Hi,

I would like to create 2 x 3-phase sine wave generator in the range frequencies from 12 Khz to 15 Khz and final voltage from +18 V up to -18 V down.

I read previous post here

I think I must adapt this circuit schema and put an inverter to have 6 phase offset identical.

I need a very "pure" sinusoidal wave form Khz : is it possible without the need of dds electronic ?

Is all of this realist ?

Best regards.
Thanks by advance.

 

If you need very PURE sine waves, you could use a good low-pass filter just above the sine frequency.

But in the end it might be easier and cheaper to just store a good sine waveform in a digital memory chip and clock out what you need to create each complete sine cycle with the phase you want. You could even just store 1/4th of a cycle and build every sine cycle from that. D-to-A converter could be as simple as power-of-2 resistors and an opamp then a low-pass filter, but DAC chips are cheap now too, probably.

Alternatively, (just thinking out loud, making this up as I go) I guess you could also CALCULATE the sine value, on the fly, in either the analog or the digital domain, with input being a repeating sawtooth (for analog; easy to make) or a repeating binary counter or an increment in a software loop (for digital). I guess "how to do it" in the digital case is fairly obvious. You might even be able to use one processor and software to make six outputs at once, if you could run it to loop at six times the needed bit-rate. For analog, maybe you could use the Taylor Series expansion of sin(x) (which would work well for software, too) and use enough of the polynomial terms to get the accuracy you need, then implement it with analog multiplier chips and an opamp summing amplifier. Wow, that might get expensive. I guess the digital memory chip (or a processor calculating the six values) is probably easiest and cheapest. Or, just use a ring modulator circuit and low-pass-filter the outputs.

Another way might be to use simple triangle wave generators (2 opamps each), and find a way to tie the thresholding circuits together to keep them 1/6th of a cycle apart. If you low-pass filter a triangle waveform to remove third harmonic and above, you get a sine waveform.

Why don't you want to use DDS chips?

 

Thanks for your answers gootee !

Why don't you want to use DDS chips?

dds seems good : I've got impress of doing really hard for simple.
My choice is stopped with 3 ad9958 however, I did not find in datasheet the way to have fine control of frequencies used.

I just found of mso (multiple sinusoidal oscillators) with op-amp and will go for further docs on this.
mso work naturally in range of Khz I wanted.

If someone is experienced with, please let me know how to avoid a non straightforward reading !

Best regards.

 

You can do simple system with a 60 ° phase shifter cell RC. But the performance will be less at appointment.
Next, the ideal is a DSP, or even a micro controller. A table in memory, and 6 analog outputs. Where it gets complicated is that the spectral purity will be determined by the rate of oversampling. Then, each output must be filtered and amplified. And to avoid dauber quality of the sinusoid generated it will make the utmost care to achieve the amp.