Temperature and calibration

Post Reply
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Temperature and calibration

Post by Peter Velez »

I have a calibration issue - I'm not sure if it is hardware or software related so I'll post both here and in the UVEX section. Feel free to respond to either.

I'm running my UVEX with a 600 lines/mm grating remotely. It produces excellent results - recognising that I can't yet adjust the instrument remotely. That said, my calibration results are proving quite sensitive to changes in temperature.

My routine is as follows:

1. Take flat frame
2. Take ArNe frame immediately prior to taking a reference star spectrum.
3. Take reference star spectra - usually no more than 10 minutes.
4. Take ArNe frame immediately prior to target star
5. Take target star spectra - total exposure time can be anywhere between 30 minutes and 2 hours depending on the target.
6. Take ArNe frame immediately after target spectra.

I record the ambient temperature (measured using the sensor in the focuser control box mounted on the side of my Planewave CDK) when taking ArNe frames and the first reference star and target star spectra. I know that the instrument is susceptible to changes in temperature and so I try and match the ambient temperature of the ArNe frame to the first frame of the relevant reference or target spectrum.

I did some work on this a few months ago and found that a change of ambient temperature of 1 degree C can result in a shift of calibration by up to 2.5 - 3 A. I found that I needed a good 1 - 1.5 hours for my UVEX to cool down to close to ambient before starting imaging. I tested this by taking ArNe frames from the time the roof of the observatory opened till at least 2 hours afterwards. I then calibrated them using Demetra (though I could equally have done this with ISIS) and compared the A0 of the polynomial. Here are the results:

14.8C - 5468.75
9.2C - 5485.42
7.4C - 5489.07C
6.8C - 5490.87
5.9C - 5493.49

The solutions themselves looked good. Reported RMS was at or below 0.1 in each instance. As you can see, the initial change in temperature and calibration is huge but settles over time.

For the UVEX users, has anyone else experienced this? I am using the Shelyak instrument with the Alpy calibration unit.

I have assumed that once it has determined the polynomial for the ArNe frame, ISIS applies it to the first target image and then adjusts the remaining spectra to match up with the first - otherwise there would be discernible drift over long exposure sets. I have run ISIS with Wavelength Registration both on and off and see little difference in results.

Recently, I have found that the calibration is still a bit off. The attached screenshot may help visualise what is going on. This is the H beta absorption line for a reference star plotted using Plot Spectra. The red line indicates where H beta should be. The 2 spectra have been processed in the same way with ISIS - the only difference is the ambient temperature at which the relevant ArNe frame was taken. The ambient temperature when the first target spectrum was taken was 10.5C and as expected, the fit is better for the 10.5C ArNe frame than the frame taken at 11.2C. But it still has an error of about 1A. The solutions produced by ISIS for both ArNe frames are near identical - the only difference is the A0 of the polynomial. I've cross-checked all of the Balmer lines and found that the difference between the 2 spectra are the same - about 2.45A.

Now the puzzling aspect of this for me is that the ambient temperature throughout much of the imaging run was in fact higher than 10.5C. I attach a graph showing the change in temperature during the course for the imaging run. You can see that the temperature ticked up before dropping.

I accept that this was a short imaging run. However, I have seen similar behaviour with much longer exposures.

Am I missing something in the ISIS processing routine which addresses this? How do I ensure that the solutions are accurate and reliable? For targets like this, I can adjust the spectrum to match my expectations - if there are Balmer lines then I can add a wavelength adjustment manually. I can also do the same with atmospheric lines in the red and NIR. I believe that Christian suggests something similar for the Lhires in his tutorial. But if I am imaging a new target without these obvious reference lines - eg when I shift to the 1800 grating - how can I have confidence in the data produced?

I'd welcome any comments/suggestions - either hardware or software.

Pete
Attachments
Screen Shot 2021-08-25 at 2.57.15 pm.png
Screen Shot 2021-08-25 at 2.25.52 pm.png
Dubreuil Pierre
Posts: 202
Joined: Sat Oct 01, 2011 12:34 pm

Re: Temperature and calibration

Post by Dubreuil Pierre »

Hello peter

What type of UVEX spectro do you use?
I have some remarks concerning the behavior of UVEX (3) and now UVEX (4) in DIY, temperature, bending ... and also remarks with the ISIS software to best overcome these generally normal constraints whatever the spectrograph .

Pierre
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Temperature and calibration

Post by Robin Leadbeater »

Peter Velez wrote:
I did some work on this a few months ago and found that a change of ambient temperature of 1 degree C can result in a shift of calibration by up to 2.5 - 3 A. I found that I needed a good 1 - 1.5 hours for my UVEX to cool down to close to ambient before starting imaging. I tested this by taking ArNe frames from the time the roof of the observatory opened till at least 2 hours afterwards. I then calibrated them using Demetra (though I could equally have done this with ISIS) and compared the A0 of the polynomial. Here are the results:

14.8C - 5468.75
9.2C - 5485.42
7.4C - 5489.07C
6.8C - 5490.87
5.9C - 5493.49

The solutions themselves looked good. Reported RMS was at or below 0.1 in each instance. As you can see, the initial change in temperature and calibration is huge but settles over time.

