Electrical nerve stimulation: re-designing,
producing and testing a
portable stimulator.
Nimesh Shah
Supervisors:
Dr. Anne Vanhoest
Professor Nick Donaldson
The Implanted Devices Group at UCL has necessity for an electrical
stimulator, and there is one that exists. This report covers how the old circuit
which produced a square pulse, and an exponential pulse was adapted. The
circuit was changed so that the square pulse had the capability of being a
stepped square pulse too. This was done by introducing a new monostable,
and extra logic components. The parameters that were preliminarily defined
were not completely adhered to. Therefore, further work still needs to be
conducted to refine the stimulator and there are suggestions concerning that.
Contents
Acknowledgement
Introduction
What nerve stim
specification
of the box
Old circuit
New circuit
Additional Circuit
Finished Stimulator
Typical Setup
Parameters
Waveforms
Periods max. min.
Amplitudes
Charge Balance.
Future Work
Bibliography
1
1. Introduction
1.1 What is Electrical Stimulation?
Electrical stimulation of a nerve is the generation of an action potential by the
application of a current waveform (in the instance of this project). Electrical
stimulation is used to restore function in neurologically impaired individuals.
Electrical stimulation can help with loss of hearing by stimulation of the auditory
nerve (Grill and Mortimer 1995), or foot drop in hemiplegic patients. (McNeal
and Bowman 1985).
It has been shown that the use of different pulse shapes have different
physiological effects (Vuckovic, Rijkhoff and Struijk 2004). It is important for the
Implanted Devices Group at University College London to have an electrical
nerve stimulator, which is able to produce different pulse shapes, and thus
differing effects.
The use of such a stimulator varies, from as a control when comparing against
another stimulator, or the ability to set one waveform and be able to see the
impact of changing the electrodes.
There is an electrical nerve stimulator which is currently in use, however, the
stimulator does not have the capabilities of producing a stepped square pulse.
This will be the scope of my project.
2
1.2 Specification of a Stimulator
In the 19th Century it was discovered that the relationship between the amount of
current needed to activate a nerve depended on the pulse width in a hyperbolic
relationship as shown in fig. 1.2.1 (Donaldson 2009)
Figure 1.2.1
Further investigation by Geddes and Bourland (1985) saw that the least amount of
current is needed when pulse widths are less than chronaxie. In humans chronaxie
occurs at about 100µs. The parameters shown in table 1.2.1, that I set out to
achieve given agree with these findings.
Table 1.2.1
Parameter
Minimum
Maximum
Frequency
5Hz
500Hz
Pulse Width
50µs
5ms
Delay
1µs
100µs
3
Figure 2.1.1
4
2. Protoboard Circuit
2.1 Old Circuit
This is the original circuit, as seen in fig. 2.1.1, that was first designed by Prof.
Donaldson and Eleanor Comi (2002). This was continued by Dr. Anne Vanhoest.
This circuit can switch between two types of pulses; a square pulse, and one that
decays exponentially as shown in figs. 2.1.2 and 2.1.3.
1
2
3
Figure 2.1.2
4
6
1
2
3
6
4
Figure 2.1.3
The desired waveform can be selected at Switch 2 on the circuit. It is important to
note that this is the logic circuit, which produces two voltage waveforms. There is
an additional circuit which is responsible for converting the voltage waveform, into
a current waveform this will be further discussed in section 2.3.
2.1.1 Clock Loop
We can clearly see the clock loop indicated in the circuit diagram in fig. 2.1.1. The
monostables involved, MONO1 and MONO2, produce the time intervals. MONO1
produces the interval 1-2, and MONO2 produces the interval 2-5. These intervals
are determined by the RC circuits R1C1 and R2C2, the two resistors are variable
and R2 is present on the front panel allowing user control of frequency. We can see
though 1-5 is the period, the user only has control over the interval 2-5. The
5
feedback into the NOR gate will be discussed in section 2.1.4.
2.1.2 Square Pulse
The formation is more easily described if each stage is considered individually. The
pulse is bound by the time intervals 1, 2, 3, 4, and 6.
