The invention relates to a liquid chromatograph comprising a multiple
piston reciprocating pump for delivering a liquid at a desired flow rate to a separating
column, each piston being arranged, during its delivery stroke, to deliver liquid
at the same rate as the others, wherein the pistons are arranged to have overlapping
delivery strokes and are driven by a motor whose speed is controlled by a control
In liquid chromatography systems, and in particular high performance
liquid chromatographs (HPLC), the pump is required to have a flow capability covering
the range of, typically, 1 µl to 30 ml/min, a minimum internal volume so that a
solvent change at the inlet to the pump rapidly reaches the column input, minimum
pulsation in flow/pressure since the stability of most detection systems is adversely
affected by such pulsations, and a capability of delivering at a sufficient pressure
to enable the required flow of solvent through all columns likely to be used.
The discharge pressure may, typically, range between 5 and 400 bar and a variety
of solvents with compressibilities typically between 50 and 150 × 10&supmin;&sup6;/bar
may be used.
A variety of pumping systems are available for use in HPLC, each
of which have their own advantages and disadvantages. This invention relates to
chromatographs in which reciprocating piston pumps are used. Such pumps can be
divided into single and multiple piston pumps. Single piston pumps have the advantage
of mechanical simplicity but pose greater difficulty in achieving low pulsation
in flow/pressure of the delivered liquid. With multi piston pumps it is possible
to arrange that, in theory, one piston is always delivering and thus pulsation
is reduced. However, due to the high pressures involved there is a significant
compression of the liquid and compressible material in the pump and hence the instant
at which any piston starts to deliver cannot be determined merely from the position
of the piston as a degree of precompression, which varies with the particular
solvent used and the system pressure, occurs before delivery commences and hence
a drop in pressure will occur when one piston stops delivering before the other
piston starts to deliver. One method of alleviating this problem is to monitor
the pressure at the pump outlet and to increase the motor speed when the pressure
falls thus minimising the period when no flow is produced by the pump and hence
reducing the drop in pressure and the magnitude of the pulsations. However this
requires a complex control system to drive the pump.
GB-A-2085980 discloses a liquid chromatograph as set forth in the
opening paragraph which includes a regulating system for continously varying the
speed of the motor so that the pressure of the liquid conveyed remains at least
approximately constant during each period.
EP-A-0215525 (published 25.03.87) discloses a liquid chromatograph
as set forth in the opening paragraph characterised in that the control arrangement
is arranged to produce a control signal a characteristic of which has a first
constant value during a first period when m of the pistons is/are delivering liquid
to the column so that the piston(s) advance at a first constant velocity and a
second constant value during a second period when (m+1) pistons are delivering
liquid to the column to cause the pistons to advance at a second constant velocity,
the second constant velocity being m/(m+1) times the first constant velocity, where
m is an integer and (m+1) is less than or equal to the total number of pistons
provided in the pump.
By arranging that the delivery strokes of the pistons overlap there
is never a period when no piston is delivering; only a first period when m piston(s)
is/are delivering and a second period when (m+1) pistons are delivering and by
providing a constant rate of advance a relatively constant flow, and hence constant
delivery pressure, can be obtained merely by dividing the rate of advance of the
pistons by (m+1)/m during the second period. Thus, if the start and finish of
the second period can be determined, a relatively simple control arrangement can
be used to control the rate of advance of the pistons, merely requiring a division
of the rate of advance at appropriate times.
This can be contrasted with the position disclosed in GB-A-2085980
where because the pistons do not advance with a constant linear velocity for a
constant angular velocity of the cam it is necessary to continuously vary the cam
velocity in order to obtain a constant flow rate. This causes considerable complexity
in the regulating system.
In the chromatograph disclosed in EP-A-0215525 the start of the second
period is detected by monitoring the pressure of the liquid delivered by said
pump, an increase in the pressure denoting the start of the second period.
This gives a precise indication of the start of the second period
when (m+1) pistons are delivering and the instant of detection of increasing pressure
can be used to divide the rate of advance of the pistons. It should be noted that
while the instant when (m+1) pistons start to advance can also be easily detected,
that instant cannot be used to control the speed of advance since at high delivery
pressures significant precompression of the liquid takes place in the cylinder
before delivery commences. The extent of the precompression will depend on the
particular solvent being pumped, which may be continuously changing, and the pressure
at which delivery takes place, which will depend on the column resistance and flow
rate. Further it is not necessary to monitor the actual pressure since it is known
that the flow rate will return to the desired value merely by dividing the rate
of advance of the pistons by m/(m+1).
However, with certain systems the compressibilty of the liquids and
the compliance of the column and other parts of the liquid system causes the pressure
increase to be relatively slow and thus the instant of speed change to be delayed
as a certain pressure margin has to be allowed to avoid the change of speed being
triggered by noise on the signal from the pressure transducer. This causes undesirable
fluctuations in the flow rate of the liquid.
It is an object of the invention to enable the provision of a liquid
chromatograph pump having a relatively simple control system which is capable of
producing an output flow of liquid having comparatively low pressure pulsation.
The invention provides a liquid chromatograph as set forth in the
opening paragraph characterised in that the control arrangement is arranged to
produce a control signal which causes the piston(s) to advance at a first constant
velocity throughout a first period when m of the pistons is/are delivering liquid
to the column and at a second constant velocity throughout a second period when
(m+1) of the pistons are delivering liquid to the column, the second constant velocity
being m/(m+1) times the first constant velocity, where m is an integer and (m+1)
is less than or equal to the total number of pistons provided in the pump; wherein
the control arrangement comprises means for monitoring the delivery pressure of
the pump in each pump cycle, and means for advancing the instant of change of
velocity of the pistons in one cycle if a pressure increase is detected in the
previous cycle or retarding the instant of change of velocity of the pistons in
said one cycle if a pressure decrease is detected on the previous cycle.
In this specification the term advanced is used to specify that the
speed change takes place earlier in the cycle and the term retarded is used to
specify that the speed change takes place later in the cycle.
The end of the second period may be detected by monitoring the positions
of said pistons. It is not necessary to detect pressure changes at the outlet of
the pump to ascertain when one of the pistons ceases to deliver liquid since this
is accurately known from the position of the piston. There is no problem with precompression
of the liquid at this end of the delivery cycle although the filling of the cylinder
during the withdrawal of the piston may be affected by decompression, both of
the liquid remaining in the cylinder and in the seals and other deformable parts.
The control arrangement may comprise a microprocessor which is arranged
to react to interrupt signals generated to coincide with given points in the pump
Thus the required calculations can be performed at times related
to given points in the pump cycle when it is known that the necessary measurements
have been made or the necessary actions taken.
The interrupt signals may be generated at the start of the delivery
stroke of each piston, at the end of the delivery stroke of each piston, and at
the instants of change of velocity of the pistons.
A shaft encoder, which may comprise a disc having a single cut-out
and an optical detector, may be driven by said motor in synchronism with the pump
and one interrupt derived from said shaft encoder.
