Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
z
This application is a divislon of our Canadian paten~ application
Serial no. 35~,~97 filed Jwle 20, 1980.
This invention relates to circ:uitry for analog-to-digital and
digital-to-analog conversion and similar operations.
The digital circuit is usually faster operating, and thus the time
required to complete each sequence of steps is largely controlled by the
response time of the analog cir~uit. Por example, if the analog circuit
includes a filter for smoothing out ripple and noise, it ma~ require a few
seconds for the output of the analog circui~ to reach a new value. When thess
few seconds are multiplied by the number of times the sequence is repeated the
lost waiting time can be considerable. We have discovered that the time lost
~aiting for the analog circuit to respond can be shortened b~ dissipating
~e.g., with a switch connected in parallel) the charge-storage device ~e.g.,
capacitor) in the analog circuit. After dissipation, the analog circuit and
charge_storage device are allowed to respond normall~.
This invention relates to light-beam deflection instruments, such
as refractometers. In such instruments, a light source typically directs a
beam through a test cell to a detector (e.gO, a photocell). A change in some
physical quantit~ (e.g., refractivit~) causes the light beam to move with
respect to the detector. A difficulty is that a change in the position of
the light source (e.gO~ from movement of the filament in an incandescent bulb)
often cannot be distinguished from a change in the physical quantity being
measured, as both cause movement of the light beam with respect to the detector.
~e have discovered that the position of the light source can be stabilized with
respect to the zone (e.g., test cell) in ~hich the measurement is made by
modulating the light beam position (e.gO, cyclicall~ sweeping it back and forth)
through a preselected amplitude which is independent of light source movernent
or other uncontrollable movemen~s of the light beam ~such as from thermal
eddies)O Light source movements only va:ry the phase of the modulation ~e.g.,
the time at which the cyclical sweeping ~tarts and finishes), and khe electron-
ics which process the detector output can easily be designed to ignore such
phase changes, such as by determining the average position at ~hich the beam
strikes the detectorO The invention provides greatl~ reduced sensitlvity ~o
light source movements, and can be inexpensively implemented in preferred
embodimentsO In a second aspect, the invention features an unfocused light
path in a direction perpendicular to the direction that the light beam moves
during a measurement. Light from many points of the light source can thereby
be spread evenly across this unfocused direction~ there making the measurement
insensitive to variations in brig'ntness of the light source along this direction.
This invention relates to devices which depend upon accurate
measurement of changes in the refractivity of a flo~ing fluid, e.g.~ in liquid
chromatographyO Because temperature affects refractivity, the temperature of
the flowing liquid must be carefully controlled. Typically in refractive
index detectors used in liquid chromatography the refractivities of sample and
reference streams are compared at a cell. One technique for equalizing
temperatures in the sample and reference streams in the cell has been to supply
the fluid to the cell after having passed through sample and reference inlet
tu~es in heat-exchanging relationship with each other and ~ith a large metal
blockO Ideally, both streams are at equal temperatures before entering the cell.
We have discovered that better control over temperature can be achieved with
a heat-exchanging relationship bet~een the sample inlet ~to the cell) and sample
outlet ~from the cell) streamsO The required hard~are is simpler and inexpen~
-2-
siveO Excellent tenlperature equalization of sample and reference fluids, and
ver~ fast warm up and cool down of the device, are made possible.
We have discovered that increased light throughout can be achieved
in the cell of a refractometer by placing an integral reflective layer on or
within the cell, to reflect the light beam back through the cell. Parallax
bet~een the flo~ cell chambers and the reflective surface is reduced. Fewer
surfaces are exposed to the ambient~ thereby reducing losses due to dust
buildup and surface reflection. Manufacturing is simplified as fewer parts are
requiredO
~e have also discovered an improved zeroing technique which is
relatively insensitive to changes in light intensity, and which, in preferred
embodiments, eliminates optical zeroing. In gener~l, ue include in the system
output a main term dependent on the difference between the two optical
measurements and an offset term dependent on a~ least one such measurement.
