INSTRUMENT SCIENCE REPORT
		CAL/FOS linearity corrections (revisited)

AUTHOR:0  Lindler and R. Bohlin	DATE: August. 1988 ABSTRACT
	Data taken during Juxie and August 1984 and April 1988 show that the
.	FOS diode non—linearity (paired pulse correction) can be modeled by:


x (1—ty)

Where y is the observed count rate (counts/sec) arid x is the true
count rate. The time coefficient t is constant (9.62 microsec) up to
52,000 observed counts per sec, which corresponds to a true input rate
of about 100,000. Above 52,000 counts, t 19 modeled by:

t=q0 + q1 (y-F)

where:
=	9.62 x io—6
=	1.826 x 1010 F = 52,000

No significant diode to diode variation in the nonlinearity is seen
below 100,000 true (input) counts/sec. Above this count rate the
observed count rate varies trom diode to diode by as much as * 10%.

The accuracy of this model in predicting true corrected count rates
should be better than 1% below 20,000 and degrading to about 5% for an
average ot many diodes at 100,000 true counts/sec.

The variation in the time constant above 52,000 observed counts may be
caused by photocathode fatigue. This variation is consistent with
both the 1984 and 1988 red tube data The blue tube lacks data above
52,000 observed counts/sec.

If the variation in the time constant above 52,000 observed counts/sec
is due to photocathode fatigue, our model may not be applicable to
point source stellar data where the apertures are not fully
illuminated.