This enables the synchronisation of the interrupt cycles with the
actual position of the pump cycle once during each pump cycle. Otherwise the interrupts
may get out of synchronism with the pump cycle due to, for example, any corruption
of stepping pulses applied to a stepping motor driving the pump. It also enables
a simple detection of a point in the pump cycle on switch on.
When the pump is driven by an electrical stepping motor, the control
arrangement may produce stepping pulses for application to a driving circuit for
driving said stepping motor, and at least some of said interrupt signals may then
be generated as a result of counting the number of stepping pulses generated since
the previous interrupt.
This avoids the need for all the interrupts to be generated by the
shaft encoder which would lead to a more complex procedure for setting up the pump
at the manufacturing stage and to a requirement for the shaft encoder to be more
accurately synchronised with each part of the pump cycle.
The control arrangement may comprise a programmable counter, the
microprocessor being arranged when each interrupt occurs to program the counter
to produce an output causing the next interrupt signal to be generated a given
number of stepping pulses later.
The interrupt signal generated at the start of the delivery stroke
of each piston may initiate the calculation of the instant of change of velocity
after the start of the delivery stroke of the next piston.
The instant of change of velocity may be advanced or retarded by
a time proportional to the magnitude of the pressure increase or decrease and when
a stepping motor is used to drive the pump instant of change of velocity may be
advanced or retarded by kMp steps of the stepping motor, where k is a constant
and Mp is the magnitude of the pressure increase or decrease at the outlet of
This enables a quicker convergence to the correct instant for changing
the motor speed. The value of the constant k may be chosen to give the best compromise
between speed of convergence and stability. High values give fast convergence but
low stability while low values give high stability but slow convergence.
The control arrangement may operate such that if the flow rate is
increased the instant of change of velocity of the pistons on the next pump cycle
is retarded by a factor dependent on the change in flow rate and if the flow rate
is decreased the instant of change of velocity of the pistons on the next cycle
is advanced by a factor dependent on the change in flow rate.
This reduces the time taken for convergence to the correct instant
for changing speed by predicting the effect of a change of flow rate.
The second period may be terminated earlier as the flow rate is increased
and later as the delivery pressure is increased.
This allows compensation for the effects of inertia which are such
that the time taken between a motor speed change (doubling) and the effect of that
change being reflected in the pressure trace varying under different operating
conditions and particularly with different flow rates and delivery pressures.
The motor may drive the pistons through a cam arrangement. The cam
arrangement may comprise a separate cam for each piston, the cams being mounted
on or formed integrally with a common shaft.
A separate cam for each piston allows the pistons to be arranged
side by side rather than being horizontally opposed. This simplifies the mechanical
arrangement of the combining means required to combine the liquid outlets of each
cylinder for feeding to the column. Having more than one piston allows more flexibility
in designing the fill stroke of each piston.
The cam(s) may be profiled such that a constant angular velocity
of the cam(s) produces a constant linear velocity of the delivering piston(s).
This enables a simplication of the control arrangement which simply
has to ensure that the motor speed remains constant at one of two values depending
on how many pistons are delivering. If the cam profiles are not formed in this
manner the motor speed has to be varied to compensate for the cam characteristic.
This could be achieved using a programmed memory, for example a programmable read
only memory (PROM) which stores a representation of the speed correction required
against cam angle to enable a constant rate of piston advance to be achieved. This
enables the control arrangement to produce a constant output signal which is corrected
for the cam characteristic by the contents of the PROM thus retaining a simple
control arrangement but requiring a set up procedure at the manufacturing stage
or if a cam is replaced.
The end of the second period may be detected by monitoring the positions
of said pistons. The position of the pistons may be monitored by means of an encoder
mounted on the shaft.
This enables the instant at which each of the pistons ceases to deliver
liquid to be accurately detected and enables a signal to be produced to increase
the speed of the motor when the end of said period is reached.
The pump may be driven by an electrical stepping motor, the characteristic
of the control signal being its frequency, and during the second period in which
(m+1) pistons are delivering liquid the frequency of the pulses applied to the
motor is divided by (m+1)/m.
When the pump is a dual piston pump and is driven by a stepping motor
the rate of advance can be halved using a simple control circuit which involves
dividing the rate of the stepping pulses by two during the second period. This
separates the control function for maintaining a constant flow rate from the control
function which sets the desired flow rate for a particular analysis.
An embodiment of the invention will now be described, by way of example,
with reference to the accompanying drawings, in which:-
- Figure 1 shows in block schematic form a liquid chromatograph according to
- Figure 2 is a perspective view of a pump suitable for use in the chromatograph
of Figure 1,
- Figure 3 is a plan view of the pump shown in Figure 2,
- Figure 4 illustrates the delivery of liquid by each piston of a multiple piston
pump constructed for use in a liquid chromatograph,
- Figure 5 shows the effect of changing the motor speed at the incorrect time
and the convergence of the control circuit to the correct time,
- Figure 6 shows in block schematic form one embodiment of a control circuit
arrangement suitable for use in the chromatograph of Figure 1,
- Figure 7 shows in block schematic fo rm an interrupt generation circuit used
in the arrangement of Figure 6,
- Figure 8 shows in block schematic form a pressure monitoring arrangement used
in the arrangement of Figure 6,
- Figure 9 shows in block schematic form a circuit for operating solvent proportioning
valves used in the arrangement of Figure 6,
- Figure 10 shows a flow diagram illustrating the generation of interrupt signals
for the microcomputer of Figure 6,
- Figure 11 is a diagrammatic cross-sectional view of one cylinder of a pump
head of a typical liquid chromatograph pump, and
- Figure 12 shows a flow diagram illustrating the generation of control signals
for the proportioning valve arrangement.
Figure 1 shows in block schematic form a liquid chromatograph which
comprises a multiple piston pump 1 for pumping a liquid through a separating column
2. The pump 1 is coupled by a shaft 3 to a stepper motor 4 which drives the pump
1. A pressure transducer 5 monitors the pressure at the outlet of the pump 1 and
also connected between the outlet of the pump 1 and the inlet of the column 2 is
a sample injector 6. The outlet of the column 2 is connected to a detector 7 which
produces an electrical output which is fed to processing circuitry 8 which in
turn drives a display device 9. The display device 9 may take any convenient form,
for example a video display unit or a chart recorder. The stepper motor 4 is driven
by a control circuit arrangement 10 which feeds stepping pulses at a desired rate
to the stepper motor 4 over a path 11. The control circuit arrangement 10 receives
a first input from an input unit 12 over a path 13. The input unit 12 may be a
keyboard to allow the desired flow rate to be entered by an operator or may be
any other arrangement which allows the operating parameters of the chromatograph
to be set up. The control circuit arrangement 10 also receives a second input
over a path 14 from the pressure transducer 5. The inlet of the pump 1 is fed from
a solvent proportioning arrangement 15 which is controlled by signals from the
control circuit arrangement 10 over a path 16.
