Precise measurement of (generic) mount tracking performance

[vlaiv]

With increasing interest for astro photography, question is often raised about suitable equipment under restricted budget.

Many people can't afford what would most consider adequate start of mount for serious AP (HEQ5 class), and there are some cheaper alternatives on the market but I think we lack proper understanding of performance of such mounts. Even higher priced mounts rarely come with graphs depicting mount tracking performance (such as obtained by high resolution encoders coupled to RA shaft).

I wondered if there is a way to qualify generic mount tracking performance, and by generic - I mean, out of the box, unloaded, with no external influences - like wind, seeing and all other things that hamper mount performance additionally. A way to measure p2p periodic error and also smoothness of tracking.

This would be considered as "baseline" mount performance, and real time usage would probably yield performance worse than that. While parameter like - "mount will perform better than ... " could be viewed as generally more useful one - I think knowing baseline is also useful - and the way of measuring it is nice "indoors" project for cloudy nights

Idea is rather simple, but it looks like it will take some creative thinking to get it to usable level.

Initially I thought of following:

Take a laser pointer (collimated light beam) and mount it on top of dovetail clamp in some way so it is secure - pointing to the side, so that beam is close to level. Take telescope and camera (equipment that most imagers have so they can readily test their mount) and place telescope with camera attached so that laser beam initially coincides with optical axis (or close enough). Since laser light is close to collimated (probably diverging very slowly) - it will act as a distant point source when focused and project a spot on sensor.

Tracking of scope is then engaged and sequence of captures is taken. Data is later analyzed and centroid of spot is calculated for each frame. If light is bright, even very short exposures in quick succession can be recorded - providing very detailed graph. Spot should move slowly across sensor - one can even calculate the rate of motion for fully collimated beam.

Sidereal rate is 15"/s, and from telescope focal length and pixel size - we can get sampling resolution. Let's take common setup - something like APS-C sized chip and short frac. Such setup often has 2"/pixel, so in one second spot should move about 7.5 pixels. If sensor has for example order of 3000-4000px across - that will let us record about 500 seconds of tracking. With use of focal reduction, most mounts can be sampled over whole worm period.

I started researching into lasers - but it looks like that might not be viable option, I'm afraid that concentrated laser beam might damage the chip! Full power of laser (even very low power one, like 5mw) will be concentrated over few pixels - extremely short exposure will be needed to avoid saturation, and I'm really worried about damage to sensor.

Does anyone have idea if laser would be feasible? Maybe strong ND filter, or even better - solar filter film - really small piece would be enough to filter out laser light to acceptable levels?

Other ideas would be - artificial star - but then there is issue of close focus - most telescopes can't focus close enough to do it indoors?

Another option would be artificial star with collimation lens? Or simple variant - eyepiece. One would place artificial star at field stop of eyepiece (involves unscrewing 1.25" nose piece) and testing against the wall - projected image should stay roughly of the same diameter - maybe diverging ever so slightly (less divergence - further out object appears to be). My worry is that in case of collimated artificial star (or simple tiny whole in some sort of screen) - there will be very large image on the sensor - whole setup will act as relay lens system with magnification (image - short fl - long fl - projection). Maybe that won't matter as we are interested in measuring center of the image as it moves - it does not matter if it's a single spot or a circle - as long as measure of position can be taken (like center of mass).

Laser sounds more interesting because filter can be employed on sensor star (laser at 532nm and Baader solar continuum for example) to boost SNR if measurement is not done in absolute dark. Question is only - is it safe to use, and if no - can we make it safe?

What are your thoughts at all of this?

 

[vlaiv]

Quick update - I decided to ditch whole laser thing and go with relay lens system.

I've got 4" F/10 achromat that will serve as "beam" and 80mm F/6 apo that will be receiver - this way I'll get image reduction. This is a bit larger setup (will hold a scope on the mount), but one can use finder as a light source and something simple as short focal length lens on camera side to the same effect.

System will be made out of simple light source (small torch) and aluminum foil pinhole - just need to fashion mounting rig for that at focal point of larger scope (duct tape to the rescue!) and do preliminary tests (two az mounts, well AZ4 and photo tripod or table top will do to test out relay system and see what kind of image I get on sensor).

[iapa]

Did this experiment progress any further?

[vlaiv]

38 minutes ago, iapa said:

Did this experiment progress any further?

Never really got around doing anything about it really.

[AstroMuni]

On 17/07/2022 at 17:24, vlaiv said:

Never really got around doing anything about it really.

Can this not be done simply by attaching a camera, pointing to a star and get drift plot results after a set period. Ensure that PA is within x arcsec of NCP and compare results?

Edited July 25, 2022 by AstroMuni

[vlaiv]

22 minutes ago, AstroMuni said:

Can this not be done simply by attaching a camera, pointing to a star and get drift plot results after a set period. Ensure that PA is within x arcsec of NCP and compare results?

It can but then you have seeing and weather to contend with.

I wanted to see if there is "lab" method that does not require certain time of day (or night rather) and is not subject to seeing noise.

[AstroMuni]

8 minutes ago, vlaiv said:

It can but then you have seeing and weather to contend with.

I wanted to see if there is "lab" method that does not require certain time of day (or night rather) and is not subject to seeing noise.

