ZP ACTIVE DAMPING and NOISE CANCELLATION

[moodz]

Just checking the specs for the SOT-223 FET its good for 3.3 watts of dissipation at 25 C and 2.25 watts at 40 C .... that is good for 2 amp peaks in the TX coil - but I dont go near that ;-)

moodz

[Mr.Jaick]

ML uses these small FETS as blocking switches they are not damping any power. ( As far as I know ).

moodz.

what I meant by that was this one:


they definitely do active damping with a FET in parallel to the coil or something like that, otherwise, how they get rid of the coil energy, some of it goes to the clamp regulation chain but there is still a lot remaining.

A few hours ?? not sure what you mean here ... the TX voltage is regulated

I meant to say what if the TX voltage is Not regulated?
because this TX voltage regulation is not a jelly bean task as you have to make sure there is always a high level of efficiency And very low noise obtained.

[moodz]

what I meant by that was this one:


they definitely do active damping with a FET in parallel to the coil or something like that, otherwise, how they get rid of the coil energy, some of it goes to the clamp regulation chain but there is still a lot remaining.
it’s possible but if they were doing the same as the method I am using you don’t need the input protection switches and if they are using bipolar pulses a single damping feet won’t work.


I meant to say what if the TX voltage is Not regulated?
because this TX voltage regulation is not a jelly bean task as you have to make sure there is always a high level of efficiency And very low noise obtained.
the tx voltage could be unregulated and compensated for voltage variation by the cpu etc I just haven’t done it that way. If it was done that way you could get more efficiency from the batteries. With the cheap linear regulator you can still get over 10 hours from a couple of 18650 cells. This design is a an easy to build demonstrator that is still very effective.

[Mr.Jaick]

if they were doing the same as the method I am using you don’t need the input protection switches

yes, that feature of yours, it's much cleaner, I tried those clunky switches once before, but didn't like them, bad THD at the beginning and some degradation of performance, they push the delay away...

if they are using bipolar pulses a single damping feet won’t work.

do they? I didn't expect them to use the bipolar method
so what do they do? any patents

I'll try your idea, but perhaps not with a TIA, I rather let the RX current go through a resistor and measure the voltage drop with a voltage amplifier, this way I won't get the slow decay at the end of the flyback.
then after the first amp, I would invert the wave and sample for the loop there.
have you tried this idea out in the field to see how it works on the ground?

[moodz]

yes, that feature of yours, it's much cleaner, I tried those clunky switches once before, but didn't like them, bad THD at the beginning and some degradation of performance, they push the delay away...


do they? I didn't expect them to use the bipolar method
so what do they do? any patents

I'll try your idea, but perhaps not with a TIA, I rather let the RX current go through a resistor and measure the voltage drop with a voltage amplifier, this way I won't get the slow decay at the end of the flyback.
then after the first amp, I would invert the wave and sample for the loop there.
have you tried this idea out in the field to see how it works on the ground?

not sure what you mean by the slow delay at the end of the flyback … if the damping was perfect there is no error signal for the damping feedback loop to operate off.

I am doing some field tests now … probably post a YouTube video.

moodz.

[Mr.Jaick]

not sure what you mean by the slow delay at the end of the flyback … if the damping was perfect there is no error signal for the damping feedback loop to operate off.

in the scope picture you provided there is an overshoot of around 6V at TIA output I think it's because the coil is shorted to AGND when the flyback falls below the diode voltage.
as I like to use +-5v supply for op-amps I need to mitigate this effect by some means if possible(without degrading the performance).

I am doing some field tests now … probably post a YouTube video.

If you could write a little report here as well it would be greatly appreciated.

Thank you for your generous responses.

[moodz]

in the scope picture you provided there is an overshoot of around 6V at TIA output I think it's because the coil is shorted to AGND when the flyback falls below the diode voltage.
as I like to use +-5v supply for op-amps I need to mitigate this effect by some means if possible(without degrading the performance).

