Not another PHD graph

[lukebl]

Sorry. I know, another PHD graph, but I just wondered if you can actually tell anything particular from this graph? This is about as smooth as I can get, and I know that I've seen some of you with lovely smooth graphs.

Is there anything in the settings displayed that can tell you what I could improve upon? Would it be smoother, for instance, if I stripped-down the mount and replaced or re-greased the bearings? I'm pretty certain balance is right, and everything is tight so differential flex shouldn't be an issue.

[Veracocha]

I suppose the true test is what are your images like, looks very good graph to me.

[George]

I'd be happy with that graph, running the PHD log through PEMPRO would tell you volumes

Also have you tried longer guider exsposures because sometimes you can be chasing local seeing........experiment is the key with PHD

[malc-c]

Yup.. I would be happy with a graph like that...

[lukebl]

Thanks guys. It's just that I'm getting a distinct creep in the stars after, say 10 minutes, and can't put my finger on it. Alignment's good, everything is tightened up, balance is good. I've run the PHD data through programes such as PECPrec (I think), but the output is meaningless gobbledegook to me, which doesn't even point me in the right direction to sort things out.

[Ant]

I clearly have some improvement to go on my graph(especially in RA), but you could improve you DEC graph a little by getting it to guide in one direction only (bottom right setting in the attached file).

I tried this for the first time the other night and my DEC immediately settled down.

Hopefully that will help, but having said that you're graph is already pretty good!

Cheers

Ant

[lukebl]

.. but you could improve you DEC graph a little by getting it to guide in one direction only (bottom right setting in the attached file). ...

Thanks Ant. That sounds interesting. But I've never quite understood how guiding DEC in only one direction helps. What if it needs to nudge it southwards, but can't because you've prevented it from doing so?

Anyway, it looks like the weather's finally broken here. The clouds have rolled in for the first time since Wednesday, so I'll probably have to wait a few weeks before I can test your suggestion!

[Ant]

it stops over correction....

Your mount only drives in RA, so PE can only affect RA. DEC errors are caused by inaccurate polar alignment.

My understanding is that your polar alignment is out in a set way, so errors in DEC will always be on one direction (which will flip when you move from the East to the West.

Because of that the mount will only need to correct the mount in one direction in my case it was South (so south was selected), then after the object had moved past the meridian it needed to be corrected in to the North (so north was selected).

Cheers

Ant

[Uranium235]

You can see which way to guide in dec by switching off dec guiding and waiting to see which way the star drifts. Most of the time I tend to leave mine on auto, not unless there is an obvious drift in any particular direction.

The key I think to this is getting your PA as good as possible, I usually end up doing at least 2 iterations of the PA routine on my mount. Balance is also important, I make my setup slightly camera heavy. Think about what is mounted, is everything placed on axis? Do you need the finderscope once the mount is calibrated? (I always remove it now). Though sometimes no matter what you do, it still wont flatten out or in my case, I start getting backlash.

After much tinkering, this is a typical guide graph (when its not windy!)

PS: I forgot to mention, it is possible to "nurse" a guiding graph by switching between dec guiding modes in response to the graph direction. Ive needed to do this a few times, but its not a fun way to spend the evening

[IvanT]

....DEC graph a little by getting it to guide in one direction only

I tried this last night between breaks in the clouds and it's the flattest I've ever seen my DEC trace. Didn't manage to get a clear trace screen shot as intermittent cloud cover kept spoiling it. I managed to guide for nearly 20 mins at one point with no obvious star movement.

I started by re-doing my PA using only the EQMOD PA routine, didn't bother with drift align for now.

One thing I did notice, if I upped the EQMOD RA & DEC sliders to x90 to speed up the calibration, it manifested itself in a very sporadic graph. If I let it cal at x20 & 3 sec exp which takes a lot longer ( > 10 iterations & < 30 I'm told is the target) then the same PHD settings produced a much flatter graph. Something else to try maybe.

Ivan

[chrisshillito]

One thing I did notice, if I upped the EQMOD RA & DEC sliders to x90 to speed up the calibration, it manifested itself in a very sporadic graph. If I let it cal at x20 & 3 sec exp which takes a lot longer ( > 10 iterations & < 30 I'm told is the target) then the same PHD settings produced a much flatter graph. Something else to try maybe.

That is what I would expect. The purpose of calibration is for PHD to learn how the system responds. Calibration does not reproduce the conditions present during active guiding and folks should never choose settings simply to speed up the calibration process. All that matters with PHD calibration is that it completes.

Choose you guiding rates according to the rates of change you expect to see when the mount is tracking. Periodic Error and polar alignment error drifts generally move quite slowly. If you select a high rate but the real errors in the system itself only vary slowly then the result will be very small period correction pulses being issued by PHD (to move a small amount at a high rate you have to move for a small period). Small period pulses are much more susceptible to timing errors an lags due to the software and hardware involved and consequently guiding accuracy will suffer.

Chris.

[themos]

You can speed up calibration in PHD by raising the value of the "Calibration step (ms)" in the "brain". It's better to do that than change EQMOD's guiding speeds.

[blinky]

Very interesting Chris, I think I will drop the sliders in EQMod as my PE is pretty good but if I remember correctly my sliders are set to X90

[chrisshillito]

You can speed up calibration in PHD by raising the value of the "Calibration step (ms)" in the "brain". It's better to do that than change EQMOD's guiding speeds.

