To reduce duplication in later schematics, and to help with modular design I decided on an interface.
Single timing resistor, 50% duty 555 Astable
This somewhat unconventional 555 timing circuit is a favourite of mine. It charges/discharges the timing capacitor through the output pin via single resistor,and thus achieves a 50% duty cycle. The frequency is far easier to calculate than other 555 astable circuits. It uses the discharge pin to pull down a signal line, so it requires a pull up resistor and it can’t source current for devices only sink it.
I found that with my water variable capacitor that the 555s were getting zapped when the capacitance was suddenly reduced. To cut a long story short, a continuous reduction of the capacitance value was resulting a voltage spike. Was possible to redirect much of the voltage spike into the positive power rail which had enough capacitance to soak it, using diode between cap+ and the positive power rail. But the voltage rises were still too fast. Needed low pass filters in series with threshold and trigger pin. This added complexity to the design, and placed additional limits on what frequencies the circuit could produce. There had to be a better solution
Looking for a better solution I came across relaxation oscillators. These often rely on simple devices which conduct heavily when a voltage exceeds a certain amount, and that cease conducting when the current passing through them falls below a threshold. this characteristic which is called negative resistance. Neon bulbs and spark gaps both do this, and would be perfect as they should clear any over voltage automatically, but their trigger voltages were fat too high for easy use here.
Then I found Programmable Unijunction transistors. Consider them as a solid state low power neon bulb, with an extra terminal for precise setting of what voltage will trigger them.
Took the PUT oscillator circuit that has a thyristor output right at the bottom of PUT relaxation oscillators. and replaced the thyristor with a transistor pull down for a signal line, and it worked brilliantly. Only faff was it’s really tricky to set the resistor to a decent value. Too small a resistor and the PUT locked in conduction. Too big and the capacitor took far to long to charge . Resolution of large value resistor sets isn’t great either. The smallish resistor set I’ve got jumps from about 500k to 1M, and typically the theoretical value needed was about 700k! Yes I could purchase resistors specifically for this task, or combined several to get the right figure, but decided to try controlling the current.
It is never as simple as it first seems! The kind of current needed is in the uA range- so well below that most regulator ICs can control. Looked up current source schematics and despite the LM317 current regulator being called a current source when current flows through it to a load (so source of conventional current) , all the schematics for current source were for arrangements that went between the load and ground, and weren’t sources of (conventional) current source but sources of electrons. So maybe it was a current sink I was looking for. Looking for them just gave the very same circuits. All very confusing, Using the terminology of “if that referred to provides something then it’s a source” then I still needed to find/develop a suitable current source as the PUT would be controlling the current out of the capacitor.
Had heard that Photovoltic devices can make decent constant current sources. So long as the output voltage is kept well below the forward conduction voltage, the current is proportional to the the amount of illumination. Illuminate a photodiode with light from a suitable LED, and the photodiode will generate a small voltage, with even smaller current. Had a few quad optocouplers in parts bin, so only took a few minutes to see if they might work. They didn’t the output voltage barely registered on my meter. Checked their datasheets and found they were photo-transistor output.
Got some individual IR photodiodes from Maplin and tested them, I got a voltage out under illumination,but couldn’t measure the current as it was so small, even with the diodes in parallel. Tried them in series as a current source for PUT oscillator, but it wouldn’t start up. voltage on capacitor rose to 3.2V, but I don’t think that was enough for PUT to conduct. (PUT fires BJT transistor so there’s the transistor diode voltage to add in to the triggering voltage?)
Then I spotted the TLP190B This has a series stack of diodes so it’ll generate up to 8V at tiny currents of about the right range. Looked perfect . Wired it up as per this schematic..
Circuit seem to work fine on 4.5V, with PUT bias resistors (R2/R3) of 6k77 and current limiting for the optocoupler LED R1=200k. Circuit is currently on a solderless breadboard so capacitance values likely to be somewhat off but with 116nF timing capacitance added the output transistor switches on/off at about 25Hz, following image is the oscilloscope trace of the voltage on the cap. 4.50V supply when running with a 50nF timing cap . Frequency about 7.89kHz Apart from when the PUT is switching on/off, the voltage increase looks reasonably linear which suggests a constant current.
I now consider this circuit to be developed sufficiently that I’d consider building it up permanently on a printed circuit board. Intend to have a try at etching a board, but maybe it would be better to wait until I’ve CNCed the table of my small milling machine, and just mill the tracks. likely need to drill holes for through hole components and that seems likely to be a major chore if done with a manual feed.