Figure 2.1.4
The first stage of the pulse occurs in the interval 1-2. This is generated by MONO1
and is indicated as the clock loop in fig. 2.1.1. The returning pulse around the loop,
instant 6 triggers the monostable on the rising edge.
Figure 2.1.5
The interval 2-5 is determined by the second monostable which is triggered by the
falling edge of instant 2 and contributes to the determination of instant 6. Instant 6
is equivalent to instant 4, (the end of the charge balancing phase of either the
square pulse or the exponentially decaying pulse) when instant 4 is greater than
6
instant 5. Or, instant 6 is equal to instant 5 when instant 4 is less than instant 5. In
this case the frequency is as set by the user. This is a condition governed by the
feedback from the circuit, which in itself is dependent on charge balance into the
flip-flop and the NOR gate. This is shown in fig. 2.1.5.
Figure 2.1.6
The pulse exiting from MONO1 not only triggers MONO2, but also MONO3 as well.
This creates the interval between 2 and 3. With the interval 2-3, and interval 2-4
being output from the flip flop, all the time intervals needed for the square pulse
can be created. The highlighted pulse on fig. 2.1.6 is created by the connection of
switch 5 to a V+ source through a resistor, during the interval 2-3.
Figure 2.1.7
Interval 3-4 is not a pulse that is output by a monostable, we can however make it
by processing pulses with interval 2-4 and interval 2-3. The resulting pulse of
interval 3-4 causes a connection to a V- source at switch 7 via a resistor. This in
turn causes the highlighted part of the pulse, shown in fig. 2.1.7.
7
Figure 2.1.8
The time interval 4-2 is grounded, this is maintained by switch 6.
The combination of these pulses, and the resulting switch connections contribute to
the production of the square waveform. The front panel houses the variable
resistors controlling the durations of the pulse width (interval 2-3), and the period
(interval 2-5). Both of these durations are user controlled.
2.1.3 Exponentially Decaying Pulse
This pulse is selected by switching at switch 2 from the square pulse circuit to the
exponentially decaying pulse circuit. This circuit as designed by Prof. Donaldson
and E. Comi, previously mentioned, remains unchanged. It is important to have the
option of an exponentially decaying pulse, as it may be preferential to avoid a
break excitation. (Frankhaueser and Widen 1956)
This circuit works mainly in the same way as the square pulse, that is, using a
combination of switches to connect to V+, ground and V- at different given
intervals. The obvious difference is in the form of the pulse. The exponential decay
is achieved with the introduction of a capacitor (C6), and a variable resistor in
parallel with switch 2, which allows user control of the decay.
8
Figure 2.1.9
Once again, the interval 4-2 is kept at ground by S3. This is shown in the
highlighted part of the pulse in fig. 2.1.9.
Figure 2.1.10
The interval 2-4 is governed by S1 and S2. During the interval 2-4 S4 is connected
allowing these switches to have an impact during this time period. This is shown in
the highlighted part of the pulse in fig. 2.1.10.
Figure 2.1.11
The inverted form of interval 1-3 connects S1 to a high voltage via R5. Although S1
9
connects to a high voltage the waveform is limited by S2, which is connected from
intervals 2-4. Therefore the high voltage is shown in fig. 2.1.11 by the square pulse
at interval 2.
Figure 2.1.12
At time interval 3 S1 becomes disconnected, and S2 connects allowing the
discharge of the capacitor (C6) across variable resistor R4. We then get the abrupt
return to ground from the disconnection of S4 at interval 4. The adjustment of R4 is
done by the user, and the controls are situated on the front panel. This controls the
decay of the exponential pulse. This is shown in the highlighted part of the pulse in
fig. 2.1.12.
The combination of these pulses, and the resulting switch connections contribute to
the production of the exponential waveform.
2.1.4 Integrator, Comparator and Feedback
These two components are responsible for charge
balancing. The integrator measures the areas of the
curves in both the square pulse and the exponential
pulse (In both cases it is shown in fig 2.1.13, with the
positive being the lighter shade, and the negative being
the darker shade). The comparator sums the 2 areas
involved in the positive and negative phase of the
waveforms and produces instant 4 when the sum of the
Figure 2.1.13
integrals equal 0.
