Target ID with simple F.E.T analogue multiplier?

[moodz]

OK, but IQ demodulation widely used is homodyne. Analog demods tend to use square waves which produces wideband demodulation, while the newer direct sampling designs use sine waves which produces narrowband demodulation. The null error ends up as a DC offset which can be cal'd out or just ignored. But even if you perfectly cal out the null error, ground loss angle can still induce a target phase error and it gets worse with depth.

We might be caught up in Atlantic Semantic .... but I have a very clear understanding of the homodyne as a Self Mixing detector.

One of the first self oscillating detectors or homodyne ( where the active device ... a valve is an oscillator and mixer ) is this one .. circa 1928

The feedback brings the oscillator just to the point of oscillation which is self synchronised to the incoming carrier. It self mixes the incoming signal and the headphones are the "low pass filter".



when the oscillator was split out to a separate valve it was called a synchrodyne however here the separate oscillator was synchronised to the incoming signal so its still "self mixing".

For some reason that is not clear the IQ demodulator is referred by some today as a homodyne even though it has two mixers and is not "self mixing" ( ie two quadrature phase signals are generated and mixed separately. )
The patents for IQ demods indicate they are a new method with improvements beyond the homodyne method.

So why IQ would be called homodyne does not make historical or actual sense. ( I would call it dfferential phase mod / demodulation )

IQdyne just doesn't quite have a ring to it does it ? LOL ... maybe that is why the term is not used.

However I will note that whereas the homodyne tries to lock to the transmit carrier ... the IQ demodulator LO only has to have nominally the same frequency as the transmit carrier ... this was the whole point and benefit in the case of radio. ( but not MDs )

moodz

[dbanner]

OK.

[moodz]

...So as your AI critique was very entertaining and right about quite a bit of the discussion ... the terminology is important or the AI .. because it is using the wrong / different terms may be at risk of error because what what may be understood by some is understood differently by others due to a difference in terminology.

[dbanner]

Yes, I simply copied and pasted the complete text of your post and asked for a critique.
I'm not going to mess with it anymore.

[Carl-NC]

By definition, homodyne means to mix to baseband in a single step by using the same frequency for the LO. It doesn't matter if the LO has been phase-shifted or whether there are 2 mixers (I & Q). Heterodyne mixes to baseband using 2 or more mixing stages. A synchrodyne is a homodyne that uses a separate free-running oscillator that gets phase-locked to the input signal. A BFO is sort-of a synchrodyne, but VLF is not. VLF designs are very clearly self-mixing as they use the exact same frequency for the LO, except for oddball cases like the Fisher Gemini or the Nokta Legend; they are both heterodynes. Passing the TX through phase shifters for the demods doesn't make it any less homodyne than directly mixing with the raw TX signal. Besides, both the raw TX voltage and the RX signal are also passed through phase shifters, you can't avoid it.

Right now I am not seeing any difference in the method you're describing and what is already widely done. Hopefully you can go a bit deeper.

[moodz]

... leaving the arguments about what dynes are what aside for now. ( its not really relevant ).

The fundamental reason ( in addition to other issues ) .. with IQ demods is this.

Below is the phase response of a phase sensitive detector at a single frequency. In an IQ demod you are combining two of these. The IQ detector quadrature signal is to ensure a "zero beat" or cancellation cant occur on the incoming signal. However cancellation is exactly what we want .. we dont want the IB leakage.

The lack of sensitivity / depth in IQ detectors is due largely because the IQ demod is demodding the IB leakage carrier which is swamping small target signals. ( having a different phase but the same frequency as the IB leakage ).
When you use only a single synchronous detector ( homodyne LOL ) ... then you can zero beat the IB leakage carrier thus effectively cancelling it near the dv/dt=0 point and now any phase disturbances will be due to weak targets that would be masked / swamped in an IQ demod.

So only a single mixer ( aka phase detector ) is required to cancell the IB leakage carrier since we have the transmit carrier reference.

At zero beat when there is no target dv/dt = 0 ie 180 phase shift degrees as per graph below. .... when there is a signal dv/dt will be something dependant on target phase shift ...but not due to IB leakage.