In operation a desired flow rate is set up using the input unit 12
which produces signals which enable the control circuit arrangement 10 to produce
stepping pulses to drive the stepper motor 4 at the appropriate speed to produce
the desired flow rate when m pistons are delivering the liquid where (m+1) is less
than or equal to the total number of pump pistons. The pump is arranged so that
the delivery strokes of the pistons overlap to such an extent that periods of non-delivery
are eliminated under worst case conditions of flow rate, pressure and compressibility
for which the system is designed. Consequently under all but the worst case conditions
there will be a period in each cycle when (m+1) pistons are delivering. If no
further action were taken the flow rate would increase by a factor (m+1)/m and
as a result the pressure at the pump outlet would increase. The outlet pressure
is monitored by the pressure transducer 5 which produces a signal which is fed
to the control circuit arrangement 10 over the path 14 and this information is
used by the control circuit arrangement 10 to cause the rate of the stepping pulses
applied to the stepper motor 4 over the path 11 to be divided by (m+1)/m when
(m+1) pistons are delivering, thus reducing the rate of advance of the pistons.
At the end of the period of overlapping delivery the rate at which the stepping
pulses are applied to the stepper motor 4 is restored to the original value.
There are various possible ways of detecting the end of the period
of overlapping delivery, for example a pressure drop at the outlet of the pump
1 can be detected by the pressure transducer 5. However, it is currently preferred
to detect when each piston reaches the end of the delivery stroke with the aid
of a shaft encoder where the pistons are driven by means of cams mounted on the
shaft 3 driven by the stepp er motor 4.
For a dual piston pump m=1 and hence when both pistons are delivering
their speed of advance is halved relative to that when only one piston is delivering.
This may be easily accomplished when the pulses for the stepping motor are derived
digitally, for example merely switching a divide-by-two stage in or out of circuit.
Where more than two pistons are provided it may be arranged that not more than
two are delivering at any one time, i.e. the situation illustrated in Figure 4f)-h),
and in this case the speed of advance of the pistons is again halved when two
are delivering relative to that when only one is delivering. The advantage of providing
more than two pistons is that the return stroke of each piston can be extended
allowing a longer period to fill the cylinder. This can be of particular advantage
when a wide range of flow rates may be required and where a more accurate proportioning
of several solvents into the cylinder is desired on each stroke. The disadvantage
is, of course, greater mechanical complexity and hence cost.
An alternative arrangement when more than two pistons are provided
is to arrange the pump so that more than one piston is always delivering e.g. with
a four piston pump it could be arranged that at one time two pistons are delivering
and that a third piston has an overlapping delivery. In this case the change of
speed is one and a half times rather than twice and can thus be more quickly achieved,
for a given acceleration, and consequently flow or pressure variations may be reduced.
The control circuit arrangement 10 is arranged to change the speed
of the motor when (m+1) pistons start delivering by dividing its speed by the factor
(m+1)/m, i.e. for a two piston pump the motor speed is divided by two. The following
description assumes a two piston pump but clearly the same principles can be applied
to a multi-piston pump having more than two pistons.
The control circuit arrangement has an input from the pressure transducer
5 to enable the pressure to be monitored throughout the pump cycle. Figure 5 a)
and b) illustrate the effect of halving the motor speed too late or too early and
how the correction of the instant of halving the motor speed progresses. In Figure
5 SST1 represents the instant the first piston starts to advance, EOD2 represents
the instant the second piston reaches end of delivery i.e. when it stops delivering
liquid, and HS1 represents the instant at which the motor speed is halved. Figure
5 a) illustrates the effect of halving the motor speed too late. This causes a
pressure peak to be produced when both pistons are delivering. The amplitude Mp
of this pressure peak is dependent on the degree of overlap of delivery of the
two pistons. The control circuit arrangement 10 detects the peak and measures
its amplitude and uses this information to generate an instruction to the motor
to halve its speed earlier in the next pump cycle. This may be by a factor equal
to kMp, where k is a constant. The value of the constant is chosen to provide
the best compromise between speed of convergence and stability. A low value would
produce a slow convergence which may be unacceptable when changing flow rates or
solvent compositions while a high value will produce an unstable system in the
presence of noise where the instant HS1 is thrown away from the desired instant
by noise spikes. As can be seen from Figure 5 a) the magnitude of the pressure
peak is steadily reduced in succeeding pump cycles until it is at the level of
noise on the pressure baseline signal. This is achieved by advancing the instant
HS1 in response to the detection of a pressure peak on the previous pump cycle.
A similar situation occurs if the instant HS1 occurs too early except
that in this case a pressure dip rather than a pressure peak is detected. The same
correction procedure is followed except that the instant HS1 is retarded rather
than advanced until the pressure dip is reduced to the noise level.
Clearly a similar procedure is carried out for the start of delivery
of the second piston, i.e. the half speed instant HS2 between SST2 and EOD1 (the
start of delivery stroke of piston two and end of delivery of piston one). If both
pistons and cylinders together with the check valves are identical then the correction
of the instant HS2 can be determined on the basis of the pressure increase or
decrease for the delivery of the other piston. However, by making the correction
dependent on the pressure during the previous cycle of the same piston the instants
HS1 and HS2 for each piston can be separately adjusted to compensate for any imbalance.
As a further refinement which is particularly useful when the required
flow rate is being changed and hence the outlet pressure (baseline pressure) is
increasing or decreasing apart from the effect of overlapping delivery, the pressure
peaks or dips and baseline pressures can be interpolated over two or more pump
cycles to produce a more accurate correction. The changing baseline pressure may
be caused by solvent changes or flow rate changes. Further, a predictive advance
or retardation may be incorporated for compensation when the flow rate changes.
The solvent proportioning arrangement 15 receives inputs from the
input unit 12 via the control circuit arrangement 10 to cause a selected solvent
or mixture of solvents to be fed to the inlet of the pump. The solvent proportioning
arrangement comprises a plurality of sources of solvents (for example four) which
are connected via respective solenoid valves to the inlet of the pump. The solenoid
valves are operated for a calculated period during the suction stroke of the pump
to provide the selected solvent composition. The solenoid valves are arranged
to be operated one at a time and it is ensured that no two valves are open simultaneously
to prevent siphoning of different solvents through paths formed by two or more
valves. An alternative possibility would be to add one way valves in each solvent
inlet line to the solenoid valves.
In order to obtain the desired solvent proportions in the mixture
the appropriate valves are operated in sequence for the appropriate proportion
of the suction stroke of the pump. Clearly if one of the solvents is to be present
in a very small proportion, for example 5 per cent, then that valve will only be
operated for less than 5 per cent of the suction stroke. For high flow rates and
small swept volume of the piston this can be difficult to achieve. As discussed
in the introduction even if the valves are very quick acting the consequent fluid
accelerations and decelerations may lead to out-gassing or cavitation.
