These spectra were created with a 2" Sodium Iodide detector and a
radioactive source, Na22. Data acquisition used a CMC100 USB
Camac crate controller and the EZCAMAC program on a Windows
The gate width was 400 nS for all spectra. The spectra shown are just
part of the total spectrum. The data were collected at a 5
trigger rate for several hours. The the minimum count in the dip
between the photopeak and the escape is about 20k-30k, so the expected
rms is less than 1% in this slowly changing region. The count at the
escape peak is about 100k.
The first spectrum used a LeCroy 2249w Camac ADC. This uses a Wilkinson
type converter with a 20 MHz clock. It has a dead time of about 110
microseconds and 11 bits of resolution. The DNL is very good,
The second spectrum used a LeCroy 4300b 11 bit Fera ADC.
was by Camac, not Fera. This is also a Wilkenson but with an effective
400MHz clock, achieved by interpolating a slower clock. The
interpolation (the same technique is used in the MTD133 tdc chip)
results in substatial DNL in the time measurement, which shows up in
the adc value. The dead time is only about 14 microseconds (including
Camac readout). This was the fastest adc available when it
introduced in the mid eighties.
The third spectrum used the CMC081, a modern replacement (and upgrade)
for the 4300b. It uses a successive approximation adc, and a
sliding scale. The resolution is 12 bits, and the dead time
the same as the 4300b (in 4300b emulation mode). The DNL is
undetectable. This adc is dual range (15 bit dynamic range), and has a
dead time less than 4 microseconds in 4 channel mode.
The last spectrum is taken with the CMC080 12 bit triple range adc (18
bit dynamic range). This is a wilkenson, using an MTD135 to measure the
rundown time. It also uses a sliding scale in the time domain
remove the MTD135 DNL (which is comparable to the DNL in the 4300b).
The dead time can be as short as 4 microseconds.
Note the complete absence of DNL in the spectrum.