Changing the calibration step will certainly result in a greater movement at the mount (and hence the star) as PHD will lengthen the length of pulse applied for each step. However, it shouldn't make calibration process quicker - it simply lengthens the time PHD will persist with calibration before giving up.

Chris.

[martin_h]

Last week my graph looked like this..(attached) This week it looks like yours...nothing has changed in the setup - obs based, a lot of small corrections might be down to iffy seeing conditions at high altitude.

[malc-c]

On these graphs, what is the scaling of the vertical axis ?

If the RA or DEC (or both !) zig-zag to the 1st line above and below the centre line, what sort of corrections is the mount actually making. Like Martin, last week I had one night where it all went quite well, able to get 5 min subs with nice pin-point stars.. OK the trace wasn't as flat as Martins one above, but it remained fairly close to that magical "25" when it did over shoot. - The following day the graphs were all over the place and I couldn't get it back on track (sorry for the punn)- like Martin, I'm observatory based and nothing had changed in the set up since the night before

[martin_h]

On these graphs, what is the scaling of the vertical axis ?

If the RA or DEC (or both !) zig-zag to the 1st line above and below the centre line, what sort of corrections is the mount actually making. Like Martin, last week I had one night where it all went quite well, able to get 5 min subs with nice pin-point stars.. OK the trace wasn't as flat as Martins one above, but it remained fairly close to that magical "25" when it did over shoot. - The following day the graphs were all over the place and I couldn't get it back on track (sorry for the punn)- like Martin, I'm observatory based and nothing had changed in the set up since the night before

The horizontal dotted lines are 1 pixel apart, the vertical lines are 25 time points apart.

[malc-c]

Thanks Martin,

So if using a QHY5 which has a 1280 x 1024 pixel CCD with 5.2um pixels and the trace is zig-zaging a pixel either side of that line - is this going to have much effect in real terms of movement on the mount and resulting images on a typical f5 telescope ?

[lukebl]

Last week my graph looked like this..(attached) This week it looks like yours...nothing has changed in the setup - obs based, a lot of small corrections might be down to iffy seeing conditions at high altitude.

Interesting. My setup is obs based too, and I know that my polar alignment is good, but still get very variable guiding results. Good some times, terrible at others. It's slightly reasurring to know that others have the same problem, even though that doesn't get to the root of the problem.

[martin_h]

Dont get to hung up on the graph, its the camera image that counts at the end of the day, last week the graph was flat in RA and DEC this week it isn't, probably due to changes in the seeing...BUT the images produced were the same - round stars and no image creep. As long as the graph is not making huge over corrections or slowly drifting across the screen all will be well - hopefully!

[Bizibilder]

The graph will, in effect, be plotting both periodic and other mechanical errors within the whole mount/drive set-up. Therefore I think you should expect variatons depending on where you are pointing the scope, for example, the position of the declination axis will have quite an effect on a not so perfectly balanced tube/camera as it is repositioned from a more "vertical" to a more "horizontal" position.

Surely the whole point of guiding is for the guiding software to compensate for this and produce an image without errors and hence pinpoint stars?

[themos]

Changing the calibration step will certainly result in a greater movement at the mount (and hence the star) as PHD will lengthen the length of pulse applied for each step. However, it shouldn't make calibration process quicker - it simply lengthens the time PHD will persist with calibration before giving up.

Chris.

I thought that my PHD setup moves the mount until it sees 24 pixels of movement in the guide star. It definitely feels it gets there faster if I increase that number.

[chrisshillito]

Yes, PHD is looking for a certain amount of pixel movement (5% of your sensors width). PHD's calibration will fail if this movement doesn't occur within 60 calibration pulses (steps). During each pulse the mount moves at the guiding rate so 5 pulses of 1 sec each will produce the same movement as a single pulse of duration 5 sec. By increasing the calibration step you increase the duration of individual each calibration pulse (step) and so a greater movement can be achieved before the 60 step limit is hit.

Chris.

[chrisshillito]

The graph will, in effect, be plotting both periodic and other mechanical errors within the whole mount/drive set-up. Therefore I think you should expect variatons depending on where you are pointing the scope, for example, the position of the declination axis will have quite an effect on a not so perfectly balanced tube/camera as it is repositioned from a more "vertical" to a more "horizontal" position.

Surely the whole point of guiding is for the guiding software to compensate for this and produce an image without errors and hence pinpoint stars?

Good points.

Also, as I understand it PHD only ever measures movement in terms of pixels not arcsecs. The amount the guide star moves in pixels, for a given angular displacement in RA (as occurs with PE), will depend on the current declination (by a facter of 1/cos(dec)). For example at DEC=60 you will observe half the amplitude of movement than at DEC=0.

Chris.

[malc-c]

Yes, PHD is looking for a certain amount of pixel movement (5% of your sensors width). PHD's calibration will fail if this movement doesn't occur within 60 calibration pulses (steps). During each pulse the mount moves at the guiding rate so 5 pulses of 1 sec each will produce the same movement as a single pulse of duration 5 sec. By increasing the calibration step you increase the duration of individual each calibration pulse (step) and so a greater movement can be achieved before the 60 step limit is hit.

Chris.

Chris, could you answer my question..

Thanks Martin,

So if using a QHY5 which has a 1280 x 1024 pixel CCD with 5.2um pixels and the trace is zig-zaging a pixel either side of that line - is this going to have much effect in real terms of movement on the mount and resulting images on a typical f5 telescope ?

Or are we all getting hung up on trying to get the perfect graph ?