Cooled enclosure for Eshel spectrograph

[Olivier GARDE]

This weekend , I worked on the modification of a cooled wine cellar to keep the Eshel spectrograph at a constant temperature.





Next step, the development and programing of a PID for temperature regulation.

[Andrew Smith]

Olivier, what temperature stability are you trying to achieve? Also what wavelength stability?

Nice work on the cooler do you have one for wine as well?

Regards Andrew

[Olivier GARDE]

Hello Andrew,

With my eshel spectrograph, at ambiante air temperature (room about 20°C), when the temperature go up about one degree Celcius, the shift of the wavelength (mesure with Thorium lamp) is about a maximum of 0,4 pixel (in the blue part of the spectrum : order 52) and about 0,3 pixel around H-alpha.

According to the dispersion of this spectrograph with my CCD, about 0,176 Å/pixel, the variation at ambiante air (no régulation) is about 0,07 Å/pixel.

I will try to do a regulation about 0,1 - 0,2 deg Celsius with PT100 sensor class A (I can have some better accuracy if I use DIN 1/10 sensor plus it's too expensive).

So I expect to have about a maximum of 0,1 pixel shift with this regulation. (0,017 Å/pixel)

The main problem is to find the right settings for the PID regulation module.

For my wine, I have a real cellar under my house with a constant temperature of 11°C all around the year
Just natural regulation here...

[Andrew Smith]

Olivier, I was many years ago a control engineer and tuned PID controllers for a living! If I remember correctly (it is 30 years ago) with the I and D set to zero you increase P until you get uniform oscillations then set it to one half this value. Unfortunately we were not allowed to use this method on distillation columns or chemical reactors! Then tune D to get a suitable response to a step change in set point. Set "I" to remove any offset over a timescale you think suitable. It is best to play with I and D by changing the set point and seeing how long it takes to settle down. If you get oscillations when increasing D or I try half the value.

I suspect the fastest change in temperature you will get is when the camera is turned on and it starts cooling. One place where a water cooled camera is a benefit. You could add mass in your enclosure and or insulate the Eshel even in the enclosure.

I hope you get the results you are looking for.

Regards Andrew

[Olivier GARDE]

Andrew,

Thanks you for your answer, my main problem is that I haven't any experience with PID programming.
I try to get some informations, and some weeks ago, I met the technician who performed regulation on SOPHIE spectrograph at OHP on the 1m93 telescope (first discovery of an exo- planet in 1995).
They also use PID (eurotherm) for regulation and 8 temperatures sensors put on different parts of the spectrograph (slit, grating, fiber output, etc..). the regulation is very accuracy (about 0,01°C at 20°C ambiante temperature).

If you have some more information about how to program a PID for regulation, don't hesitate to send me it.....

Olivier

[Andrew Smith]

Olivier - My old books are long gone but i had a quick "Google" and found this that looks straight forward and covers some of the implementation issues http://www.scilab.ninja/doc/b4/pid.pdf I hope this helps.

I have been doing some tests on my home made echelle and looking at the Ar lines (I use a lamp starter as reference) in order 40 and the O2 lines cropped from order 33 I get mean shifts of 0.023 +/- 0.3km/s and 0.000 +/-0.5km/s respectively for real world observation of 30 minutes with calibrations taken at the beginning and end of the target exposure. The figures are the mean and standard deviations from 8 sets of observations and the standard deviations represent 0.03 and 0.05 of a pixel respectively. I used ISIS and it's CCF routine to measure the shifts.

I may follow your lead and look at a temperature controlled enclosure at some point in the future but currently I have gone for a high thermal mass and thick insulation with the CCD cooling unit outside the enclosure.

Regards Andrew

PS when I originally posted this I had forgotten that I had been used to a real time operating system that took care of the interrupts need for keeping the strict timing of the algorithm. The paper does not cover these issues. I think I would go for a hardware controller myself as it is 40 year since I did any serious programming which was in FORTRAN on a PDP11 before that it was an Alpha with real ferrite core memory!

[Martin Dubs]

Hi Andrew and Olivier,

I have some experience with echelle stability. The biggest problem I found with the heating produced by the camera cooling. I solved it by bolting a water cooled alu plate to the back of my camera, a SBIG 8300. For good thermal contact I inserted a sheet of tin foil.

echelle water cooling.jpg (103.42 KiB) Viewed 7682 times

For the cooling I used a big pot of water, which kept the temperature stable within 1 degree during the night. Even better would be to put some block of ice in, a very simple temperature stabilization.
For the sensitivity of my echelle I measured a shift of 0.015A/°C at 6257 A, corresponding to a radial velocity shift of 750m/sec/°C. Compare this to a change of radial velocity by the index change of air of about 300 m/sec/°C or by air pressure of 80 m/sec/mbar.
Another point to consider for high precision radial velocity measurements is a shift of calibration lines to the star spectrum if the f-ratios of the two light sources are not identical. This gave in my case errors of up to +/- 500 m/sec within one order, checked by cross correlation of water lines in the H-alpha region, in addition to errors caused by temperature drift. The cause for these shifts are coma errors of the lens, in my case a Canon 85mm lens, which is also used in the eshel. For details about this see:
http://arxiv.org/abs/1001.0794

Regards,

Martin

[Olivier GARDE]

Andrew, Martin

Thank you for sharing your experience.

I will try this weekend to test the stability of my spectrograph inside the cellar.

[Olivier GARDE]

Here's my spectrograph room that I just finished today

Equip with an industrial computer for the spectrograph CCD and to control the calibration modul.
A Nanodac PID regul the temperature of the wine celar witch was customized for the e-shel spectrograph

All power supply for mount, CCD and telescope are here and remotely control (with IPX800 Module)
The celar is filled with bottled water and ice pack to obtain a thermal inertia

And the computer screen is fixed on the wall .