found between events in one CCD frame and much lower energy events in the following CCD frame. M any contiguous pairs of CCD frames were analysed in order to determine the extent to which a low energy event was located in exactly the same position as an event in the previous frame. After finding th a t at least 10% of events had a low energy secondary event associated w ith it in the next frame, it was decided to find the cause of this phenomena.
The fact th a t the energy associated w ith an event was sometimes spread over two contiguous CCD frames implies some form of storage mechanism being involved. There are only two mechanisms th a t could possibly cause this affect:
1. The CCD. Some charge could, in principle, be left behind in each CCD pixel after a readout. However, this can be discounted as a huge charge transfer inefficiency must be associated and would be immediately noticeable in the profiles and am plitude of
th e events of the video data. This is not seen.
2. The o utpu t phosphor of the image intensifier. The phosphor m ust be causing the additional counting as no other known mechanism remains. The phosphor is a P20 and the decay time has been measured as 2/is to 10% implying th a t there should not be a problem. The hypothesis here, though, is th a t there is a low level, long tim e period, tail to each event as it appears on the o u tp u t phosphor. This tail, when integrated over a frame period could then accum ulate enough charge in the associated CCD pixel(s) of the next frame to create a false event. If this false event were then large enough to lie above th e photon counting threshold a form of double counting could take place.
In order to quantify the degree to which double counting was taking place an experi m ent was carried out in order to measure the intensifiers’ phosphor decay curve.
T h e I n te n s if ie r s P h o s p h o r d e c a y C u r v e C h a r a c te r is tic s
The detector was set up so th a t a LED was flashed once every two frames. In the flash frame, frame A , photon events were recorded in the form of accum ulated charge on the CCD. In the next frame, frame B , in which there were no photon events, smaller am ounts of charge were seen to have accumulated in the same place as photon events in frame A; these residual ‘events’ were caused by the energy em itted by the longer secondary decay components of the phosphor. The two frames were analysed to find correlations between residuals in B and photon events in A . Having found a correlation the com puter used a 3 X 3 array of pixels around the photon event and calculated the energy in fram e B as a percentage of the energy in both frames. An average was found for two hundred correlations.
The energy of the residual depends upon th e tim e of arrival, t, of the photon event w ith respect to the end of frame A . Photon events were simultaneous w ith th e LED flash and so by altering the tim e of the flash, which was known, it was possible to obtain a graph of percentage residual energy, E, versus t.
In order to deduce the decay constants and relative intensities of the secondary com ponents it is necessary to consider the area under the decay curves, which are assumed to be exponentials. This assum ption is valid since m any phosphor decay curves are well
modelled by exponentials or their sum [Smit}}? et a1\. Let r,- be th e decay constant of the component and let ki be its intensity a t t = 0. If there are n com ponents the energy em itted by the phosphor in frame A is then
1=1
(4.8)
A fter the end of frame A there is a frame transfer period, ft, during which the entire frame is transferred to the storage area. Any photon events arriving in this period wiU not be counted and nor will their residuals contribute to the mean residual energy since they will not be correlated. Frame B begins at the end of tt and ends one frame period,
i f y after the end of frame A . The relationship between t, tt and t f is shown schematically in F ig 4.13.
FRAM E A F R A M E S
LED Flash
t = Integration tim e in frame A
t f = Frame period
tt = Frame transfer period
t f — tt = Integration tim e in frame B
Ea = Energy integrated on each event in frame A
Eb = Energy integrated on each event in frame B
Figure 4.13 The relationship between and t f and betw een Ea and F&.
Frame transfer is always sufficiently long th a t the 2//s m ain com ponent completely decays before the sta rt of frame B . The energy em itted by th e phosphor in frames A and B can then be w ritten
E,
dt
(4.9)*+*/, _ _ L
ki€ ^
1=2 •'*+*»
dt (4.10)
From these equations Beilis [FordharrA et aî\ was able to At the d a ta w ith exponential decay curves which could describe the phosphors secondary decay components. A best fit to the d ata, which is shown in F ig 4.1 4 , was derived using three decay components. The long term decay components, curves 2 and 3, are characterized by their m agnitude and tim e constants, r,- and ki where T2= 340/x s , kifk2=570; r3=7300//s and A;i/A;3=23000.