Any change in light intensity will substantially equally affect the main and
offset terms, so as to maintain zeroing accuracyO
~e have discovered that increased light throughout can be achieved
in the cell of a refractometer by incorporating an integral curved surface
hith the cell) to ac* as a lens for focu~ing the light beam. Parallax
bet~een the flo~ cell chamber and the lensing surface is reduced. ~ewer
surfaces are exposed to the ambient; thereby reducing losses from dust build-
up and surface reflection. Manufacturing is si~plified as fewer parts are
required. In a second aspectJ our invention features incorporating an opaque
mask within or on the flow cell.
~ccording to a broad aspect of the inventlon there is provided ln
apparatus for measuring changes in the refractivity of a fluid flowing
.
through a temperature~controlled cell, tha~ improvement wherein air inlet and
air outlet tubes for respecivel~ carrying sample fluid to and from said cell are
mounted in a heat-exchanging relationshiy with each other.
The structure and operation oE a preferred embodiment of the
invention will now be described, after first briefly describing the drawings.
Drawings
Figure 1 is a perspective view of said embodiment.
Pigure 2 is a partiall~ cross-sectional view of said embodiment.
Figure 3 is a cross-sectional view of 3-3 of Figure 2, showing the
photocell end of the optical bench.
Figure 4 is a cross-sectional view of 4-~ of Pigure 2, showing the
flo~ cell end of the bench and the outer insulating cylinder and shields, with
internal heat shield/light baffle 77 removed.
Figure 5 is a cross-sectional view at 5-5 of Figure 2, sho~ing the
light source.
Figure 6 is a schematic of the heat exchanger plumbing.
Figures7a and 7b are cross~sectional views through the sample and
reference heat exchangers~ respectively.
Figure 8 is a cross-sectional view at 8~8 of Figure 4, showing
construction of the flow cellO
Figure 9 is an elevation view of the back surface of the flow cell
at 9-9 of F~gure 8~
Pigures lOa and l~b are diagrammatic views of the optical path
through said embodiment.
Figure 11 is a block diagram of the electronic~ that process the
outputs of the photocells.
_~ _
3Z
~igures 12a and 12b are sche~latics of the electronic clrcuits that
- null the photocell output and process the nulled output for display and
integration.
Structure
Turning to Figure 1, optical bench 10 is supported inside an oven
on four insula~ing posts 12 attached to floor 14 of the ovenO I.ight source 16
for the bench is positioned belo~ the bench and outside of the oven. Piber-
optic cable 18 carries light from source 16 to the benchO Sample liquld from
the outlet of a chromatographic column ~not shown) positioned inside the
oven flows into the optical bench through inlet tube 20 (0.009 inch ID), and
out through outlet tube 22 ~0~0~0 inch OD). A small diameter sample inlet
tube is used to minimize band spreading in the chromatogram. Similarly,
reference liquid flows into and out o the bench through inlet tube 24 ~0.020
inch ID) and outlet tube 26 tO.040 inch OD~ All four tubes are stainless
steel and haYe 1/16 inch outside diameters. The outlet tubes have larger
internal diameters than those of the inlets to lower backpressure and its
effect on refractivit~u Outlet tubes 22, 26 are connected together ~ownstream
of the optical bench to equalize sample and reference pressures a-t the flow
cell. Electrical wires 28 from photocells 52 (Figure 2) lead from the bench
to processing circuits shown in Figures 11, 12a, and 12b.
Turning to Figures 2 through 4, optical bench 10 consists of an
inner cylinder 32, ~hrough which a light beam B is passed, and a concentric
outer cylinder 34, which provides an insulating air gap 36. Two flat shields
38 ~Figure 1) retard radiation of heat to and from the ~ench, and act as legs
~Figure 4) to support the cylinders, via four bolts 40, on posts 12. End
caps 4~j ~4 close each end of outer cylinder 34, and end caps 46, 48 each end
. ' :
'3?.,
of inner c~linder 32. End cap 46 supports elongated outlet 5D (0.050 inches
wide b~ 0.35 inches high) of fiber-optic cable 18 and photocell 52. End cap
48 supports flo~ cell 54 via cell bridges 56, 58, which are attached to the cap
and each other by screws and epoxyO Sample inlet and outlet tubes 20, 22
terminate at bridge 58; reference tubes 24, 26 terminate at bridge 56. Recess
60 in end cap 48 behind the bridges contains about four coils of sample inlet
tube 20. Notches 62, 63 in inner c~linder 34 provide entrywa~s for the tubes.
