Automatic Toy-Traffic-lights

Inspiration to and definition of the project
         At some point younger boys become very interested in everything concerning traffic and means of transportation. Their attention is by far restricted to the vehicles, but includes anything related to these, such as rails, flashlights and the common traffic light seen at urban crossroads.
         Making an automatic toy-version of the latter is not as simple as first thought, but a digital solution to the problem is rather obvious and here a simple dual input EXOR-gate is put together from 4 dual input NAND-gates to provide for the 3 interdependent outputs.
         In order to acquire some suitable input-signals I resorted to an old article on CMOS-triggers. One of these was designed to generate a kind of triangular waveform at a very low frequency and the other detected this shape at some appropriate points on its slope. The resulting two digital signals posses the wanted phase-shift and, fed to the EXOR-gate, provide the able outputs.
Experiences from the one-and-only prototype:
         As is obvious from the scheme, the actual construction became a hybrid between CMOS and TTL-logics. The reason for this is the high impedance of the CMOS-gates, whereas the TTL-gates provide the current large enough to drive the LEDs.
         Since the theoretical considerations for the CMOS-triggers were originally based on a 5 V. powersupply, the introduction of a TTL-device did not, in that respect, change anything.
Concerning the "calculations", I simply imported the resistance values from my article on CMOS-triggers and made sure, that the oscillator had a hysteresis bigger than the one found in the curveshaper. Also I had to take precautions to outrun the "leaky" electrolytic capacitors -- in fact using a rather smaller charge/discharge-resistance.
         The resulting running-speed became somewhat faster than considered realistic. This was expected, so the circuit has got a couple of points for the attachment of an additional, much bigger capacitor.

Proposals for improvements of the design:
         It turned out, that the comparatively small hysteresis combined with the very big capacitor resulted in an unreasonably lengthy first charging time. This meant, that it took several minutes before visible oscillation actually occurred. The nuisance is best overcome by changing the ratio between the two resistors, which are responsible for defining the hysteresis. In this case the 680k can be replaced by 560k and the 47k increased by a factor 2 or more.
         The TTL-IC needs a stable 5.0 V. powersupply, which rarely is available with childrens toys. In case the TTL-IC is replaced by a CMOS-IC (4011) any DC-voltage between 4 and 12 V. will become acceptable -- provided the replacement includes three output-transistors for driving the LEDs. These are best employed as PNP emitterfollowers with a suitable (bigger) resistor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .