The detrimental effects of turbulence are well known when the goal is high resolution imagery.  The following image shows a recording of star trails at the focus of a long focal length telescope.  It was obtained by effecting a continuous read of the Audine CCD while keeping the star stationary (apart from turbulence).  Compared to the traditional drift method, in which the telescope drive is stopped, this technique allows one to study the state of turbulence over long periods.  The object observed is the double star b1 Lyrae.  It is clear that atmospheric agitation displaces the two stars in exactly the same manner at any given instant, since they are only a small angular distance apart.  The coherence of the turbulence field is shown over several seconds of time and tens of seconds of arc.  The undulations along the track show that the temporal coherence is itself extremely small.  To freeze the the image fluctuations, the exposure must not exceed a few milliseconds unless one is patient and waits for a lull in the turbulence which will allow an exposure of tens of milliseconds.  These are the brief moments of good seeing which we are going to track thanks to a particular acquisition mode permitted by the Audine camera.  We are going to acquire a long sequence of images with very brief exposures from which we will later select only the best, which we will combine to increase the signal-to-noise ratio.

Turbulence disturbs the two components of a double star.

This acquisition mode does not need a mechanical shutter, which would here have to be particularly delicate, for the vibrations must be negligible, robust, for the number of operations is considerable, and finally expensive, since the exposures must be brief.  Besides, the the time taken to save the images between each exposure is a potential waste when one is aiming to profit from brief calm periods of the atmosphere.

The video mode of the Audine camera in fact relies on:

To do this, one performs a judicious reading sequence of the CCD, in which one interlaces phases of rapid reading of the CCD lines, phases of digitisation and phases of waiting corresponding to the exposure time.  The technique is basically similar to the half-frame mode.  In that mode, the object (star or planet) is placed in the upper part of the CCD.  At the end of the exposure, the information in the upper half-frame is rapidly transferred to the lower half, a zone assumed to lack bright objects.  The signal recorded by the upper half can then be digitised without being polluted by the presence of a luminous source.  The difference with the Audine video mode is that the integration, rapid charge transfer and digitisation phases are carried out continuously, allowing as many images of the object to be recorded on the same CCD image as desired.  Here is the sequence of operations:

1. Rapid read of X lines (without digitising the signal)
2. Exposure (may be very brief, limited by the computer speed)
3. Rapid read of X lines
4. Digitising of X lines

The cycle is then repeated from (1) as many times as desired, each cycle corresponding to the acquisition of one image.  This image has the dimension X in one of its axes.  The width in the other dimension can be reduced by digitising only those columns in the immediate neighbourhood of the object, allowing the reading rate to be increased.  Generally, the final image looks like a narrow band containing a large number of images of the same object along its length.  It is like a digital version of cine film.

The object must not be placed just anywhere on the CCD.  Only one band X pixels in height is usable for every two inactive bands. With a little experience, it is easy to know where the object is with respect to the active bands.  A slight nudge of the telescope will often suffice to position the object correctly on the CCD.

The images below are an extract from a typical video mode recording.  They are several consecutive images of a single 3rd magnitude star observed at the focus of a telescope with 7.4m focal length.  The complete record includes several hundreds of images.  Because the band is narrow, its size in bytes is not prohibitive.  For example, those presented here are extracts from an image which is 60 pixels wide and 20,000 pixels long, representing more than 300 separate images of the star.  Careful inspection of this figure shows some signal linking the star images.  This parasitic signal is recorded during the rapid reading phase, the star always illuminating the sensitive surface during this process.  Of course, the parasitic signal is the less important the more quickly the charges are shifted under the star during the rapid reading of the CCD.  This is the phenomenon of smearing.  With the Audine camera and a 200 mm telescope, experience shows that smearing is almost negligible when imaging point sources fainter than magnitude 3.

Two visualisations with different contrasts from a sequence of video-mode images.  In the high-contrast image at the bottom, one notes smearing between the images of the 3rd magnitude star.  With such a star, the smearing signal has 2% of the peak intensity of the star image.  The random movement of the field caused by turbulence is easily seen here (the smearing tracks do not line up exactly from one image to the next).