The end caps, cylinders, and shields are all made from aluminum, to speed warm
up of the bench while also insulating the bench by virtue of air gap 36
bet~een the c~linders.
Turning to Figures 8 and 9, 1OW cell 54 has two hollow chambers
70, 72, for the sample and reference llquids, respectively. Each chamber has
a triangular (about 45x45x90 degrees, 0O062 inches on each short side) cross
section (Figure 8), and is connected to its respective inlet and outlet tubes
b~ internal passages 73. The height (or vertical dimension in Figure 9) of
the chambers is about 0050 inches. The flo~ cell is manufactured by fusing
together, without adhesive, pieces of borosilicate glass. Teflon ~rade~ k)
seals 75, compressed against the cell by the bridges, provide a seal between
the sample and reference tubes and internal passages 73 of the cell. The
front surface 74 of the flo~ cell is ground to provide an integral lens that
has curvature in horizontal but not vertical planes. The back surface of the
cell has a reflective surface coating 7$ of gold to provide a mirror to reflect
light back through chambers 70, 72 to photocell 520 The focal line of the
lens is positioned at photocell 52, and the spacing bet~een mirror 78 and
photocell 52 is about 6.0 inches. As shown in Figure 9, the mirror coating 78
~s limited to approximatel~ the area directl~ behind chamber 72, thereb~ to
limit reflectlon principally to llght p~sslng ~hrough the ~riangula~ chan~bers.
Other light is c~sor~ed by ~lack epox~ coatlng 76 applied over and around the
mirror coatingO The mirror coating :Is slightl~ larger than the chambers to
accommodate variations in the internal size of chambers 70, 72. The coating
stops short of the ~op of chambers 70, 72 ~igure 9) so as not ~o re~lect light
passing through the top of the chambers, where bubbles might form.
To reduce radiant and convective heat transfer to the flow cell
from within the optical bench, a blackened disk 77 with rectangular light-beam
aperture 79 ~just large enough to expose the flow cell) is positioned ahead of
the flow cell. This disk also serves as a light bafle, and is tilted down
10 ~igure 2).
Sample and reference liquid are brought into the flo~ cell through
sample and reference counterflo~ heat exchangers 9O, 91 (~igures 1, 6, and 7),
each of ~hich are formed by bonding corresponding inlet and outlet tubes
together inside a tubular jacket, Each bonded pair is then routed along a multi-
~one path beginning outside the bench and ending at the flo~ cell. As shown in
Figure 7, sample heat exchanger 90 i5 constructed by placing tubes 20, 22
inside a tubular copper braid 80, heat shrinking a polyeth~lene tube 82 over
the outside of the copper braid, and illlng the interstices between the braid
and the inlet and outlet tubes with a lo~-viscosity, moderatel~-heat-conductive
epoxy 84 ~StycaSt 3051) ~-Trade ~ark). ~eference tubes 24, 26 are bonded without
a copper braid b~ inserting the tubes inside a Teflon tube and filllng the tube
~ith the same low-viscosity epox~ as used for the sample tubes. The braid is
omitted because less efficient heat transfer is needed for the reference, as
it does not flo~ during measurement, but onl~ during flushing between measure~
ments
`:'
:,
. '
312
Turning to Figure 6, the multL~zone path ~ollowed by the heat
exchangers is sho~n diagrammaticall~, The first zone for both sample and
reference heat exchangers begins outside the optical bench and extends along
the outside length of the bench bet~een outer c~linder 3~ and shields 38 ~total
zone length abou~ 8 inches). The sample heat exchanger 90 is positioned on
the side of the bench closer ~o the center of the oven, where temperatures
are better controlled. At end cap 42, both heat exchangers turn 180 and
enter gap 36 between cylinders 32, 34, through a slot ~not sho~n) in the end
cap. The second zone for both sample and re~erence extends along gap 36 ~total
length about 7 inches~0 The reference heat exchanger goes directly from gap
36 into cylinder 32 through notch 62 in the end of cylinderO Inside cylinder
32J the reference inlet and outlet tubes are brought directly to flow cell 54
via bridge 56.