The control circuit 10 and solvent proportioning arrangement 15 are
constructed so that the proportioning can take place over a plurality of pump or
piston cycles when the operating time of any solenoid valve is less than a given
value or is likely to become less than the given value. In one arrangement this
is determined by the selected flow rate produced by the pump and once a flow rate
above that value is selected the solvent composition is averaged over a plurality
of piston cycles. Thus if one solvent is required in a small concentration, for
example 1 per cent, this may be achieved by providing 5 per cent of that solvent
during one return stroke but only providing that solvent every fifth return stroke.
An alternative method of deciding whether to average over a plurality of piston
cycles is to calculate the required valve operating times to produce the required
mixture in one suction stroke and to cause averaging to occur if any calculated
value for a valve opening time is less than a given value.
The arrangement may also make a calculation of the actual suction
volume, that is a calculation of the suction time corrected for the time taken
for the unswept volume of liquid and other parts to decompress so that pressure
inside the cylinder drops to atmospheric pressure. Clearly unless the solv sources
are pressurised no suction exists until the pressure in the cylinder drops to atmospheric
pressure. At the delivery pressures normally involved in HPLC liquids and plastic
parts are compressed by a significant amount. The decompression time is calculated
from a knowledge of the precompression time on the previous delivery stroke of
that piston. Once the decompression time has been calculated a precise instant
for the start of suction can be determined and hence the total suction time can
be accurately found. This enables a more accurate proportioning of the solvents
to be achieved.
It is not essential that a multipiston pump is used as far as the
invention relating to proportioning the solvents is concerned. It is equally applicable
to single piston pumps and may have even more utility in such pumps since the suction
time is likely to be a smaller proportion of the total pump cycle. One reason
for the use of multipiston pumps is to allow a greater suction time without unacceptably
increasing the flow pulsation on delivery.
Figure 2 is a simplified perspective view of a pump 1 and motor 4
suitable for use in the chromatograph of Figure 1 and Figure 3 is a plan view
of the pump 1 and motor 4 shown in Figure 2.
The pump 1 comprises a pump head 20 which is clamped between two
side panels 21 and 22 made from sheet metal. The pump assembly has cross pieces
24 and 25 to provide a rigid structure onto which the motor 4 and various component
parts of the pump are mounted. A cross piece 23 is allowed to float between the
side panels 21 and 22 so that tubes 30 and 31 which slide in bearings 32,33,34
and 35 are not constrained by any misalignment of the bearings. The pump 1 is provided
with two pistons 26, 27 mounted in respective rods 28, 29 which pass into the interior
of two tubes 30, 31. The tube 30 is slidably mounted in bearings 32, 33 in cross
pieces 23 and 24 while the tube 31 is similarly mounted in bearings 34 and 35.
Two cams 36 and 37 are mounted on the shaft 3 of the motor 4 and the tubes 30 and
31 are provided with respective cam followers 38 and 39. Coil springs 40 and 41
bias the cam followers 38 and 39 against the cams 36 and 37 by way of projections
(not shown) on the rods 28 and 29, the rods 28 and 29 bearing against transverse
members in the tubes 30 and 31. A transverse portion 42 of a bracket 43 provides
a bearing surface for the other ends of the springs 40 and 41. Two circlips 44
and 45 are provided on the rods 28 and 29 to retain the pistons 26, 27 in the head
20 when the head is dismantled from the rest of the pump assembly. Two tubes 46,
47 take the outlets from each cylinder of the pump head 20 and combine them in
a manifold 48 having an outlet 49 which forms the pump outlet. A shaft encoder
50 is attached to the shaft 3 of the motor 4 and a detector 51, which may be an
opto electronic detector, is carried by a bracket 52 attached to the side panel
21 of the pump.
In operation the motor 4 is supplied with stepping pulses by the
control circuit arrangement 10 (Figure 1) and causes the shaft 3 to rotate at a
desired speed. As is known, by providing controlled currents to drive the stepper
motor windings and ensuring the correct ratios between them the motor rotation
can be incremented by several sub-steps or microsteps between each fall step.
The design of a suitable stepper motor drive circuit to achieve this is well known
to those skilled in that art. By utilising such a system the rotational position
of the motor and its instantaneous speed can be more precisely controlled. Consequently
the cams 36 and 37 cause the pistons 26 and 27 to advance and retract in accordance
with the cam profiles which are designed to cause the pistons to advance on their
delivery strokes at a constant linear velocity when the cams have a constant angular
velocity. The design of such a cam profile is well known to those skilled in the
art. As has been described with reference to Figure 1 the pistons are arranged
to have overlapping delivery strokes, i.e. there are two periods of time during
each revolution of the motor when both pistons are advancing. Figure 4 illustrates
the delivery of liquid by each piston and the motor speed at various points during
a pumping cycle. Figure 4a shows the cam angles. Figure 4b shows the pressure in
the pump head due to the motion of piston 26 and shows that at point Ax the delivery
stroke begins, i.e. the piston starts to advance, but that precompression of the
liquid and plastic parts is occurring and no actual delivery of liquid takes place
until point Ay. From point Ay until point Az delivery of liquid at a constant rate
takes place since the piston is advancing at a constant linear velocity, point
Az being the point at which the piston reaches the end of its delivery stroke and
thus liquid can no longer be delivered. Figure 4c shows the pressure in the pump
head due to the motion of piston 27, the points Bx, By and Bz corresponding to
the same points on the delivery cycle of piston 27 as points Ax, Ay and Az of piston
26. It can be seen that there is an overlap of the delivery strokes of the two
pistons and that consequently unless further action is taken there will be periods
during which the flow rate will be doubled. This undesirable occurrence is prevented
by halving the motor speed during the overlapping delivery periods as shown in
Figure 4e which illustrates the motor speed the lower level S/2 being equivalent
to half the speed of the upper level S. The instant when simultaneous delivery
by the two pistons commences is predicted by monitoring the pressure at the output
of the pump as is shown in Figure 4d. When simultaneous delivery commences the
pressure rises and this pressure rise is used by the control arrangement 10 to
control the motor speed on the next pump cycle so that when simultaneous delivery
is predicted to start on that cycle the motor speed is halved and the pressure
returns to the original value since the combined delivery of the two pistons is
equal to that of the original single piston delivery. The end of the period during
which both pistons are delivering is detected with the aid of the shaft encoder,
and at that instant the motor speed is doubled to regain its original value.
Figure 4 f) to j) illustrates in a similar way to Figure 4b) to e)
the situation where three pistons are provided in the pump and the cam profiles
are such that at any one time either one or two of the pistons is/are delivering
liquid. The advantage provided by the use of three pistons operating as illustrated
in Figure 4 is that a longer period is available for filling each cylinder with
the solvent which is to be pumped, i.e. the period A′z to A′x is longer
than the period Az to Ax. This may be of importance when several liquids are being
serially fed into each cylinder to give a desired solvent mix for pumping, particularly
at high flow rates when the fill time becomes shorter since the total pump cycle
time becomes shorter. Thus it becomes advantageous to make the fill time as large
a proportion as possible of the pump cycle time. Of course more than three pistons
could be used giving even longer fill times but every added piston increases the
mechanical complexity and hence cost.
