Vacuum/volume 32/number 5/pages 257 to 263/1982 0042-207X/82/050257--07503.00/0 Printed in Great Britain Pergamon Press Ltd

A u t o m a t i c cont ro l of gas pressure in a v a c u u m system J Lucas, C S Smith and M M Meadows, Department of Electrical Engineering and Electronics, University of Liverpool, PO Box 147, Brownlow Hill, Liverpool L69 3BX, UK

received 14 April 1981; in revised form 3 August 1981

This paper describes the development of a microprocessor based system to be used for the automatic setting of the gas pressure within a vacuum chamber. The requirements of the system have been presented by five modes of operation which are easily set up by thumbwheel and on/of f switches on a panel. The microprocessor system automatically balances the rates of the pumping system and gas inlet to set and maintain the requested gas pressure. The gas inlet needle valve is controlled by either an ac or dc motor. The use of computer software allows the system to be readily adapted for use with the commonly used needle valves, pressure gauges and pumping systems. It allows for non linear transducers and mechanical wear and additional modes of operation may be easily added when required. The system is intended for use as a subsystem of a microprocessor controlled vacuum system for laboratory or industrial use. The technique has been illustrated by controlling the pressure over the range 1-20 tort in a conventionally pumped vacuum system.

1. Introduction

For research into gaseous electronics more and more components of a vacuum system are being controlled by a microcomputer system 1. One important requirement is the automatic setting of the gas pressure 2 in both a sealed and in a continuously pumped system. The gas from the gas cylinder enters the vacuum chamber via a needle valve and the gas pressure is measured using a pressure gauge having an electrical transducer. The chamber may be continually pumped or sealed by using an electromagnetically operated valve and both methods of operation are required for electron swarm investigations 3. The measurement of swarm properties in expensive inert gases (e.g helium, krypton) require a fixed pressure sealed system 4 whilst the operation of a heat pipe using a metal vapour and inert gas (e.g. Cs/Ar) requires a fixed pressure system which is continually pumped in order to remove the impurities produced in the gas by the heat 5. In both cases it is useful to simply type in the mode of operation and the required pressure and to allow the microcomputer to automatically set and maintain the pressure.

The work described in this paper describes the design of a low cost, universal system which is independent of such parameters as the volume of the vacuum chamber, the time constant of the pumping system, cylinder gas pressure and the backlash on worn needle valve mechanisms. Although stand alone systems (e.g. Granville-Phillips, Automatic Pressure Controller) have been produced and sold for many years this current design may be incorporated as a low cost attachment to a micrometer based system.

2. System requirements

The system is illustrated in Figure 1. The vacuum system has four valves which are relay operated and have a digital output to sense their open and shut positions. There is a single valve to isolate the chamber from the pumping system and three valves connected to the gas inlet of the needle valve in order to select the connection to either the gas cylinder, or to atmospheric air pressure or to a rotary pump. The pressure of the gas in the chamber is measured by a gauge with an electronic readout. The reading is converted

••4Pu•aps and gas supply J

, T°: '2 GAmins (¢Ylhnder pressure

[ Pressure gauge ( Ovn/rostfe.~t L~

I A * ° ° c ° n v ' " l II .

Figure I. The automatically operated system.

257

J l, ucas, C S Smith and M M Meadows: Automat ic control of gas pressure in a vacuum system

into a digital value by using an analogue to digital converter. The digital values of the pressure and valve positions are transmitted via a bus structure to a microcomputer and may be displayed. The microcomputer controls the gas pressure in the chamber by opening or closing the needle value by the motor. An optically generated counter gives positional information about the motor to the microcomputer.

2.1. Modes of operation. The system has five modes of operation defined as follows. The pumps are continually operated and the valve settings are given in Table 1. Mode 1 allows the chamber to be fully evacuated. Mode 2 allows the chamber to be brought to atmospheric air

pressure. Mode 3 opens the needle valve by a preset number of counts (N)

and then produces an equilibrium (dynamic) pressure for continuous pumping.

Mode 4 sets up the required pressure (p) for continuous pumping.