The sample heat exchanger 90 continues into a third zone beyond the
end cap, where it is bent into coil 92, consisting of four turns ~total coil
length about 24 inches) positioned in the space behind end cap 48. The last
coil is adjacent to the back of the end capO From the coils the sample heat
exchanger enters cylinder 32 through notch 63. Inside the cylinder, sample
outlet tube 22 is connected directly to the flow cellO Sample inlet tube 20
is ~ound in another coil 94 ~total length about 12 inches) before entering the
flow cellO Coil 94 i5 positioned in recess 90, and potted with a heat-
conductive epox~ to provide good conductivity with the end cap and cell bridges.
Turning to Figures 2 and 5, light source 16 includes an incandescent
bul~ 100 ~Phillips 6336, H3 base, 6 V, 55~, operated at 4.8 V) with vertically-
extending filament lOl, a concave light-focusing mirror 102 ~gold-coated glass),
and a rotating prism 1040 rhe prism is rotated at about 50 to 60 rpm along an
axis parallel to thc filament axis ~ a shaded~pole AC motor 1060 ~h- motor
also drives a fan 1~8 ~hich supplies cooling air to the bulb. Prism 104 is
about 0.37 inches high, is made of glass~ and has a rectangular cross-section.
Two opposite surfaces 110 of the prism are clear and about 0O3 inches wide.
The other two surfaces 112 are opaqued ~ith a white opaque silicone rubber,
and are about 0O25 inches wide. Fi~er-optic inlet 11~ is round (about 0.150
inches in diameter) and is positioned opposite the prism from the bulb. Mirror
lQ2 is positioned so as to focus an image of filament 101 on the face of inlet
114. Bulb 100 has a peak output in the near infra~red spectrum at a wavelength
of about 1000 nanometers.
~iber-optic cable 18 is broken internally into sub-bundles and the
sub-bundles are intentionally disordered at one end to randomiæe the light path
between inlet 114 and outlet 50.
Photocell 52 has two adjacent triangular dual photo_voltaic cells
180, 182 ~gold-bonded silicon) arranged so that their long dimensions extend
horizontall~, which is the direction of movement of the light beam. Each
triangle is about 0.150 inches long and 0O05 inches high. The spacing between
the triangles is about 0.008 to 0.010 inchesO The shunt impedance at operating
temperature ~about 150 C) is maximized, as is the sensitivit~ to long wave-
lengths.
The oven in which the optical bench resides is heated by proportion-
ally~controlled electrical resistance elementsO Within the oven, temperatures
can var~ as much as 5 to 7 C from point to point~ but by much less ~e.g., 0.3 C)
at the same point over timeO The time period during which the resistance
elements are on is varied in proportion to the difference between the actual
oven temperature and the desired temperature and in proportion to the integral
z
of this differenceO To make oven temperature less sensitiue to variations in
AC line volatge the time period is also made inversel~ proportional to the square
of the line voltage, as the heat generated by the elements is proportional to
the square of the line voltage. The elements are SCR controlled, and are
turned on and off only at zero crossings of current.
Figures 11, 12a, and 12b sho~ the electrical circuits that process
the outputs of photocell 52. Figure 11 sho~s the overall circuitry in block
diagram formO Panel inputs 118 ~e.g., recorder gain) are fed to a central
processor 1200 The central processor ~CPU) initiates the automatic electrical
zeroing (nulling) of the photocell outputs, and sends signals via buffers 122
and gain latches 124 to circuitry sho~n in Figure 12b to set the gain for
display of the chromatogram on a recorderO An analog data acquisition voltage
~D.A.V.) is converted to digital and sent by the processor via the input
buffers to a panel displa~ 126.
