In document A Mössbauer study of spin relaxation of ⁵⁷Fe ions (Page 171-174)

Mössbauer spectrometer


Ground 0 1 2 3 4 5 6 7 8 9 10 It 12 13 14 1.5 0 - r ! ' v ! ’ > 4 'S,i G round' " Display range 0 II 2 3 4 § 1 5 : 1 6 3 7 Z S ' 8 Z r\j 9 10 II 1 OP B IT S DR IIC « 2 T ransfer 0 d>.§! , c r o ; ,p.D I ' C ,r0 .0,?,' T 9 P Ob'! S.hP I ■ h?,1 I ■DiS -

ci°; s

D.e .l I % I o ^ . 0 W j 'S«!o I m CT 1

Y analogue c irc u it X analogue c irc u it

ure 5 The connections between the D AC and DR 11C ts for the velocity reference (X) and the display (Y) rials

contents o f this array are sequentially moved into the X register.

I he overall reference waveform is given by: Linear ramp

V — —512 + a for x= 0 to 1024.


V —512— 1024v + (32768/3077-) {sin (2ny) — 5/32 sin (4tti ) + 1 /256 sin (877-1)} where

y = ( . v - 10241/512 and ,y= 1024 to 1536. The actual reference signal is shown in figure 6, together with the error signal achieved using a velocity servosystem. Note the absence o f any spikes or oscillations which usually accompany discontinuities in other waveforms.

Figure 6 (a ) The velocity reference waveform generated by the computer and (/>) the error signal from the velocity servoamplifier with the transducer operating to ± 13 mm s_1

splay) signals. In programming, allowance lias been made I variations in processor speeds o f ± 10°/, (as quoted by : manufacturer), and for the fastest and slowest times ougli the controlling test loop. In the case of a double >p, computer input and output must occur in the interval -5 0 /is and 50 100 ps with respect to the clock, that is, 10 /<s and 40 90 ps with respect to the computer. Including

p processor variation these times become 0 36 ps and 81 /<s.

To ensure that the counters remain in synchronism with the :mory locations, the two binaries within the sequencing cuit are set once every sweep period using bit 15 of the (put register of DR I 1C #2.

( Analogue reference signal

lis signal (X) is used to control the velocity servosystems on cli experiment, and good long-term stability is required. I11 e data accumulation period the linear ramp is produced incrementing the output register o f the D R IIC # 2 and nsisls of 1024 steps. The flyback waveform is given by a liple Fourier series, the eoellieients being such that there 7 no discontinuities in the first seven derivatives o f the total crencc waveform. The values o f the flyback were calculated a desk computer, and read in as an array on paper tape iring program initialization. During the llyback period, the

3.4 Analogue display signal

The contents o f the displayed group were read in during the first part of the linear sweep period into the output register o f D R 1IC # \ . Display ranges o f 2,r\ 214, 212 and 210 full scale were achieved using a 12 bit four way multiplexor between this register and the digital to analogue converter (figure 5), that is. 12 bit accuracy on all ranges, except for the 2 " ’ range where only ten bits were available.

It is helpful to have some form o f marker on (he display, and the simplest method o f obtaining this is to use logic to pick every multiple o f 16 oil' the X digital buffer register. This pulse is used to brighten these channels on the oscilloscope display.

3.5 Input o f

To display or print out a particular spectrum, the computer must ascertain the starting location and the step between channels within the memory. The display group and grouping selector switches arc wired to represent a five digit binary number, and this binary number is obtained at the start o f the llvback cycle. I he number is placed in the upper byte o f the input register of DR 11C #1 at the start of the llyback period by using bit 15 o f the X register (this signal resets the sequenc­ ing binaries). This binary number is used to locate an entry in a table stored in the computer, giving the starting location

B Window, B L .Dickson, P Ronteliffe and K K P Srivastava and the steps for the setting o f the two front panel switches. This table was stored in the computer during program initialization.

3.6 Cleaving o f a spectrum

A selected spectrum can be cleared within one flyback period using the starting location and steps defined in §3.5. The ‘ Request A ’ bit of the D R 1 IC # I was set by a push button and initiated the clearing cycle.

3.7 Overflows oj"the memory

The use o f a 16 bit word length requires a method of determining the number o f times the counts accumulated in each channel exceeds the memory capacity, that is, 65535. Double length words could be used to increase the memory capacity, but with an increase in complexity o f programming, and in the time to punch out a spectrum. Recause o f the limited time available in the sweep, it is not possible without increasing the sweep time to just detect the carry from the lower word and increment the upper word. An alternative involving no time penalty within the sweep would be to scan the 2048 words in memory over a number o f flyback periods, and if the most significant bit is set. clear it and add one to the corresponding element in the array giving the upper words. This gives a memory capacity o f (2:tl I).

fn practice, the most significant bits will be the same for many of the data points in a spectrum, and the storage and processing of these extra digits for every number is inefficient. In the majority of cases it is only necessary to record how- many times the contents o f a single channel have exceeded the memory capacity o f 2lf\ and punch out this number. This is the method we have used.

The contents of the channels 9 16 arc scanned in the flyback and if an overflow' has occurred, the corresponding element of the block 1-8 is incremented. Thus the first channel o f a block o f 256 channels contains the number of times that 65536 should be added to the second channel to give the true total of counts. A block o f 512 channels has two such points.

One problem this method docs not solve is that o f spectra becoming ‘wrapped’ over a number of full scales o f the memory, and care must be taken to ensure that this ‘ wrapping’ does not exceed the capability o f a processing computer to unravel it.