An ideal TIA has an input impedance of zero ohms so when the diode stops conducting the only thing the coil sees is a dead short to ground ... so you are correct the coil is "shorted" to ground as the flyback falls to zero - the input impedance of the TIA is close to zero.
This is not a design fault this is intentional. Traditionally pulse induction detectors use voltage amplifiers to measure the induced target voltage ( across the coil / damping resistor ) however the TIA in this case is amplifying the current and converting it to a voltage at the output. The TIA sees only the reactive impedance of the coil itself and the coil sees only the very low input impedance of the TIA.




If you could write a little report here as well it would be greatly appreciated.

Thank you for your generous responses.

[moodz]

I got logged out before I finished typing the previous response ...

An ideal TIA has an input impedance of zero ohms so when the diode stops conducting the only thing the coil sees is a dead short to ground ... so you are correct the coil is "shorted" to ground as the flyback falls to zero - the input impedance of the TIA is close to zero.
This is not a design fault this is intentional. Traditionally pulse induction detectors use voltage amplifiers to measure the induced target voltage ( across the coil / damping resistor ) however the TIA in this case is amplifying the current and converting it to a voltage at the output. The TIA sees only the reactive impedance of the coil itself and the coil sees only the very low input impedance of the TIA.

1. Straight up this results in a huge reduction in noise ... as the input noise is proportional to the connected source resistance.

2. The presentation of a "short" across the coil shorts out any capacitance across the coil .... another benefit. Its more complex than this but it is a benefit.

3. It also becomes obvious that any damping resistance will also damp target responses. A "shorted" coil will actually extend target induced currents. This is what we are trying to detect ... target induced currents ... so damping them is not what we want ... we just want to measure those target currents.

4. I am working on a report.

moodz

[Mr.Jaick]

1. Straight up this results in a huge reduction in noise ... as the input noise is proportional to the connected source resistance.
moodz

let's say we have 1mA of current coming from the coil, if TIA is used with a 1k feedback resistor, we get 1V at the output, if a 1k damping is used at the voltage amplifier input we would also get 1V the only difference is, in the first approach coil sees AGND and in the second one sees a 1k to the ground, IMO both 1k resistors would generate the same amount of noise, wouldn't they?
as the current has to pass through them anyway.

2. The presentation of a "short" across the coil shorts out any capacitance across the coil .... another benefit. Its more complex than this but it is a benefit.

this could be something nice, I think it will help to some degree I have to test and see for myself.

3. It also becomes obvious that any damping resistance will also damp target responses. A "shorted" coil will actually extend target induced currents. This is what we are trying to detect ... target induced currents ... so damping them is not what we want ... we just want to measure those target currents.

this is correct, but the devil is in the ground, how are we going to distinguish between a long-ground VRM response and a gold coin?
I know the loop will compensate for some of that ground(slowly) but what about hot rocks or little bits and pieces of iron embedded in some clays?
they are not all over the place, we just hit them once in a while.
I'm not sure if the subtractive GEB method would work with ZPD or not...

4. I am working on a report.

Nice.

[moodz]

let's say we have 1mA of current coming from the coil, if TIA is used with a 1k feedback resistor, we get 1V at the output, if a 1k damping is used at the voltage amplifier input we would also get 1V the only difference is, in the first approach coil sees AGND and in the second one sees a 1k to the ground, IMO both 1k resistors would generate the same amount of noise, wouldn't they?
as the current has to pass through them anyway.
the fb resistor is effectively in parallel with the coil resistance and damping fet in the TIA case…only a few ohms … so the noise is lower.


this could be something nice, I think it will help to some degree I have to test and see for myself.


this is correct, but the devil is in the ground, how are we going to distinguish between a long-ground VRM response and a gold coin?
I know the loop will compensate for some of that ground(slowly) but what about hot rocks or little bits and pieces of iron embedded in some clays?
they are not all over the place, we just hit them once in a while.
I'm not sure if the subtractive GEB method would work with ZPD or not...

the gold coin is conductive …. The VRM is not asfaik. That is key to the operation.
I have plenty of hot rock samples that ML cannot reject … it will detect bits of iron they are conductive targets.


