything? Feels very smooth in both directions.

The chart really would make sense if it was a backlash / balance problem. I've tried balancing in all directions, but I haven't been particularly precise about it, other than having tried "perfect", east heavy to an arbitrary degree and the same for west... maybe there's a sweet spot that I've missed.

Shiraz

14-09-2014, 05:29 PM

lock the clutches to stop the thing from moving, put on a scope and then see if you can feel any slight rocking (a sort of "clonking" feel) as you try to move the RA by pushing/pulling gently on the end of the scope. If there is any noticeable back-forth movement at all with the clutches locked, you have backlash. You need a very small amount to stop the gears from binding, but if you can feel some slop, it is too much and it would be worth tightening the worm up a bit.

If the backlash is OK, it would be worth checking that there is not too much preload on the RA bearings - which would make the movement a bit stiff - smooth, but a bit sticky to get moving. The last dozen or so panels in http://www.astro-baby.com/EQ6%20rebuild%20guide/EQ6%20RA%20rebuild.htm show what you need to do to check that - just take off the polar scope surround, loosen the grub screws and rotate the nut to tighten/loosen the axis bearings. Tighten the bearing nut until you can just feel that the rotation is stiffer and then back off the nut until the rotation frees right up. Then redo the grub screws and replace the surround.

mithrandir

14-09-2014, 10:51 PM

For those who don't follow the Open PHD Guiding Google group the max duration warning should show up sometime soon. Andy agreed it was a good idea and posted a draft of how it would look a few hours ago.

codemonkey

15-09-2014, 05:41 PM

Thanks mate :-)

Checked the backlash and couldn't notice any movement, certainly no clunking back and forth. Looks like the next step down this path requires opening 'er up... not sure I'm game just yet, I'll see how I go.

I'll have one more crack in the next couple of days I think, try a couple of long shots and if that fails then I might have to open her up.

Thanks again!

Amaranthus

15-09-2014, 05:55 PM

I'd try VS-PEC first. A simple fix it if works. It takes about 45 min to go through 5 worm cycles of training, but you can choose anywhere from 1-5 depending on the robustness of the final function you're willing to accept. You can have this running whilst you are imaging except if you are dithering.

Since you are using EQMOD already, it will be easy to do this. It's literally a 1-button operation -- nicely called 'AutoPEC'. Try it!

codemonkey

15-09-2014, 06:03 PM

Ah yeah, forgot about that, thanks for the reminder :D

AndrewJ

15-09-2014, 09:06 PM

Gday Andrew

As you are a member there, just another thought re warnings.
The mount i got the earlier plot from had a lot of underlying drift as well as very large PE ( even tho it was smooth ).
In some of the runs we grabbed, the "unguided" data for DEC looked horrible, but the error was semi masked when it was guided.
I traced this back to ( very ) bad calibration data, and what appears to be the way PHD converts between sensor and sky coordinates.
I see PHD2 calculates 2 calibration angles to convert the sensor X,Y into RA/DEC.
These should always be 90deg apart ( ie X to RA and X to DEC ), but in some cases we got ( IIRC) up to 90+/-35deg difference between them.
PHD then converted the data using the 2 separate angles and as such both the displayed data and sent pulses were wrong as the RA PEC was contaminating DEC data.
Not saying thats the case here, but it is another possible cause for oddball guiding.
I have no idea why PHD2 uses 2 discrete angles, as there is no way to properly measure the DEC angle due to the PE in the RA drive????

Andrew

mithrandir

15-09-2014, 11:20 PM

Andrew, I think you'd do better asking Andy Glasso this one. (Yes another Andrew.) Either join the group (https://groups.google.com "open phd guiding") and send the debug and guide logs, or if you want to avoid that send me the logs and I'll submit your question.

If you are using ASCOM to do pulse guiding PHD2 is supposed to get the current RA and Dec from the mount and use the Dec to recalculate the pulse lengths as you change position.
If you are using ST-4 the recent versions can use "On camera" for guiding and have an "Aux mount" connection to get the mount position. I haven't had a change to try this yet.

Garbz

16-09-2014, 07:14 AM

If you don't use 2 discrete values then then how can you account for a mount that isn't perfectly polar aligned? If the angles aren't 90deg apart then it's a good sign your alignment is wrong.

Also I can guess that during calibration the PE of the RA axis would be so insignificantly small compared to the step sizes used for calibration that it probably doesn't matter.

