Digital time-averaging

In the past, except for the low-temperature range, the uncertainties of noise thermometry were not comparable to those of the gas thermometry due to the non-ideal performance of detection electronics. Up to now, the most successful technique is the switched input digital correlator proposed by Brixy et al. in 1992 [89], In this method, the noise voltage is fed via two separate pairs of leads to two identical amplifiers whose output signals are multiplied together, squared and time averaged (see Fig. 9.10). 

When the commercial OMA (Model 1205, Princeton Applied Research Corporation, Princeton, NJ) became available, we recognized its potential as a replacement for the photomultiplier detector. The vidicon detector surface was divided into 500 channels, the image could be seen on a cathode ray tube (CRT) monitor in real time, the intensity profile was available in digital form, the profile could be time-averaged for any desired number of video scans, and the final profile was stored in internal memory for transfer to an external output device. Not only had a considerable amount of work gone into its development and the verification of performance, but its potential for use for a variety of physical techniques would ensure the construction of enough units to support further development of the system. Moreover the need for low-light-level detectors for other purposes would lead to further improvements in detector devices. 

For some mass spectrometers, such as TOF mass spectrometers, for reasons of data transfer speed and data storage capacity, the data acquisition system needs to accumulate data for a period of time in the summing memory, and forward the accumulated data to the data system. Each spectrum is added to the sum of the previous spectra so that a continuous summation process takes place. This type of data acquisition system is called digital signal averaging (DSA) or integrating transient recorder (ITR). Figure 3.8 illustrates the principle of mass spectrum acquisition with this type of system. 

Fig. 6-8. Time evolution of the pyrene anion (A ) and triplet pyrene ( A ) of the reaction (6-21) in methanol measured in the absence and presence of a magnetic field of 50 mT. For each wavelength and magnetic field, 8-10 transients were averaged with a transient digitizer (time resolution 0.5 ns). (Reproduced from Ref. [23b] by permission from The American Institute of Physics)...

The combustion air and fuel gas flow to the gas burner were measured with flow meters. The flow rates were corrected for temperature and pressure pertaining to experimental conditions. The waste feed rate is calculated from the speed of the feeder motor. Two digital temperature indicators are used for continuously monitoring the temperatures in the furnace and exhaust duct, respectively. The furnace temperature is measured using a type R [Pt—Pt/13% Rhodium] thermocouple of wire size 0.01 inch while the exhaust gas temperature is measured using a type K (Cr—Al) thermocouple having 0.025 inch wires diameter. A microprocessor-based data acquisition and analysis system was used to measure the instantaneous and time-averaged temperature in the combustion chamber and exhaust duct. 

Table 3-4 and Fig. 3-51 make evident that the dramatic effect of noise reduction drops off after about 50-70 cycles beyond that the noise reduction is relatively low considering the amount of effort required. Therefore, a combination of time averaging and digital smoothing is often more favorable. Note that the noise also decreases by /n when the number of points in smoothing operations increases. For this reason one comes... 

The best solution of the problem of digital filtering may be true arithmetic time averaging of as many independent scans (or other signals) as possible. A minimum of eight accumulated scans are recommended. If the SNR is small, then 32, 64, or more scans are necessary. The optimal number can be determined by independent repetition of this procedure. If the results are congruent, the best solution is given. 

NMR spectra were collected on a 40-MHz spectrometer of conventional continuous wave design employing 35-Hz magnetic field modulation. The spectrometer was interfaced to a Digital Equipment Corporation PDP 8/S computer that was used to time average the spectra. The spectra were punched on paper tape and further processed on an IBM 1130 computer. 

In the beginning a tape recorder was used to record the high speed transducer data. However, because of high noise level in the system, the data collection was interfaced with a modified PDP-15 dual processor digital computer. Comparing the signal-to-nolse ratio for both schemes, the former had a 14 1 ratio while the latter had a 250 1 ratio. The precision has been improved from about 10 torr for the tape recorder scheme to 0.3 torr for the computer scheme without time averaging the data. 

---