ISIS slit width temperature dependence

• Introduction
• Measurements
• Results
• Implementation of the correction
• Troubleshooting

In February 2012 the 4MS system controlling the ISIS instrument was replaced by a PLC (Programmable logic controller) system that is integrated to the existing instrument control system. Due to the new PLC controllers, among others, a new absolute zero point of the ISIS slit width had to be defined. During a preparation work for new PLC controllers and a definition of new ISIS slit-width zero point, the dependence of the ISIS slit width on the temperature was noticed for the first time. Only during 2012 the dependence became more obvious with more measurements made at different telescope temperatures, but results showed some outliers from the linear trend dependency. At that time, we had only telescope truss temperature (TEMPTUBE) available from the image headers and we actually did not know the temperature inside the ISIS slit assembly, which could have been an explanation for the outlier measurements. After the ISIS is mounted to the telescope it can take up to several days for the temperature inside ISIS to equalize with the ambient temperature inside the dome, and during this time we could not relate the ISIS slit-width dependence with the actual temperature inside ISIS.

In order to confirm the dependence of ISIS slit width on the temperature a new temperature sensor was mounted in January 2013 close to the ISIS slit unit (see picture below).


The temperature sensor readings are saved in PLC interface and labeled as Instrument_Temperature (see picture below).


The origin of the slit-width dependence on the temperature was investigated, to exclude the possibility that this behavior is caused by PLC controllers. We checked many arcs from 2011 and compared the expected versus measured resolution in relation with the telescope truss temperature from the image headers. Interestingly, the dependency of the ISIS slit width on temperature is also apparent except of few measurements which most of them were taken on the day when ISIS was mounted to the telescope or up to two days afterwards. Thus we can exclude that PLC controllers are the cause of this temperature dependence. To measure the actual value of the slit width at given temperature we took a series of tungsten lamp flat fields, all with the same exposure time, and slit width ranging from 3 to 0.4 arcsec. From each flat a mean bias level was subtracted. We checked that overscan region correction did not have any influence on the resulting flux level of our flat fields, and thus this correction was not applied. We measured a mean level of flux in each flat in two different regions, and we plotted this flux versus demanded slit width (the one shown in headers). We then fitted a linear trend, and the intersection of the fit with x axis is the actual slit-width offset (SWO) with respect to the demanded slit-width position. To get the final value of the actual slit width offset for each temperature, we made an average of values coming from two different regions. These two values were always consistent and very similar to each other, which is also an independent check of our measurements. An example of such a fit is shown in the figure below, where the interception of both linear fits with x axis is 0.102 arcsec. This means that the actual slit width was 0.102 arcsec smaller than demanded slit width, i.e. less flux than expected is coming through the slit. The telescope truss temperature was 3.39 deg and the instrument temperature was 5.54 deg during the measurements shown in the figure below.


From each series of flats we obtain one value of the SWO between the absolute and demanded slit width. The dependence of the SWOs versus temperature based on measurements in different days and times of the year is plotted in the figure below. It is apparent that only for temperatures colder than 10.9 deg the slit is in fact narrower than expected, and for temperatures hotter than 10.9 deg we do not see any such behavior. In the figure the red line is the linear fit to points with temperature less than 10.9 deg (red dots), and the green line is the constant fit to points with temperature more than 10.9 deg (green dots). Individual measurements can be found in Table below, where also a telescope truss temperature (TEMPTUBE) is shown for a comparison.

Measurements of ISIS slit width offset (SWO)
Date	TEMPTUBE	Instrument_Temperature	SWO
	deg	deg	arcsec
20120620	12.8		0
20121230	8.3		0.05
20130201	9.68	10.78	0.016
20130209	5.98	7.22	0.093
20130212	7.76	8.8	0.086
20130318	5.48	8.95	0.033
20130323	3.39	5.54	0.107
20130323	4.56	6.00	0.095
20130324	3.03	5.09	0.125
20130326	8.58	9.56	0.031
20130330	11.05	11.41	0.006
20130331	10.13	12.03	0.019
20130615	14.63	15.01	0.025
20130616	14.46	15.16	0.037
20130824	19.79	20.00	0.013
20130825	14.79	16.92	0.030
20130825	15.97	17.10	0.004

As of 2nd December 2013, the slit width offset (SWO) is applied to the slit width using a linear fit or a constant offset, as presented in section Results, and depending on the temperature inside the ISIS slit unit = Instrument_Temperature (see Introduction). The correction for the SWO takes place every time when the slit width change is requested, or a new command for setting the slit width is issued, e.g.:

SYS@taurus> slitarc 1

The Instrument_Temperature is implemented in ISIS Mimic, so that the observer can monitor the temperature during the run and issue the new slit width command if needed, i.e. if the temperature changes by considerable amount. The quick temperature changes by several degrees may happen e.g. in winter when ISIS is moved from the telescope focal station to the cold dome and it takes several days for the temperatures to equalize.

Since 2nd December 2013 there are also two new header keywords in ISIS fits headers:

SLITEMPL= '10.21 ' / Slit unit temperature last (Celsius)
SLITEMPA= '10.21 ' / Slit unit temperature Average (Celsius)

SLITEMPL is simply the last temperature reading of the Instrument_Temperature. The new value is read every 10 s.

SLITEMPA is calculated using the following: SLITEMPA=0.9 * SLITEMPA + 0.1 * SLITEMPL

In order to avoid the slit width mechanism being stacked at very low values, the lowest possible value for the slit width was defined as 70 microns, which equals to 0.32 arcsec. Given the pixel scale of 0.2 and 0.22 arcsec/pixel for the EEV12 and REDPLUS CCDs, respectively, this limitation is not a problem for observations.

During December 2013 and January 2014 we did several tests to check the performance of the slit width correction. Each time, we took data as described above in the section Measurements, for both slit width temperature correction applied and not applied. The example can be seen at the figure below, which is zoomed in for better clarity. The red (blue) points are measurements with (without) a temperature correction applied, and red and blue lines are corresponding fits to the data. The red fit gives a slit width offset of 0.021 arcsec, whereas the blue fit 0.141 arcsec. The temperature during the measurements was 4.7 deg, thus a correction of 0.12 arcsec was applied by the software. This nicely agree with the observed difference of 0.12 arcsec between the corrected and uncorrected data, which is really re-assuring. We conclude that the slit width temperature correction is working well, and even though there are residuals (e.g. 0.021 instead of 0 arcsec slit width offset) they are much smaller than in a past.


ISIS Mimic shows 'Error' for both temperatures:
Probably the communication with the PLC has been interrupted. In this scenario you will probably see other ISIS related errors in the Mimic.

ISIS Mimic does not refresh the temperature values:
It could happen that the software reading the Instrument_Temperature has been killed for any reason. In the instrument control system, type:

SYS@taurus> IsisSlittempPublisher restart

The engineering guide containing more details about the software reading the Instrument_Temperature can be found here.