Precision Slew - so close but

I have been using Voyager to automate my spectroscopy rig. I use the array set-up (but this post is not specific to array). (Leonardo, feel free to move this to another sub-forum).

The spectroscope has a slit mirror - the target needs to be positioned on the slit so that light passes through to the spectroscope - the image from the mirror is diverted to the guide unit like an off-axis guider. I use my guider for focussing and plate solving as well as guiding. The spectroscope is set up as a slave and commences imaging when the master (the guider) has placed the target on the slit.

Unlike other programs, it accomplishes all that I need very well - with one reservation. I’ll post a separate report about my successes separately but I can say that the system is brilliant.

The reservation - I fear I am at the edge of the capabilities of my mount which limits what I can practically do. I suspect that there is nothing that can be done in Voyager to overcome this (though I am happy for any suggestions to the contrary) - if there are any users of Paramount mounts that have tips, I’d love to hear from you.

The FOV of the guider is small - 15 x 9 arc minutes. Voyager and TSX plate solves and undertakes precision slews easily (which I personally find amazing). The slit is 4 pixels wide - my plate scale is 0.84 arc seconds per pixel. For bright targets, target stars are wide enough to cover the slit and I can gather enough data to generate a useful spectrum. However, fainter targets present a challenge - I need to have the target centred on one of the 2 pixels in the middle of the slit. If the target is perched on the edge of the slit, there is insufficient light transmitted to generate a reliable spectrum.

This is an issue in RA only - the slit is aligned with the Dec axis and so an error of as much as 10 arc seconds is fine. But if my pointing is out by 2 pixels in RA, then I am off the slit and the spectrum image is unusable. I can be out by 1 pixel in RA in one direction but not in the other.

As you can tell - precision in slewing is critical. (Once correctly positioned, guiding is rock solid and my scope is F8 so the focus generally can be set at the beginning of an evening and left as is. I also have the calibration regime set up and it runs smoothly)

Now I am using a parameter of 1 arc second in the Mount set up tab for precision slewing. Looking at the logs, after 4 attempts, the actual reported error for precision slews is generally in the range of 0.5 - 1.5 arc seconds with most around 1.3 arc seconds. I can’t seem to get the error down below that. If I am manually controlling the mount, I can jog the scope enough to get the final correction to place the target on the slit. But with Voyager, I can see from the logs that the instructions to the mount are not moving the mount enough to achieve the required precision.

I’m able to automate spectroscopy reliably with targets down to about V mag 9 - 10 which is brilliant. Recently I managed V mag 11.3. Operated manually, I can gather useful data with my set up down to V mag 13.5. I’m very keen to close the gap as all the really interesting targets are in that range - usually transients in outburst.

I use a Paramount MX controlled by TSX. I have refined my T Point model and adjusted the polar alignment. Slewing speeds are reduced to 50% to improve pointing accuracy and I have the 10 second delay after slewing set in Voyager.

Does anyone - especially Paramount users - have any other suggestions?


Not a Paramount user (two AP mounts) but the precision you are looking for is pretty high even for a premium mount. I could tell you to change the mount but 1.3 pixels at your image resolution is touching your seeing conditions unless you are doing this in Chile?
You may get an improvement if you change your astrometric catalogue and increase the magnitude of your plate solve by lengthening the plate solve image.

Thanks Roberto

you are absolutely right - I am at the bleeding edge with this.

I was tempted to upgrade the mount but the next step up for me is a Planewave LX at A$25,000 which is a bit much for my budget.

I’m based at Siding Springs Observatory in NSW, Australia. Not quite Chile but still probably as good as can be found in Australia.

Happily the plate solving is fine - I’m using the Gaia catalogue with TSX. It works very well - better than the Atlas catalogue with Pinpoint.

At this stage improvements may be at the margins but worth experimenting with


Have you tried pinging a message to Patrick Wallace on the SB forums? He’s a really nice guy and about as knowledgable as it gets when it comes to telescope pointing.

What is the RMS in your plate solve? Not from the Voyager log but from Pinpoint or whatever standalone software you use with the Gaia catalogue? You are unlikely to achieve accuracy beyond 2x that.


How short an exposure is useful on the spectroscope? Would it be possible to write a routine which takes relatively short exposures, issuing guide pulses in RA in between exposures to fish around for the highest received light level in the spectrograph, before starting normal guiding?

I imagine that would be fairly complex to achieve but it would more or less automate what you would be doing manually now, precise pointing to whatever tolerance is possible with the mount then an automated sniff around to find the best received light level (With limitations on the total RA movement allowed, to try to ensure it does not latch on to the wrong star)


TSX reports of RMS varies between 0.7 and 1.2 arc seconds in RA for my plate solves so you are right that this will be a key limiting factor.

Seeing at my site is generally between 1.5 and 2.5 arcseconds. I accept that seeing will contribute to the error but it should also result in bloated stars which may work in my favour.

I’m operating on the assumption that the limiting characteristic is the ability of the mount to take an instruction to move from Voyager and to reliably move by that amount. I suspect that the error associated with an instruction to move a large distance is modest (as a percentage of the required move) but much larger (again as a percentage) for smaller move instructions.

