723

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 31, NO. I , JULY 1990

Design for a Constant Peak Current Defibrillator JOHN WILLIS MACHIN. JULIE BROWNHILL,

Abstract-A damped sine wave defibrillator that is capable of delivering a constant peak discharge current over a wide range of patient thoracic impedance is shown to be theoretically possible. This is realized in practice by the technique of storing an amount of energy above that required for delivery, and dumping the excess charge when the discharge current has reached a predetermined level, the dumping being triggered by a current sensor. A larger than normal inductance assists in swamping the effect of varying impedance and produces a smooth waveform. The preselected peak current can be delivered with a variation of about +6% to 95% of a normal patient population. The energy expended in a simulated patient circuit is somewhat lower than for conventional defibrillators.

I. INTRODUCTION N THE PAST decade or so the consensus view has developed that the effectiveness of countershock treatment for defibrillation is closely related to the current density within the heart. In transchest defibrillation, it is related, although not simply, to the peak current attained during discharge [ 5 ] ,[21, [121, [81. Defibrillators are commonly calibrated either in joules of energy initially stored in the capacitor or in energy which would be dissipated in a load of 50 Q . This value may be readily related to peak discharge current if the patient’s thoracic impedance is known and is constant. In practice, the clinician can usually do no more than make an estimate based upon past experience, taking into consideration the physical attributes of the patient. Geddes et al. [6]and Kerber et al. [7]described experiments using a high-frequency current to obtain a precountershock measurement of impedance with some accuracy, but even this value can be to some extent modified by variations in selected charge energy and by repeated discharge. The problem can be resolved if a constant, selectable peak current can be delivered for the range of patient impedance generally encountered. Such a system has the added advantage that the electrode-patient interface, which has been shown by measurement on dogs [ 4 ] , [13]to give an impedance variation between 14 and 19%, ceases to have any significance. Observations of peak current variation, taken over patient population samples of around 150 discharges, seem

I

Manuscript received October 11, 1988; revised July 20, 1989. This work was supported by British Heart Foundation Grant Number 83/79. J . W . Machin and J . Brownhill are with the Department of Electrical and Electronic Engineering, Staffordshire Polytechnic, Stafford ST18 OAD, England. A . Furness is with Staffordshire Design and Technology Center, Staffordshire Polytechnic and the University of Keele, Keele, Staffordshire, England. IEEE Log Number 903445.

AND

ANTHONY FURNESS

to indicate that the problem will be easier of solution if a fairly high initial charge voltage, say about 7 kV, is used [lo]. 11. CONSTANT PEAK-CURRENT DEFIBRILLATION DESIGNS

Three candidate designs were considered for the implementation of constant peak current. These will be briefly described.

A. An Active Element Constant-Current Regulator The basic circuit is shown in Fig. 1 in which the peak current level is controlled by the zener diode Z and the resistance R . The shape of the falling edge of the waveform is modified by the diode D and resistance r. The necessary high voltage and current transistor however, is not presently commerically available, nor is the waveform ideal.

B. A Machine Using Energy Stored in an Inductor Instead of in a Capacitor An approach on these lines has been proposed and investigated [ 111 and enables a peak current to be setup with a considerable degree of accuracy, largely independent of patient impedance. The system would however appear to be impracticable as a portable unit. C. A Charge-Dumping Circuit with a Larger than Normal Inductance From the candidate designs this was the system chosen for further investigation. The theory and practical details are now described. 111. DESIGNTHEORY When a charged capacitor ( C ) is discharged through an inductance ( L ) in series with a resistance (I?), the resulting current ( i ) after time t seconds is expressed in the following equation:

i

VO

= - exp

[x] -Rt

.

sin wt.

WL where V, is the initial charge voltage. The terms C , L , and R are also implicit in the w term,

LC R2 4L

4L2

and if

1 -

LC

> 7the circuit is oscillatory.

0018-9294/90/0700-0723$01.OO O 1990 IEEE

124

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 37, NO. 7, JULY 1990 R

Patient

Fig. 1. Active-element constant current regulator.