Nice.

[Carl-NC]

I think it will help to some degree I have to test and see for myself.

Here is a simple test circuit you can use with any metal detector:



The coil is grounded on one end and virtually grounded on the other end. Yet is still produces a response. I designed a Magnetic Field Probe using this because it shows the actual magnetic field waveform, not the derivative as a voltage-mode coil does. The opamp can be anything, I use an LT6231 for raw bandwidth.

It works because the coil has resistance:


So the gain is RFB divided by the DC resistance of the coil.

[Mr.Jaick]

the gold coin is conductive …. The VRM is not asfaik. That is key to the operation.
I have plenty of hot rock samples that ML cannot reject … it will detect bits of iron they are conductive targets.

VRM is magnetic right, so you say the idea does not detect any magnetic response, for example only the conductive property of iron can be detected but no magnetic.
if what you say is 100% correct then this could be the world's best GEB technique.
as you say it elongates the eddy responses, now we should be concerned about wet salt.
in my experience, it can be quite hard to have a robust GEB on (wet-salt + magnetic) soil.

I think we can use this for the RX of an IB coil as well (without using a damping resistor for the RX coil).
but the coil's voltage never gets above the diode voltage, so it's always grounded.

[Mr.Jaick]

Here is a simple test circuit you can use with any metal detector:



The coil is grounded on one end and virtually grounded on the other end. Yet is still produces a response. I designed a Magnetic Field Probe using this because it shows the actual magnetic field waveform, not the derivative as a voltage-mode coil does. The opamp can be anything, I use an LT6231 for raw bandwidth.

It works because the coil has resistance:


So the gain is RFB divided by the DC resistance of the coil.

so let's say this L is the RX coil for an IB configuration, no damping needed?

[moodz]

VRM is magnetic right, so you say the idea does not detect any magnetic response, for example only the conductive property of iron can be detected but no magnetic.
if what you say is 100% correct then this could be the world's best GEB technique.
as you say it elongates the eddy responses, now we should be concerned about wet salt.
in my experience, it can be quite hard to have a robust GEB on (wet-salt + magnetic) soil.

I think we can use this for the RX of an IB coil as well (without using a damping resistor for the RX coil).
but the coil's voltage never gets above the diode voltage, so it's always grounded.

Yesterday I took the detector to the beach went up to the waters edge and found half a dozen bits of metal rubbish in 5 minutes ... so it works fine in salt water / sand.

As for your other concern ... a picture says a thousand words.

Look at the PIC below ... the blue trace is no target .... the green trace is large ferrous VRM ground and the red trace is small gold coin near the coil ( 50 mm distance ).
Take particular note of the where the zero volt reference is the polarity of the responses and also how the voltage level does not return to zero volts.

The coil is damped in 1.7 microseconds.

Its should be clear that there is a clear signal distinction between target and ground :-)

moodz

[Carl-NC]

VRM is magnetic right, so you say the idea does not detect any magnetic response, for example only the conductive property of iron can be detected but no magnetic.

There are two magnetic responses: instantaneous and viscous. PI detectors normally ignore instantaneous magnetic responses (ideal ferrite) because it dies faster than the decay. Viscous magnetic responses are time-delayed decays so they are not ignored. But the decay curve is not exponential, it is t^-n where n is typically between 1 and 1.1. This is what the 2-point subtractive GB removes. I suspect that ZP damping doesn't change this at all.

so let's say this L is the RX coil for an IB configuration, no damping needed?

Not necessarily. I use a 1k damping resistor in my probe design, otherwise the probe can ring a bit. That's because I'm using a 215MHz opamp.