To give an example, here is an extract from a log of precision slews from last week:
2022/03/23 00:11:48 199 - Solved (J2000) => RA 13 12 14.648 DEC -48 59 05.67 PA 270.5 Res. 0.84 [as/px] FL 2870.77 [mm] Star/s 54
2022/03/23 00:11:48 294 - Pointing Error Spherical => 00 00 18.19
2022/03/23 00:11:48 307 - Pointing Error Position Angle => 332 38 19.80

2022/03/23 00:12:11 916 - Solved (J2000) => RA 13 12 14.160 DEC -48 58 48.75 PA 270.6 Res. 0.84 [as/px] FL 2877.62 [mm] Star/s 58
2022/03/23 00:12:12 010 - Pointing Error Spherical => 00 00 03.63
2022/03/23 00:12:12 036 - Pointing Error Position Angle => 257 59 33.46

2022/03/23 00:12:35 655 - Solved (J2000) => RA 13 12 14.106 DEC -48 58 48.54 PA 270.5 Res. 0.84 [as/px] FL 2870.77 [mm] Star/s 60
2022/03/23 00:12:35 737 - Pointing Error Spherical => 00 00 03.16
2022/03/23 00:12:35 750 - Pointing Error Position Angle => 252 16 56.25

2022/03/23 00:12:59 826 - Solved (J2000) => RA 13 12 13.862 DEC -48 58 48.52 PA 270.5 Res. 0.84 [as/px] FL 2870.77 [mm] Star/s 56
2022/03/23 00:12:59 911 - Pointing Error Spherical => 00 00 01.16
2022/03/23 00:12:59 933 - Pointing Error Position Angle => 211 39 34.37

2022/03/23 00:12:59 991 - A Last Best Performance Limits Retry will be done

2022/03/23 00:13:24 118 - Solved (J2000) => RA 13 12 13.772 DEC -48 58 48.18 PA 270.5 Res. 0.84 [as/px] FL 2867.36 [mm] Star/s 60
2022/03/23 00:13:24 214 - Pointing Error Spherical => 00 00 01.35
2022/03/23 00:13:24 226 - Pointing Error Position Angle => 168 22 10.36
2022/03/23 00:13:24 238 - Retries finished … assume this is the Best Performance Pointing for this time !!

The target location is RA 13 12 13.6 Dec -48 58 50.5.

You can see that the first corrective slew took the error down from 18.19 arc seconds to 3.63 arc seconds. This is an excellent result. Voyager then took 2 more slews to get the error down to 1.16 arc seconds. However, the really difficult part is to get the final slew to have an error of less than 1 arc seconds - my thinking is that such a small nudge is hard to achieve and that it is the limitations of my mount (ie hardware not software) that prevent me getting to where I need to be.

The other limitation is that the precision slew feature makes 4 attempts to get the target where it needs to be. I have been working in getting the initial pointing as good as possible with polar alignment adjustments etc so that I don’t use up the 4 attempts with large slews. So if the initial error is say 5 arc seconds, then I get more value for my 4 precision slews. It might assist if I could increase the number of slew attempts but I don’t see how Voyager can accomplish this and if the issue is hardware driven then additional slew attempts won’t get me there - unfortunately.

I know I am operating at the margins here. But given the potential of this (automation is the holy grail for spectroscopists like me) then its worth tinkering


thanks for your suggestions. It all depends on the brightness of the target. If I am imaging a mag 6 star, a 60 second sub gives me excellent SNR - that said, I don’t need that level of precision to get the star right in the centre of the slit. If my target is say mag 11, I need 300 second subs to generate a meaningful image and then stack 12 of them to achieve the required SNR.

A routine to “hunt” for the target would be great. I had played with ACP using a custom action to drag a nearby star onto a designated pixel. The issue is that the target would disappear into the slit when well aligned which upset the computer. The Voyager system is so much better as I am able to guide on any bright star and can set the parameters so that bright stars are avoided - a bright target might otherwise be selected to guide on and guiding goes haywire when the guiding software tries to drag the star across the slit. There are other approaches and yours is an excellent one - but I suspect that it would work better with a more sensitive instrument than mine.


To provide more context of what I am looking to achieve, here is a screenshot of a guider image taken by Voyager when completing a precision slew. The slit is the horizontal line above the crosshairs. The target is the relatively bright star to the left of centre. You can see that it is perched on the edge of the slit. If it were another pixel lower, it would be very well centred on the slit and would yield an excellent result with the spectroscope.

Feel free to laugh at me for trying!


Hi, if you are using PhD as the client guider you might like to see if you could interrogate your primary imager for the spectrum intensity values associated with your best solved position and the star you are looking to guide on and issue sticky guider correction commands to PhD to move the guide box small amounts resulting in low speed corrections to home in on the guide star, transiting the slit across the target and then going back to the optimal position.
I know the ability to move the guide box is available manually, I’ve used it myself in my low-res spec setup.