10
Interval 4 is created by the comparator; the flip flop assembles this with interval 2,
which is outputted from MONO2. This produces a pulse interval 2-4 from the non Q
output. This is then fed back into the NOR gate of the clock loop.
The NOR gate within the clock look sets the time interval 6. This NOR gate sets T6
as either T4 or T5, working on the condition:
T6 = T4 if T4 > T5
T6 = T5 if T5 > T4
Therefore, the instant 6 is either set by the user by way of adjusting the period
(T6=T5) or, is extended until the condition of charge balance is achieved (T6=T4).
2.2 New Circuit
The existing circuit, as previously described, produces a square pulse, and an
exponentially decaying pulse. Changing the square pulse to a stepped pulse has
been shown to allow the use of up to 30% lower charge (Vuckovic, Rijkhoff and
Struijk 2004).
It is then obvious that this would be a nice option to have. The change in waveform
that is necessary is shown below in figs. 2.2.1 and 2.2.2.
Figure 2.2.1
Figure 2.2.2
11
Legend for Figure 2.2.3
Resistors
Table 2.2.1
Resistor Number
Value (Ω)
R1
2k
R2
120 k
R3
200 k
R4
50 k
R5
200 k
R6
47
R7
47
R8
47
R9
200 k
R10
200
R11
-
R12
8k
R13
10 k
R14
1k
Capacitors
Table 2.2.2
Capacitor Number
Value (nF)
C1
10
C2
100
C3
47
C4
22
C5
50
C6
47
C7
1
12
13
Figure 2.2.3
There has been an introduction of a time interval, the interphase delay. The new
circuit is shown on the previous page in fig. 2.2.3.
The greyed area labelled “Square Pulse”, and the fourth monostable are the
changes which I have completed, they are shown in fig. 2.2.4. We previously saw
in section 2.1 how the original wave was generated.
Bibliography
1. Different Pulse Shapes to Obtain Small Fiber Selective Activation by Anodal Blocking - A
Simulation Study. Vuckovic, A, Rijkhoff, N and Struijk, J. 5, 2004, IEEE, Vol. 51, pp.
698-706.
Figure 2.2.4
14
The pulse with time interval 2-3 is generated from MONO3, this triggers the pulse,
with time interval 3-4 from the new MONO4. The period in which Sw5 is connected
to a high voltage is not affected by the new circuit, as the controlling pulse is
coming from MONO3.
The durations in which Sw6 and Sw7 are connected are altered. Sw6 was
previously connected from intervals 5-2, providing the ground state outside the
positive and negative phases. Sw6 is now connected to ground between the newly
formed interval 3-4 as well as seen in the pulse entering Sw6 in figure. The variable
resistor on MONO4 allows an adjustment of the duration of the delay.
As I introduced the new delay, we also had to change how S7 was connected, as
interval 3-4 is being used in the delay, we needed to create a consecutive interval
4-5. This is done by using the pulses from MONO3, MONO4 and the feedback
(due to charge balance)
As a late addition (not shown on the circuit diagram) the RC circuits on
monostables 3, and 4 were connected via a fixed resistor (R3F = 2kΩ, R4F = 430Ω)
to high voltage. This ensures the resistance never falls to 0.
The combination of these pulses, and the resulting switch connections contribute to
the production of the stepped square waveform as shown in fig. 2.2.5.
Figure 2.2.5
15
As we have introduced a new time interval the conditions of when interval 7
change to:
T7 = T5 if T5 > T6
T7 = T6 if T6 > T5
2.3 Additional Circuits
16
The circuit I have been working on is the logic circuit which is produces a voltage
waveform. The circuit shown in fig 2.3.1is responsible for converting this voltage
waveform into a current waveform. The voltage waveform is initially split into two
proportional waveforms, there is then a voltage to current conversion stage, which
results in the current waveforms for the anodes.
