|home > History Making robots > Grey Walter's Tortoises > Circuits||22 March 2015|
STRUCTURE AND MECHANISM
Despite the complexity of the tortoises behaviour, the mechanism which produces such extraordinary results is surprisingly simple. It is all contained in an opaque Perspex carapace, supported by a tricycle pneumatic-tyred undercarriage. The front wheel assembly, consisting of a periscope-like photo-electric cell, the driving motor and its associated gearing, is rotated on a vertical axis through 360° by a separate suitably geared scanning motor on the main chassis. A 6 v accumulator carried above and behind the rear axle supplies the current for these motors.
[only Elmer and Elsie had pneumatic-tyres]
So much for the electro-mechanical workings of the robot. The electronic control circuits, however, are rather more complicated. The vital components comprise the photo-electric cell already mentioned and two pentode valves with a switch relay in the anode circuit of each. As soon as the H.T. supply from the underslung 45v battery is switched on, a considerable current flows through the anode of the first amplifier valve because the bias on its grid is strongly positive. This current closes the switch relay in the anode circuit, but the anode load -- mainly the resistance of the relay coil -- reduces the voltage to a minimum. The potential of the screen of the second valve is correspondingly low, as it is connected to the anode of the first, and consequently the current through the anode of the second valve is also very small. Its associated relay is, therefore, open.
[two pentode valves - Elsie is described (GW-LB) as having a triode and a pentode. The 51 machines two pentodes.]
The position of the relays is now such that the small headlight is switched on, the driving motor comes into operation at half-speed and the scanning motor at full speed. The front wheel of the tortoise is then driven forward while it slowly rotates through 360° and the animal proceeds to explore its surroundings in a continuous series of cycloidal movements, ready to respond to suitable stimulus.
Should this be provided by a light of moderate intensity duly picked up by the photo-electric cell, the relative positions of the relays are immediately affected. The bias on the grid of the first valve is suddenly reduced. This naturally brings about a similar reduction in its anode current which, although not sufficient to open its relay, raises the screen potential of the second valve, applies a transient signal to the grid, raises the anode current and thereby closes the second relay. The headlight and scanning motor are switched off and the driving motor reaches its maximum speed. The photo-electric cell, now pointing at the light, is fixed to face the direction of drive, and the animal moves off towards the source.
The rear wheels, however, tend to exert a trailing action that is often sufficient to swing the photo-electric cell away from the light. In this event the relays return to their original position and scanning begins again until the light is again picked up. This sequence may be repeated many times before the tortoise nears its objective. As the light intensity gradually increases with proximity, the bias on the grid of the first valve steadily falls and the anode current with it until the first relay opens. The current through the second valve has, of course, been increasing proportionately but its relay is already closed. The effect of opening the first relay is to turn the scanning motor half on, so that the tortoise turns away before a collision can take place. If, however, it is hungry -- that is to say if its battery is running down -- the supply may be insufficient to operate the scanning motor and the animal retains an 'interest', particularly in the bright light shining within its hutch.
The sense of touch is supplied by a multi-vibrator circuit feeding the output of the second valve back to the grid of the first via a condenser. Collision with an obstacle tilts the animals shell slightly and closes a stick and ring switch that completes the circuit. The relays then open and close in turn and the tortoise feels its way clear.
1951 - Original circuit for the Festival of Britain (FoB) Tortoises - BNI archives
This circuit shows the relays with their contacts in the normally closed position, i.e. with no power to the coils.
P.E.C. - the Photo-Electric Cell
V1 - the DL93 valve for relay RL1, the grid is the lower dotted line and the screen the upper dotted line
V2 - the DL93 valve for relay RL2, the grid is the lower dotted line and the screen the upper dotted line
Behaviour Pattern E - Exploration
Dark conditions - little or no light on the P.E.C.
RL1 on, RL2 off
Slow driving and fast turning/scanning
The P.E.C. has high impedance and the grid of V1 is pulled high by the 22M ohm resistor so V1 conducts and RL1 turns on.
The grid and screen of V2 are pulled low by the 68K + 1M resistors and by the anode of V1, so V2 stays off and RL2 is off.
The Turning motor is fed directly from LT+ (6v) so the Tortoise scans fast.
The Drive motor is fed through the lamp resistor combination so the Tortoise drives slowly.
(No point in driving fast if you don't know where you are going.)
Behaviour Pattern P - Positive phototropic response
Light detected by the P.E.C.
RL1 on, RL2 on
Fast driving and no turning
The P.E.C. has medium impedance and pulls the grid of V1 lower but not sufficient to turn it off and so RL1 stays on. However the anode voltage of V1 is now higher.
The higher anode voltage of V1 causes the grid voltage of V2 to rise sufficiently to turn on V2 which turns on RL2.
RL2 switches from supplying the Turning motor and supplies current direct to the drive motor so the lamp and resistor have no effect and the drive motor runs at full speed.
Behaviour Pattern N - Negative phototropism
Strong light detected by the P.E.C.
RL1 off, RL2 on
Fast driving and slow turning - will turn away from light
The P.E.C. now has low impedance and the grid of V1 is pulled low enough to turn it off so RL1 turns off so the contacts supply power through the lamp and resistor to the Turning motor so it turns slowly.
The grid of V2 is pulled high through RL1 and so V2 turns on keeping RL2 on.
Inhibited Behaviour Pattern N to enable charging
Grey Walter asserted that when the charge in the 6v lead-acid LT and motor battery dropped, the gain of the circuit would be reduced so that V1 would never turn off in strong light, RL1 would remain on and Behaviour Pattern N would not occur and the Tortoise would head for the light to recharge. There is no record of this working, it wasn't observed in the restored #6 nor has it been achieved by the FoB replicas. Probably it just needs some fine adjustment of the relay contacts and the right level of ambient light.
There is a second explanation in the article by Fletcher K.H. quoted above "..if its battery is running down -- the supply may be insufficient to operate the scanning motor..". Did Fletcher observe the effect and presume a cause?
Behaviour Pattern O - Obstacle avoidance
Collision or tilt detected by the Shell touch contact
RL1 on and RL2 off alternating with RL1 off and RL2 on every second
Repeated bump then turn
The output of V2 is connected by the Touch Contact to the input of V1.
The output of V1 is already connected to the input of V2 forming an oscillator.
The output changes about every second and the timing is governed by the resistors and capacitors of the circuit.
While the circuit is oscillating it is not responsive to the light falling on the P.E.C..