There is an automatic adjustment of the pressure to allow for small changes in temperature, outgassing, system leakages etc. in order to keep the pressure within a fixed percentage of the required pressure. This automatic operation is activated when the pressure becomes 'out of range' and it resets the pressure back to p. An indicator is provided to show the operator when the gas pressure is 'in range'. Mode 5 sets the required gas pressure (p) in a closed system. An

indication is given when the gas pressure has stabilized within the requested pressure range.

All these modes of operation were obtained by using the same hardware and simply selecting the appropriate software program from the control panel. Thumbwheel switches are used to set the mode (1 to 5), time constant (0 to 10 min), percentage range (0 to 10), requested pressure (0 to 100 torr) and requested number of motor counts (0 to 1000). There are BCD displays for both the actual pressure and the actual number of motor counts. There are

several push button switches controlled by a 'reset" which is used to cease the present operational mode. If the'manual' operation is chosen then the counter may be reset and the motor operated in either a forwards or backwards direction. In the 'automatic' operation the information given by the thumbwheel switches is applicable when the command 'go" is given.

2.2. System programs. The automatic control of pressure changes may only be made in times which are comparable with the time constant of the system.

, Since the size and pumping speed of systems vary it is necessary to allow for a wide variety of time constants. The advantages of using a microcomputer is that the time constant is readily varied by using either hardware or software control. The time constant for the system is T= V/(d V/dt) where Vis the volume of the system and d V/dt the volume pumping rate. For the system used for this paper the time constant was the order of 0.5 min so that this is the order of the speed pressure changes can be made.

The requirements of the programs will be considered by discussing both the mode 4 and mode 5 operation.

2.2.1. Description of mode 4. To set the required pressure let us assume that the initial pressure was zero and therefore below the required pressure. The needle valve starts in the fully closed position and is opened by the motor. The opening of the valve may have an hysteresis effect caused by a worn valve and this gives an offset (N o) in the relationship between the number of motor turns iN) and the gas pressure (p). It is also anticipated that the relationship could be non-linear. The motor is therefore only initially operated until the pressure rises to 3". of the required chamber pressure (p). After the motor is stopped at N~ turns the chamber pressure will still rise but an equilibrium pressure p~(=f~p) will be established. The wait period is set equal to the system time constant. The valve is then opened until a 10",, change in pressure (i.e. 0.1 p) is detected then the motor is stopped at N 2 turns and the chamber pressure allowed to stabilize at a value

Table I. The system valve settings

Gas valves 1 2 3

Pump Needle Gas Chamber Mode valve valve Pump Atmosphere cylinder state

1 Open Shut Open Shut Shut Open loop Fully evacuated

2 Shut Open Shut Open Shut Open loop Atmosphere pressure

3 Open Open Shut Shut Open Open loop Dynamic pressure N counts

4 Open Open Shut Shut Open Closed loop Dynamic pressure P

5 Initially Shut Open Shut Shut Open Closed loop Finally Shut Shut Open Shut Shut Static

pressure P

258

J Lucas, C S Smith and M M Meadows: Au toma t i c cont ro l of gas pressure in a vacuum system

On Motor

Off

Pressure

" - F . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .- In range--p required . . . . . . . . . . " " V ~ o ~ J ' l - . . ~ . ~ ' ~ . . . . . . . . . . . . . . . . . . . . . . . . . - - 1 ~ . . . . G, Pa

, w o ~ t , /

3

Woit wait I.~.I /

PI Time Motor

3% I0% 50% 50% 50% Fine On/off : 1 : : i " =I ~ " 1 : = 1 : : ~= range

I I i-I 1 I [~1 Pl R fl T,me N t N 2 N s Motor

Cou~ts

Figure 2. Mode 4 operation.

pz(=J;p). This procedure is illustrated in Figure 2. If a linear relationship existed between N and p with an offset No then the number of turns required to attain equilibrium pressure p would be N r where

- N t

However to anticipate a non linear relationship, a needle valve (N3) setting was calculated so as to achieve a midway pressure of IP + p2)/2 and the valve opened appropriately. After the pressure has stabilized to P3 the motor is again operated to give a valve reading of N4 corresponding to the midway pressure (p + p392 is achieved. This procedure is repeated until we achieved within say 10",, of the required pressure and then a fine control routine is used to avoid hunting. This simply moves the valve in one direction by one motor revolution until the pressure (P) is attained. The pressure is now continually monitored to check that it remains fin range'. If it falls out of range then the motor opens (or closes) the needle valve by single steps until the required system pressure (p) is obtained again.