Figure 2.3.1 shows the circuit responsible for this conversion, also in the circuit are
stages that deal with voltage conversion, discharge of the blocking capacitor,
isolation of the signal and saturation. However this is something that I have not
worked on, and I will not focus on it. It is however, important to acknowledge it as it
is an important part of the stimulator.
Figure 2.3.1
3. Finished Stimulator
17
3.1 Typical Setup
Power
Generator
Oscilloscope
Circuit
Figure 3.1.1
Figure3.1.1illustrates the typical setup of the protoboard, and the instruments.
18
3.2 Parameters
As mentioned in the previous sections the parameters that we set out to achieve
are shown in table 3.2.1.
Table 3.2.1
Parameter
Minimum
Maximum
Frequency
5Hz
500Hz
Pulse Width
50µs
5ms
Delay (square pulse)
1µs
100µs
The parameters that the stimulator actually produces are shown in tables 3.2.2.
and 3.2.3.
Stepped Square Pulse
Table 3.2.2
Parameter
Minimum
Maximum
Period
215µs
10ms
Frequency
4600Hz
100Hz
Pulse width
50µs
9ms
Delay
16µs
1.1ms
Exponentially Decaying Pulse
Table 3.2.3
Parameter
Minimum
Maximum
Period
200µs
41.5ms
Frequency
5000Hz
24Hz
Pulse Width
100µs
9.5ms
Decay
≈0s
41ms
The implications of these results are discussed in section 4.
19
3.3 Waveforms
To demonstrate the capabilities of the stimulator I have set various values and
explained them below.
3.3.1 Stepped Square Pulse
3.3.1.1 Frequency
Figure 3.3.1
The adjustment of the variable resistor in the RC circuit of monostable 2 allows the
control of the time interval 2-6 i.e. a control of frequency. We have previously seen
the condition of:
T7 = T5 if T5 > T6
T7 = T6 if T6 > T5
20
Altering the time interval 2-6 clearly will then alter the frequency of the waveform.
Fig. 3.3.1 shows a high frequency, in this particular case the period of the
waveform is 650µs, which corresponds to a frequency of approximately 1.5 kHz.
Figure 3.3.2
Alternatively, we are also able to adjust the resistor so that the time interval 2-5 is
very large, reducing the frequency markedly.
In this instance, the period of the waveform shown in fig. 3.3.2 is 4ms,
corresponding to a frequency of 250Hz
21
3.3.1.2 Pulse Width
This is specifically the duration of interval 2-3. As explained previously the
integrator and comparator produce instant 5 when the condition of charge balance
has been met.
Figure 3.3.3
Fig. 3.3.3 shows how the pulse width can be adjusted. This is done by changing
the value of the variable resistor in the RC circuit of monostable 3. In this instance
we can see the pulse width is at 100µs, the delay is 150µs and the duration until
the next waveform is 200µs. The period of the waveform is 550µs.
22
Figure 3.3.4
The pulse width approaching instant 7 can be seen in fig. 3.3.4, which in this case
is instant 6. In this instance the pulse width is 200µs, and the delay is 150µs. the
period of this wave is also 550µs.
23
Figure 3.3.5
Fig. 3.3.5 demonstrates the conditional control over instant 7. Increasing the pulse
width, increases the duration of the negative pulse and forces the integrator and
comparator to delay instant 5. This increase in pulse width, results then, in a
decrease in frequency. In this specific example the pulse width shown is 500µs,
and the delay is still 150µs. The period of this waveform is 1150µs, perfectly
showing us how the frequency decreases on increase of pulse width.
24
3.3.1.3 Delay
Figure 3.3.6
Fig. 3.3.6 and fig 3.3.7 show us the limits of the delay. The duration of the delay
can be altered by changing the value of the variable resistor in the RC circuit of
monostable 4. Fig 3.3.6 is showing us how long the duration of the delay can be.
This picture shows us that the maximum delay is 1.1ms.
25
Figure 3.3.7
The scale on the oscilloscope has been adjusted in fig 3.3.7 so we can clearly see
the duration of the delay. On setting the delay to as low as possible the waveform
appears to be a square wave with no step, however, further investigation shows us
that a delay is still present, with a duration of 16µs.