The control circuit arrangement 10 shown in Figure 6 comprises a
microprocessor (CPU) 100, for example a Zilog Z80 Central Processing Unit, which
is interconnected over a data bus 101, an address bus 102 and a control bus 103
with a memory 104, an interrupt generation and timing circuit (IGTC) 105, an I/O
decoder (I/O DEC) 106, a keyboard and display unit (KDU) 107, a pressure monitoring
unit (PMU) 108, and a solvent proportioning control unit (SMC) 109, (the unit 109
corresponds to the arrangement 15 of Figure 1). The interrupt generation and timing
circuit 105 is connected to a stepper motor drive circuit 110 which drives the
motor 4 over line 11 as shown in Figure 1. In addition the circuit 105 has an input
to which the opto-detector 51 (shown in Figure 3) is connected over a line 111,
and a further output which is connected over a line 112 to the pressure monitoring
unit 108 and the solvent proportioning control unit 109. The I/O decoder 106 is
connected over an I/O select bus 113 to the interrupt generation and timing circuit
105, the keyboard and display unit 107, the pressure monitoring unit 108 and the
solvent proportioning control unit 109. A clock generator 114 is connected over
a line 115 to the microprocessor 100 and over a line 116 to the IGTC 105.
The I/O DEC 106 receives address and control signals from the microprocessor
(CPU) 100 and decodes them to produce appropriate control signals for the I/O circuits
which connect the blocks 105, 107, 108 and 109 to the CPU 100 and for the memory
104. The I/O DEC 106 may, for example, comprise a number of one out of eight decoder/demultiplexer
chips type 74138.
The IGTC 105 is shown in Figure 7 and comprises a clock input 120
which feeds a divider chain 121, 122, 123 which may be formed from a programmable
interval timer (PIT) which is sold by Intel Corporation under the type reference
8254. These devices receive data from the CPU 100 over the data bus 101 under the
control of the address, control and I/O select buses 102, 103 and 113. This data
comprises a number to which the counters are preset so that a given division ratio
is obtained. A desired flow rate for the solvent is selected by the operator by
means of the KDU 107 and the CPU 100 calculates appropriate division ratios for
the counters 121 and 122 so that pulses of the correct frequency to drive the stepper
motor are fed to the stepper motor drive circuit 110, these pulses being derived
from the output of counter 122 and being available at output 124. A divider 125
which is formed by part of a Zilog type 8430 counter timer circuit has its clock
input connected to the output of counter 123 and is programmed to produce an interrupt
signal on control bus 103 when the counter 123 reaches zero count. The line 111
which connects the opto-detector 51 to the IGTC 105 is connected to a divider 126
which is also formed as a Zilog Z8430 counter timer circuit and is programmed by
means of signals on the address, control, data and I/O select buses to produce
an interrupt signal immediately a signal from the opto-detector is received. A
further divider 127 has its clock input connected to input 120 which is fed with
a clock signal from the clock generator 114 over the line 116. The divider 127
which is also formed from a Zilog type 8430 counter timer circuit is programmed
to divide the clock signal so that an interrupt signal is generated every 1 msec.
and is fed to the control bus 103 over line 128 and over line 112 to the pressure
monitoring unit 108.
The pressure monitoring unit 108 is shown in block schematic form
in Figure 8 and comprises a pressure transducer 130 which is connected to monitor
the liquid pressure between the pump outlet and the column input. The pressure
transducer 130 feeds a voltage to frequency (V/F) converter 131 either directly
or through a current to voltage converter depending on whether the pressure transducer
provides a current or voltage output. The output of the V/F converter 131 is fed
to a first input of an AND gate 132 whose output is connected to the clock input
of a counter 133. The counter 133 is part of an INTEL 8254 PIT and has inputs
connected to the control and I/O select buses 103 and 113 and an output connected
to the data bus 101. The 1 ms pulses on line 112 are fed to a divider 133 whose
output is connected to a second input of the AND-gate 132.
The operation of the control circuit arrangement shown in Figures
6 to 8 will now be described with the aid of the flow diagram shown in Figure
The interrupt HEDINT is generated by the opto-detector and interrupt
generator 126. When HEDINT occurs the microprocessor 100 sets a number into the
counter 123 which corresponds to the number of microsteps of the motor 4 between
the instant of HEDINT and the start of stroke of piston 26 (ST.STRINT1). This
number may be generated either from a calculation based on the designed cam profile
or from an initial calibration at the manufacturing stage and stored in the memory
104. The counter is clocked by the output of counter 122 (that is by the microsteps
fed to the motor 4) and caused to count down to zero (DEC.CTR). The number set
into the counter 123 is such that the counter 123 reaches zero when the start of
stroke of piston 26 occurs. When the zero count is reached an output signal is
fed to the divider 125 which generates an interrupt (STRINT1) on line 103. The
interrupt STRINT1 causes two procedures to be initiated. First a number is set
into the counter 123 (ST.HSINT1) which represents the number of microsteps to
HSINT1 (that is the instant when the speed of the motor is to be halved) and secondly
a calculation (CLC.HSINT2) is made to produce the number to be fed into the counter
123 at the corresponding point in the pump cycle for the other piston to generate
HSINT2. The counter 123 is decremented (DEC.CTR) by the motor pulses generated
by counter 122 and at the same time the number of motor microsteps between HSINT1
and EODINT2 is calculated, that is the number of microsteps between the instant
at which the motor speed is halved and the instant at which the piston 27 ceases
delivering liquid when the motor speed has to be restored to its original value.
The calculation merely requires the substraction of the number of microsteps between
STRINT1 and HSINT1 from the known number of microsteps between STRINT1 and EODINT2,
this number being known from the cam calibration. When the counter 123 reaches
zero count the interrupt HSINT1 is generated and the microprocessor 100 responds
by increasing the divisor of the counter chain 121 and 122 by a factor of two
to cause the rate of microstep pulses at output 124 to be divided by two and sets
(ST.EODINT2) the calculated number of microsteps to EODINT2 into the counter 123.
The counter 123 is decremented (DEC.CTR) by the motor pulses generated by counter
122 until the interrupt EODINT2 is generated when the counter 123 reaches its
zero count. The interrupt EODINT2 causes the microprocessor 100 to restore the
divisor of counters 121 and 122 to its original value since this marks the end
of overlapping delivery by both pistons and to set (ST.STRINT2) the number of
microsteps to the interrupt STRINT2 into the counter 123. Again this number is
non-destructively accessed from the memory 104 in which it was initially entered
from the known cam profile. The number corresponds to the number of microsteps
of the motor between piston 27 reaching end of delivery and the start of the delivery
stroke of piston 26 which is a function of the cam profile and remains constant
for a given pump. The counter 123 is decremented (DEC.CTR) by the microstep pulses
until the zero count is reached when the interrupt STRINT2 is generated.
The interrupt STRINT2 initiates the calculation of the half speed
instant for piston 26 (CLC.HSINT1) and causes the microprocessor to set the number
of microsteps to the half speed instant of piston 27 into counter 123 (ST.HSINT2).