This enables a detector which is more sensitive than a PI simply because it has a much better SNR. ( ie the TX never turns off )

moodz

[dbanner]

"the distinction between homodyne and IQ demodulation can blur in modern implementations. Here’s the nuanced breakdown:

---

1. Shared Foundation: Both Are Homodyne

• Literal Definition:
  "Homodyne" means mixing the input signal with an LO at the same frequency (no IF stage). By this definition, IQ is a subset of homodyne—it uses two homodyne mixers (I and Q) with quadrature LOs.
• Modern "Direct Sampling":
  Newer designs digitize the RX signal directly and perform digital homodyne/IQ demodulation (e.g., with CORDIC algorithms). This is still homodyne—just implemented in software/DSP.

---

2. Key Differences

LO Signals	Single LO, phase-locked to TX	Two LOs (I/Q), 90° apart
Phase Knowledge	Must know TX phase exactly	Tolerates unknown phase
Output	Direct baseband (X/R)	I/Q → Polar conversion
Nulling Leakage	Analog cancellation (PLLs)	Digital suppression (DSP)

---

3. Why IQ is Called "IQ" and Not "Homodyne"

• Historical Context:
  • "Homodyne" originally referred to single-mixer systems (e.g., analog AM radios).
  • "IQ" emerged in communications for blind phase recovery (e.g., SDRs), emphasizing quadrature processing.
• Practical Naming:
  Engineers reserve "homodyne" for phase-coherent single-mixer systems (e.g., lock-in amplifiers), while "IQ" implies quadrature math.

---

4. Metal Detector Implications

• Narrowband vs. Wideband:
  • Classic homodyne is narrowband (optimal for single-frequency detectors).
  • IQ can be wideband (handles multi-frequency/mineralized ground).
• Your Point About Sine Waves:
  • True! Modern "direct sampling" detectors use:
  • 
  • 
    
    This is digital homodyne but still called "IQ" due to the quadrature math.

---

---

Conclusion

• IQ is technically homodyne (same-frequency mixing), but the term "homodyne" is reserved for phase-locked single-mixer systems.
• Naming depends on implementation:
  • "Homodyne" → Analog, phase-coherent.
  • "IQ" → Quadrature (analog or digital).

[dbanner]

... leaving the arguments about what dynes are what aside for now. ( its not really relevant ).

The fundamental reason ( in addition to other issues ) .. with IQ demods is this.

Below is the phase response of a phase sensitive detector at a single frequency. In an IQ demod you are combining two of these. The IQ detector quadrature signal is to ensure a "zero beat" or cancellation cant occur on the incoming signal. However cancellation is exactly what we want .. we dont want the IB leakage.

The lack of sensitivity / depth in IQ detectors is due largely because the IQ demod is demodding the IB leakage carrier which is swamping small target signals. ( having a different phase but the same frequency as the IB leakage ).
When you use only a single synchronous detector ( homodyne LOL ) ... then you can zero beat the IB leakage carrier thus effectively cancelling it near the dv/dt=0 point and now any phase disturbances will be due to weak targets that would be masked / swamped in an IQ demod.

So only a single mixer ( aka phase detector ) is required to cancell the IB leakage carrier since we have the transmit carrier reference.

At zero beat when there is no target dv/dt = 0 ie 180 phase shift degrees as per graph below. .... when there is a signal dv/dt will be something dependant on target phase shift ...but not due to IB leakage.

This enables a detector which is more sensitive than a PI simply because it has a much better SNR. ( ie the TX never turns off )

moodz

And what to do about the dynamic ground conditions? And the leakage?

You solved these, yes?

[moodz]

I think so ... its a full digital system. ie Theres a TX connected to the TX coil and theres an ADC connected to the RX coil. No analog processing.
Once the data is captured by the ADC .. you can do this scheme. The IQ scheme. PI as well all in parallel or save it to a USB key and run it at home.

:-)

[moodz]

To answer your ground query more fully ... there is magnetic susceptibility and there is conductance.

The ground can have both.

