field test unit no 001 "model T"

[Qiaozhi]

A PI with early sample and late sample which is then subtractively integrated is really quite complex. As has been said, if the two samples are equal i.e closely spaced in comparison with the variation of the interfering field, be it earth's field or powerline, then good attenuation is achieved. The longer the integration time the better, as more samples are processed. However, there is a limit depending on the response speed required of the detector in passing over an object. Also a factor is the number of pulses per second that the whole TX and sampling cycle operates at. However, the rejection is badly compromised when interfering signals are sychronous, or nearly so, with the TX/sampling rate, or the intersampling pulse spacing. Filter response is more like a comb with little attenuation for synchronism and high attenuation for asychronous. That is why PI detectors benefit from a frequency control to shift the TX/sampling cycle out of synchronism. Serious synchronous interference can occur with a low frequency radio signal that is within the bandwidth of the detector i.e. 200kHz. A small adjustment of the frequency control and it is gone.

Ground balance is different, and one way of doing it is to have two pairs of sample pulses and two receiver channels. The second pair is time shifted from the first but each channel gives earth's field and low frequency rejection. The integrated results are then gain balanced to be equal and then subtracted. Hence earth's field rejection is not compromised.

Eric.

Eric.

Very good explanation.

I've actually experimented with ground balance using two pairs of sample pulses and two receiver channels, so it all makes sense to me. Also, the part on rejecting synchronous interference is interesting, as I've never really considered this as a major problem.

Now we just have to wait for Aziz to tell us how he's going to achieve the same functionality by using a simple high-pass filter at the input ... hopefully without speaking in riddles.

[Ferric Toes]

Too much ado... wouldn't the whole shebang perform even better with a separate Rx and bipolar pulsing?

Bipolar pulsing will require just a single sample pulse to achieve EFE and powerline rejection, but you will still have the synchronous/asynchronous noise problem, and will benefit from a TX frequency control. Ground balance will need a second later pair of samples and another channel; unless there is another way of doing it . Bipolar is easy with Moodz centre tapped bifilar coil and that is something I might investigate.

Eric.

[Ferric Toes]

Also, the part on rejecting synchronous interference is interesting, as I've never really considered this as a major problem.

This problem also shows up if you are working close to another detector, either PI or continuous wave. The adjacent detector has to be far enough away so that there is no overloading of the pre- sampling circuits, and then again, a small shift in frequency of either one will get rid of the problem. In industrial units where sometime several PI detector were in close proximity we had a master/slave system which either locked the detectors to pulse together, or to pulse sequentially. This could be done by radio link in some situations.

Eric.

[Davor]

OK, and what about pre-conditioning the search head so that the pre-digested signal is fed forward? In such case the first approximation of target vs ground, EF, EMI etc. is sifted through the search head geometry. Say, a differential coil.

In case a search head is capable of reducing the undesirable effects by an order of magnitude, and a clever pulse system by another ... there'd be very little the rest of Rx should do to achieve superb performance.

BTW, a separate Rx with constant impedance and referenced to the ground (any ground) should in fact attenuate EF and mains by means oh a high pass filter. In such system an integral of a signal over a whole period should be zero - unlike the systems with switched input impedance that in fact act as rectifiers.

I'd say it is worth further examination.

[Carl-NC]

Also, the part on rejecting synchronous interference is interesting, as I've never really considered this as a major problem.

Also with security walk-through designs.

[Aziz]

Let's make it clear, so that we can all understand what you're claiming:

Are you suggesting that a simple PI detector (that has only a single sample) will have the capability of eliminating the Earth field effect (EFE) by simply adding a high-pass filter at the input?

Yes or no?

Yes, would be enough. But ground balance (GB) is an another issue of course (and isn't the topic of the discussion yet. For GB'ing you would need at least two samples with different frequency response. ) .

Just consider the following fact:
Low frequency EMI +
pulses per second (PPS) modulated (target) response
= full band response (0 Hz .. upper frequency range)

High-pass filter with cut-off frequency set to PPS region -> (attenuation of the low frequency EMI)
-> remaining bandwidth of the detector

The low frequency EMI isn't modulated with the pulse frequency. The front-end can be capacitively coupled to the back-end.

The practical proof:
I can attach a simple PI front-end with an active high-pass filter pre-amp directly to the input of the sound card (AC coupled) and can process the whole response (up to 48 kHz with my sound card). The active high-pass filter just blocks the low frequency EMI and doesn't overload the pre-amp.

Aziz

[Carl-NC]

Filter response is more like a comb with little attenuation for synchronism and high attenuation for asychronous.

A comb filter is what I was thinking of. Aziz, I think you will see this if you did a more detailed selection of frequencies.

[Ferric Toes]

OK, and what about pre-conditioning the search head so that the pre-digested signal is fed forward? In such case the first approximation of target vs ground, EF, EMI etc. is sifted through the search head geometry. Say, a differential coil.

In case a search head is capable of reducing the undesirable effects by an order of magnitude, and a clever pulse system by another ... there'd be very little the rest of Rx should do to achieve superb performance.

I'd say it is worth further examination.

Maybe bifilar centre tapped TX coil to facilitate bipolar pulsing. Bifilar centre tapped double D coplanar RX coil (two D's front to front in antiphase) for earth's field cancellation, ground cancellation, and EMI cancellation. The Rx would be a smaller diameter than the TX optimised for additional ground rejection (reverse signals between the coils, particularly close to the ground). Moodz true differential preamp.

Haven't worked out the sampling yet (if any). Do that tomorrow first thing .

Eric.

[Aziz]

A comb filter is what I was thinking of. Aziz, I think you will see this if you did a more detailed selection of frequencies.

I have taken the worst case response scenario. One can't know the phase lag of the disturbing EMI and the subtraction method delivers even more attenuation on "good" phase lags. But again, you will have multiple frequencies of the EFE and low frequency EMI (wide band) and have to take the worst case scenario into account.

[moodz]

Let's make it clear, so that we can all understand what you're claiming:

Are you suggesting that a simple PI detector (that has only a single sample) will have the capability of eliminating the Earth field effect (EFE) by simply adding a high-pass filter at the input?

Yes or no?

I can answer that ..... the answer is yes .... its a straight radio problem

[UrbanFox]

I can answer that ..... the answer is yes .... its a straight radio problem

Moodz, radio transmissions are sinusoidal??

PI is not sinusoidal.

[Davor]

priceless

[UrbanFox]

priceless

Pray tell, Davor, does capacitive coupling treat a sine wave differently to a non sine wave?? You tell me (smartarse )!!

Experience with a scope's AC coupling function will demonstrate that a slope will appear in the otherwise horizontal parts of a square wave. True or False???

[moodz]

Moodz, radio transmissions are sinusoidal??

PI is not sinusoidal.

ROFL ... "every" time varying waveform can be broken down to a series of sinusoids.

PI is a wideband time varying signal ( but can be described/analysed as a bunch of sinusoids)

Filters (except maybe some types of active filters ) dont care if the input signal is wideband or narrowband they just filter according their transfer characteristic.

A high pass filter will knock out lower frequencies in the stopband and let through higher frequencies in the passband.

[moodz]

...if a slope appears it means your CRO is cheap LOL