This is the value which was calculated in response to the interrupt STRINT1. The
counter 123 is then decremented by the microstep pulses (DEC.CTR) and when the
count reaches zero the interrupt HSINT2 is generated. In parallel with this the
microprocessor calculates (CLC.EODINT1) the number of microsteps of the motor
between HSINT2 and EODINT1 which corresponds to the number of steps between the
half speed point and the instant at which piston 26 ceases delivering liquid.
When the interrupt HSINT2 is generated the microprocessor 100 responds
by increasing the divisor of the combination of counters 121 and 122 by a factor
of two to cause the motor speed to be halved and setting (ST.EODINT1) counter 123
to the calculated count (CLC.EODINT1) to the end of delivery interrupt EODINT1,
that is the instant when piston 26 stops delivering liquid. The count er 123 is
decremented by the microstepping pulses from counter 122 (DEC.CTR) until it reaches
zero count when the interrupt EODINT1 is generated by counter 125. This interrupt
causes the microprocessor 100 to set counter 125 to its maximum count. The reason
for this is that the next interrupt which should be generated is HEDINT which is
generated in response to the output of the opto-detector 51. Consequently by setting
counter 123 to its maximum count it is ensured that this counter does not cause
another interrupt signal to be generated before the next HEDINT occurs.
It would be theoretically possible to set a number into counter 123
which corresponds to the number of stepping pulses between EODINT1 and STRINT1
but using an encoder disc ensures a positive location of one instant in the pump
cycle. The interrupts EODINT1 and EODINT2 which correspond to the end of delivery
points of pistons 26 and 27 where the motor speed is doubled may be compensated
according to the current flow rate and delivery pressure of the pump. This is necessary
since the time taken between a motor speed change and the effect of that change
reflected in the pressure trace varies under different operating conditions.
These compensations are carried out in the following manner:
- 1. Flow rate compensation.
- For a maximum flow rate no adjustment is made.
- For a zero flow rate a given number of microsteps are added to each EOD position.
- The relationship of flow rate to compensation is assumed to be linear and
to a linear interpolation between zero and the given number based on the actual
flow rate is carried out.
- 2. Pressure compensation.
- For a pressure of 0 bar no adjustment is made.
- For a maximum pressure a given number of microsteps are added to each EOD position.
- The relationship of delivery pressure to compensation is again assumed to
be linear and consequently a linear interpolation between zero and the given number
based on the actual delivery pressure is carried out.
In some applications of high performance liquid chromatography (HPLC)
it is necessary to change the composition of the solvent or mobile phase in a controlled
manner during the analysis. For example, it may not be possible to choose a mobile
phase which will enable all of the sample components to be separated and eluted
in a reasonable time. This problem can be overcome by the use of a technique known
as gradient elution which is analagous to temperature programming as used in gas
Normally gradient elution involves starting the analysis with a mobile
phase consisting of one particular solvent and then adding progressively increasing
amounts of a second solvent during the analysis. The composition change required
may involve a linear increase in the concentration of the second solvent with
time or a more complex gradient may be required. However, it may also be necessary
to add more than a second solvent and in some instances third and fourth solvents
may be required to be mixed to produce the desired mobile phase.
It is also sometimes required to have a constant solvent mixture
which may contain a small percentage of one or more particular solvents. To achieve
this very short valve operating times may be required, particularly at high flow
There are two main methods to obtain the desired composition of the
mobile phase when using a reciprocating piston pump to produce the flow of mobile
phase to the column. The first is high pressure proportioning, where the high pressure
outputs of two or more pumps are combined together before being applied to the
column. The individual pump flow rates are selected to give the desired composition
of the solvents while the sum of their flow rates gives the desired total flow
rate. The second method is to use low pressure proportioning where the solvents
are proportioned by a set of solenoid valves or similar devices which are switched
to give the desired mix composition. This switching or proportioning is made to
happen during the suction period or periods of a single high pressure pump and
the switching device is fitted in the inlet line of the pump.
Both methods have advantages and disadvantages. When low total flow
rates are in use together with low percentage mixes, high pressure proportioning
demands that one pump is running at a very low flow rate which is often difficult
to achieve in a reproducible and reliable manner. Thus if the mix required is 99%
of solvent A and 1% of solvent B then the pump supplying solvent B will be running
at approximately one hundredth of the rate of the pump which is supplying solvent
A. Further, high pressure pumps are frequently controlled by means of pressure
measuring devices connected at the outlet side of the pump. This can make it very
difficult to connect such pumps together for high pressure mixing because of the
problem of identifying which pressure measuring device is controlling which pump.
Simple synchronous methods will not work if gradient elution is employed where
the solvent mix composition varies as an arbitrary function of time during the
Low pressure proportioning which is used in this embodiment has the
disadvantage that the system delay volume (volume between the mixing point and
the head of the chromatographic column) is larger since the whole pump volume
is involved. This method is not limited at low flow rates since it becomes easier
at low flow rates to proportion the solvents at the inlet to the pump. However,
as flow rates increase the time allowed for proportioning the solvents into the
pump is steadily decreased. Further, in order to reduce pulsations of flow at the
outlet of the pump, the suction time of each piston is normally made a small proportion
of the total pump cycle. Thus the time for proportioning the solvents into the
inlet is correspondingly reduced. As a result the valves used to proportion the
solvents to the inlet of the pump have to be very quick acting. Minimising the
pump suction time increases the inlet flow rates and the consequent fluid accelerations
and decelerations. This may lead to out-gassing, cavitation, or the cylinder failing
to fill completely due to inertia or compressibility effects on the liquid. Any
of these effects will cause errors to be produced in the solvent composition delivered
at a particular flow rate.
In this embodiment the chromatograph is arranged to achieve a desired
solvent composition by allowing small amounts of the different solvents into the
liquid stream in succession and allowing the system volume to mix them. This is
achieved by switching solenoid valves at the appropriate times to allow the required
volumes of each solvent into the pump head during the return stroke.
In order to determine precisely the times at which the valves have
to be switched the suction time has to be accurately determined. Theoretically,
this can be accurately determined from a knowledge of the cam profile and the motor
speed, i.e. the suction time will correspond to the period of the return stroke
of the piston. However, in practice at the pressures used in HPLC liquids and
plastics compress to a significant extent and a certain pressure differential needs
to be established across the check valves before they will operate. Thus the start
of the suction time does not correspond accurately with the start of the return
stroke of the piston. Consequently unless a more accurate determination of the
start of the suction time can be achieved a significant inaccuracy in the solvent
mixture will occur, particularly at high pressures.
Figure 11 is a diagrammatic cross-sectional view of one head of a
liquid chromatograph pump which comprises a cylinder 200, a piston 201, plastic
seals 202 between the piston and cylinder and input and output check valves 203
and 204. As shown in Figure 11 the piston 201 is fully retracted and the dotted
rectangle shows its position when fully advanced.