3.8 Printing

The basic cycle time o f the spectrometer (13 Hz) is fast enough to allow one character per flyback period to be punched without the punching time for a spectrum becoming excessive. The program docs the following:

(i) Punches a leader. (ii) Punches a tape identifier.

(iii) Decodes the binary numbers by repeated subtractions to decimals, and punches the corresponding ASC II codes. Before the first number and after every eighth a carriage return, line feed is punched: alter every other number a space is punched. The punching o f a 256 channel spectrum takes two minutes.

The control of the punching involves the most complex programming and comprises the major portion of the program. Punching is initiated by a push button w hich sets a flag on the ‘ Request B’ bit of DR 11C # I . On completion of the punching, bit 14 is set under program control to clear the flag (figure 4). A number of safeguards arc incorporated into the punching sequence. If the displayed group, which is the group being punched, is altered during punching, the punching is stopped.

The punch flag also causes the counts to the particular sub­ group to be blocked for the entire punching period (figure 4).

3.9 Power fa il interrupt

One major advantage o f a minicomputer not shared by m ulti­ channel analysers is the power fail restart feature. I f the mains voltage drops below a certain level, the computer interrupts the program and allows time before switching off to ensure that when the power is restored, the program will recommence at an appropriate location. The memory contents are fully protected against disturbance from the power failure so that there is no loss of data.

4 Discussion

There are a number of features which, because of the versati­ lity of the computer based system, can be incorporated with little additional cost. There is ample time during the flyback period to perform further computations and control opera­ tions. Except when a group is being cleared, only 32 of the available 512 cycles of flyback are currently used, and even in the longest clearing operation, over 100 cycles are still available. The Mossbaucr program requires very little of the available 8 k o f computer memory; in fact, only 512 words are required for the program and 2-7 k words for the storage of data and other variables. This leaves 4-8 k words available for further computation and data storage.

Some o f this available computing power could be used to control experiments by sensing the experimental conditions, and to print out spectra at preset times. However, these modifications would considerably complicate the design and destroy one o f the spectrometers most valuable assets, its ease o f operation. At present, there is incorporated one easily provided facility; the computer by counting the sweep periods delivers a pulse on a spare bit o f the X register at a fixed time interval ( ~ 1 5 h) to initiate tasks such as the transferring o f liquid nitrogen into dewars.

This spectrometer also difTcrs from most other systems (except those from AERE Harwell) in that the reference signal is a ramp with flyback instead o f a symmetrical sawtooth. The only disadvantage of this system (not to mention its many advantages) is that 1/3 of the available counting time is wasted by the flyback period. To put this in perspective, if these extra counts were collected, that is, no time wasted in flyback, the relative error in the data would lie reduced by a factor of only I 2. A shorter flyback period could be used to slightly improve the signal to noise. As mentioned earlier, only 32 flyback loops arc needed to perform the necessary computation (aside from clearing spectra), and hence the flyback period could be made not 1/2 but 1/32 of the sweep period. In this case, the clearing would have to be spread over a number o f flyback periods, similar to the printing operation. DilTcrent waveforms to provide olfsct sweeps, again with no discontinuities, could also be read into the flyback waveform.

Because o f the sequential accessing o f the four sets of buffers, each of the four groups will have different centres for the velocity reference signal. For example, the centres of A, C, E and G w ill be separated by 1/4 o f a channel, and will be the same as for groups B, D, F and //, respectively.

5 Conclusions

The computer based Mossbaucr spectrometer satisfied the design objectives, and in three months' operation has proved to be exceedingly reliable anil easy to operate. Over 200 spectra have been accumulated without one occurrence of an obvi­ ously incorrect channel. Copies o f the program are available.


: advice of Dr Iain McLeod of the Department of Engin- ng Physics concerning design and construction of the :uit is gratefully acknowledged.


an A, Shoshani A and Montano P A 1970 Nucl. Instrum. 't h . 89 21 6

inshaw T E 1964 Nucl. Instrum. Meth. 30 101-5

odman R H and Richardson J E 1966 Rev. Sei. Instrum.


lvius G M and Kankeleit E 1972 Mössbauer Spectroscopy

/ its Applications (Vienna: IAEA) pp 9-84

►urnal of Physics E: Scientific Instruments 1974 Volume 7 inted in Great Britain © 1974

J I ' lm\ ( >„ v.» /i4/<. |9*’f*,.A '» l pp 4 4 ' Pcrp.tmpn Pn sv I’n ntci) in ( ir t ;il H iit.iin APPENDIX B


B. L. Dickson and K. K. P. Srivastava

Department of Solid State Physics. Research School of Physical Sciences. Australian National U niversity. Canberra. A C T. 260(1. Australia

tKecciied 17 April 1975: accepted 30 June 1975)

Abstract— Mössbauer effect spectra of Fc (Utilised into powdered synthetic spinel. MitAI 0 4. exhibit paramagnetic h\perline stiucture. 1 he iron was determined to be equally distributed between the tetrahedral and octahedral sites, with the tetrahedrallv coordinated iron having a longer spin lattice relaxation time. Analysis of the spectra taken in small applied fields between 30 and MUM) Oersted at 4 2 and I -5 K y ielded the electronic zero held splitting parameters and showed both sites to have significant dcpartuies from axial symmetry.

In document A Mössbauer study of spin relaxation of ⁵⁷Fe ions (Page 171-174)