|
3. Design Solution | 5. Testing of the Device | |
4.1 Detectors
4.2 Pre-amplifiers
4.3 First difference amplifiers
4.4 Automatic gain control (AGC)
4.5 Final difference amplifier
4.6 Power supply
The individual parts of the block diagram of figure 7 will now be described in more detail, paying particular attention to the Automatic Gain Control (AGC) circuits, which are complex in comparison to the other circuit elements, and require sensitive adjustment.
The two position sensing detectors are the type PIN SC/1OD lateral effect transistors. These transistors have a common connection and four side connection pins to the sides of the active area, as illustrated in figure 8. If a narrow beam illuminates the transistor, then the output from each connection is proportional to the distance of the light spot from that side of the detector area. Thus the difference between outputs from two opposite side connections is linearly dependent on the horizontal position of
the light beam on the detector, ignoring changes in beam intensity. Because the beam intensity is liable to change during the muscle contraction, this must be compensated for.
Movement of the beam in the vertical direction was found to alter the output at the horizontal connections. Therefore it is recommended that in use, the light beam is either focused onto the central horizontal area of the detector surface using a Fresnel lens, or the detector covered except for a narrow horizontal slit (See fig 9).
[Diagram 1 shows the main circuit diagram]
The circuit around IC1 is a simple operational amplifier circuit in negative feedback configuration. This amplifier biases the lateral effect transistor and increases its output to a level which is suitable for further processing. IC2-4 are similar circuits, to amplify the other inputs from the detectors.
In these high gain amplifiers it is important to use the offset-null facility of the type 741 op-amp, in order to get an accurate representation of the input signal, with no d.c offset. VR1-4 are multi-turn presets used for this purpose. To adjust the offset null of a particular op-amp, the input should be connected to ground through a low resistance such as 100 ohms, and the output of the op-amp (pin 6) monitored with an accurate voltmeter. The preset is then adjusted until the output is zero.
The amplified signals from the two opposite sides of each detector are fed to the circuit around IC5 and R5, 6, 11 & 14. This op-amp is configured as a unity-gain difference amplifier, and subtracts the amplified signal from input 2 from that of input 1. A similar circuit, consisting of IC6 and R15, 16, 20, 24 takes the difference between inputs 3 and 4.
As mentioned earlier the intensity of the two beams incident on the detector is likely to vary during muscle stimulation. Therefore it is necessary to design a circuit to compensate for this variation, on a time scale so fast that the final output will give a true indication of the beam separation, undistorted by intensity variations that can occur on a time scale of the order of 5mS.
Considerable detail will be presented here, as the design of this crucial part of the circuit occupied by far the most time during this project. Also, it is an interesting and innovative circuit, in comparison to the fairly standard op-amp configurations used elsewhere in the design.
At first sight this may seem a formidable task; however a reasonably simple solution was designed using a transconductance operational amplifier. The type LM13700 chip is a dual transconductance op-amp; the difference between these amplifiers and the usual, straight-forward op-amps is as follows. In a normal op-amp the amplifier gain is determined by the combination of resistors used in the negative feedback loop, whereas in a transconductance op-amp, the gain is governed by the current through an extra
input connection (pins 1 and 16)
The original idea (illustrated in diagram 3) was to sum the signals from the two sides of the detector: this sum is independent of beam position and varies solely as intensity. This signal is passed through a transconductance op-amp, and the result compared with a preset value using an ordinary 741 op-amp configured as a comparator. The output of this comparator charges up a capacitor when the incoming sum signal falls below the preset value. The voltage across this capacitor, when fed back to the gain control input of the transconductance op-amp acts to increase the gain, thus compensating for the fall in intensity.
After a small interval (determined by the time constant of the R,C combination), the gain will have increased too much, and the amplified sum will now be above the preset value, causing the comparator to change state and start discharging the capacitor again. This reduces the gain, until such a level that the whole process starts again. Thus whatever the input sum signal, the gain of the amplifier always adjusts itself to bring its output up (or down) to the preset value. The same signal that alters the
gain of this transconductance op-amp can also be used to control another, this one amplifying the difference signal. In such a way, the difference output is amplified if the intensity is low,
attenuated if the intensity is high (relative to the preset intensity) This signal is now a 'true' difference, independent of the intensity of the incident beam.
In order to ensure that the gain fluctuations are small enough not to cause a significant oscillating signal to modulate the amplified difference, thus distorting it, the time constant of the R/C combination must be relatively large. Unfortunately, when the time constant is increased, this increases the response time of the circuit. Since the intensity can change on a time scale of only milliseconds, the response must be faster than this. It was found to be impossible to have a low enough time constant to allow fast response, and yet at the same to time have a large enough time constant to keep distortion of the amplified signal below a reasonable level.
The solution is in a deficiency of practical op-amps. The ideal, perfect op-amp would have the following characteristics:
It would:
i) Draw no input current,
ii) Have a closed loop gain dependent only on the feedback elements of its host circuit,
iii) Drive a low resistance load,
iv) Have constant gain with respect to frequency,
v) Be able to follow indefinitely sharp pulses independent of amplitude.