50 T3 40 a o s l/i 30 •S
I
g 20 (U cio S3 0 1000 2000 3000 4000In tegration tim e in first fram e (fxs)
Figure 4.14 Sm ooth curve fitted to real data using three exponential components: ri=0.8686/is; r2=340/is, li/lr2=570; r3=7300/is, ti/t3 = 2 3 0 0 0 . Error bars extend <r
above and below each point.
Although the decay tim e of the P20 phosphor has been m easured as 2/is to 10% the two long term components decay at a much slower rate. W hen the energy associated with these long term components is integrated over a frame period it is large enough to create the residual events seen previously.
H o w R e s id u a l E v e n ts C a n A ffect T h e D e te c to r s D y n a m ic R a n g e
The am ount of residual energy th a t an event deposits in a CCD frame (in this case. Frame B) is dependent upon the to tal event energy (Ea+ Eb) and the tim e a t which it arrived
w ithin the first frame (frame A). The greater th e to ta l event energy and the closer the event arrives to the end of the first frame, the more energy is deposited in th e second fram e (Fram e B).
The way in which energy can be deposited in the next frame can take one of three forms;
• B y a s in g le e v e n t; An event arriving close to the s ta rt of a CCD fram e period will deposit almost all of its energy in th a t fram e (F ig 4 .1 5 a). The phosphor intensity will have decayed to such a low level by the tim e the next frame period starts, th a t virtually no light will be captured in th e next frame. If an event arrives towards the end of a frame period (as in F ig 4 .1 5 b ), then the phosphor will still be em itting some light at the sta rt of the next frame period, and this will be captured in the next frame (Frame B). This energy, Eb will only be a small fraction of the total light em itted by the event, so although energy is captured in the second frame, it is unlikely to produce an event above the event detection threshold.
E v e n t I D ecny C u rv e FRAME A FRAME B D ecay C urve Bfl FRAME B FRAME A F ig 4 .1 6 a S in g le e v e n t a rriv in g a t th e b e g in n in g o f a fra m e p e r io d . F ig 4 .1 6 b S in g le e v e n t a r r iv in g a t t h e e n d o f a fr a m e p e r io d .
• B y tw o o r m o re c o in c id e n t e v e n ts ; If two or more events are spatially coincident and arrive in the same frame period, the intensity of light from the phosphor will approxim ate to the sum of th a t from each individual event. If bo th events arrive close to the beginning of a frame, then the phosphor will not be em itting much light at the sta rt of the next frame (F ig 4 .1 5 c). In this case, energy captured in the second
frame is unlikely to produce an event above the detection threshold. However, if both events arrive towards the end of a frame period (as in F ig 4 .1 5 d ), then the phosphor will be em itting a much greater am ount of light at the beginning of the second frame, and will be captured by it. The energy Eb could be large enough to be above the detection threshold and thus be counted as an event. This causes an increase in detector linearity because two events are counted when two (coincident and otherwise indistinguishable) events are captured.
E v e n t 2 E v e n t 1 D ecay C u rv e FRAME A tt FRAME B E v e n t 2 E v e n t 1 D ecay C urve FRAMED FRAME A tt *2 *1 *2 F ig 4 .1 5 c T w o e v e n ts a rriv in g c lo se to t h e F ig 4 .1 6 d T w o e v e n ts a r r iv in g c lo s e t o th e b e g in n in g o f a fra m e p e r io d . e n d o f a fr a m e p e r io d .
• B y a n e v e n t a r r iv in g in t h e f r a m e tr a n s f e r p e r io d ; If an event arrives whilst the image is being transferred to the storage area, it will not be detected in th e first frame. A proportion of the energy will be captured by the CCD as th e image is being transferred from the image area to the storage area but this will be smeared along a CCD column. Once the second frame period begins, all rem aining energy will be captured as an image. Because the event has arrived so close to the beginning of the second frame, a large fraction of the energy will be deposited in th e second frame, and from the LED tests, it was shown th a t most events are actually detected.
E v e n t 1
D ecay C u rv e
FRAME A tt FRAME B
Fig 4.15e Single event arriving in the frame transfer period.
By having slow phosphor decay components, the bright lim it of dynamic range is increased by detecting an otherwise coincident event (or an event arriving in th e frame transfer period) by capturing a large proportion of its energy in the next frame. The easiest way in which to judge its effect on dynamic range is to incorporate it into the dynamic range com puter simulations.