INSTRUMENTATION Modest computers, in the chemist's price range, would find many applications, making available information more useful
T^OR some strange reason, acoustical analysis, based upon t h e velocity of sound t h r o u g h a system, h a s n o t attracted as much attention as it deserves. I t is a n ancient a r t ; it was used in G e r m a n y prior to World W a r I in the "whistling-analyzer" wherein nitrogenhydrogen mixtures were blown through a n organ pipe. T h e resultant pitch, measured b y comparison with a variable-speed siren, was a function of gas composition. This is based upon t h e easily demonstrated principle t h a t a husky quarterback is bound to sing soprano if he fills his lungs several times with hydrogen or helium. M o d e r n techniques use t h e "ringing" or feedback principle. If a pulse is sent down the sample tube, picked u p b y a transducer, a n d t h e n after passing t h r o u g h a unit gain amplifier, is reintroduced to t h e front of the tube, the system will set u p a recurrent pulsing, t h e frequency of which is determined primarily b y the sound velocity in the specimen. An improved version of this idea is given b y Wilson [Rev. Sci. Instr., 2 5 , 927 (1954) ] in a n acoustic gas analyzer particularly suitable for the analysis of anesthetic (air, ether, oxygen) mixtures. Wilson's improvem e n t consists primarily in t h e mode of measuring t h e frequency. T h e frequency of t h e gas cell oscillator circuit is mixed with a secondary frequency in a n electronic mixer a n d filter which produces a difference frequency t h a t is measured b y a frequency meter. T h i s yields greater precision t h a n t h e a t t e m p t t o measure t h e absolute frequency. Computers Once more, we wish to bring u p t h e m a t t e r of computers, m a t h e m a t i c a l machines, or computing elements. F r o m all newspaper accounts, one m i g h t infer t h a t t h e subject is well in hand, VOLUME
26, NO.
t h a t all one m u s t do is describe his needs a n d t h e a p p r o p r i a t e "magic b r a i n " can b e t a k e n down from t h e shelf and p u t in operation. I n a sense this is true, b u t there is one small drawback. P r a c tically no one can afford these devices. W i t h few exceptions, these machines are in t h e " m e g a b u c k " category; their use is limited to vast programs on weapons developments, actuarial studies, a n d other beneficiaries of governm e n t a l largesse. Despite the expense, political " d o p e s t e r s " h a v e tried t h e m on two occasions in polling early election trends, with amusing consequences. T h e latter application illustrates two of t h e l a y m a n ' s principal misconceptions a b o u t these devices. T h e first is his unscientific reluctance to face facts, if t h e y are c o n t r a r y t o his preconceived notions or desires, a n d the second is the failure t o realize t h a t t h e most perfect computer can be no b e t t e r t h a n t h e reUability of t h e information which it receives from its h u m a n masters. One encouraging aspect of t h e p r o b lem is the increasing availability of components from which simple computing devices can be assembled. Despite this, no concerted effort has been m a d e to develop modest computers for t h e use of t h e chemist. Where would these things be used? A machine to solve linear simultaneous equations requires n o high-pressure salesmanship t o recomm e n d itself to the users of infrared spectrophotometers or mass spectrographs. These things h a v e been in use for a long time, b u t there are countless others which could find use and justification. Anyone can see the need for a "giant b r a i n " in computing t h e trajectory of a guided missile. T h e same goes for computations on nuclear energy levels. B u t let us look a t a few simple things. After all, it is only t h e simple questions in n a t u r e for which we h a v e no accurate
12, D E C E M B E R
1954
by Ralph H. Müller
information, only a constantly improving terminology. Our handbooks of physical a n d chemical d a t a , critical compilations, a n d reference works are crammed with information, most of it in poorly assimilable form. T h e r e are dozens of empirical equations relating two or more variables, let us say t h e density of a liquid as a function of its t e m p e r a t u r e . I t does not require a very complicated computer to handle expressions of this sort and, for small additional cost, it can p r i n t results, interpolated to a n y permissible degree of subdivision, on t h e pages of a book directly, in order t o avoid errors of transcription. This has been done before. During t h e war, elaborate navigational tables were recomputed from improved formulas a n d furnished improved d a t a as well as eliminating old errors. Again, without approaching t h e "diamond-inlaid" models, one could devise computing devices t o bring more light t o bear upon some of these empirical expressions. Making
Data
Useful
I t is a t once astonishing how m u c h scientific information is available in our compendia of d a t a , and also the high degree to which m u c h of it is in an awkward and unassimilable form. A n y protracted research problem, involving physical a n d chemical d a t a , is bound to encounter delays wherein some required information from t h e literature m u s t be recalculated t o t h e conditions u n d e r which it is to be used. T h e r e are n o table exceptions, of course. F o r one 35 A
Your Spectrophotometer is ot its
mt when input line is regulated by Sorensen! A scientist, working with a spectrophotometer in a government laboratory, recently asked us to cure a condition of instability which was preventing his making accurate or reproducible analyses with the instrument. Despite the use of another type of regulator, line voltage variations were causing iandom fluctuations of the spectrophotometer meter. This was readily understandable. The light source of the spectrophotometer was an incandescent lamp, and lamp intensity is an extremely sensitive function of voltage, being expressed by the formula Lamp Intensity = KE 3 · 5 The existing regulator arrangement allowed voltage variations of ± 1 % , resulting in lamp intensity variations of approximately 7%—sufficient to seriously affect readings. Sorensen's fully satisfactory solution was to replace the other regulator with a Sorensen Model 500S AC Line Regulator. This instrument has a regulation accuracy of ± 0 . 1 % with nearly instantaneous recovery time, and effectively eliminated line voltage variations as a source of significant error. A comparison of the spectrophotometer operations under the two input arrangements is plotted below.