The rate of current increase is given by

[x]

Vo -Rt di = - exp dr WL --

[w cos wt - - sin wt 2L

From (2) it is seen that when (3) which is independent of the resistance R so that the initial gradient of the discharge curve is determined by the charge voltage. For purposes of defibrillation we require to know the maximum current attained during discharge. This value 2 occurs at the instant when d i l d t = 0, i.e., when w cos at = R/2L sin wt or tan w t = 2wL/R. If the time for this state to be reached is t l then t,

= W

[

tan-'- 2 i L ] a n d

and 90 Q , and the peak current found to vary by about +4.0% on the middle value at 60 Q . On this basis the system appears to be workable, but it suffers from the practical disadvantages of a large and heavy inductor and a very high voltage on the capacitor, approximately 14 kV for a 400 J charge. These results are shown in Fig. 2(b). This circuit is oscillatory but can be made unidirectional by connecting a diode as shown in Fig. 2(a). The shape of the curve after time t , can be modified by varying the resistance I-. An alternative approach is to make use of the information in (3). If Vo is made larger than normal and the circuit is discharged using three different values of R (say 30, 60, and 90 Q as before), with C of 16 p F and L of 90 mH, waveforms shown in Fig. 3 are obtained. At point A the three curves differ only slightly, the difference diminishing if Vo is further increased. The circuit is now modified so that when A is reached a small resistance is connected across the capacitor to dump the remaining charge. The system at this point approximates to an LR circuit and the current decays as shown by the broken line in Fig. 3. The maximum value of current I reached is not greatly altered as R is varied, provided that the switching action at A is very prompt. The duration of this pulse will be somewhat shorter than that of the unmodified circuit. It was thought that a satisfactory compromise might be obtained between the two systems represented by (3) and (5) by using the charge dumping technique coupled with some increase in the value of inductance. The experimental investigation of such a system is described below.

A. Computer Simulation VO I = - exp WL

-Rtl

[ T I sin wtl.

(4)

Considering the case where L is large, 1 I L C R2/4L2, w = W C and

>>

a JLC 2wL tan-' - = n / 2 or t I = n/2w or R 2

Thus, Rt,/2L exp

=

nR/4.

Jc/L which

is small, so that

[2]

= 1, also r--

a t , = a / 2 and (4) becomes 1 =

v, = WL

It can be seen from (5) that I depends only on V, provided the condition mentioned above can be fulfilled. In practice of course this condition cannot be obtained. However, an experiment was carried out at low voltage (30 V) using an inductance of 633.35 mH with a capacitance of 4 pF as shown in Fig. 2(a) to examine to what extent the ideal could be realized in practice. The load resistance R (representing the patient) was set at 30, 60,

Fig. 4 shows the circuit used in the computer simulation studies. Rs represents the subject impedance, which is taken to be purely resistive, Ra is the resistance of the charge dumping circuit, and r is the internal resistance of the defibrillator. The capacitor is charged to the required voltage and switch SI is closed. When the current in Rs has reached a predetermined value, switch S, closes and the remaining charge is dumped through the resistance Ra. The simulation was undertaken using SPICE circuit analysis package supported on a VAX computer. A range of component values were considered, to provide a suitable pulse duration and allow, by experiment, a choice to be made for the best value of dump resistance. For this purpose the mean value of impedance was again taken as 60 Q (a value associated with adult male subjects) with variations between 30 and 90 Q , these being approximately two standard deviations on either side of the mean, assuming normal distribution. The results obtained suggested that the following values might provide acceptable constant peak current performance: Capacitance

18 pF

Inductance

200 mH

725

MACHIN et al. : DESIGN FOR CURRENT DEFIBRILLATOR

>

Patient Impedance

R

30n-gon

40r

\

(b) Fig. 2. Large inductance machine

80-

-

Normal discharge;

a: 60n, b:30n,c : 9 0 n

-

60

---- Discharge with damping

a L

40-

0

0 a

20-

----

0-

I

8

Ib

-

-

ms

Fig. 3. Effect of charge dumping

Internal resistance 29 Q (dictated by the available inductance)

Dump

25 Q

The results for this configuration can be seen in Table I. These results for the 9 kV case are shown graphically in Fig. 5 .

Fig. 4. Circuit for computer simulation.

