Star for PE recording

[Demonperformer]

I am looking to do a PE recording using PERecorder for analysis by the EQMOD prog.

In all my reading about this, I have not encountered anything that restricts what star you should use, so I am thinking of Capella - nice and bright, so easy to get on the tiny chip of my webcam (it seems PERecorder can only be used with a webcam).  However, it has a declination of 38 degrees N, and I am wondering if I need to use something nearer the equator.  Other possible options from my position at the moment are:

Aldebaran, 16N, 0m87

Menkar, 4N, 2m54

Mintaka, 0.3S, 2m25

Any thoughts on my best option?

Thanks.

[chrisshillito]

So long as you tell PECPrep to compensate for the star declination when you import the PHD you don't have to choose a star at DEC 0. However if there is a bright star around the equator that you can track then that would represent the optimum choice as it will produce the greatest apparent deflection in RA - and so you get better resolution of the RA data. The more data and higher resolution that PECPRep gets fed the better its analysis (so use a high frame/sample rate as well).  The pixel movement in RA is affected by  factor of cos(dec) so at 38 degrees north you will see a pixel deflection in RA that is around 0.8 of what you would see at  0 degree - so you really wouldn't loose much resolution by choosing Capella over Mintaka. I'd go for the brighter star as this gives you more options with regard to getting a perfectly exposed star image for PHD's centroid algorithm to lock onto.

Of course the process of PE capture using PHD requires guiding to be off and so the downside of a higher resolution / greater pixel movement in RA is that it may be harder to keep the star on chip during the capture period. For a decent analysis you will need a minimum of three complete worm cycles of data - preferably five.

Chris.

[Demonperformer]

Thanks for the reply, Chris.

I don't have PHD, but presumably your 'PHD' comments apply equally well to PERecorder.

I don't know where I got Capella being 38N from ... it's 46N.    But cos(46) is still only 0.695.

Knowing that I don't need to use an equatorial star, and thinking a little more about it, I think I will start with Capella.  At this stage, my mount is going to be at its worst, so my priority has to be to keep the star on the chip for 40 minutes (it's an EQ6 mount & neximage webcam).  Once I have done that and processed the results and am applying them to the mount I can see how that goes.  I really don't know what magnitude of improvement I should expect, but I would hope that halving any errors is not an unreasonable expectation.  That being the case, it will be easier to keep an equatorial star on the chip using this first analysis than keeping Capella on the chip without one.

And in another month, Procyon (5N, cos 0.996) will start being reasonably placed for me, which would make a decent bright, almost equatorial star for a second, more accurate, recording to add to the first, I reckon.

Thanks.

[chrisshillito]

Sorry, I'm not sure why I brought PHD into this - probably confusing this topic with another. Anyhow, Perecorder has a slight advantage over PHD in that it automatically applies the compensation for declination to the data it logs (as does metaguide). Other than that the priciples at play remain the same, i.e. the closer you get to dec=0 the larger the observed pixel deviation in RA and so the more resoltuon the resulting RA PE data has. Essentially for the UK stars from the zenith down to the celestrial equator should be fine but avoid PE capture around the pole.

Chris.

[Demonperformer]

Thanks, Chris.

A couple of (possibly silly) follow-up questions:

Does the law of diminishing returns apply?  What I am thinking is that, certainly for my initial recording, it will be all I aim to achieve that night (trying to do any imaging without the benefit of the PE data I have just collected seems a bit futile!).  This being the case, I could as easily set it up to run for 2 hours as 40 minutes and get 15 rather than 5 cycles.  But will this produce a result that is 3x better, or only a little bit better (for some reason I want to say the (square root of 3)x better)?

Allied to this is what sort of size file does PE recorder produce (per minute/per cycle)?  I would hate to run out of file space when I had got 14/15 cycles and then have it crash and lose it all.

