RX Flyback Blocking for Best Op-Amp Performance

[Carl-NC]

But let me take advantage to ask the professor about the Johnson noise. I had the understanding that this is of very wide spectrum and includes the 1/f noise. Could you dispel my doubts and explain in more detail?

Johnson noise has constant power per BW, and does not include 1/f noise. It is usually due to the inherent random motion of electrons. Most 1/f noise in circuits is due to flicker noise, which is cause by flaws in the silicon lattice which trap & release electrons randomly, but with a 1/f response. MOSFETs have the worst flicker noise, BJTs the best. Carbon resistors also have flicker noise whereas metal film not so much, which is why we prefer metal film for the damping R.

- Carl

[Monolith]

Johnson noise has constant power per BW, and does not include 1/f noise. It is usually due to the inherent random motion of electrons. Most 1/f noise in circuits is due to flicker noise, which is cause by flaws in the silicon lattice which trap & release electrons randomly, but with a 1/f response. MOSFETs have the worst flicker noise, BJTs the best. Carbon resistors also have flicker noise whereas metal film not so much, which is why we prefer metal film for the damping R.

- Carl

Hi Carl,
many thanks for the explanation.

Monolith

[Monolith]

Is that an IB coil on a PI detector, or VLF detector?

-SB

I had difficulties building a PI detector for such small targets, so I made a bast'ard, spawned from a VLF and a PI, using IB coils.
Then I scaled the size of the coils and the corresponding power and surprisingly the high sensitivity remained to scale.

You could also call it a hybrid.

Monolith

[simonbaker]

I had difficulties building a PI detector for such small targets, so I made a bast'ard, spawned from a VLF and a PI, using IB coils.
Then I scaled the size of the coils and the corresponding power and surprisingly the high sensitivity remained to scale.

You could also call it a hybrid.

Monolith

I was wondering about the large bandwidth in the pre-amp which makes me think it is basically a PI type detector. Still not sure how switching "noise" factors into all of this but I'll keep trying to follow along.

-SB

[mikebg]

...For very small targets I need up to 300kHz bandwidth in the preamp. For large targets, 50kHz is good enough. I use IB, by the way.
But where does the sweep speed come in? with perfect Ground Balance the detector only "sees" the difference between the Ground and the Target and this difference happens during the time the coil sweeps over the target...

Monolith, in theory all targets need only 12 Hz bandwidth in the preamp. The filter block (illustrated in the above block diagram) is bandpass. It should suppress frequencies below 3Hz because they represent modulation of GND signal, and frequencies above 12 Hz because they are 1/f noise of RX electronics.
http://www.findmall.com/read.php?34,129427
Excitation and receiving frequency spectrum above 25kHz contains too large GND signal generated by conductive soil. In time domain that means you should avoid sampling at small delays. However metal detectors for demining can operate at 300kHz because they suppress GND signal using two RX loops. Landmines with low metal content are as "very small targets". Note that each RX loop is in induction balance with TX loop.

[moodz]

The proof is in the pudding .....

This is a very good discussion and I was down the beach when I saw it on the phone ... so on the way home I thought about Monoliths kind words regarding the differential coil .... there is no reason the technique I am using should not work on a "real" monocoil. So into the workshop grab one of my early monocoils ... connect to the UNIPI chip breadboard and "unbelievably" ... I had to check it twice ... the same technique works with monocoils.

I have posted the result below .... I can get rid of / attenuate the AIR signal. I cant get rid of the flyback ... that is a physical property of the coil and the sudden cessation of current at turnoff.

I think it was Mikebg who mentioned that the signal to noise was awful on the monocoil because it is masked under the AIR signal.

Well .... ( Aziz ... are you paying attention ) ... not only is the signal to noise on a monocoil quite good when the AIR signal is removed but you also get the excellent descrimination signal as well.

Not only that but to proove a point ..... I set the gain of this preamp to 1 ( unity ) . OK the targets are large ( 4 inch square copper & 4 inch steel ) but not only is there bags of signal but also bags of descrimination signal as well. ( 1 volt without amplification !!! )

The coil I am using is nothing special ... the fastest sample time I could ever get on it was 18 microseconds on a conventional PI front end. You can see how fast it is now

When this gets out ..... I think a new era in MD technology will start.

cheers from moodz.

[moodz]

... point of clarification.

In the previous picture where I note the pulse width change the spikes are not the fly back pulse ... the whole pulse is the flyback pulse.

....and Aziz ... watch out ... when I done with the monocoil I am going to give the IB a work over as well.

Regards,

moodz.

[Aziz]

Hi moodz,

you might measure parasitic coupled flyback voltage.
The flyback voltage behaves similar to off-resonance detector type, where it's voltage varies due to energy loss of the coil.
This can be observed clearly if the flyback voltage is not clipped (mosfet does not break down the flyback voltage).

Check this possibility please.

Cheers,
Aziz

[moodz]

update ... clarification.

A new pic just to make sure that its clear that the monocoil gets the following benefits from this technique in addition to benefits I already listed ....

6. No false descrimination ( thanks Mikebg )
7. Faster sampling due to removal of AIR signal
8. Descrimination ...
9. Big improvment in signal to noise


.... and since I had a celebratory vino .... Aziz ... you can use your IB coils as pizza trays now ... LOL ..... ( no offence intended )

Please accept that as a challenge not an insult ...

moodz (half pi$$ed )

[moodz]

Hi moodz,

you might measure parasitic coupled flyback voltage.
The flyback voltage behaves similar to off-resonance detector type, where it's voltage varies due to energy loss of the coil.
This can be observed clearly if the flyback voltage is not clipped (mosfet does not break down the flyback voltage).

Check this possibility please.

Cheers,
Aziz

Aziz .. observe the clean transition from flyback falling edge to target signal .... ( in the centre of the pic ) .. these preamps I am using have huge bandwidth at unity gain and beautiful recovery .. this one is better than 13 Mhz. The Mosfet is not breaking down and the technique counteracts the parasitic capacitance ... that is why there is no ringing ... think opposite of resonance ( anti-resonance ??? )

Inductive circuits have a "Q" factor ... I have discover how to apply negative "Q". Thus emliminating the tendancy of the circuit to "ring" or oscillate. It is so beautiful I am going to have another drink ....

Hic.

Regards,

moodz.

[Aziz]

Hi Moodz,

look at the peak flyback voltage!!!

Aziz

[moodz]

Hi Moodz,

look at the peak flyback voltage!!!

Aziz

OK already ... I have to turn the CRO on ....

moodz

[moodz]

... the picture is a bit blurry.

... so is my eyesight after this vino

50 volts / div vertical
2 us / div horizontal

about 400 volts peak ??

The MOSFET breakdown is 500 volts ( probably 550 in reality )

The coil is 16 turns 18 inch diameter.

moodz

[mickstv]

Hi Moodz, you've got me and maybe everyone else stumped on how you've got negative "Q". It will be very interesting to maybe one day see your design.



Mick

[Aziz]

Moodz,

you didn't understand me.

Make the measurement for no target, copper and iron. And look at the peak flyback voltage. If you can zoom around the peak voltage, the better we could see the difference.

Cheers,
Aziz

PS: You had one drink too much according to the picture.