~igure 12a sho~s the circuitr~ for electrical zeroing. The current
outputs ~AC signals~ oE photocells 180, 182 are ~rought via shi01ded cable to
current~to-voltage converters 130. The AC voltages A, B produced by the con-
verter are summed and amplified b~ a gain of 202 at amplifier 132, to form the
expression -202~A+B), ~hich is called SUMo Amplifier 134 subtracts voltage A
from voltage B, and adds to the difference the sum of three voltages: SUM,
FINE ZERO~ and COARSE ZEROo The latter t~o voltages are produced b~ multi-
pl~ing SUM by a negative scale factor. Thus the output of amplifier 134
~ZEROED W TPUT~ can be expressed as
~B - A) ~ 2~2~o33 ~ o67KC - oO033K~) ~A + B)
~here KC is the coarse zero scale fact~r and Kp ~s the fine zero scale factor.
~cale factors Kc, ~ are set bet~een about zero and about one by the digital
-10-
circuitry of block 136, ~henever a signal is; sent across the AUTOZERO CO~MAND
lead. Normally zeroing would he done ~efore a chromatogram was generated, but
can be done at any time.
The above expression for the ZEROED OUTPUT can be presented in
simplified form as
~ B - A) -K~A ~ B)
l~here K is the overall scale factor. The expression is independent of
variations in the overall brightness of the light beam striking photocell 52
because the zeroing term ~K~A ~ B)) is not a constant, but, like the difference
term ~B - A) is proportional to beam brightness. For example, if the brightness
~ere to rise by 10%, both the difference term and the zeroing term would
similarl~ rise b~ 10%, and thus the ~hole expression would still remain equal
to zeroO ~hen beam deflection does occur~ as the result of refractivity
changes, the zeroing term remains roughly constant because of the complementary
shape of the t~o cells 180, 182, ~hich at anr horizontal location have roughly
the same combined vertical heightO
T~o successive-approximation registers 138, 140 drive a pair of
digital~to_analog converters 142, 144 to form the FINE ZERO and COARSE ZERO
signals~ Each of converters 142, 144 multiplies the SU~ signal by a scale
factor set by the digital output of registers 138, 140. Registers 138, 140
follow a conventional successive approximation algorithm to select the digital
outputs or scale factorsO About once a second, the registers receive a clock
pulse from chip 148, ~hich produces a slo~ clock from the much faster processor
clock signalO At each clock pulse, the output of a register is adjusted in
res;ponse to the output of comparator 146 ~hich indicates ~hether the applied
~INE~COARSE ZERO signal is too large or too small. The input tQ comparator 146
is the DC OUTPUT, produced at filter amplifier 150 (~igure 12b). A ~I~rER
RESET connection ~et~een the zeroing circuitry and filter amplifier 150 is
used during the zeroing proces~s to discharge capacitors in the filter and
reset the DC OUTPUT to zeroO This, allows for a more rapid autozero sequence.
Register 138 works first to set the coarse scale factor Kc, and then register
140 to se~ the fine scale factor ~ , I'he AUTOZERO COMMAND is used by the CPU
to start the autozero sequenceO The AUTOZEROING signal is used to alert the
central processor that the refractometer is autozeroing.
Turning to Figure 12b, there is shown circuitry for processing the
ZEROED OUTPUTo Amplifier 152 raises or lo~ers the signal level in response
to command signals 151 from the central processor 120 via the data latch 124.
Demodulator 154 ~ith the help of phas,e computing block 153) converts the AC
signal to DC, and filter amplifier 150 smooths the PC signal. Switching block
156 operates during zeroing to turn off the RECORDER and INTEGKAT~R s,lgnals.
