Kipp&Zonen BSRN Scientific Solar Monitoring System - Manual (Page 56)

44
5.0 Data Acquisition
5.1 Introduction
Installing and m aintaining the network data acquisition system (s) is crucial if consistent high quality
radiation data is to be sent to the archive. W ithin this m anual data acquisition system (DAS) m eans
those electronic devices (including the controlling software) and the connectors, which carry out the
process of m easuring the signals em anating from the radiation and ancillary m easurem ent devices
(transducers).
A DAS designed for operating and recording data from autom ated laboratory equipm ent is generally
suitable for radiation m easurem ent. Usually, these system s consist of four com ponents:
(1) the m ultiplexer sequentially switches across a num ber of input channels, each of which is
connected to one of the transducers that are to be m easured.
(2) the analog-to-digital converter (ADC), that converts the analog signal (e.g., voltage, resistance)
into a digital signal.
(3) the recording system , which m ay be a com bination of internal and external, buffers and
perm anent storage locations
(4) the controlling com puter(s), both internal and external, that handle sending control signals
to the m ultiplexer, the ADC and the storage based upon the user’s com m ands.
These m ay be com bined into a unit, m ay be separate units connected, for exam ple, by a General Purpose
Interface Bus (GPIB) instrum ent bus (e.g., HP, Fluke), or m ay be on a card that plugs into a PC that
is used as the overall control unit. Each of these arrangem ents has their own advantages and
disadvantages. The com bined system is m ore com pact and the program m ing m ay be easier. A system
with a separate com puter m ay allow for easier data analysis by allowing it to be accom plished on the
sam e com puter. Those that are totally separate norm ally have higher accuracies than com bined or
PC card system s and can be m ore easily updated if required.
Although the num ber of data acquisition or data logging products on the com m ercial m arket is enorm ous,
m any do not m eet the exacting specifications required by the BSRN. Annex F lists the nam es and
addresses of com panies that can provide system s that will m eet the general requirem ents im posed
by the BSRN. The m ost stringent of these requirem ents is the accuracy requirem ent of 1 :V on a 10
m V signal (0.01%) and the need to m ake 60 m easurem ents per channel per m inute. W ithin the system ,
the two m ajor com ponents that m ust be carefully considered with respect to accuracy and tim ing are
the m ultiplexer and the ADC.
Multiplexing is accom plished either by m agnet operated relay contacts or by sem iconductor switches.
Relay m ultiplexing is better for radiation m easurem ent because the relays contribute very little noise
(1-2 :V). Unfortunately, som e relay equipped systems are slow. Conversely, sem iconductor m ultiplexing
system s are m uch faster, but the noise or offset voltage m ay be greater than 15 :V. In either case,
settling tim e is required before the m easurem ent can be m ade.
In considering the ADC, the type and tim e of integration, the num ber of bits of resolution and the linearity
m ust all be considered. High-end, bench-type digital m ultim eters (DMM) are now capable of 24-bit
resolution and uncertainties of 10's of ppm under stable operating conditions. More rugged system s
usually consist of either 12 or 16 bit ADC. The form er does not provide the resolution, without regard
to the accuracy, required for the BSRN, while the latter still m ay not m eet the accuracy requirem ents.
In cases where the resolution is fixed to a 5 V scale, but the accuracy is adequate with respect to full
scale, the addition of a high quality instrum ent pre-am plifier at the transducer end of the signal can
increase the m agnitude of the signal to a level where the DAS can m eet the BSRN requirem ent. In
such cases though, the overall accuracy of the system is a linear com bination of the uncertainties of
both the DAS and the pre-am plifier. (A cautionary note: resolution does not equal accuracy).
Another m eans of reducing uncertainty for system s that otherwise m eet the resolution and tim ing
requirem ents is by calibrating individual data acquisition system s and then correcting for any nonlinearity
etc. found within the DAS. This process, while possibly saving capital funds, can be labour intensive
and requires that the DAS be calibrated under the conditions associated with the m easurem ent regim e.