My telescope is a 21 cm Newtonian, and the photometer employs a 1P21 family PMT and DC amplifier. The amplifier has a time constant of 0.3 seconds, and the output goes to a V/F converter and from there to a frequency counter. Manual photometry was performed by centering the star in the aperture, removing the flip mirror, walking to the desk (which took a few seconds during which the amplifier output stabilised and any vibrations in the telescope died away), pressing the reset button on the counter, noting the time and writing down the first three readings. Logging data by hand in this fashion was easy, but entailed a great deal of typing at the computer the following day !
Three main improvements have been made. First, the periodic error in the telescope drive has been reduced. The drive uses a mains synchronous motor to drive a worm, which rotates once every three minutes. Originally, the motor was driven from a variable frequency oscillator and the periodic error could be as much as several tens of arc seconds - which could have disastrous consequences when I was using an 80 arcsecond diameter aperture. In the present system, which was entirely home-built, the drive is operated in 'learn' or 'replay' mode. In 'learn' mode, the telescope is made to track a star, using the fast and slow buttons as necessary to compensate for periodic error. This operation is recorded on an static RAM, so that after switching to 'replay' mode the telescope will track the star properly taking care of periodic error by itself. The fast and slow buttons can still be used to adjust the drive speed and to centre an object in the photometer aperture. (As an aside, many telescopes must have worm and wheel drives with synchronous motors, and I was suprised that PEC systems were not commercially available for such drives). Fine adjustment in declination is provided by means of a tangent arm which is operated by hand.
Second, it was necessary to read data directly into the computer. To do this, I bought an 8 bit to RS 232 interface kit from Maplin Electronics and built a simple 8 bit counter which has a 40 millisecond gate time and which is triggered by the Maplin interface. This combination samples the prescaled V/F output at 9 Hz. Given the noise in the signal, sampling the signal at 8 bit resolution 9 times a second is quite satisfactory. The computer sits in my study in the house, and is connected to the photometer by a 20-m long cable which runs out of a window and out across the garden.
Third, I added a guiding telescope. This time I opted to make life easy and bought something ready made - an 8-inch f/6 tube assembly from Orion Optics in Crewe. This was fitted to my original telescope so as to allow small adjustments of the two telescopes with respect to each other. A star central in the photometer aperture on the 8½ inch is arranged to be on the cross wires of the guiding eyepiece of the 8 inch.
This system was working in time for the predicted occultation of a 10th magnitude star by 162 Laurentia on 1999 September 8. It was a beautiful moonless night and the equipment worked fine, but the path missed Lymm. Perhaps some other time ...
On 1999 September 15-16, I monitored EV Lac for 70 minutes with a B filter without detecting any flares. On 1999 September 16-17, I observed the star again. After an hour, I was rewarded with the spectacular flare shown in the figure. I was away from the telescope during the rise part of the flare, but on returning, I was surprised to find the signal much higher than usual. I looked into the guiding eyepiece, and EV Lac did appear brighter than it had done several minutes previously. That said, I might add that I was very tired by that time, and would not relish the prospect of trying to detect flares visually. The figure shows the measurements binned to 1 second time resolution.
EV Lac brightened by almost a magnitude in 1 minute, with most of the brightening in about 20 seconds. There was then a slower fading, followed by a long 'tail' during which the brightness returned to normal around 20 minutes after the flare had started (not all the tail is shown in the figure). The gap in the lightcurve at 00:15 UTC is a guiding correction.
Unfortunately, this record is not perfect, which is a pity given the rarity of the event. The telescope tracking is good enough to be left unattended for several minutes at a time (just long enough to go indoors and make a cup of coffee!), but a small misalignment of the polar axis requires occasional manual adjustments in declination. I initially assumed the flare would be short lived, and was reluctant to break the record and manually move the telescope. The direction of drift was such that the companion star would reach the edge of the aperture before EV Lac itself. In fact, the flare lasted longer than I had expected, and I appear to have left the correction too late, for when I finally did apply a declination correction (at about 00:15:00) the signal increased by a level consistent with the companion star re-entering the aperture. The previous step in the lightcurve (near 00:13:20) is of a similar size, and therefore also somewhat suspect, although it could also represent a secondary flare.
We can estimate the energy radiated by the flare by calculating the area under the light curve. According to Leto et al, the quiescent luminsosity of EV Lac is 3.49x1022 Watts in B. This flare radiated about 1.5x1025 Joules in B alone. Leto et al show the energy distribution of flares they observed: only 10% radiate 1x1025 Joules or greater in B ! For comparison about 1x1025 Joules are radiated in large solar flares, about a quarter of this appearing at visible wavelengths (Phillips 1992, p187). My EV Lac flare was thus an order of magnitude larger than a large solar flare.
G Leto, I Pagano, C S Buemi and M Rodono. Astron. Astrophys., 327, 1114-1122 (1997). K J H Phillips, Guide to the Sun, Cambridge, 1992. I have also made use of the SIMBAD database operated by CDS at Strasbourg, France.