2.2.2, Description of mode 5. There are two ways of implement- ing this requirement. The system could be operated under mode 4 conditions and then the chamber may be quickly isolated from the pump and gas inlet by using solenoid valves when the required equilibrium pressure has been achieved.

An alternative method, illustrated in Figure 3, is to isolate the chamber from the pump and then to allow in gas until the required pressure is achieved. A quick way of achieving this is to open up the needle valve with the motor running continuously until a gas pressure of ½p is detected. The motor is then reversed and the needle valve closed again by running the motor continuously. For an ideal needle valve whose opening and closing properties are identical then the resultant pressure will be p. Because of backlash in the needle valve and delay in reversing the motor direction it was decided to only initially open the needle valve until pressure of x/2 p(x < 1 ) was detected. When the needle valve was closed then the resultant pressure Pn was higher than xp but still well below the required pressure p. The process was repeated again until the pressure increased until a pressure P l + x ( p - P 1)/2 was attained. The valve was then closed and a resultant pressure P2 obtained. The sequence was then automatically repeated until the pressure

i-E ron0e__ . . . . p required . . . . . . . . . .

Woit

Woit

reverse I i I I I I I ~ t ~ nu ~ Time

Figure 3. Mode 5 operation.

came within the range requested. After n such repetitions the pressure will be

p n = p ( x + (1 - x ) x + (1 --x)2x + . . . (l - -x)"- Ix)

i.e.

p , = p ( 1 - ( l - -x)")

The fractional setting from equilibrium is

R A N G E - p - p " = (1 - x ) " P

If RANGE= l°,~ and x=0.6 then further analysis gave n=5, which indicates that the requested pressure can be attained by five successive opening and closing of the needle valve.

3. Hardware description

The functions of the hardware were as follows: (I) To convert the analogue output of the pressure transducer

(0-10 V) to a digital signal for entry into the microcomputer. (2) To allow the microcomputer to drive and set the direction of a

motor, either a 240 V ac motor (or a 6 V dc motor) for controlling the needle valve position.

(3) To enable the microcomputer to read the position of the needle valve by using a counter driven by optically generated signals. Each revolution of the motor produces one count.

(4) To provide a facility to allow manual control of the system at any time.

(5) To operate solenoid valves required by the pumping and gas supply system as described by Lucas et al x.

The hardware components have been listed in Figure 1 and the mechanical components are illustrated in Figure 4. These will be discussed individually.

3.1. The needle valve. The needle valve used was an Edwards High Vacuum Ltd LB2B, panel mounted, bellows type. It had an 18 turn, 100 division dial and a revolution counter. The basic valve is illustrated in Figure 4. To allow the valve to be driven by a motor, a few modifications were made. A worm and worm-wheel was mounted onto the dial shaft having a ratio of 1:40. To avoid overdriving the valve, either when fully open or fully closed, a slipping clutch arrangement has also been fitted, whose torque is adjustable. This also still allows hand operation to be achieved.

259

J Lucas, C S Smith and M M Meadows: Automatic control of gas pressure in a vacuum system

To chamber Needle valve 3.5. Microcomputer system. The Quarndon microcomputer system was used. This is based upon the Intel 8080A micropro- cessor and the computer memory had 2~ K bytes of RAM and sockets for 3 K bytes of EPROM (INTEL 2708).

Cascaded gears (1:3-1:4-1:6)

Figure 4. The automatically controlled needle valve.

oy

3.6. Motor drive circuits. The needle valve position may be controlled by either an ac or a dc motor and both types have been tried out.