AndrewJ

16-09-2014, 07:40 AM

Gday Chris

For the given step sizes, any polar misalignment error will be negligible,
and the scope was relatively well polar aligned ;)
The angles in question are used to convert the sensor axes to the true RA/DEC axes. The calibration process tells the scope to move in an EW direction and notes the angle and direction that makes on the chip.
It then moves in a N/S direction and again notes the angle and direction.
The EW error angle/direction should become the absolute datum and then the NS angle will be 90deg to that and in the "direction" measured by the cal,
ie RA and DEC cannot be anything other than 90deg apart,
certainly not over 20deg, which is what we saw.

Maybe, maybe not.
The mount we were testing had an RA drift of 10-15arcsec/min plus a large PE that sometimes matched this. ie In some spots we were seeing local RA error rates of up to 30arcsec/min
Its the only reason i can see that the cal results were so far away from 90deg apart.
Anyway, irrespective of this, i now know that PHD2 converts "for display" based on its cal angles, as I wrote an app to convert the raw X,Y to RA/DEC based on RA being the master angle ( with DEC normal to RA ) and the plots start to make more sense.
I am just suggesting that this is something that you should look at if the data looks suss. I also recon PHD2 should throw a warning if an oddball cal gets registered, as usually, noone looks at that data.

Andrew

codemonkey

18-09-2014, 07:33 AM

I had another shot last night. I noticed that my filter wheel was a bit loose when connecting to the OAG, and no matter how I tried to tighten it, it would tighten and then come loose again. After switching a few things around there I got that locked up tight, so I removed one potential area for problems.

I didn't bother correcting polar alignment as it was within a couple of arcminutes which I expected to be good enough for these purposes.

I reset all PHD2 settings back to default (as best I can remember) and used Auto-PEC on EQMOD. Actually, I lie, I did set it all back to default but then reduced RA aggressiveness to 80 as it was going nuts. It was still going nuts either way.

Anyway, PEC did more harm than good. I ran it through the full five cycles and had it auto apply. The result was that DEC went off the charts bad, as in it was reading over 5, when I've never seen it go much above 0.7. I quickly disabled this.

This time, for whatever reason, both DEC and RA were jumping all over the place. This was true even after I finished the PE recording and tried to configure it how it was before when only RA was acting up. Disabling guiding on DEC helped, which suggests it was partially software related, but didn't eradicate it.

I think my real problem is that the stars are bouncing around even when guiding is off. I think PHD2 is rightly just trying to correct for problems, but since the star is bouncing around from frame to frame, it's a happy accident when it actually gets it right.

I also screwed around with RA balance, making it varying amounts of east heavy to very west heavy and watched for changes. No apparent pattern to the changes were observed.

At this point I'm reasonably confident that my problems lie with the movement of the star that's present even when not guiding. The question is, what could cause this? I can definitely see some "seeing" issues as well as the stars will get more or less defined. Maybe the bouncing is just seeing as well, but if that's the case I'm permanently screwed because in the weeks I've been trying this the movement has been consistently bad.

Based on previous results I think that the bouncing mainly occurs on one axis, the RA, but I'm no longer sure after last night. If it does mostly occur on RA it definitely suggests to me a mount problem to me. If it was seeing I'd expect it to be more random and showing as much on DEC as on RA.

My plan of attack now is to try and reduce the problem space. The following have already been eliminated from the list of possible primary causes:

1) Balance (adjustment made no observable difference)
2) PHD2 settings (bouncing observed when guiding disabled)
3) Connection between OAG and filter wheel

Tonight I'll try removing the OAG and filter wheel and put the guide cam directly into the OTA and see if it's still bouncing around.

Any ideas on how I could easily confirm / eliminate the RA axis from the list of potential causes? Would pointing near the meridian reduce the impact of RA issues? All of my tests have been done with the scope pointed fairly low on the eastern horizon, as that's where my planned targets have been.

Only once I've figured out what the actual problem is can I then hope to start on the road of fixing it..

SamD

18-09-2014, 08:49 AM

You could try close to 90 deg declination in Octans, and even turn off tracking on the mount. This should show how much PHD wobble there is for stars which really are stationary in the sky.

You would also need to enter calibration data in PHD (if you haven't already calibrated somewhere else), and set it not to send guide corrections to the mount. This way you would just use PHD to plot the unguided movement of a guide star for you.

Shiraz

18-09-2014, 10:06 AM

Crikey Lee, you are having big problems.