For the UVEX users, has anyone else experienced this? I am using the Shelyak instrument with the Alpy calibration unit.
I am not surprised that 3D printed versions have stability problems but if this is the commercial version I would say that is unacceptable and would expect an answer from Shelyak. Stability, both thermal and mechanical are vital parameters for a spectrograph, particularly if used remotely. Some are good (ALPY), others are poorer (LHIRES.LISA) but manufacturers of amateur spectrographs never quote figures in their specification. Olivier Garde talked about this on the AAVSO forum and suggested the (commercial version of) UVEX as a good stable choice for remote spectroscopy but there are no figures.
https://www.aavso.org/remote-imaging-lisa

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Temperature and calibration

Post by Peter Velez »

Pierre and Robin

Apologies for the delay in responding – I was keeping an eye on my post in the UVEX Project subgroup and didn’t see your posts till late last night.

In my view there are 2 issues to consider here – hardware and software/wetware ie does the UVEX calibration shift with temperature and is it a problem (can it be addressed via software or appropriate observation technique).

Hardware

The answer to the first question for my instrument is yes. I have a first release Shelyak UVEX – I assume this is based on the UVEX3 model Pierre but I don’t know that for sure. There are 2 other users who have each run tests on the response of the UVEX to changes in temperature and they confirm what I am seeing. As temperature changes, there is some change in the instrument which affects the spectrum produced.

The UVEX is very linear – I see no change in the dispersion of the instrument with temperature change or grating angle. What I do see is a change in the A0 element of the polynomial – as temperature changes, the produced spectrum is shifted along the x axis. The shift is linear and, interestingly, doesn’t appear to affect resolution so the UVEX is not flexing in a way that alters the internal focus. The shift is the same whether measuring ArNe or target images and is not correlated to changes in position of the mount – I have confirmed this by taking spectra of sunlight at regular intervals at dusk, just after the roof of the observatory has opened as the temperature in the observatory drops rapidly. The scope is stationary and I am not activating the calibration unit. Given this, we can rule out flexure.

I’ve shared my data with the Shelyak team and they are doing some tests. I believe the 2SPOT guys are looking at putting a UVEX on their rig in Chile. I expect they will be doing similar tests to me given our set ups are similar (though I don’t have as pristine a desert environment as they do).

The engineers among us can probably determine the source of this change from this description. I’m happy to share the data I have if someone is keen to investigate. Its also possible to attach a temp sensor to the unit itself to see how it reacts to changes in ambient temperature. I use the ambient temperature recorded by the EFA box that controls the Planewave focuser as a proxy for the temperature of the unit – that may be a poor proxy to use.

Similarly, a hardware-based solution might involve some way of altering the temperature of the unit itself – wrapping it in a blanket, adding a dew heater, fans etc etc.

My post was primarily directed at finding out whether other users of the UVEX has encountered this issue – particularly users who have 3d printed their instruments. It would be good to hear from anyone about this as I have only exchanged notes with other users of the Shelyak instrument.

Software

I am uncertain whether this is really a problem and would defer to the more experienced among us.

I’ve processed my data in both Demetra and ISIS and achieved the same results. I would have expected that, as the UVEX contracts as temperature drops, the spectra produced from a long series of exposures would see deteriorating results. If the position of a bright Ha emission on the sensor shifts over time, the Ha line would (I expect) appear flat topped and broadened as the data is smeared across the sensor. I don’t see this. Also, I don’t see much difference in resolution between 10 minutes or 2.5 hours of integration time.

This makes sense if the software sets the first image as the reference image and aligns subsequent images to it before stacking. I understand that the Wavelength Registration feature in ISIS does this. Shelyak confirms that Demetra does not do this and, I suspect, ISIS in standard mode does not. I may not be using the Wavelength Registration tool properly, but it doesn’t make any difference to the result and the shift reported in the log is 0A even when I know that the ambient temp has changed.

What is important though is to make sure that the ArNe image is taken at the same temperature as the first image in a spectrum. If there is no shift at the beginning of the sequence, then the calibration will be good and the spectrum fine. In my experience anyway.

In the past I have been guilty of taking ArNe images infrequently – I might take an ArNe frame, then 10 x 1 minute exposures of a reference star, slew to the target and get started expecting that the ArNe frame taken 10 – 15 minutes earlier will be fine. With the UVEX, it likely won’t. I’ve seen changes in calibration with temperature shifts of 0.2 – 0.3 degrees C.

So long as my understanding of the way the software works is correct, I have no issues if I take a calibration frame immediately before imaging a target. It becomes more complicated if the roof closes for a bit mid-sequence. I am playing around with the possibilities for automation (a subject for another post) and recognise that I need to accommodate this in any observing plan.

The other step I have taken is to make sure that I allow sufficient cool-down time for the UVEX to stabilise. My rig is in a shed with a roll off roof – the roof rolls open around dusk. The ambient temperature of the observatory drops for 2 reasons - quickly when the roof is opened as hot air in the shed is released and less quickly as the sun goes down and the outside air temperature cools. The UVEX will not respond as quickly. If I allow an hour after sunset for the instrument to stabilise, the changes in calibration are modest. Previously I was calculating my cool down time from when the roof opened and was – a bit hasty perhaps.