AC motor. The motor used was a 240 V ac capacitor start motor. This type of motor has two coils and a capacitor is connected in series with one of them in order to provide the phase displacement necessary for the motor to start. This is illustrated in Figure 5(a). By changing the position of the capacitor (i.e. in series with either the main winding or auxiliary winding) the direction of the motor can be changed. A relay is used to change the position of the capacitor. For convenience a mains driven relay was chosen so that two identical mains drive circuits were built, one to drive the motor and the other to drive the relay. These circuits were built around the zero voltage switch (RS 305-800) so that switching only takes place when the voltage is zero and electrical interference is absent if the load is resistive, i.e. current and voltage in phase. However with an inductive load (as in our case) the current and voltage are out of phase by up to 90 ~ and a net resistive load obtained by placing a correction capacitor in series with the load. The switching circuit is illustrated in Figure 51a).

Two types of motor have been used. Initially a Fracmo single phase permanent split capacitor induction motor was used and this runs at 56 rev/min, with gearing. More recently a Montage-Beispiel 6 V, 2 A dc motor and this operates at 83 rev/min. The dc motor is intrinsically safer because of the low voltage rating. The motor direction is obtained by using a relay.

DC motor. Power switching is done by a PNP Darlington power transistor. The devices used (LAMBDA 13K40) have an 8 A continuous rating and a minimum current gain of 800. The collector emitter saturation voltage is approximately 1.5 V at full load. Drive from the microcomputer is stepped up by open collector drivers (RS307-109). This is illustrated in Figure 5(b).

3.2. Pressure gauge: The pressure gauge used was an MKS Baratron type 170M (MKS Instruments, Inc). This is a capacit- ance manometer type and is used with an electronic unit which provides excitation to the head and converts its output to a proportion dc output of 10 V full scale. Three decades of pressure measurement are possible of 1000, 100 and l0 torr full scale with an accuracy of _+ 1 ~/o and the pressure is given on a digital display.

3.3. Optical switch and counter. An ir source and sensor housed in a slotted moulding (Radio Spares, RS 306-061) is used to detect a revolution of the motor which is related to the needle valve position. When the beam is modulated by a hole cut in the disc mounted on the motor shaft a pulse is given to a counter (CMOS 4 decade counter 7217) and displayed on a four digit, seven segment display (RS 587-024).

3.4. Analogue to digital converter. The converter was constructed from a standard device with a voltage comparator. The clock was provided by a 555 timer set to 40 kHz. A convert command pulse initiates the conversion and a status flag is examined to indicate when the conversion is completed. The output is an 8 bit binary word and the quantization interval is 10 mV.

260

Software

The software was written in modular form and the program was just under 800 bytes long and was stored in a 1K 2708 EPROM. The program also required a data storage area of 11 bytes of RAM. The software was developed using the lntellec System.

The executive program is effectively the interface between the human operator (via the switches and displays) and the actual control routines. It allows the required parameters to be entered via the thumbwheel switches. These are: (1) Mode of operation either 1, 2, 3, 4 and 5. (2) If mode 3 then the number of motor counts (N) has to be set

up. (3) If modes 4 and 5 then the required pressure (p), its range [r)

and the system time constant (T) have to be set up. The operating instructions are listed in Appendix 1. The executive performs modes 1 and 2 which either fully

evacuate the vessel or bring it to atmospheric pressure. Mode 3 opens up the needle valve for a required number of counts. If modes 4 or 5 are required then the required value of p must be entered. When in either of these modes, it is necessary to use the 'reset' to return to the executive routine because they require a continuous monitoring of the dynamic pressure. The operating system for modes 4 and 5 have been given in Section 2.

J Lucas, C S Smith and M M Meadows." Automatic control of gas pressure in a vacuum system

240 V O.C.

Main winding

• 00000 ~ Auxiliory winding

l Capacitor

T

¢ . m _

16 15 13 12 I1 I0 9

i'" RS 305-800 0 ==.,fl .

Zero voltage switch

2 3 5 6 7 8 MT2

T 25V

N 0 240V o.c.