Your original graph showed that the RA is not correcting properly - it gets into a no-mans-land where phd2 is generating lots of correction signals but the mount is not responding. Provided phd is generating normal corrections signals and your mount autoguide speed is set at 0.5x, that still suggests backlash to me (ie the motor is moving, but the gearing is not transferring that motion immediately to the wheel). If you cannot feel any worm slop, it could be due to the internal transfer gears. As others have said and you have tried, you should be able to manage this by offsetting the weights to make it east or west heavy, but the offset weight must be sufficient that it overcomes stiction (ie the mount moves freely under gravity when the clutch is released). If that doesn't help, there is not much left mechanically, apart from bearing preload. If the bearings are too tight, you could get jerky motion - I recently improved my DEC tracking by backing off the bearing preload a bit.

If you are looking low down east, you will be viewing through lots of turbulent atmosphere. The turbulence could have an RA-worse component due to slowly rising plumes of warm air (they don't move sideways, so have less effect on DEC). Maybe do a test up near the zenith and see what happens when you are looking through less air. Imaging below 45 degrees will be a lost cause most of the time anyway due to turbulence, unless your system is heavily undersampled.

Amaranthus

18-09-2014, 11:48 AM

PEC should only effect RA, not DEC! Strange. I assume you were guiding when you recorded it - maybe there is an iterative/circular problem here...

Amaranthus

18-09-2014, 09:01 PM

I've been mulling over this problem Lee and something Ray mentioned, has led me to offer another suggestion. The fact that you're seeing weirdness even when PHD is not active suggests that maybe your mount is tracking too quickly (or slowly)...

You've tried changing the maximum duration of the guide pulse in RA to no avail, but what about the minimum? That is, how about adjusting the Guide Speed (0.nn x sidereal rate) downwards, to say 0.25x or even lower the sidereal rate in EQMOD. After all, guiding on the RA axis works by either increasing or slowing the standard guide rate, so as to avoid issues with backlash (DEC works differently, which is why an algorithm to correct for directional shift, such as resist switch, is appropriate).

If this doesn't work, try ramping up the sidereal rate to 0.75 or 1 (or more) -- assuming that you've currently got it on the default of 0.5x. Experiment, and see if this helps...

codemonkey

18-09-2014, 10:14 PM

Thanks mate :-)

That's very interesting actually. I've definitely been below 45 degrees except when drift aligning az. I did try tonight and all kinds of angles and directions, including near zenith and at -90 dec but I still see stars bouncing around.

Yeah I thought that was one of the more interesting things!

Thanks mate, I'll look into that as well :-)

Here's some more charts captured over the past couple of days. I just saw a thread where people in SE QLD were complaining about bad seeing lately, so maybe that actually is a large contributor to my problems. Hope so!

First graph is when I was recording the PE the other night, with mostly default PHD2 settings.
Second is close to the zenith with guiding/tracking enabled.
Third one is close to the zenith with guiding disabled tracking enabled.... note the scale on this one!
Final one is with the QHY5L-II directly in the focuser, not in the OAG... best one, but still not good.

Amaranthus

18-09-2014, 10:44 PM

Lee, I see from your ASCOM settings that your pulseguide rates are 0.1x sidereal! I think this may be the problem... Try changing these values to 0.5x (or more), and see if that helps.

mithrandir

18-09-2014, 10:48 PM

Lee, if you want to run your logs through PHDLab you will have to edit the mount name in the logs. PHDLab can't cope with the "/" in the name. Change it to any other character - "_" is good. http://countingoldphotons.com/phdlab-downloads/

Shiraz

19-09-2014, 07:09 AM

good call Barry - that would fit the symptoms :thumbsup:

Octane

19-09-2014, 07:22 AM

Yep, increase the guide rate to 0.5x sidereal.

This should improve guiding. Then, lower aggressiveness until you get a flat-ish graph.

And, yep, anything lower than 45 degrees and you're trying to guide through turbulent crap. I have set CCD Commander to only start imaging once an object rises past 45 degrees and to cease imaging/move to next target once an object has set past 45 degrees.

Hope you get it sorted out!

H

codemonkey

19-09-2014, 07:31 AM

Thanks for the tip! Didn't know I could actually change the name, that was just the default.

I don't recall ever touching that setting until last night; it was already on 0.1 and when I changed it to 0.5 it resulted in an immediate, significant worsening of the symptoms... chart went off the scale so I set it back to 0.1 very quickly.

