Oscillating signal splitter into 3 output

If the atomizers don't all have exactly the same resonant frequency, it doesn't seem possible to lock on the resonant frequencies and still maintain a 60 degree phase shift.

Why do you need to have a fixed phase difference?

I think it might be possible to design the phase lock circuits to have 60 degree differences without the splitter, but they couldn't lock to different resonant frequencies at the same time. 

Hi,

It's good you have a diagram, not many people make that effort, but it's bad it doesn't make sense (to me at least). From left to right:
Regarding 'oscillator circuit at resonance', you have a resonant circuit as part of your oscillator, so of course it will resonate. It's like a tautology.

Next, what you're calling a 'signal splitter' isn't a splitter. It's some sort of delay circuit with op-amp buffers.
Finally, what's the point of the PLL, where is your VCO, it's absent, and what would that achieve in any case?

Trying to guess the intent, it seems you might wish to adjust each output frequency slightly and thought the third stage in your diagram would do that? As Doug mentions, how would you maintain the phase difference if each frequency is different? I think you need to reconsider that.

Personally I think you started with a bad design, because if you tweak the first stage oscillator, all your phase shifted outputs are messed up. Then you somehow wish to fix or alter that with the PLL.

You've made this more complicated than it needs to be, and ended up with what will be a drifty circuit which likely won't meet your needs anyway, if you try to tweak the frequency. Maybe it's best to start from scratch. I don't know what you're familiar with. If you want to do it in logic, then use an FPGA (or even a CPLD, or discrete logic), and generate three digital outputs. If you want to do it in software, then there are microcontrollers with timer modules. Infineon XMC series would be good at it, perhaps google "infineon XMC phase shift" or similar. Other microcontrollers could manage it too, so it is best to use what is familiar to you. Anyway, that will be rock-solid, and phase shift maintained regardless of the output frequency, leaving you free to fine-tweak that frequency and it would apply to all outputs.

Once you have the three outputs, if you really need it to be a sinewave, then either low-pass-filter it, if that's good enough, or use three PLLs if you prefer. Your transducers will have to be identical, maybe even matched from within a batch of them. I have no idea what you wish to do with them and if you need amplitudes matched or not, these transducer outputs are not my knowledge area, but I know you're starting off wrong with the circuit/block diagram you have.

I was thinking about this design, self oscillating at a frequency of choice (110 kHz) with the 60° 120° and 180° available:

Ignore the R and C values - I just entered them to get LTSpice running.

You'd need to select the capacitors high enough, so that the capacity of the signal-tapping circuit doesn't alter the behaviour. And the signal will be attenuated at each point, so you'll need tapping circuits with different amplification.

That approach would work too. I think you or I or others could make that function, maybe even place varicap diodes in there so that they become ganged with a single control to trim to maintain phase shift during any desired slight frequency changes, or to be sure of the resonant frequency beforehand. I don't know if the OP has all the detail up-front to be able to do this though. 

I also don't get the purpose and complete requirements of it all, especially since the diagram makes no sense, so I can't really be sure what's a good approach, just that the initial approach isn't good : )