Seems like an interesting general question, so I'll cc wfd-list.
> What would you say if I told you that the current (unmodifed)
> discriminator is between 50%-80% efficient for detection of spes?
I'd say: "what's your definition of an spe?".
> This seems to be what we are observing. We are firing LED
> pulses at tube in the dark box here, and comparing the signals from a
> PMT fanout channel to the output of the 74AC08 (or of the plus and
> minus discriminator) and we are clearly no discriminating on some of
> spes that are visible in the fanout channel.
>
> We have taken pictures of waveforms with a digital scope
> showing pairs of pulses at the same amplitude, where one discriminates
> and the other doesn't.
>
> I have to run outside now, but I was hoping you could give me
> a quick rundown on the testing procedures used to check this, and the
> caveats pointing to what we might be doing wrong.
I made spe like pulses with the pulse generator and fed them in.
Verified by hand that they turned on about where the threshold is. I
did this in situ at LNGS.
To discriminate, a pulse has to be somewhat above threshold. In other
words, if the threshold is at -2.5 mV, I would expect very few -2.5 mV
pulses to pass, more at -3 mV, more at -3.5 mV and (hopefully) all at
-4 mV. Half the spe's are supposed to be over higher amplitude than -4
mV. Keep in mind that a DAC setting -2.5 mV threshold is probably
+-.25mV due to component accuracy (we can check this). Whether a pulse
discriminates is also dependent on the width of the pulse. A -3 mV
pulse that is very narrow might not discriminate (at -2.5 mV) but a -3
mV pulse that is wider and above -2.5 mV longer probably will.
Another factor is any unobserved +-.5 mV long wavelength modulation of
the baseline... that might lift some pulses over threshold and depress
others below. This baseline might wiggle differently in the
discriminator circuit than in the recorded waveform. Our little
ferrite donuts suppress a large -50 mV 10-100 kHz oscillation, but
probably do not supress it to a perfectly flat line. (By the way, I
very much hope we either don't disturb or improve our noise supression
when we uncable to fix the fanouts... if it gets worse, we have a
problem).
The bottom line is spe's are represented by a continuous spectrum and
exactly how many and how often we trigger on a train of them is best
determined based on an overall light level using the LED system. I'd
like to know what you find out about single pe triggering, what you
think our nominal threshold is etc, but I wouldn't spend to much time
getting too precise about it.
Ed