This last characteristic is particularly important here. A practical, real-world op-amp has a 'slew rate' usually measured in V/µS, which is the maximum rate at which its output voltage can change, due to its internal circuit. A typical value is 1V/µS, which is close enough to ideal in most situations, as to be neglected.
However, in this design, it was found that this deficiency of the practical op-amp can actually be used to advantage here. Removing the capacitance altogether and coupling the comparator output directly through a resistor to the gain control of the transconductance op-amp, the time constant is now so small (only determined by the op-amps internal capacitance) that the comparator op-amp cannot change its output fast enough to keep up with the changes at its inputs. Hence its comparator function is no
longer a simple fully-positive-if-greater-than, fully-negative-if-less-than output, but rather a voltage somewhere in between, that depends on how much is the difference between its inputs. It should be noted that this is not just another form of difference amplifier: here there is no negative
feedback etc, the op-amp is configured as a comparator, but due to the finite slew rate, its output cannot change fast enough. A normal difference amplifier cannot be used here.
This novel usage of a component's deficiencies, rather than the usual constant attempts to minimise them, solves both problems with the automatic gain control circuit: firstly, since there is no actual switching between fully positive and fully negative, the gain of the transconductance op-amp does not oscillate strongly as before; secondly, because there are no external time-constant components to slow down the reaction time, the reaction time is extremely fast, in fact limited only by the fourth op-amp deficiency listed above: the finite frequency response of practical op-amps. In this application, the response time is far less than that required by the time scale of intensity variations. This was demonstrated during later testing.
With reference to the main circuit, diagram 1, the automatic gain control is implemented in the circuit around IC7a, and IC9, with the automatic gain control signal routed to IC7b which amplifies the primary difference signal. The output of this amplifier is the intensity independent difference between the
inputs 1 & 2.
IC6, IC8a, and IC8b form a duplicate circuit providing AGC for the difference between inputs 3 & 4.
It should be noted that as the beam intensity changes, the AGC circuit will compensate for changes in the total light incident on the detector, hence any ambient light in the lab will, if unshielded, cause the device to operate inaccurately. Therefore the detectors must be well shielded from ambient lighting, or the experiment conducted in a darkened room. If this is likely to be a problem, op-amp subtracter circuits could be added directly before the inputs to the AGC circuits, i.e at pin 4 of IC7a and IC8a, to remove the ambient light contribution to the signal by subtracting a predetermined amount representing the light. However, if the ambient light is not constant, for instance if it is altered slightly by people moving around in the room, or if it is directional so that it has more effect at one side of the detector than the other, the output will still be affected. Hence it is recommended that the ambient light is prevented from reaching the detectors, either by shielding or by use in a darkened laboratory.
The AGC circuits are adjusted for the required range of intensity conditions using the presets VR5 & VR6. To make the adjustment, the detectors should be set up as in fig 10, using a laser and diffraction grating. A filter should be inserted in the beam path immediately after the laser, chosen so as to reduce the intensity of the beam to roughly the same as that expected from the muscle under investigation. The voltage at pin 5 of IC7 should be monitored with a high impedance (>20 M-ohm voltmeter or oscilloscope, and VR5 adjusted until this voltage remains constant whatever light level is incident on the detector. The light level can be varied by holding a polaroid filter in front of the
detector: at a certain angle the filter will allow most of the light through, then as it is rotated the light will be reduced until after 90 degrees, it is stopped entirely. The circuit obviously cannot compensate indefinitely to cope with this situation, however in tests it was found that the light level can be reduced by a factor of at least 20 before the AGC circuit starts to fail to compensate. This is easily sufficient in this application. Once VR5 has been adjusted to give a constant voltage over
a range of light intensities, the voltage at pin 5 of IC8 must be monitored in a similar way, and VR6 adjusted until this voltage equals that previously measured, at pin 5 of IC7. Now both AGC circuits will be correctly calibrated to compensate for changes in intensity over a wide range, and in addition any possible differences between the gain of the two lateral effect transistors is compensated for.
Note that this adjustment of the AGC circuits also has an effect on the magnitude of the final output of the device. In this application this should not cause any difficulties, however if problems do arise, another non-inverting op-amp amplifier could be connected, following the final difference amplifier.
The relative difference between the beam position on the two detectors is found by IC11, configured as a difference amplifier. The values of resistors R28,40-42 are necessarily large, as the output of the LM13700 transconductance amplifier is unbuffered, and hence cannot source more than a tiny current without introducing severe distortion. The capacitor C1 is included to filter out any small high frequency oscillations that could be produced in the AGC circuits, and to eliminate high frequency noise.
The circuit of the power supply is shown in diagram 2. This is a standard transformer, rectifier and regulator arrangement; the two secondary windings of the transformer are both utilised, to provide + and -15 volts supply to the op-amps. Rectification is accomplished by bridge rectifiers, and the output heavily smoothed (C4-7); this is followed by voltage regulators, type 7815 (IC12) for the positive supply, and type 7915 (IC13) for the negative supply. Further smoothing is effected by capacitors C2
and C3.
|
3. Design Solution | 5. Testing of the Device | |