T

I

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 31, NO. 7 , JULY 1990

126 TABLE 1 Patient Impedance R,

60 R

90 Q

30 R

9 kV 7 kV 5 kV

40 A 30 A 20 A

-3.9% -3.5% -3.6%

+6.1 F +5.6% +7.8%

C

Patient

5c

40

I

30

‘6on

-

2 30A 3-9OA

Cur rent A

input from 20

IineA

(Figure 6 )

dump resistance

10

0

Fig. 5 . Digital computer results

120kA

0 5w

B. The Experimental Circuit Using the component values suggested by the computer simulation the basic circuit shown in Fig. 6 was constructed and evaluated. SI represents the high-voltage relay and S2 and Rd represent the dump or “crowbar” circuit. Ri and R, provide the current and voltage output signals. The simulated patient impedance was made up of 10 X 10 Q , 11 W wirewound resistors connected to allow cancellation of a slight self-inductance. The need for fast operation of the “crowbar” circuit (Fig. 7) precluded the use of a mechanical relay. A hydrogen thyratron, capable of withstanding 9.5 kV and able to pass a transient peak current of up to 300 A, was used for achieving the fast ‘‘crowbar” function. The thyratron was triggered, via an isolating interface and grid pulse shaping circuit, by a comparator tripped at the required current level with a signal from R;. The dump resistors used were noninductively wound. However, a small, saturable inductor was needed to limit the initial rate of rise of current when the thyratron fired. Since the “crowbar” circuit was meant to simulate a closed switch, conduction had to be possible in both directions. This was achieved by “flywheel” diodes and an associated potential grading network. By including extra resistance in series with the diodes a certain amount of pulse shaping was possible. The energy dissipated in the load during the discharge was measured by integrating i2 R, over the period of the discharge. An AD532 integrated circuit multiplier was used to square the current signal which was then integrated using a chopper-stabilized operational amplifier

1 Fig. 7. “Crowbar” circuit.

(7650). It was necessary to include a large resistor ( 3 . 9 M Q ) in parallel with the integrating capacitor to prevent drift when no input was applied. This circuit is shown in Fig. 8. The procedure for operating the system for experimental evaluation involved charging the capacitor to a voltage slightly higher than that required, whereupon the excess charge was allowed to leak away through the potential grading resistors connected across the “flywheel” diodes. When the voltage had drifted down to the required value the discharge was initiated. The maximum charge voltage was limited to 9 kV to avoid any problems with insulation requirements. The maximum stored energy was then 745 J. A charge voltage of 9 kV was required to provide a peak “patient” current of 40 A; 7 kV was used to provide 30 A, and 5 kV to yield 20 A. The trip level of the comparator circuit was adjusted until the required peak current, with a 60 Q simulated patient impedance, was obtained. With the same trip level the discharge was then observed for impedance values of 90 and 30 Q . The output from the comparator circuit (Fig. 9) was displayed with the current waveform to aid measurement

MACHIN et ol. DESIGN FOR CURRENT DEFIBRILLATOR

727

'

0.5n

41

,

100 kn

,I0.1pF

... *.. II

3 . 9 ~

IOOkn

55

lookn

P

1,

/

Crowbar

Voltage V

I f

Voltage

"

Current A

10

0 1

1

1

1

1

1

0

1

2

3

4

5

1

6

1

7

1

,

8

9

Fig. 9 . Discharge waveforms.

1

1

1011

1

1 2 m s

lOOnF

728

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 31. NO. 7, JULY 1990

of the time taken from the start of the discharge to the operation of the thyratron. The output indicates that the dumping procedure produces a voltage transient in the current sensing circuit. The current flowing in the dump resistors was measured with a ring-type current transformer with a ratio of 300 : 1. A 0.5 Q resistor across the secondary provided the voltage signal source for observation purposes. The various waveforms were displayed on a Nicolet 309 1 oscilloscope having movable cursor and digital display facility.