At working pressures encountered in liquid chromatography (typically
up to 400 bar) liquids and plastics compress to a significant extent. This causes
volume loss as the rigid piston advances on its delivery since the liquid and plastic
compress as the pressure in the cylinder is increased to a sufficient value to
open the output check valve 204 and deliver liquid to the column at the system
pressure. Similarly no suction occurs on the return stroke until the pressure within
the cylinder drops to atmospheric pressure to allow the inlet check valve 203 to
Let it be assumed that the pump head has a stroke volume v, that
is the piston swept volume denoted by the dotted rectangle, an unswept liquid volume
Vl, and the plastic seal 202 a volume Vp. Further, let the
liquid compressibility be Kl and the plastic compressibility be Kp.
If we now consider precompression when the piston starts its delivery
stroke we have
at zero pressure, the head volume V(0) = Vl + v + Vp
at pressure P,
liquid volume ((Vl+v)e-KlP) + plastic volume (Vpe-KpP)
Volume loss on precompression (PL) = V(0) - V(P)
∴ PL = (Vl+v)(1-e-Klp + Vp(1-e-kpp)
If we now consider decompression then the piston starts its return
stroke we have
at pressure P,
liquid volume (Vl+Vp (1e-KpP))+
plastic volume (Vpe-KpP)=V'(P)
At zero pressure,
liquid volume ((Vl+Vp(1-e-KpP)) eKlP)+plastic
Volume loss on decompression (DL) = V'(0) - V'(P)
DL = (Vl+Vp(1-e-KpP))(eKlp-1)
The correction required is PL- DL
= v(1-e-KlP)+Vl(2-2 cosh klP)+Vp(e(Kl-Kp)P-e-KlP-e-Kpp+1).
when KlP and KpP<<1
Then PL -DL≙ v klP ...................... 1
Typically the maximum value of Kl or kp is 150x10-6/bar
and the maximum pressure is 400 bar and therefore KP≙0.06. In expression
1 the stroke volume (v) is known precisely for a given head and the system working
pressure P can be measured precisely. Kl is unknoun for mixed solvents
but typically lies between 50x10-6/bar for water and 150x10-6/bar
for Heptane. Setting Klto be equal to 100x10-6/bar gives
a satisfactory result.
PL is the volume of the stroke from the start of stroke until delivery
commences and can be accurately determined since, as described hereinbefore, the
start of delivery can be accurately measured and is equal to the stroke between
the appropriate STRINT and HSINT times multiplied by the piston area. This stroke
can be determined from the cam profile and number of microsteps of the motor between
STRINT and HSINT.
It should be noted that, in particular, Vl, Vp
and Kp do not need to be known. Also if the plastic parts do not behave
as ideal compressible solids this is compensated for. Thus, provided that the
volume of stroke lost on decompression can be determined, DL can be calculated
from expression 1 for any given pressure P.
Therefore, since the decompression volume can be determined the proportion
of the return stroke, or suction time, taken for the unswept volume of liquid and
the plastic parts to decompress can be determined since the time taken for the
piston to retract a distance equivalent to the decompression volume can be determined
from the known piston area, the rate of application of microstepping pulses to
the motor, and the known cam profile. Thus the actual start of suction can be determined
by calculating the number of microsteps from EODINT (1 or 2) which represent the
The decompression volume and hence the actual suction time is preferably
determined separately for each piston since unless both pistons and cylinders are
identical, which is difficult to achieve in practical manufacturing processes,
a different precompression and decompression volume will occur in each piston/cylinder
combination. This is easily achieved when the HSINT is calculated separately for
each piston as has been described hereinbefore.
Figure 9 shows in block schematic form the solvent proportioning
circuit 109 of Figure 6. It comprises eight programmable interval timers 150 to
157 having inputs connected to the data, address, control and I/O select buses
101, 102, 103 and 113. The clock inputs of the timers 150 to 157 are fed with the
microstepping pulses derived from the counter 122 in the IGTC 105 over a line
158. Four set-reset bistable circuits 160 to 163 have respective set inputs connected
to the outputs of timers 150, 152, 154 and 156 and respective reset inputs connected
to the outputs of timers 151, 153, 155 and 157. The outputs of the bistable circuits
160 to 163 are fed to solenoid valve driver circuits 164 to 167, respectively,
while the outputs of the drive circuits are connected to respective solenoid valves
170 to 173.
In operation the microprocessor sets a count into the timers 150
to 157 and when the timers are clocked to a zero count their outputs set or reset
the corresponding bistable circuits 160 to 163 and hence activate or de-activate
the solenoid valves 170 to 173 at instants which depend on the initial value set
into the relevant timer.
Figure 12 shows a flow diagram illustrating the generation of the
numbers to be set into the timers by the microprocessor 100. Step DET.DL comprises
the calculation of the compression lost volume where the microprocessor 100 reads
the numbers of microsteps from STRINT to HSINT for the delivery stroke of the piston
(26 or 27) and converts this to a volume using a look-up table of cam shape stored
in the memory 104. It then calculates the value v KPwhere v is the
total piston displacement volume, K is the mean liquid compressibility (assumed
to be 100×10&supmin;&sup6;/bar), and P is the measured pressure. Then the
volume DL=PL-vKP is calculated. Step DET SV comprises the calculation of the suction
volume which is equal to the piston displacement minus the decompression volume
Step DET.SSU comprises determining the instant of the actual start
of the suction which is obtained from the calculated value of DL converted into
microsteps which is added to EODINT.
Step CALC.VAT comprises calculating the time for which each valve
must be open to produce the necessary solvent mixture. This is calculated from
the suction volume, the flow rate (that is the speed of travel of the piston),
the percentage of each solvent requested, and the start of suction time calculated
in step DET.SSU.
Step ORD.SOL causes the order of operation of the valves to be selected
so that the largest proportion solvent is split into two equal parts and the corresponding
valve arranged to open at the start and the end of suction while the other solvents
valves are opened in sequence between these two openings. The opening of the valve
for the largest proportion solvent at the beginning of the suction stroke may take
place before suction begins sinc e the inlet check valve will be closed. Similarly
the closing of that valve at the end of the suction stroke may occur after the
end of the suction stroke.
Step CAM.CORR causes the valve timing, which is in terms of numbers
of microsteps, to be corrected for the cam profile by means of a look-up table
stored in memory 104. Step PRO.PITS comprises the setting of numbers into the programmable
interval timers which correspond to the calculated valve opening and closing times
for each solvent.
This procedure is satisfactory for low flow rates, that is for the
decision NO to the question FLOW > MAX?, but as the flow rate is increased so
the suction time decreases and if the pump displacement is small, for example 30
microlitres, at flow rates of 1ml/min and above considerably less than one second
is available for proportioning the solvents. This requires extremely fast acting
solenoid valves to achieve mixtures having only a small proportion (< 5%) of
A solvent proportioning system according to the invention overcomes
this problem by setting a minimum value of opening time for any valve and proportioning
the solvents over a plurality of piston cycles, for example sixteen piston cycles.