My concern with this issue may stem in part from the fact that I am taking spectra of objects where there may well be redshift or blueshift – RS Oph is a prime example. I am very keen to dispel my own doubts about my results by making sure my calibration is rock solid.

Pierre, I’d be delighted to hear from you on ways to refine my results with ISIS. I’m also keen to hear from anyone else using the UVEX and how they address this. There are plenty of users now and I’d be surprised if they have encountered this.

A final observation. The UVEX is a wonderful instrument – the ability to adjust focus and the grating angle remotely will make it so much better for observers like me. The Shelyak version is solid and well made and I’m very happy with it. But I am keen to run this issue to ground so I can trust my own results – otherwise I’m just collecting spectra that look nice but have no scientific value and for that I should just go back to astrophotography.

Pete
Tonis Eenmae
Posts: 9
Joined: Thu May 14, 2020 5:18 pm
Location: Tõravere, Estonia

Re: Temperature and calibration

Post by Tonis Eenmae »

Peter,

it's really good to hear that aluminium version of UVEX performes very well!

As Robin has pointed and Pierre hinted, there is almost no thermally STABLE spectrograph that is directly attached to the telescope. At least not in professional astronomy field. All those instruments experience large temperature changes during a night and constantly changing gravity vector... The question is how unstable they are. Often they are only very slightly and predictably unstable (but enough to notice that) and proper observing and data processing procedures solve those issues.

Dispersion curve itself shouldn't change - that's the property of a grating - too much, as you also described. When spectrograph shrinks because of falling temperature, zero point (A0) may change (and usually does that a bit). The worst case is when spectrograph itself becomes out of focus or some kind of tilt appears between perfectly collimated state and thermally distorted state. E.g. the focus may become "soft" at one end of the recorded spectrum. In that case also dispersion curve coefficients would change. As you wrote, that's not the case of UVEX and I think this is actually a very good sign.

There is one additional possible source of shifts - exact geometry of providing wavelength calibration light into the spectrograph and if that geometry may change because of temperature. I don't know if that applies to ALPY calibration module or not. The general idea is: imagine a slitless spectrograph and a spectrum of point source of an arc lamp. When the location of the image of that point source wanders on the focal plane of the telescope, physical location of the 0th order (and spectrum) on the slitless spectrograph camera sensor will vary. That also happens with slit spectrographs when slit is not illuminated completely uniformly or calibration light is focused on slit from telescope side and that exact (possibly a tiny point of) point source image location is changing perpendicular to the slit.

If you observe in regions where there are (stronger) atmospheric lines present or with long enough exposures to catch atmospheric emission spectrum, you can use them to double-check or even correct your wavelength scale zero point. I personally do that (and sometimes use them as the only wavelength information) to achieve the highest accuracy for radial velocity measurements with low-to-medium (up to R=5000) resolution.

Best wishes,
Tõnis
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Temperature and calibration

Post by Robin Leadbeater »

Peter Velez wrote:
I’ve processed my data in both Demetra and ISIS and achieved the same results. I would have expected that, as the UVEX contracts as temperature drops, the spectra produced from a long series of exposures would see deteriorating results. If the position of a bright Ha emission on the sensor shifts over time, the Ha line would (I expect) appear flat topped and broadened as the data is smeared across the sensor. I don’t see this. Also, I don’t see much difference in resolution between 10 minutes or 2.5 hours of integration time.

This makes sense if the software sets the first image as the reference image and aligns subsequent images to it before stacking. I understand that the Wavelength Registration feature in ISIS does this. Shelyak confirms that Demetra does not do this and, I suspect, ISIS in standard mode does not. I may not be using the Wavelength Registration tool properly, but it doesn’t make any difference to the result and the shift reported in the log is 0A even when I know that the ambient temp has changed.
It is certainly strange that you are not seeing any blurring effect in the spectra. ISIS does not align the spectra unless the box is ticked and I do see a broadening of narrow lines with my LHIRES which can shift several pixels during a long exposure (mainly due to flexure). As standard I take lamps at start and end and sometimes in the middle of an observation and average them. The shift in the lamp spectrum is clear just viewing the images. (Note that the ISIS alignment tool uses a lamp at the start and only works well if there are good features in the spectrum to lock onto.) Is is possible somehow that your lamp spectrum is shifting but the star spectrum is not. (It would be difficult to explain how though ) What do you see if you just bin the individual spectra at the same Y location and overlay them without calibrating? Do they line up perfectly (also the same for lamp spectra at start and end)

Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Temperature and calibration

Post by Robin Leadbeater »

An example with the LHIRES showing the same shift in lamp and star from start to end of run (2 pixel shift in 45 min (15x 3min exposures)
binned in the same Y position using Visual Spec.
lires_lamp_star_shift.png
lires_lamp_star_shift.png (71.74 KiB) Viewed 4101 times
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Temperature and calibration

Post by Peter Velez »

Thanks Robin

I have a few tests to run when the weather clears. Will report back shortly

Pete
Post Reply