26
3.3.1.4 Amplitude
Figure 3.3.8
As a late addition (which is not shown on the circuit diagram) I added a variable
resistor between the low voltage and the switch. This allows the negative phase to
be “shallower”. This feature is also demonstrated on the front panel design shown
in section 4.2 as both the positive and negative voltage can be set by the user. Fig
3.3.8 shows how the charge balance is maintained when the low voltage is
changed.
27
Figure 3.3.9
The response of the circuit to a variation of the positive voltage amplitude is shown
in fig. 3.3.9, the maximum amplitude previously was 5V relative to ground. Here the
amplitude is 10V, there is charge balance though the low voltage is still -5V, which
explains the lack of symmetry.
28
Figure 3.3.10
Fig 3.3.10 shows the breakdown of the waveform at 3.50V when the positive
voltage was reduced. However, this is due to the fact that some components were
not receiving enough power at 3.50V to produce the pulses. The voltage supply
was the same to both the components and the V+ that determines the amplitude of
the waveform.
29
3.3.2 Exponential Pulse
Figure 3.3.11
Though nothing has been altered in the exponential pulse, I made the circuit just to
check the durations of the time intervals. A typical exponentially decaying
waveform can be seen in fig. 3.3.11.
30
3.3.2.1 Pulse Width
Figure 3.3.12
As with the square pulse we are also able to reduce the duration of the pulse width
in the exponential pulse. The pulse width was reduced to 150µs as shown in fig.
3.3.12.
31
Figure 3.3.13
Fig. 3.3.13 shows the effect of increasing the duration of the pulse width to 1ms.
32
3.3.2.2 Exponential Decay
Figure 3.3.14
By altering the variable resistor across which C5 discharges, we can alter the time
taken for the decay. In this situation the capacitor discharges so fast it almost takes
on the appearance of a square pulse, this is shown in fig. 3.3.14.
33
Figure 3.3.15
Conversely, we can change the value of the resistor so the time taken for the
capacitor to discharge is longer. Here it takes 2.6ms for the charge to be balanced
and a return to ground it is shown in fig. 3.3.15.
34
4. Future Work
4.1 Parameters
We can see from the protoboard circuit that the circuit is functional, and is
producing charge balanced waveforms. As seen before in section 3.2 there is not
perfect agreement in the parameters that I set to achieve, and the ones that were
then consequently attained. As we have seen the time intervals are set by the RC
circuits attributed to the various monostables. Future work will consist of altering
the values of the resistors and capacitors, to achieve the parameters defined.
Monostables 1 and 2 follow equation(see Appendix B):
Two = K.Rt.Ct
(1)
Where:
Two = Output pulse width (s)
Rt = External timing resistor (Ω)
Ct = External timing capacitor (F)
K = 0.42 for VDD = 5v
Monostable 2 is responsible for the interval 2-5, the main interval responsible for
period, and therefore frequency.
The periods necessary for the frequencies are 2ms – 0.2s. The R2C2 circuit can be
adapted to fit these values better. Working from equation (1) I could recommend a
variable resistor value of 100kΩ, and a capacitor value of 47µF. This gives you a
variable period of 0.198ms to 0.198s.
The pulse width conforms quite well, however the upper range may need to be
tapered. The delay, doesn’t fit perfectly either, the R4C4 circuit could also be better
optimised. This monostable has equation:
35
tW = K.REXT .CEXT
(2)
Where:
tW = output pulse width in ns;
REXT = external resistor in kΩ;
CEXT = external capacitor in pF;
K = constant = 0.55 for VCC = 5.0 V
The duration of the delay that we would like is 1µs to 100µs. the values of R4, R4F,
and C4, according the formula allow the delay to go as low as 5µs. This however,
is not reflected in the measurement, this could be due to the tolerance in
manufacturing of the resistors. The delay can be lowered to 1µs, with a decrease in
R4F.
Currently, all the parameters are adjusted using variable resistors. In the future, it
might be worth considering the use of variable capacitors, in cases where a
variable resistor is impractical.