,Mains fuse dA

-tEffective load I i Power factor = correction C

! I I I t Load motor/re oy

I . -J Triac tr i 400 -0 .35

IA-Relay G M' I ' I IOA- Mo?or

From output port

Figure 5(a). AC motor and driver circuit.

Op'to- isolator (RS 307-979)

+ 0

t Main 6V d.c. winding

- 0

6V 0

o - t > o - - MIC roFro°mputer O°d I;?vg:r° n

RS 307- I0~

Logic I (On) 0 (Off)

PMD 13K40

t lOK~

126K,~, Motor

*-1 - J

Figure 5(b). DC motor and driver circuit.

The flow chart for operation in mode 4 will only be discussed and this is given in Figure 6. The control routine begins with the assumption that the vessel is fully evacuated, i.e. the pressure is 0 torr and has stored in memory the required pressure (p), the system time constant (T) and the required range of accuracy (r). First of all the program opens the needle valve (N1 turns) until it

detects a pressure of ~p(=3.1%). It allows the pressure to stabilize to Pl. The needle valve is then opened to N2 turns at which the pressure is detected as ~p(= 12.5%). The pressure is allowed to stabilize to P2. The next setting of the motor (N3) is the setting required to give a midway pressure p+pJ2 to the equilibrium value by using equation 11) which gives

N 3 = N 2 + f2-~--~-I (2)

The system is allowed to reach equil ibrium and the process repeated unti l the vessel pressure is within ~p(= 12.5%) and then the motor is single stepped unti l the final pressure p is attained. This pressure wil l init ial ly be slightly on the high side of p but within the requested range r. The pressure is now continually monitored and only if the pressure falls out of range wi l l i t be reset back to its requested value p by single stepping of the motor with wait intervals. I f the pressure is greater than p + r then the motor direction will be reversed.

5. Results

5.1. Open loop response. The open loop response shows the relationship between the dynamic pressure (p) and the needle valve position (N) when in equilibrium. This relationship is dependent upon the gas cylinder pressure (P) at the needle valve and the pumping rate of the system. The results for the manual opening of the needle valve are shown in Figure 7 for two cylinder

261

J.Lucas, C S Smith and M M Meadows: Automat ic control of gas pressure in a vacuum system

Enter from executive program ) I

/ Display 'out of range' led / I

J Calculate 3.1% of P I I

J Run motor until pressure is set Slore / motor position NI J I

II wai t for T sec I I I

/ / Read pressure P and display =t /

J " Calculate _~.5% o fP J

J Run motor until pressure is set. Store motor position N z J

II Wait for Tsec I I

J Read pressure P2 and display Jt

J Calcu late N3 corresponding to mid pressure (P; P~)--~J

J Run motor unt i l N 3 is set [

11 Wait for T sac II

,/t Read pressure P and d,splay evaluate error as ERR= I~--~ I ~

SetP2= P3 I . No T~PP~ And P, =p~ i-

/

No i

/ II

• NO

J Run motor I rev J I

I1 Wait T sec II I

Read pressure PI and display. Calculate ERR = 1 ~ - ~ 1 - - - - 7

Read pressure and display it. Display 'in range' led / / i

Keep pressure within range J J

(, Return to executive program )

13 r

1 2 - -

IO--

Figure 6. Flow diagram for mode 4 operation. (Known parameters are pressure (p), time constant (T) and range (r).)

gas pressures (P). These curves show a reasonably linear relationship up to the maximum opening of the valve at which the levelling-off occurs. The particular valve illustrated had a lot of backlash and this is clearly shown by the offset which occurs when the valve was closed. The automatic system described in this paper is totally independent of the relationship between p and N provided the required gas pressure is attainable.

,-7 =-

== Q)

(-9

Closing valve

~pening valve

0 20 40 60 80 I00 120

Valve position N counts

Figure 7. O p e n loop operat ion.

8 - -

7 - -

6 - -

Mode 3 - - r

I

I 5

~ 5

o. 4

~ 3 IiJ a.