It also is used to change the polarity of the DC signal in res,p~nse ~o a
POLARITY signal from the central proces,sor 120 via data latch 124. Downstream
of hlock 156 the DC signal is processed by amplifier 158, and supplied to an
integrator output leadO The DC signal is also processed by attenuator 160,
under control of the central processor via signals 162. The attenuator
produces a recorder output 164, ~hich is supplied to a recorder output terminal
and to amplifier 166, and a data acquisition voltage (D.A.Vo~, which is
supplied to the central processor for panel displayO Block 168 supplies a mark
signal for the recorder in re~ponse to the AUTOZERO COMMAND, to indicate on the
chromatogram the point at which the sample injection occurs. The CPU issues
the AUTO~ERO COM~AND at the time of sample injectionO
Operation
5`2
In operation, the oven surrounding the optical bench and chromato-
graphic column i~ turned on, and about an hour and one half warm up period is
allowed for temperature equali~ation ~Yithin the bench. After warm up, solvent
is pumped through the sample and reference circUits wlthin the bench. When
solvents are changed, sufficient time is allowed for flushing both circuits.
Flo~ is then stopped in the reference circuit ~but reference chamber 72
remains filled ~ith reference liquid), A sample is then injected into the
s~mple column. The electrical output of the refractometer is zeroed b~ init-
iating the automatic zeroing sequence described above. Sample passes through
the chromatographic column and into the optical bench. Generally speaking,
variations in refractivity of the sample cause movement of the light beam wlth
respect to photocell 52, and thereby change the electrical output, ~hich is
plotted against time on a chart recorder, producing a chromatogram.
Temperatures ~ithin chambers 70, 72 of the flow cell are maintained
within about 0.0001 G of each other during operation to minimize error. A
temperature difference bet~een the t~o flo~ cell chambers results in a refrac-
tivit~ difference. Temperature equalization is achieved by providing good
thermal insulation around the flo~ cell, in the form of air gap 36 bet~een the
inner and outer cylinders~ shields; 38, and blackened disk 77; surrounding the
flo~ cell ~ith a thermal mass, in the form of bridges 56, 58 and end cap 48;
and directing incoming sample flo~ through a very efficient counterflow heat
exchanger to bring the temperature of the sample to the flo~ cell temperature.
Incoming sample upstream of the heat exchanger is typically as much as 1 C
~and possibly 2 to 3 C) different in temperature than ~he floN cell because
of spatial differences in oven temperature and because of heat generated by
viscous heating inside the inlet tube. This difference in temperature is
-13-
Z'
graduall~ reduced ~long the le~gth of th~ heat exchanger b~ thermal conduction
between the inlet and outlet tubes. At t:he end of the heat exchanger, whatever
very small temperature difference remains is minimized by heat transfer
between end cap 48 and coil 94 just prior to entr~ into the flow cell.
The sample heat exchanger is divided into three zones to improve
its efficiency, with each successive zone being more thermally stable and
closer to the temperature of the flo~ cell. The construction of the heat
exchanger provides good thermal conduction between tubes but very low conduction
along the flow direction of the tuhesO There is significant heat transfer
bet~een the tubes and the surrounding air; thus thermal interaction between
the heat exchanger and the region surrounding it must be considered. The first
zone, between outer cylinder 34 and shield 38, provides a gradual approach in
temperature before the heat exchanger enters the optical benchO The length
of this zone is greater than 10% of the length of the sample inlet tube within
the bench. Without the first æone, iOe , if the inlet and outlet tubes were
joined jus~ outside the entry to end cap 42, there would be a steeper
approach in temperature along the heat exchanger, and much of this approach in
temperature would occur along portions of the heat exchanger inside air gap
36, thereby undesirably transferring heat to or from the bench. With the
preferred arrangement of a first zone outside the bench, the heat exchanger
temperature is closer to that of the bench when entering the air gap.
The heat exchanger enters the gap at the photocell end of the benchJ
thereby assuring that whatever heat transfer to or from the bench does occur is
at a location ~ell separated from the flo~y cell.
This same concept of routingthe heat exchanger through increasingly
more thermall~ stable regions is also applied to the second and thîrd ~ones.
-14-
z
In the s.econd zone, the sample heat exchanger is directed along air gap 36
from the photocell end to the ~low cell end, where temperature stability is
highesta In the third zone, the sample heat exchanger is coiled behind the
flow cell end cap, with each successive coil being closer to the end cap and
flow cell.