If I have understood your previous replies correctly, the data is based on how many pixels the star moves.  If this is so, presumably I could carry on getting better and better data by using increased focal length?  My initial run(s) will be with a 660mm f/l, but presumably I could (as better data is applied during capture) work up to my 8SE with a 3x barlow at 6m f/l, which, with 9x the f/l and therefore 9x the pixel movement, would give me a 9x more accurate PE correction?

Apologies if they are silly questions .

[chrisshillito]

Does the law of diminishing returns apply?  What I am thinking is that, certainly for my initial recording, it will be all I aim to achieve that night (trying to do any imaging without the benefit of the PE data I have just collected seems a bit futile!).  This being the case, I could as easily set it up to run for 2 hours as 40 minutes and get 15 rather than 5 cycles.  But will this produce a result that is 3x better, or only a little bit better (for some reason I want to say the (square root of 3)x better)?

The law of diminishing returns certainly applies and much ultimately depends on you reasons for collecting the PE data in the first place. If you want to do an in depth mechanical analysis of your mount then the more data to have for the analysis tools to work with the better. It's like imaging you can't ever have too many subs. Out of the box NEQ6 or HEQPro mounts have gearing arrangements that conspire to produce an RA axis tracking movement with a period that is much longer than an individual worm cycle and to precisely  analyse such a signal you need a data set that covers several of these longer periods. In addition, these are not finely engineered mounts and each worm cycle itself may have its own unique profile due to the specific tooth on the RA axis being driven. If PEC is your eventual aim then in my experience it really isn't worth recording more than 5 worm cycles of data. The best PEC can do is produce and average response and over a five cycle period you will usually capture a spread of data that is at least representative of the typical variations present.

Allied to this is what sort of size file does PE recorder produce (per minute/per cycle)?  I would hate to run out of file space when I had got 14/15 cycles and then have it crash and lose it all.

The file size will depend on the capture rate/sample rate of your webcam. If you are able to use a bright star and can use an exposure of as 0.1S then per minute you are going to get 600 lines of data added to your file. A perecorder line is around 70 characters long so that's an increase in file size of around 42K bytes per minute. 15 cycles of an EQ6 would therefore require around 5MB.

If I have understood your previous replies correctly, the data is based on how many pixels the star moves.  If this is so, presumably I could carry on getting better and better data by using increased focal length?  My initial run(s) will be with a 660mm f/l, but presumably I could (as better data is applied during capture) work up to my 8SE with a 3x barlow at 6m f/l, which, with 9x the f/l and therefore 9x the pixel movement, would give me a 9x more accurate PE correction?

PErecorder measures logs theapparent movement of the star in pixels and then coverts this to arcseconds. If say you have a 30 arcsecond Peak to Peak PE and at the meridian your imaging setup produces a movement of 30 pixels then clearly PERecorder will be able to record movemets with a 1 arcsec resolution - image up at DEC=60 degrees and the same PE and imaging system is only going to produce an movement of 15 pixels and so the movement the PERecorder is able to resolve is now 2 arcsec. If one of your gears is producing a 1.5 arcsecond wobble then you're not going see it in the data recorded at dec=60 but you might at dec=0.

Using a higher focal length will give you a better data set in terms of resolution. But it comes with potential compromises elsewhere, for instance the guide star will be dimmer and you may need longer exposures so what you gain in arcsec resolution you loose in time resolution. Also keeping the star on chip in the face of non periodic errors such as linear drifts due to polar misalignment or wind disturbance etc. becomes harder.

So if PE analysis is your aim then it is more likely to make sense to capture the most data, at the highest resolution practical as this may reveal mechanical characteristics that would otherwise be lost. If you only interested in PE capture for PEC then it really isn't worth worrying about detecting these small level signals - PEC on these mounts will never be optimal and will only act to reduce the large error signals.

Chris.

[Demonperformer]

Thank you, Chris, for your detailed response.  I now have a much better understanding of where I am going with this.

Certainly, at this stage, PEC is my concern, so I will stick with the 5 cycles on Capella, followed (possibly) by 5 cycles on Procyon using the correction provided by the first recording.  That seems like a reasonable compromise.

Thanks.