3530' 25-

Current A

2015-

105-

0-

IV. RESULTS The waveforms for peak currents of 40, 30, and 20 A are shown in Fig. 10. The best results on the basis of peak current were obtained with a dump resistance of 25 Q and are shown in Table 11. The maximum current available from this circuit, with a 60 Q simulated patient, was around 44 A. However, for the higher patient impedance of 90 Q the thyratron tripped almost at the peak of the waveform and this produced a distortion as shown in Fig. 11. It will be further observed from Figs. 10 and 11, and from Tables I1 and 111, that a change in patient impedance affects the pulse duration (f,,) to some extent, and also the energy ( J , ) dissipated in the patient; the former increasing with reduced patient impedance and the latter decreasing. Although 95% of patient impedances appear to lie between the two values of 30 and 90 Q , measurements were taken at 20 and 100 8 to determine whether the performance would deteriorate at extreme limits. The percentage difference between peak currents was from +8.4 to -7.8%. From Table I1 it can be seen that, for a given comparator setting, the dumping action does not occur at quite the same level of discharge current ( I , ) . This is due to the fact that while the time taken for the trip to operate is constant, the rate of rise of current is greater for a 30 Q load than for one of 90 Q . If the interval between the operation of the comparator and the firing of the dumping device can be reduced the variation of peak current will be reduced also. Referring again to Fig. 9, the scaling factors of the integrator circuit were arranged so that the output measured had only to be multiplied by the patient impedance to provide the energy dissipated. Table I1 shows the energy values obtained from this circuit along with a few available corresponding energies from two commercial defibrillators obtained from earlier in vivo experiments [ 101. It appears that, with the experimental defibrillator, less energy is expended in the patient circuit for approximately equal current. The current in the crowbar circuit is also shown in Fig. 9. The negative part of the waveform is caused by the reversal of current through the dump resistors. The max-

30-

Current

25-

A 201510-

5-

0-

251 20

Current A

Fig. 10. Waveforms for peak currents of 40, 30, and 20 A

imum current in this circuit was 238.2 A; this was at a peak current of 47.12 A in a 30 Q patient. In an attempt to obtain greater peak currents a different inductor was substituted. This had fewer turns which reduced the resistance from 25 to 17.2 Q, but a laminated magnetic core was needed to bring the inductance up to 200 mH. The protocol was as before and the results obtained are shown in Table 111. The reduced resistance improved the maximum peak current that could be obtained (around 50 A ) but the percentage difference between the peak currents was greater than before. The period of the discharge waveform was less (typically down to 7.46 ms from 8.07 ms), and the rising curve was slightly concave. These effects are caused by the magnetic core becoming saturated at high levels of current and thus reducing the effective inductance in the circuit.

V . DISCUSSION OF RESULTS The results from the experiments confirmed the computer predictions that, within limits of around f 6 % , it is possible to deliver a preset constant peak current to rep-

729

MACHIN er al.: DESIGN FOR CURRENT DEFIBRILLATOR

TABLE 11

C = 18.4 p F J , = Energy dissipated in patient Cjoules) L = 200 mH (no core) Rd = 25 Q r=29R HP = Hewlett-Packard 78670 A CR = Cardiac Recorders 61 A I, = Current at tripping level t, = Time before operation of trip r,, = Pulse duration (taken at 0.1 of maximum current)

U, 60 90 30 60 90 30 60 90 30 60 90 30 60 100 20

I

1

3

V,, kV

Imax A

9 9 9 7 7 7 5

40.08 37.78 42.46 30.02 28.46 32.06 19.90 18.78 21.26 44.26 41.50 47.12 40.00 36.90 43.38

5 5 9 9 9 9 9 9

4

s

6

7

a

9

io

1, A

t, ms

tp ms

37.68 36.76 38.06 27.84 27.24 28.38 17.76 17.50 18.12 43.32 41.42 44.36 37.64

1.28 1.42 1.16 1.18 1.30 1.08 1 .oo 1.08 0.94 1.67 2.30 1.45 I .25

8.07 6.66 11.07 7.98 6.51 10.75 7.78 6.13 10.44 8.40 7.37 11.08 8.01

1'1

112

1'3

14

lams

TABLE 111 C = 18.4 p F L = 200 mH (with core) r = 17.2 R Rd = 25 R V

Design for a constant peak current defibrillator.

A damped sine wave defibrillator that is capable of delivering a constant peak discharge current over a wide range of patient thoracic impedance is sh...
619KB Sizes 0 Downloads 0 Views