The ground signal ( for magnetic susceptibility ) is carried by the leakage IB carrier ... if you null the leakage carrier then you have nulled the ground. ( ground conductivity requires a second strategy )

Importantly the leakage carrier is nulled by mixing ( AKA zero beating ) the leakage IB carrier with itself ( true homodyne ) not the raw TX phase. The RX leakage phase is close to the TX phase but is not the TX phase.

This also leads to an interesting contention that a single coil VLF TX should be possible where the receiver circuit cancels the TX signal by self mixing ( true homodyne ) leaving only the target signals. This has already been done in a related field so should be possible.

moodz

[Carl-NC]

The lack of sensitivity / depth in IQ detectors is due largely because the IQ demod is demodding the IB leakage carrier which is swamping small target signals. ( having a different phase but the same frequency as the IB leakage ).
When you use only a single synchronous detector ( homodyne LOL ) ... then you can zero beat the IB leakage carrier thus effectively cancelling it near the dv/dt=0 point and now any phase disturbances will be due to weak targets that would be masked / swamped in an IQ demod.

At zero beat when there is no target dv/dt = 0 ie 180 phase shift degrees as per graph below. .... when there is a signal dv/dt will be something dependant on target phase shift ...but not due to IB leakage.

The ground signal ( for magnetic susceptibility ) is carried by the leakage IB carrier ... if you null the leakage carrier then you have nulled the ground. ( ground conductivity requires a second strategy )

I still don't see anything you're describing as being different than what IQ demodulation does. Assuming that your term "IB leakage" refers to the lack of a perfect coil null, then it's the Q demod that is normally phase-adjusted to null this signal. But not quite, because we also have to deal with the ground's loss angle. This is usually close to the null angle (~0°) but can be as high as 10°. In other words, it is rarely the case that you can precisely cancel both the residual coil null and the ground at the same time, using one demod. So we pick one, and it's always the ground angle because its amplitude varies considerably with coil height, whereas the null error tends to be rather static. In fact, the null error is largely a static DC offset on the output of the demod so it really has no effect on target ID. I say "largely" because a bad coil null can lead to lift-off effects, which you can hear when bobbing the coil over really bad ground, especially with a concentric coil.

There is another "leakage" term but I think you are not talking about it, which is LO leakage in the demods. This also results in a static DC offset so, again, it doesn't really matter.

It is not at all true that if you null the leakage carrier then you have nulled the ground, even if the ground has zero conductivity. That's because viscous ground has a higher loss angle than, say, magnetite and the loss angle increases with frequency, and this has nothing to do with ground conductivity. Again, this is why we null one of the demods to the ground and not the coil error. Furthermore, if you only have a single demod, I can't see how you can resolve a phase angle. The output of a single demod will depend on both the phase angle and the target strength, and there is no way to separate the two. Even considering dv/dt doesn't help because that depends on the phase and the sweep speed and, again, you can't separate the two. That's why we add the I channel.

[moodz]

I still don't see anything you're describing as being different than what IQ demodulation does. Assuming that your term "IB leakage" refers to the lack of a perfect coil null, then it's the Q demod that is normally phase-adjusted to null this signal. But not quite, because we also have to deal with the ground's loss angle. This is usually close to the null angle (~0°) but can be as high as 10°. In other words, it is rarely the case that you can precisely cancel both the residual coil null and the ground at the same time, using one demod. So we pick one, and it's always the ground angle because its amplitude varies considerably with coil height, whereas the null error tends to be rather static. In fact, the null error is largely a static DC offset on the output of the demod so it really has no effect on target ID. I say "largely" because a bad coil null can lead to lift-off effects, which you can hear when bobbing the coil over really bad ground, especially with a concentric coil.

There is another "leakage" term but I think you are not talking about it, which is LO leakage in the demods. This also results in a static DC offset so, again, it doesn't really matter.