I feel that if you ask several folk to attempt this at night and give them the parameters to stick to, you could collect the data and get an averaged out data per type of mount. I dont reckon seeing noise is going to impact this as we are just seeing how well the mount stays on track with a particular star. So as long as the star is visible in all frames then we are good 🙂

[vlaiv]

2 hours ago, AstroMuni said:

I feel that if you ask several folk to attempt this at night and give them the parameters to stick to, you could collect the data and get an averaged out data per type of mount. I dont reckon seeing noise is going to impact this as we are just seeing how well the mount stays on track with a particular star. So as long as the star is visible in all frames then we are good 🙂

You can already do this with different tools that do PE Analysis and PEC curves - and they all have incorporated filters to remove high frequency motion to combat the seeing.

Problem is - this high frequency motion can also come from mount roughness - and there is no way of telling as it will be masked by seeing fluctuations.

It is really tug of war between exposure length which averages seeing effects and loss of high frequency detail. It is similar to guiding. I use 4s to 8s exposures to average the seeing - and it works, but I have no idea how my mount performs in those 4-8s. Is it smooth running or has some sort of vibrations?

[markse68]

I’d have thought that with lower priced mounts the variation in manufacturing would make it pretty difficult to generalise from a test of one of them? Most issues with guiding that I have are down to getting the worm mesh right- too loose and you have excessive backlash and too tight and it will bind, and external factors like wind or vibration from cars going by.

Mark

[AstroMuni]

8 hours ago, markse68 said:

I’d have thought that with lower priced mounts the variation in manufacturing would make it pretty difficult to generalise from a test of one of them? Most issues with guiding that I have are down to getting the worm mesh right- too loose and you have excessive backlash and too tight and it will bind, and external factors like wind or vibration from cars going by.

Hence my suggestion to get data from many users with same mount and collect the stats. And at the end of the day we need to know how mount performs in a real situation, not lab conditions. Its a bit like rated MPG vs actual 🙂

Edited July 26, 2022 by AstroMuni

[ONIKKINEN]

On 26/07/2022 at 11:07, AstroMuni said:

Hence my suggestion to get data from many users with same mount and collect the stats. And at the end of the day we need to know how mount performs in a real situation, not lab conditions. Its a bit like rated MPG vs actual 🙂

You can sort of extrapolate this kind of metric from forum posts and pictures taken with said mounts. Nobody seems to be doing high resolution work on EQ3 class mounts, few are doing so with EQ5 class ones and some are doing OK with EQ6 class. Pictures taken with mounts bigger than EQ6 or tuned EQ6 mounts seem to be consistently better than all the rest.

[vlaiv]

I'll probably revisit this idea in near future as I now have all the ingredients needed to try.

Have artificial star, have 3d printer to print adapter to attach that to long FL scope (4" Maksutov).

Have lower focal length telescope - 80mm f/6 that provides about 3.25 degrees on diagonal - enough for 770 seconds of tracking - which is more than one worm period of my HEQ5.

If need be, I can switch to 85mm lens for recording, but probably need to model geometric distortion of that lens first.

[900SL]

Is there a simple way to calculate peak PE from a single long exposure frame on a bright star, at a low gain/ISO or through a filter so the LP doesn't swamp the sensor? 

For example:  

Deliberately misalign polar alignment

Focus on a high star near the meridian (will drift faster I assume)

Record a star trail over several minutes

View at pixel scale and calculate drift in pixels (the jagged edges where it departs from the straight line)

Back calculate peak PE in arc seconds from known image scale?

[vlaiv]

13 minutes ago, 900SL said:

Is there a simple way to calculate peak PE from a single long exposure frame on a bright star, at a low gain/ISO or through a filter so the LP doesn't swamp the sensor? 

Peak PE is quite easy to measure from single exposure. You need to create exposure that is as long as period of your RA drive and measure length of resulting streak.

For best results - align RA to say X axis of sensor and then just measure length of streak in X axis (any Y movement will be due to polar alignment error). Alternative - you can guide but only in DEC and disable RA output.

After converting from pixel to arc seconds - this will be P2P PE.

 

[wimvb]

Assuming good polar alignment, you could take a single exposure long enough to cover one worm period, while slewing the mount in DEC at less than or equal to sidereal rate. The star trail not only shows p2p periodic error, but also its form.

[Nik271]

How about this method: Attach the laser pointer to the mount saddle and position a white paper screen about 1.5 meters away to capture the dot. In 4 minutes the mount should rotate 1 degree which I calculated would result in the dot traversing about the width of a crop sensor on the screen (25mm).

Now take a camera with a crop sensor with say 6000 pixels width and use a macro 1:1 lens to focus it on the path of the laser dot on the screen.

Run the mount for 4 minutes and take a photo of the screen every second. Compare the width of the lengths of the segments to see if there is variation. 

1 degree is 3600'' so an arcsecond of devation should correspond to about 6000/3200, about 2 pixels, which is possible to spot.

Maybe use the center of the image only to avoid possible lense distortion. Using shorter times and longer distance of screen from the mount can achieve even better precision but may not capture the full length of the worm period.

I'm tempted to try this with my EQ5. Just need to buy a laser pointer.