2

Mode I

I0

Time,

Figure 8. Modes 1 and 3 operation.

lode 3

15

min

)=2 bar

valve

P = I. 5 bar

140 160 tSO

Mode I

1% I 20 25

5.2. Mode 1 and mode 3 results. The mode 1 and mode 3 curve is given in Figure 8. These curves were obtained by connecting the analogue output of the pressure transducer to the Y input of a chart recorder with the time base on the X input. The needle valve was opened by mode 3 until the valve position registered 64 counts and the pressure allowed to stabilize. Mode 1 was then operated and the needle valve was fully closed. The process was repeated using a needle valve setting of 132 counts and then mode 1 was operated. The majority ofthe time taken was the time required to open or shut the needle valve at the rate of 56 counts/min.

5.3. Mode 4 and 5 results. Since the operating mode of both 4 and 5 are similar only the results of mode 4 will be illustrated and these are given in Figures 9 and 10. In Figure 9 the pressure (p) is initially at 0 torr and increases in a series of steps towards the required pressure 3.8 torr (__2%). The time constant was set at 0.5 min and this equals the wait time between switching on and off the motor. The pressure was within range after about l0 min and was set at the requested pressure of 3.8 torr. The motor was switched off until the pressure became 2% out of range, i.e. 3.9 < p < 3.7 torr, at which the pressure was reset to 3.8 torr. Once

2 6 2

J Lucas, C S Smith and M M Meadows: Automat ic control of gas pressure in a vacuum system

Tn range Range p required . . . . . .

o .

P

I I I I I I I 5 I0 15 20 25 30 35

Time, rain

Figure 9. Mode 4 operation.

transducers. The control panel is easy to operate and a choice of five modes of operat ion is available. The use of the microcompute r will allow other modes to be readily added and the use o f ' look up' tables in the computer memory allows non l inear t ransducers to be readily used. This control ler is intended for use in the research labora tory and the many industrial processes using vacuum apparatus.

A c k n o w l e d g e m e n t s

The authors wish to thank Professor J H Leck for his invaluable discussions of the overall problem.

,I

o

4

73 2

O.

I 5

e I P3

I I I I I I0 15 20 25 30

Time, min

Figure 10. Mode 4 operation with three requested pressures.

R e f e r e n c e s

i j Lucas, I Griffiths and C Goodwin, Vacuum, 30, 159 (1980j. -' P Watkinson and W E Austin, Vacuum, 22, 261 (19721. 3 L G H Huxley and R W Crompton, The Diffusion and Drift of Electrons in Gases. Wiley-Interscience, New York (1974). "~ J W Limbeek and J Lucas, lEE Solid St Electron Devices, 2, 161 (1978). s H T Saelee and J Lucas, J Phys D, 12, 1275 (19791.

the pressure was in range it remained cons tant and in fact required only a further count of 1 over a period ofapproximate ly 20 min. In Figure 10 is shown the response of the control system to a change

in the requested pressure which is initially lower and then higher than the actual pressure. Initially the pressure rises from zero to the required pressure of 6.7 torr (_+ 2 0,,) achieving equil ibrium in about 8 min. Once the pressure was stabilized the required pressure was changed to a lower value of 3.8 torr (_+2"0) achieving equil ibrium in about 5 min. After the pressure had stabilized the required pressure was increased to 6.4 torr (+_ 2 °,,) and this was again achieved in about 5 min.

6. S u m m a r y

The automat ic pressure control ler is designed to be fully adaptable to a variety of vacuum chambers, pumps and pressure

A p p e n d i x I

(1) Press 'reset'. (2) Press either "automatic" or "manual'.

If automatic (3) Check that thumbwheel switches are set correctly. These refer to:

(a) Mode. (b) Time constant. (c) Range °o. (d) Requested pressure or requested motor counts.

(4) Press "go'. (5) Press "reset" to cease automatic control.

U manual (3) Press three operational switches.

(a) Reset counter. (b) Motor put into forward or reverse. (c) Motor on or off.

(4) Press 'reset" to cease automatic control.

263