Dear Robert,

We looked at this carefully some years ago.

The digitizers for the MAT 252 are called nonius counters.  Signals from 
the FCs are directed to voltage to frequency converters (VFC), which put 
out an alternating signal with frequency proportional to the voltage (ion 
signal).  Separately, there is a clock running at a constant ~4 MHz.  When 
the time bin is set to 0.25 s, the nonius counts peaks from the VFC for 
0.25 s.  Once 0.25 s elapses, a gate drops and the clock continues to run 
until the next count arrives.  So the actual data for an interval consists 
of an integral number of counts (numerator) in a time interval 
(denominator) that is variable and always greater than the nominal bin size 
(0.25 s).  In this way, the precision of the measurement is limited by the 
clock speed (Mhz); if exactly 0.25 s were used, the precision would be 
limited by the VFC rate (variable depending on signal, kHz or less so not 
high precision).  At high count rates, the last count arrives soon after 
the 0.25 s nominal bin size, whereas at low count rates, the last count 
arrives later, so that the bin size is 0.26 s.

Merritt & Hayes, Anal Chem 1994,66, 2336-2341 (see p 2337) describe this in 
a paragraph.

Tom Brenna

At 03:15 AM 4/23/2008, you wrote:
>Hey,
>
>we are using a Finigan MAT 252 mass spectrometer to measure the 13C
>isotopy on CO2. When I started to reproduce the results obtained by the
>intern isodat software, I had a look to the raw data. Thus, to the
>mass44, mass45 and mass46 signals in time. I recognized, that the time
>intervals are not equidistant, although they are set to 0.25s. It seemed
>to be dependent on the actual signal. If the signal is high, like when
>measuring a peak, 0.25s is the normal time interval between two points.
>If only background values are measured, the time interval often changes
>to 0.26s.
>So, I would like to know, how the time column might be produced. Is it
>an intern effect of the farraday cups and how often they can be read in
>a certain accuracy, or does the computer-speed maybe influence the read
>out time? I just want to figure out in which way I can trust this time
>column and how the raw data are basically produced at all inside the
>mass spectrometer.
>
>Any help would be useful.
>
>Thanks in advance.
>
>respect,
>
>Robert