Ok thanks, I'll make sure not to do that in future.

Octane

19-09-2014, 07:37 AM

It's all a mystery!

You need to buy a Paramount ME II. :P

H

DJT

19-09-2014, 08:03 AM

Graph 2 is interesting. You have an FWHM of over 6 on your guide star but your guide exposures are just 1 second. Chasing seeing? Your other graphs have fwhm between 3 and 4. The intent is to make as few adjustments as possible so take longer guide exposures? Also gives all your other settings less to deal with.

Shiraz

19-09-2014, 08:25 AM

you need to recalibrate phd when you change the guide rate - otherwise it will not compensate by the right amount and you will see wild swings.

Amaranthus

19-09-2014, 11:04 AM

Yep, as Ray notes, you MUST recalibrate PHD after making a change in the sidereal rate - otherwise the corrections will be 5 times too large compared to its training movement! I'm still convinced this is the problem.

Task:
-- Set guide rate at 0.5x sidereal in ASCOM (leave other settings as is).
-- Set exposure to 3 sec.
-- Set noise compensation to 2x2 mean.
-- Set RA aggr to 80 and Hyst to 15.
-- Set DEC to resist switch.
-- Set min movement of both to 0.25 pixels.
-- Set max RA and DEC movements to 1000 ms.
-- Point at DEC 0 and maybe 5 degrees east of the meridian.
--(Re)Calibrate PHD. Make sure that calibration used at least 8 steps (if not, adjust the pulse duration in the Brain settings).

Do all this, and report back :)

mithrandir

19-09-2014, 12:05 PM

Lee, you can't change it in EqMod. You have to edit the log before PHDLab can cope.

codemonkey

19-09-2014, 05:15 PM

haha. That's it! I'll let the wife know what to get me for christmas ;-)

I suspect this is actually it. I still see stars bouncing all around the place even when guiding is disabled so this suggests to me either:

1) Seeing
2) Mount problems
3) Physical instability somewhere

I saw a thread saying that my area has had terrible seeing lately, so maybe that's all it is. I've done exposures up to 5 seconds long to try and offset seeing, but I was still seeing the star bouncing around from exposure to exposure.

Yeah, while that's obvious I'm not actually sure if I did that or not lol. I'll try it again.

Thanks mate! I'll try that tonight if the weather holds :D

Ah ok, that explains why I haven't seen a setting for it. Thanks :-)

Peter.M

19-09-2014, 05:48 PM

Not to beat a dead horse here, but did you actually take an image with your main camera yet? The last graph that you said was "best one but still not good" shows +- 1 arc second (assuming that your graph scale is set to arc seconds and not pixels, and assuming you have entered your focal length in phd2)

That would be quite acceptable to me on most nights.

codemonkey

19-09-2014, 06:01 PM

I did take some exposures last night and the stars were malformed and much larger than they should have been (I didn't save them, though maybe I should have in retrospect).

That one was taken with the guide cam through the focuser rather than the OAG; the OAG seems to make it worse, presumably due to star shape.

That graph also ranges from +2" to about -1.5" which gives me a range of ~3.5".

Peter.M

19-09-2014, 07:41 PM

Tonight I am taking some exposures, I will screenshot my graph for you and post an image.

codemonkey

19-09-2014, 09:22 PM

Thanks mate, appreciate the effort :-)

Good news! I'm taking some 4 minute exposures of the helix nebula as I write this, and the stars are (mostly) round! Interestingly on one side of the frame they're consistently less round, but that's a topic for another day. Probably something not quite square I guess.

So, I followed Barry's advice to the letter but I still had a not-so-great looking chart. I ended up reducing the numbers significantly, now to 40% aggression, 15 hysteresis, 0.4 min move on RA and 0.5 on DEC, which reduced it somewhat but it's still looking pretty damned horrible.

One thing of note is that I managed to tweak the OAG until I got almost round stars in PHD2, which I've never been able to do before.

So I think Peter was on to something there. Even though the charts looked terrible that doesn't mean the results were... although sometimes they have been and now tonight, it's not.

The stars are still bouncing around, which I think is the real cause for the horrible chart, and I'm putting that down to seeing at the moment.

So now I should be able to grab a couple of hours of luminance on the helix before I have to pack it up and wait for the next clear night to get some RGB... which looks like it may not be coming soon judging by the BOM.

Thanks everyone for your help, I really appreciate it!