Sine wave multi frequency

[Olly]

Would there be any advantage to generating a multi frequency tx drive using mixed sine waves rather than the square wave generators used by most modern systems.

Any number of sine waves could be generated and the amplitude of each could be adjusted before mixing.

just a thought!

[Carl-NC]

If you transmit simultaneous sine waves by mixing them then you cannot resonate the coil. This means you have to brute-force drive the coil which increases power consumption. It also makes it harder to create the TX signal especially if you want to alter the frequencies, either for mode selection or for EMI rejection. The square wave drive addresses all of this. By using an H-bridge with a large supply cap energy is efficiently recycled, similar to a resonated single frequency drive. And the frequencies chosen are a simple matter of altering a PWM in software. The energy wasted in harmonics doesn't amount to much, maybe a few percent.

White's had a couple of patents on the use of sequential sine waves (US5642050 & US565463 and had built a 4-frequency prototype called "East Coast Beach". It worked by switching in different resonant caps, but every frequency transition required time for the TX to settle, which creates dead time in the RX and reduces sensitivity.

[Olly]

Thanks Carl, makes perfect sense when you explain it like that....

Cheers

[mateatek]

Hello!

I built such a machine. It was very complicated to build. The machine changed frequencies 128 times per second. When changing frequencies, one had to wait for the vibration to build up, and only then could the signal be examined. When changing frequencies, the resonance of both the transmitter and receiver coil had to be changed. My machine went at 8 and 16 kHz. Actually, I recommend building a multifrequency machine driven by a square signal. They are much better.

[Olly]

So what is the actual logical algorithm used when combining 3 square wave frequencies into one complex signal?

[mateatek]

If you make a machine where you use a single frequency square signal, you can extract all kinds of information from that signal. Everything you can. For both small and large targets. There is one small problem with this machine and it is caused by a poor signal-to-noise ratio due to the low operating frequency, since the input of the machine is wideband.
That's why it's good to have a square driven machine where you put the transmitter coil in an H bridge. You drive one half of the bridge with low frequency and the other half with high frequency. High-frequency components provide a good signal-to-noise ratio due to dense repetition. These higher frequency components will ensure metal detection. You'll be able to distinguish results with high confidence from lower-frequency components.
What should be the 3rd frequency component? What additional information could it provide?

[PolarisBlitzX]

Could you provide more details on what you're referring to? Are you discussing generating a sine wave with multiple frequencies, signal processing, or something else? Happy to help if you're looking for insights on waveform generation, modulation, or practical applications!

[Olly]

We were discussing whether there was any merit to using sine waves as opposed to square waves when generating the transmitter waveform of a simultaneous multi-frequency VLF detector.

I'm using an FPGA so there's no problem to generate and combine any number of sine waves, but the general consensus is that square waves are more efficient from a current consumption point of view so I was interested in the algorithm one would use to combine multiple square wave frequencies into a composite for driving the external H-bridge.

[Carl-NC]

...so I was interested in the algorithm one would use to combine multiple square wave frequencies into a composite for driving the external H-bridge.

I doubt there is any one ideal way to do this. For example, if you want a 1x and 3x frequencies you could do it this way:



The top waveform is coil current. You can adjust the short ramplet duty cycle to change the amount of TX energy between the 1x & 3x frequencies. This is probably best done by watching an FFT.
Here's another way of producing 1x and 3x frequencies:



It's not necessary that the ramplets be symmetrical. Here's the DFX TX waveforms:



It's similar to the first waveform above but with two extra reversals per cycle for a 1x & 5x frequencies.
The short of it is, create some waveforms and see what they look like in an FFT. A digital oscope makes this easy; here's the FFT of the V3 TX, you can see that 7.5kHz is the strongest while 22.5kHz is the weakest.

[Olly]

Thanks for the detailed response Carl, lots of food for thought.
I suspected that there might not be a turnkey solution;
As with most complex engineering problems - there seldom is...

[mateatek]

A picture of the square multifrequency machine when it was built. It has operating frequencies of 5 and 20 kHz. Completely self-designed. Simple little circuit.

[Carl-NC]

I suspected that there might not be a turnkey solution;

One of the guys at White's had a spreadsheet for optimizing the MF waveform but I don't know what he was doing.

[Carl-NC]

A picture of the square multifrequency machine when it was built. It has operating frequencies of 5 and 20 kHz. Completely self-designed. Simple little circuit.

Is this sequential multifrequency?

[mateatek]

Is this sequential multifrequency?

Continous. One half of the H-bridge goes at 5 kHz, the other half at 20 kHz.

[Carl-NC]

I've never tried it that way. But that explains your oscope waveform; sometimes the coil is driven, and sometimes it's effectively shorted.