Focus based on guide star proflie


I was just reading on CN and found a post that got me thinking about this. I think the idea has some merit.

Here is a link to the thread where I replied.

The idea is that IF having a slightly defocused guide star is beneficial in some way, and I have heard some say it is but I don’t know why, then you could use the guide star profile to essentially emulate the idea behind the Optec FocusLock. What they do is use astigmatisim introduced with a piece of glass at 45 degrees in the light path of the off or on axis guider to create a cross bar like this “+”.

Through the calibration routine they can monitor the position of one of the lines of the cross bar to determine which way it moves in relation to perfect focus and then bring the bar back to its starting position. So if the telescope cools during the night and the optics change, the bar will move and they can bring it back as it happens. This makes focus stay tight even during an exposure.

The idea presented in the thread I linked depends on having the guide star slightly out of focus.

You would start with your CCD in perfect focus. Then get your guide star also in perfect focus by manually and physically adjusting it. Get it close, but not perfect.

Then a calibration routine, built into SGP, will take a CCD exposure then a guide exposure. Maybe take a user set number of each and average them. Then it would move the automatic focuser X steps IN and take a photo with the guide camera. Does the star profile improve or get worse? Then move the focuser IN X more steps and measure it again. Now go back to the starting position and reverse it. Move the focuser X steps OUT. Do it again.

You should be able to see the star profile, its sharp point on the graph, its HFR, any and all other ways to measure the quality of the star either improve or degrade as a direct correlation to the focuser position.

But because you know the actual focuser position that had the CCD in perfect focus, you can then determine that if the guide star is slightly worse, you need to move some steps to put it back where it was. If the guide star is slightly better, you move it the other way.

The work flow I see is as follows.

  1. Run the focus control to get sharp focus in SGP with a nice V curve and all that. Be happy with the result.
  2. Get your guide camera in best focus using PHD2 for example by monitoring the HFD and the star graph. How sharp the point is.
  3. Slightly defocus the star so the HFD is a percentage worse than perfect. This is to be determined by people that understand this much better than I do. :slight_smile:
  4. SGP would have a tool drop down like the flats calibration wizard. Where it would focus the main focuser and determine the best step position.
  5. SGP would pull information from PHD2 on what the stars HFD, ADU, FWHM or whatever information is available.
  6. SGP would defocus the CCD by a number of steps as entered in coarse and fine step size times a number or percent.
  7. SGP would pull information from PHD2 on the same above information and determine the new values.
    8-10 would be further testings and information gathering on that same side of focus
    11-15 would be further testing on the opposite side.

The information would be used to adjust focus LIVE to maintain the the best focus position.

But again this all hinges on the ability to guide with a slightly defocused star. I googled “how critical is a sharply focused guide star for autoguiding” and found this link:



indeed this would be great. Simply using the picture that is updated everx X seconds to adapt the focuser position. I also already thought about something like this. Especially this could be implemented without buying extra equipment.

This really would be true “real-time” autofocusing!




Ah yeah;

as for autoguiding on a defocused star. I am not pretty sure if this still is the case. PHD and other guider applications calculate the centroid of the given star, so in theory the star could also be well focused IMO.




This “property” should be present on all OAG or some mod would be required ? The prism is already at 45.º in relation to the light path.


I just had that “wow” moment…
This is indeed clever, no requirement to physically affect the light path, all is done based on the max HFD of the star a run it at a defocused state to know the direction of change.


This kind of thing works very well with the ONAG as you’re guiding on the center of the image circle where little distortion exists. I would assume it works “less good” the farther to the edge you get when using an OAG. I have yet to use my Lacerta, but it’s on the todo list, I think it will be interesting to see how it performs very near the limits (if not beyond them) of my smaller scope (SVQ100). I have a MMOAG which means the pickup is a ways from the center of the image and likely distorted by the optics already.

But yes, I think something like this could work, but setup might be a bit “fiddly”.