As a final step, the sample inlet tube alone is coiled in recess 60
of end cap 48 to minimize ~hatever small temperature difference remains between
the ;ncoming sample and the flo~ cell.
Because the reference solvent does not flow during a measurement,
the reference heat exchanger is less sophistlcatedO It lacks the third coiled
zone, and has no copper, heat-conductive braid to surround inlet and outlet
tubes. Limited heat exchange is; provided on the reference side to maintain
raugh ternperature equalization during flushing of the reference circuit,
thereb~ shortening the period needed ta stabilize temperatures after flushing.
The optical elements of the re~ractometer are shown diagrammatically
in Figures lOa and lOb. For clarit~ the optical path is sho~n unfolded, with
mirror 78 treated as a ~indo~O Figure lOa sho~.~ a horizontal sectlon through
the optical path; Figure lOb sho~s a vertical section.
Turning to Figure lOA, a single light ra~ B is shown to illustrate
beam movements. Lens surface 74 on flow cell 54 focuses the light emerging
from fiber-optic cable outlet 50 onto photocell 52. The focused image on the
photocell is shown diagrammatically in the views on the left side of the Figure.
To illustrate the effect caused b~ rotation of prism 104, four views ~A through
D) of the prism in different angular positions are shown along ~ith the
corresponding positions of the light beam on the photocell.
Light pass;ing through chambers 70~ 72 is bent in proportion to the
-15-
~ 2
difference in the refractive index of the liquids in the two chambers.
Referring to ~igure 8, the chambers- are conventionally constructed so that
surface 190 in chamber 70 is parallel to surface 188 in chamber 72 and,
similarly, so that surfaces 184 and 186 are parallel. These four surfaces are
the four at ~hich light is bent by refraction. If the liquid of the same
refractive index is in both chambers, light will be bent by the same amount at
each of the corresponding parallel surfaces, and will emerge from the flow
cell along a path B which is essentially unaffected by changes in refractivity
c~mmon to both chambers. If liquid ln the two chambers differs in re-
fractive index, light ~ill be bent differentially at these parallel surfaces,
and ~ill emerge along a path skewed from the equal-refractivity path. Such a
condition is illustrated in Pigure lOa by light ray Bl~ The amount by which
the light beam is skewed or bent at the flow cell is measured by detecting the
position of the image of the beam at photocell 52~ The difference between
the electrical outputs of the two triangular cells 180, 182 can very finely
resolve the horizontal position of the light beamO Imperfections in alignment
of the photocell with the flo~ cell and other tolerances in the system typically
cause these electrical outputs of the two cells to be unequal even when sample
and reference liquids have the same refractive index. This initial electrical
difference is nulled by the automatic ~eroing procedure described above.
Ideally, the light beam location on the photocell 52 should only be
a function of the difference in refractive index between sample and reference
(and not a function of the location of bulb filament 101). To achieve this,
the light intensity distribution across the fiber optic outlet 50 must be
spatially stable over the time period of chromatographic interest ~1 second to
several hours)O This requires that the light intensity distribution into the
-lG-
t 9Z
fiber-optics be stable. As viewed from thle fiber-optics inlet 114, the apparent
position of bulb filament 101 varies due ~o filament distortion and thermal
eddies in the air path bet~een the filament and the inlet. ~ilament movements
along the length of the filament ~vertical in ~igure 2~ are relatively non-
criticalO Similarly, changes in the filament distance from the fiber-optics
inlet are not observable and thus are noncritical. Along the third axis of
movement ~vertical in ~igure 5) the apparent filament location as viewed by
the fiber^optics inlet may be spatially stabilized for the beam location at the
photocell to be independent of filament locationO To achieve stabilization, a
Spatially Homogenizing Optical Modulator (SHOM) in the form of rectangular
prism 104 is employed in the light path between the filament and the fiber-
optics inlet. ~he prism provide~ an optical path offset ~hich is a function of
its rotation position. When the prism rotates, the filament optically appears
to s~e~p across the face of the fiber-optics inlet 11~ In position A, the
prism is so oriented that the light rom fila~ent 101 is bent outside the
acceptance angle of the fibers in cable 18, and negligible light is transmitted
to the benchO In position B, the prism has rotated sufficiently for light to
- be transmitted through at least some of the fibers in the cable. In position
C, the prism has swept the filament image across the face of the fiber-optics
inletO In position D, the prism has moved the inlet to a position beyond the
acceptance angle of the fibers, and again negligible light is transmi~ted. As
the prism rotates further, the beam irst reappears beyond the acceptance angle
of the fibers, as in position A, and then another s~eep begins~ The sweeping
action, including the period of negligible light transmission, occurs two times
during each revolution of the prismJ or about 100 times per second~
If filament 101 moves or appears to move~ this has the effect of
-17-
. .