The preceding image, along with the following ones, were taken with a Takahashi CN-212 used in the Cassegrain mode (212mm primary mirror).  As we are seeking maximum spatial resolution, the potential problem of sampling has been solved by inserting a Barlow lens to increase the effective focal length of the CN-212 to 7407mm.  One pixel then subtends an angle of 0,25 seconds of arc, which allows the finest theoretically-observable detail to be sampled with a sufficient number of pixels.

Examination of the recording shows that in several images it is possible to discern an annular structure around the star.  Here, we are observing the Airy pattern, which is the diffraction pattern of the circular aperture of the telescope.  But note that to observe this clearly, it is really necessary to freeze the turbulence.  The system in question allows us to observe this bright 3rd magnitude star with an exposure of only 10 milliseconds!

The following figure shows a single high-quality image, greatly magnified and in negative, taken from a sequence of exposures of the double star b2 Lyrae.  The exposure is here 35 milliseconds.

The double star b2 Lyrae.  The ticks on the axes represent seconds of arc.  The inner circle marks the position of the first dark ring of the Airy pattern which has a diameter of 1.66 arcsec with this optical system.  The outer circle shows the position of the second dark ring, which has a diameter of 3.05 arcsec.

Between the two circles, the first bright ring of the diffraction pattern can be seen in its theoretical position.  This ring is visible despite imperfect collimation of the telescope: slight coma is present with its axis oriented towards the top of the image.  Moreover, the CN-212 used has slight astigmatism.  It has been possible to reduce this by loosening the attachments which hold the primary mirror to its support, but unfortunately not to eliminate it entirely.  These optical faults are always more accentuated in the presence of turbulence since atmospheric movement transfers the energy of the central maximum towards the rings, which enhances the coma, for example.  The central obstruction also enhances this degradation in the presence of turbulence (without obstruction, the central peak of the Airy disk contains 84% of the signal, but only 70% with an obstruction of 0.33, as in the case of the CN-212).  It goes without saying that if you wish to record the diffraction pattern of your telescope, you must have it perfectly collimated and focused.

On this extract from a sequence of images of b2 Lyrae one can clearly see the damage caused by turbulence.  The exposure is 0.1 seconds and the turbulence particularly high.  Only the central image will be retained for the final composite image.  For average turbulence, only 4% of the raw images are usable.

Composite image of the double star b1 Lyrae, using 12 selected exposures of 30ms.  The components are of magnitude 5.0 and 6.1 and their separation is 2.60".  On the left, we have the negative image (the graduations represent 1 arcsec).  On the right, we have a 3-D view of the double.  The FWHM (full width at half maximum,=seeing) is 0.75", while the FWHM of the theoretical diffraction pattern is 0.66".

The star b2 Lyrae with components of magnitude 5.1 and 5.4 and separation 2.28".

The double S2751 with an angular separation of 1.6" and magnitudes 6.1 and 7.1.  The exposure is
140ms.  The result was synthesised from 14 quality images extracted from 640 acquired images.

The double S2780.  This time, the components are definitely closer: 1.0" for magnitudes 6.0 and 7.0.  One approaches the theoretical limit of resolution, noting that there is still a little further to go.

The Audine video mode can be used on the planets, as shown in the following figures.

Extract from an image sequence of the planet Jupiter acquired directly at the Cassegrain focus of the CN-212 (F/D=12) in I band.

Images of Jupiter in several colours obtained using the video mode.  At top left: B-band, exposure 450ms, top right: V-band, exposure 350ms, bottom left: I-band, exposure 350ms, and bottom right: methane band at 900nm, exposure 7 seconds for this last image.  The interference filter had a passband of 10nm (a filter centred on 890nm would give better results).  Each image in each band is a composite of about 10 raw images.  Conditions were very unfavourable for high resolution imaging during the run, with Jupiter passing just over roofs heated by the sun during the day and seeing estimated at 2.5 arcsec.

The technique described shows how it is possible to select images singly to attain high spatial resolution.  The design of the Audine camera facilitates this exotic mode of reading the CCD, since the clocks are generated by the PC software.  Modifying the timing takes only a few minutes.  At the moment, the video mode is implemented in a special version of QMiPS32 (non yet available) and in Pisco.