It is not at all true that if you null the leakage carrier then you have nulled the ground, even if the ground has zero conductivity. That's because viscous ground has a higher loss angle than, say, magnetite and the loss angle increases with frequency, and this has nothing to do with ground conductivity. Again, this is why we null one of the demods to the ground and not the coil error. Furthermore, if you only have a single demod, I can't see how you can resolve a phase angle. The output of a single demod will depend on both the phase angle and the target strength, and there is no way to separate the two. Even considering dv/dt doesn't help because that depends on the phase and the sweep speed and, again, you can't separate the two. That's why we add the I channel.

I think it will come down to this ... you have not been able to explain to me why an IQ demod is better than a single demod since the magnitude and phase of a signal can be perfectly determined using a single dmod when the transmit reference frequency and phase is also known. Its called synchronous demodulation.

If you have a composite phase vector made up of say ground and residual carrier you can do some calculations to determine a solution for the constituent phase vectors.
When a new target vector phase comes along you recalculate to determine the new vector constituent. If you can cancel one or more of the "constant" vectors using zero beat then the calculation is simplified.



I will also note that a good phase detector is insensitive to amplitude changes ... so varying the amplitude all you want ( eg bobbing the coil ) wont have any affect on the phase cancellation.
Also noting that the ADC is a spectrum capture from DC to 1.25 Mhz .. so any number of additional demod channels can be added up to the limit of the processing capability /chip real estate.

noting also there is no LO leakage as the LO is a software oscillator and the mixer being implemented in software also is mathematically ideal.

moodz.

[Carl-NC]

In order to use a single demod to determine phase you need to rotate the LO phase until the signal is nulled. This is how the old TR discriminators worked. But when the signal is nulled for phase, you lose the amplitude information and you have to then rotate the LO by 90° to determine amplitude. IQ demods give you both answers at the same time, without the need to rotate the LO. However, you are correct that IQ demods with fixed-phase LOs are not as precise at determining target phase (worst at the 45° points) than if you actually find the null response by rotating the LO.

So it seems a best solution is to rotate IQ demods until the Q in nulled; this gives you the phase and the I channel gives you the amplitude. But when you rotate the LO to find the target null, you lose the null on the ground. So you could then run a third demod that always maintains a null to ground (or, more precisely, ground+coil imbalance) and use that signal as the "all-metal" targeting signal. And this is exactly what many VLFs do.

Yes, a good phase detector is insensitive to amplitude changes, but it's common for a signal phases to vary with amplitude. Take the case of the coil null combined with ground. The coil null is typically very close to 0° and its amplitude is constant. The ground response is typically at a higher loss angle and its amplitude varies considerably with coil height. The combined responses will therefore vary in both amplitude and phase as you bob the coil over the ground:


This is an exaggeration because the coil error is usually tiny compared to ground, and the angles are not so large. But it illustrates that the incoming phase can change with amplitude and even a perfect phase detector can't fix this. It also illustrates why a good coil null is needed.

Another problem is that the target phase varies with depth in mineralized ground. The loss angle of the ground alters the phase of the TX signal with depth and then does the same thing with the target return signal. So as targets get deeper the amplitude drops and you also get more apparent phase error. This has nothing to do with the accuracy of the demod.

[dbanner]

A little humor.

[Skippy]

Quote: "Also ... a good phase detector is insensitive to amplitude changes, so varying the amplitude all you want ( eg. bobbing the coil ) won't have any affect on the phase cancellation"

As Carl stated, bobbing the coil doesn't just affect the amplitude of the ground signal, it varies the phase, too. Based on personally taken measurements: as the coil is raised, the ground phase will change from, say, 5 degrees, towards 90 degrees [ a 'saltwater' angle ] when well clear of the ground. It's clearly hard to measure angle accurately with the coil 50cm off the ground, as the signal is greatly reduced in amplitude, but that's what is seen. This is for a DD coil, I haven't tested a concentric ( or any other type ).

Quote: "This is an exaggeration because the coil error is usually tiny compared to ground"

Agreed. Even fairly mild ground has a strength much greater that the residual null of a well-made coil.
It's also easy to 're-null' a coil with great precision ( using tiny ferrite bits, and small very-low time-constant non-ferrous items ) , and no observable benefit results -- such as cleaner/more stable target ID, ability to 'hit' on a weak target more consistently, etc.