AC LINE VOLTAGE
W i t h scientists' and technicians man-hours at a premium, the small investment required to provide a stable input line to power delicate and costly laboratory instruments represents true economy. W e invite your inquiry concerning the application of Sorensen electronic power controlling devices to your particular problems. Write directly to Sorensen & Co., Inc., 375 Fairfield Ave., Stamford, Conn. In Europe, correspond directly with Sorensen A. G., Gartenstrasse 26, Zurich 2, Switzerland.
SORENSEN SORENSEN & CO., INC., 375 FAIRFIELD AVE., STAMFORD, CONN. For further information, circle number 36 A on Readers' Service Card, page 41 A
36 A
INSTRUMENTATION practical reason or another, extensive tables have been available for the potential of hydrogen, glass, and quinhydrone electrodes over a wide temperature range. The same is true for the e.m.f. of various thermocouples as a function of temperature. In many cases, these extensive tables have been prepared by instrument companies as a useful and time-saving adjunct to the instruments which they manufacture. It might be useful to consider some ways in which a general approach to this question could be set up. In the first place, it may be assumed that more and more of our newer instruments will be designed in such a manner that assimilation of data will be an inherent feature and that, as far as possible, the information which they furnish will require no further computation or correction. This leaves us, nevertheless, with the large mass of empirical information already available but requiring translation into more directly useful form. Conventional business computing machinery, of which IBM machines are the outstanding example, present a completely satisfactory solution, as far as the most exacting computational requirements are concerned. The one qualifying consideration is the fact that company policy, as far as these machines are concerned, is that they are rented, at carefully computed rates, rather than sold outright. It is an obvious consequence that such machines must be used more or less continuously to justify their cost. If a program were set up to reduce any set of tabular data to the most useful form, with interpolated values computed to the least permissible value, it could all be done with such machines. This includes the automatic typing of data in final form, from which electrotype plates could be made without fear of transcription errors. An alternative approach could be achieved with mechanical or electrical analog computers. These do not possess the almost absolute precision of the more elaborate digital computers, but, to a limited degree, this defect could be minimized if they were employed for short-range interpolation. Most of the simpler mathematical functions can be duplicated electrically: addition and subtraction by parallel cathode followers; multiplication, division, and ratios by a servo-driven Wheatstone bridge; while potentiometers with specially shaped cards can provide logarithmic, sine and cosine, and numerous ANALYTICAL
CHEMISTRY
INSTRUMENTATION
N·M·R
Completes Your Spectroscopy Team Perplexing chemical problems which resist solution by other methods ma) often be solved by applying the revolutionary High Resolution n-m(nuclear magnetic resonance) technique — a powerful new member of the spectroscopy team. Varian n-m-r Spectrometers are being used daily ir many laboratories for detailed functional-group analyses of organic anc petrochemical mixtures. And in such rapidly expanding fields as silicon boron and fluorine chemistry, n-m-r is providing penetrating insight intc molecular structures. The newly-styled Model V-4300A High Resolution n-m-r Spectrometer nov, incorporates an exclusive Varian "sample spinner." By mechanically rotat ing the sample during signal observation, tiny magnetic field inhomoge neities can be averaged out. Result: the "resolving power" of the Spectrometer is extended to a higher degree than ever before.
DATA
C) 18 mol percent hexametnyltrisiloxane Signals observed: H l Frequency. 30 mc. Sweep Rate: 10 milligauss/sec.
Zero of reference: H22u0 eterence.n
H
FOR C O M P L E T E
INFORMATION...
O n Vorîan's V-4300A High Resolution n-m-r Spectrometer, as well as the Model V-4200 Variable Frequency n-m-r Spectrometer, and associated magnet systems, contact the Special Products Division for:
THE MARK OF LEADERSHIP
V A R I A N PALO
A L T O
MICROWAVE
4,
· · ·
Technical Bulletins Dafa Sheets n-nw Tabie
associates C A L I F O R N I A
TUBES — S P E C I A L
PRODUCTS
For further information, circle number 38 λ on Readers' Service Car·", page 41 A
38 A
other functions. Combinations of these elementary devices can handle more complex functions. For our purpose, the principal problem would not be how to solve any given equation, but rather to examine physical data which are to be collated and see what simple computational array could be set up to handle them. It is more than likely that a large class of data can be handled by one simple class of analog computer. Thus, a large mass of data could be processed at low cost. Another class of simple computer might be expected to be suitable for other groups of data. The moment one asks for complete versatility, he is likely to run into high cost. Storage Although simple functional relationships are easily set up with analog devices, even the simplest calculations often require temporary storage of numerical information. Storage is one of the important factors in all large and expensive computers, and there are many practical solutions to the problem. There are not so very many cheap solutions to this problem, and what we have been discussing might well require further investigation in order to provide economical storage elements. Of the various storage elements, such as magnetic tape, magnetic drums, cathode ray tubes with external wire gauze electrodes, and ring elements of special magnetic materials, the latter arc the most promising. They consist of very tiny rings or toroids made of Deltamax or related magnetic alloy material. These are characterized by an almost rectangular hysteresis loop and require extremely small magnetizing force to reach saturation. If such a loop is used as the core of a small transformer, then a negative pulse through one of the windings will induce an output pulse in the other winding and also saturate the core. Additional negative pulses will produce no further output and, in effect, one "bit" of information has been stored in the core. At any later time, this information can be "read out" by interrogating the system with a positive or reading pulse. This will produce one, and no more, output pulse, and the core has been returned to its original state—ready at any time to receive another bit of information. The discovery or development of other schemes for information storage is a large and promising field of investigation in itself. ANALYTICAL
CHEMISTRY