Preferably when proportioning over a number of piston cycles a solvent mixer such
as that disclosed in our co-pending application No. .......... (PHB33297) should
be used to prevent unmixed slugs of one solvent of the mixture reaching the column.
If the decision YES is reached to the question FLOW>MAX?, i.e. a high flow rate
is requested by the operator, for example greater than 1ml/min then the step SEL.MCM
is taken. This is a procedure which causes the solvent proportioning to occur
over more than one piston cycle. The procedure follows the steps DET.PMIX in which
the microprocessor determines what mixture of solvents is present in the system
as a result of the previous fifteen piston cycles. This is calculated from a knowledge
of the valve opening times during the previous piston cycles and is calculated
as a running average; CAL.NMIX in which the microprocessor calculates the proportions
of each solvent which needs to be taken up during the next suction period to bring
the mixture to the selected value; and CALC.VOT in which the required opening times
for each of the valves is calculated to provide the proportions just calculated
in CAL.NMIX. A decision is then taken ANY < MIN? to determine whether the calculated
proportions require a valve opening time less than a preset minimum. If so, the
smallest component is set to zero (DEL SC) and the new mixture recalculated. If
not, the step CALC VAT is entered and the programmable interval timers are set
Figure 13 shows a flow diagram illustrating a second method of generating
the numbers to be set into the timers by the microprocessor 100. The step CALC.TAR
% comprises the calculation of the percentage (target percentage) of each solvent
which is required in the solvent mixture in accordance with the requirements set
by the user. This may involve a specific constant percentage of each solvent selected
by the user by means of the KDU 107 or may be calculated from a programmed gradient,
i.e. a changing solvent composition with time which may be automatically calculated
for each suction period by the microprocessor 100 using data entered by the user
and a knowledge of the previous state of the system. The step CALC.REQ.% comprises
the process of calculating the percentage of each solvent required in the next
suction period to bring the current average percentage of each solvent to the value
calculated in the step CALC.TAR.%. If the composition of the solvent mixture is
being averaged over n piston cycles then
Required % = n × target % -(n-1)x average %.
Step FD.MIN.% comprises the calculation of the minimum percentage of any solvent
which can be introduced in a suction period. This depends on the minimum valve
operating time (which is set to a constant value whi n the valves selected for
the system) and the flow rate. Thus the lower the flow rate the smaller the minimum
percentage can be since the suction period is correspondingly increased. The step
FD.LST.SOL comprises the action of sorting the required solvent percentages into
order with that requiring the lowest percentage first and in ascending order of
required percentage. A decision is then made (%.<1/2 MIN?) as to whether the
lowest percentage required of a solvent is less than half the minimum value calculated
in step FD.MIN.%. If this is not the case a further decision is then taken (1/2
MIN < % < MIN?) as to whether the lowest percentage solvent is between the
minimum value and half the minimum value. If this is not the case then the required
percentage of that solvent is assigned to that solvent (ASSN.REQ.%) and stored
for subsequent use in the next suction period. The step FD.NLST.SOL selects the
solvent having the next lowest proportion. A decision is then taken to determine
whether this is the final solvent of the mixture (FIN.SOL?) and if so the step
ASSN.REM.% is taken and the remaining percentage of the suction period is assigned
to the final solvent. The step CONV.%.µST comprises coverting the percentage suction
periods assigned to each solvent into microsteps of the motor. The actual suction
period calculated as described hereinbefore can be converted into microsteps of
the motor and vice versa from a knowledge of the flow rate and hence the motor
speed. It is convenient to work in terms of motor microsteps since the suction
period will be equivalent to a given number of microsteps dependent on the construction
of the motor and pump.
The procedure just described shows the sequence of operations for
solvent proportioning when the flow rate and proportions are such that each solvent
can be proportioned within a single piston return stroke. However at high flow
rates or with small proportions of one or more solvents this may not be. In these
cases the solvent composition is achieved by averaging the individual solvent percentages
over two or more piston cycles. Thus if the lowest percentage solvent requires
a valve operating time of less than half the minimum valve operating time the
answer YES is given to the decision % < MIN? This causes the step ASSN.0% to
be taken which step causes the valve associated with that solvent to remain closed
for the next suction period. The next lowest percentage solvent is then found
(FD.NLST.SOL) and unless this is the final solvent the decision % < MIN is again
taken for the next lowest percentage solvent. This procedure is repeated until
the answer NO is given to the decision % < MIN?
If no further solvent proportion is less than half the minimum a
further decision is taken as to whether the lowest percentage proportion is between
the minimum and half the minimum value. If the answer is YES then a further decision
is taken as to whether this is the first solvent for which that applies (1ST.SOL?)
and if the answer is YES then this solvent is assigned the minimum valve operating
time (ASSN.MIN.%) for the next suction period. If the answer is NO then step ASSN.0%
Thus once any solvent percentage is required which needs a valve
operating time of less than the minimum value the arrangement will cause the solvent
composition to be averaged over more than one piston cycle. The number of piston
cycles over which the averaging takes place will vary according to the proportions
required and this will take place automatically by means of the target and required
percentage calculations. The microprocessor 100 keeps a record of the assigned
percentages of the solvents in each piston cycle so that the averaging can take
place. The minimum number of cycles over which the averaging can take place will
depend on the number of previous piston cycle assignments stored and the actual
number of cycles over which the averaging takes place will depend on the actual
It should be noted that, depending on the number of microsteps of
the motor, the percentage required may not be precisely obtainable in whole numbers
of microsteps. This error can also be averaged over a plurality of pump cycles
to increase the precision of the solvent composition.
It may also be desirable when performing gradient elution to correct
for the fact that solvent compositions can only be changed at fixed instants whereas
a linearly changing composition may be desired. Again this can be achieved in the
CALC.TAR.% and CALC.REQ.% steps.
Thus if during a suction period it is required to proportion a solvent
into the pump and the pump is operating at such a speed that it is not possible
to open the proportioning valve for a short enough period to give x%, then it is
decided to put x times y% into one piston suction period and then again y piston
cycles later and so on to give an average percentage of (x times y)/y=x%.
A running average is used to monitor what has been put into the system
and then to work out what is needed to add to the average, during this piston cycle,
in order to achieve the required solvent composition. If any of the required proportions
are found to be less than the minimum that the proportioning valve can be operated
for then it is excluded from the piston cyclefor which the calculation is made.
The requirement for this proportion will increase automatically each piston cycle
until it is large enough to be included.
By using this running average method, proportioning over more than
one piston cycle is automatically catered for, for any system flow rates, valve
switching times are required percentages.
The suction time and suction volume are related by the rate at which
micro stepping pulses are applied to the motor and the calculations can be performed
in terms of volume, time or microsteps since the speed of the pistons will be related
to the rate of application of micro stepping pulses to the motor by a fixed relationship
dependent on the construction of the motor, the cam profiles and the piston dimensions.
It is convenient to use a stepping motor and perform the calculations
in terms of microsteps but clearly, with appropriate modifications to the calculations,
other types of motor could be used, for example d.c. motors.