changing the time at whlch the bea~ s~arts and inishes its sweep across the
fiber-optics inletO That is, only the phase of the beam movement is altered by
movement of the filament. The electronic:s described above compute the average
or middle position swept b~ the imageO The electronics are insensitive to such
phase or time shifts, and thus the undesirable effects of filament shift are
minimizedO
The a~parent l~ght s.ou~ce pos~tion is further stab~lized b~ using the
randomized fiber-optics bundle 18. In a perfectl~ randomized fiber-optics
bundle, adjacent fibers at one end of the bundle are randomly distributed at
the other end. Therefore, increasing the light on one side of the bundle in-
put while decreasing it on the other side results in no change in the light
distribution across the fiber-optics output end~ In actual practice) the
randomization in a bundle i~s not perfect, and some change does occur at the out-
put end. But using the randomized fiber-optics doe~ further decrease the effect
of filament motion on ~eam movement at photocell 52.
As can be seen in Pigure 10b, the optics do not focus the beam onto
the photocell in the vertical direction, as done in the horizontal direction.
Instead, light emerging from outlet 50 of cable 18 remains unfocused in vertical
planes, thereb~ producing for each point of light at the outlet a vertical line
of light at the photocell. The vertical height of this line is limited b~ the
vertical height of mirror 78, which acts as a mask. Light ra~s from individual
points, eOgO, points X and Y, on the cable outlet 50 fan out, but onl~ rays
inside of limit rays Xl9 X2~ (Yl, Y2 for point Y3 reach the photocell~ (Other
ra~s are not reflected through the photocell). The vertical heights of mirror
78, photocell 52, and cable outlet 50 and the s~acing between the flow cell and
photocell ends of the bench are all selected so that the limit rars for all
~18-
points. on the ca~le outlet strike full~ ove and ~ully belo~ tr~angular cells
180, 182 of the photocell. Limit rays for point X and point Y, at the top and
bottom extremities of the cable outlet, are shown in ~igure lOb. Thus each
point on the cable outlet produces a line of uniform intensity at the photocell.
And these lines all overlap over the photocell, thereby assuring a uniform
vertical intensity across the photocell tlO matter ~hat vertical variation in
intensity may exist at the cable ou~let ~e.g., due to varia~ion in filament
intensity in the vertical direction). The end result is the light intensity
profile sho~Yn at the left side of ~igure lQb. Across the vertical height of
the photocells the intensity is uniform outside the photocells the intensity
falls off to zeroO Vertical uniformity of light intensity at the photocells
i5 needed to linearl~ determine the horizontal light beam location on the
triangular-shaped cells 180, 1820 (A vertical variation in intensit~ would
be indistinguishable from a horizontal movement of the light beam).
Other Embodiments
Other embodiments are possible. For example, reflective coatings
other than gold could be used ~e.g.J aluminum, silver~ or a multilayer coating);
an anti-reflection coating could be substituted for black epoxy coating 76,
with a light trap positioned behind and external to the cell to absorb light
passing through the coating; quartz glass and the like could replace the
borosilicate glass used for the flow cell; and the glass pieces of the flow
cell could be joined together by diffusion bonding or with adhesiveO
--19-
'