4.2 Box Design
The next step of this project is to make the box. the design of the old box was not
intuitive, and I have therefore redesigned the front panel with approximate
dimensions as shown in figure 4.2.1
I wanted the new front panel to be intuitive, that someone would be able to work
the stimulator who had a knowledge of electrical stimulation, and not necessarily
experience in working with this specific stimulator.
This new front panel is designed in vertical panels. The top left box, indicates the
initial control of the box where the power, triggering, and trigger coupling is set.
Below that is where the leads for the power are inserted; they are sufficiently far
away to not interfere with the operation of the box.
To the control panels right is where the user is able to set the amplitude of the
36
wave. This is show in fig. 4.2.1. To the right of the amplitude control are the
waveform parameters, where the user is able to set pulse width, period, delay, and
decay durations.
Furthest right is the adjustment for the BNC sockets. The BNC sockets go to an
oscilloscope, they are separated from the circuit by isolation amplifiers that are
powered by batteries. The top control knob selects which of the BNC sockets,
outputs a signal. The LEDs associated let the user know when there is a current
pulse, and when there is saturation.
The bottom control knob allows user control of the ratio of I1/I2 from the 4mm
sockets going to the electrodes.
250mm
150mm
Figure 4.2.1
37
The designing of the box runs in parallel with the making of the PCB, which also
has yet to be done. Once the PCB is made, populated and tested. The box can be
finally assembled.
38
Appendix A
Logic Gates
Some of the time intervals are created by the processing of other time intervals
through logic gates. The truth tables for the logic gates that I have included in the
circuit are drawn below.
NOR
p
q
p↓q
0
0
1
0
1
0
1
0
0
1
1
0
NAND
p
q
p↑q
0
0
1
0
1
1
1
0
1
1
1
0
EXNOR
p
q
0
0
1
0
1
0
1
0
0
1
1
1
Inverter
p
¬p
0
1
1
0
39
Appendix B
Monostables
There are 4 monostables that have been used in this circuit, they are triggered on
either a rising, or falling edge depending on how the chip has been wired. On this
edge the monostable produces a pulse of standard time, which is determined by
the RC circuit. Monostables 1 and 2 are on chip HEF4528B. They follow equation:
Two = K.Rt.Ct
(1)
Where:
Two = Output pulse width (s)
Rt = External timing resistor (Ω)
Ct = External timing capacitor (F)
K = 0.42 for VDD = 5v
The datasheet for chip HEF4528B can be found on:
http://www.datasheetcatalog.com/datasheets_pdf/H/E/F/4/HEF4528B.shtml
Monostables 3 and 4 are on chip 74HC/HTC123. They follow equation (when
values for CEXT exceed 1nF):
tW = K.REXT .CEXT
(2)
Where:
tW = output pulse width in ns;
REXT = external resistor in kΩ;
CEXT = external capacitor in pF;
K = constant = 0.55 for VCC = 5.0 V
The datasheet for 74HC/HTC123 can be found on:
http://www.digchip.com/datasheets/parts/datasheet/364/74HCT123-pdf.php
40
Bibliography
Donaldson, Nick. "Stimulation Basics." Lecture 18 MPHY3013, Medical Electronics. 2009.
Frankhaueser, Bernhard, and Lennart Widen. "ANODE BREAK EXCITATION IN
DESHEATHED FROG NERVE." Journal of Physiology, 1956: 243-247.
Geddes, and Bourland. "The Strength Duration Curve." IEEE transactions on bio-medical
engineering, 1985: 485.
Grill, Warren M, and J.Thomas Mortimer. "Stimulus Waveforms of Selective Neural
Stimulation." IEEE Engineering in Medicine and Biology, 1995: 375-385.
McNeal, D R, and B. R. Bowman. "Selective activation of muscles using peripheral nerve
electrodes." Medicinal & Biological Engineering & Computing, 1985: 249-253.
Vuckovic, Aleksandra, Nico Rijkhoff, and Johannes Struijk. "Different Pulse Shapes to
Obtain Small Fiber Selective Activation by Anodal Blocking - A Simulation Study."
IEEE, 2004: 698-706.
41