NEW !!! Full differential PI front end.

[Stefan]

Hi all,

I would take two same mosfets and drain diodes with single coil power supply. The capacitive load to the coil ends should be almost same. A simple passive adjustment circuit could correct any inbalance.
As the transmit drive would be ignited alternately on the coil ends, a bipolar magnetic field emission occurs automatically even the coil is driven with single power supply.

You can make the design either with P or N channel mosfets. P channel mosfets make the MCU design a bit easier, as the center tap of the coil can be put to ground. With N channel mosfets, ground must be set to the battery +. The analog front end should have a bipolar power supply (like +5/-5V or something similar).

Aziz

Hi Aziz,

The dual supply design has some advantages as the maximum voltage at each resistor is half the max voltage seen across the coil. The two diodes will isolate the coil and if there are imbalances then this can be trimmed out as you have indicated.. The two damping resistors centre tapped to ground would have to be matched very closely and tied to the same common point to ground. The second advantage is that there is a better choice of MOSFETS as the voltage does not have to be as high . A 200 v MOSFET will give me 400 volts prior to avalance.

The Instrument amp would need to be dual supply and the INA163 is a good choice. As I will be using MOSFET for signal input switching there is still a good choice of 250 V signal mosfets to choose from.

Now have just to build the beast. I don't expect it to be easy.

Regards,

Stefan

[Aziz]

Hi all,

here is, what I mean. If you don't use the bipolar transmit pulses (alternating coil current), one can even balance the coil driver with a second switched-off FET stage (see below). The capacitive load of the coil ends to ground should be almost same. The FET below may also avalanche. It adds some capacitive load into the coil.

The differential stage will "kick" the target stronger compared to single ended design.

Aziz

[Aziz]

If you don't use bipolar transmit pulses and second switched-off FET stage, just add a high voltage adjustable capacitor parallel to the below coil half to achive capacitive load balance. Below part can then be omitted.

But the symmetrical FET stage has a better CMRR and PSRR. If the coil driver is made for true bipolar transmit pulses (totally symmetric coil driver), this should have the best performance. It will also degauss the ground!!!

Aziz

[porkluvr]

... If the coil driver is made for true bipolar transmit pulses (totally symmetric coil driver), this should have the best performance. It will also degauss the ground!!!
Aziz

I was having a hard time following this discussion, but now I am totally lost. I can understand how two opposing magnetic fields would degauss the ground when they are temporally displaced, and there could be a usable signal to process. But if these two opposing fields are transmitted at the same time, then it seems to me like they would cancel each other.

So, when the net residual magnetism of the ground equals zero after swinging the search coil, it wouldn't be because of alternating fields wiping the other's residual, it would be because of opposing fields cancelling each other.
There would be no residual magnetism to degauss in the first place.

If the two opposing coils and their transmitted fields were precisely the same but opposite, wouldn't the net target excitation be zero? This train of thought might be the result of my ignorance and two dimensional thinking, but to me it seems very rational.

[moodz]

If you don't use bipolar transmit pulses and second switched-off FET stage, just add a high voltage adjustable capacitor parallel to the below coil half to achive capacitive load balance. Below part can then be omitted.

But the symmetrical FET stage has a better CMRR and PSRR. If the coil driver is made for true bipolar transmit pulses (totally symmetric coil driver), this should have the best performance. It will also degauss the ground!!!

Aziz

Hi Aziz ... I tried something like this but found that the capacitance on the 'balance' or turned off fet does not discharge enough between pulses thus affecting the balance .... this cap needs a leakage resistor or a quick discharge pulse on gate.

moodz

[moodz]

I was having a hard time following this discussion, but now I am totally lost. I can understand how two opposing magnetic fields would degauss the ground when they are temporally displaced, and there could be a usable signal to process. But if these two opposing fields are transmitted at the same time, then it seems to me like they would cancel each other.

So, when the net residual magnetism of the ground equals zero after swinging the search coil, it wouldn't be because of alternating fields wiping the other's residual, it would be because of opposing fields cancelling each other.
There would be no residual magnetism to degauss in the first place.

If the two opposing coils and their transmitted fields were precisely the same but opposite, wouldn't the net target excitation be zero? This train of thought might be the result of my ignorance and two dimensional thinking, but to me it seems very rational.

Hiya Pork .... umm I think that we are talking about alternate pulses being of opposite flux not the same pulse or they would cancel out. Some mine detectors use this type of AC pulse as a monopolar pulse can detonate some types of antitank mines that are sensitive to magnetic fields ..... at least I think thats what we are talking about

moodz

[moodz]

for reference to the current discussion ... credit to Aziz ....
... http://goldprospecting.invisionplus....l=differential
... and repost because I think Aziz lost the original ...

[moodz]

Hi all ... here are bench results for v600a schematic ..

PIC 1 balanced flybacks at either end of coil.


PIC 2 o/p from diff amp ( green ) vs MOSFET gate drive ( white )


PIC 3 o/p from diff amp ( green ) with target vs MOSFET gate drive ( white )


As you can see you dont need fancy bipolar flux capacitors to get good results. In fact the balance circuit consists of two 10K resistors.

moodz

[porkluvr]

Hiya Pork .... umm I think that we are talking about alternate pulses being of opposite flux not the same pulse or they would cancel out.
moodz

Whew. Aye, of course that's the only way that it could make sense. That might have been obvious to you all along but this thread has been a bit over my head for some time now (er, from the beginning). I'm not trying to throw barbs but only to understand as much as I can. Thanks for clearing that up.

Some of the ideas here seem to have been over-simplified, and I might take other ideas to be more complicated than what is needed. It's hard to know the one from the other.

This discussion is interesting to me because I have a design that needs two transmit FETs, to either feed into two separate coils, or else to feed the two ends of a differential coil. Your starting this thread came at a good time, but the ideas can be hard to follow.

Actually, I would not need two seperate FETs except for a desire to degauss. Unfortunately, my two transmit channels have different pulse streams, so there will not be equal but opposite fields, and demagnetization will not be complete. I would like to find an EASY way to remedy that, but I don't see it happening any time soon.

I have not been following this thread too closely (I get tired of the drowning sensations) but still I try to grasp the concepts presented.

[Aziz]

Hi Paul,

thanks. Indeed, I lost this picture and the corresponding spice model. Anyway. The picture shows the conventional difference amplifier with low input impedance. There is an inherent attenuator, which allows early sampling.
A true instrumentation amplifier design should increase the input impedance without attenuating the signal.

The idea behind making the coil driver symmetric is the following: The leakage current through the FET and drain diode will be compensated (eliminating offset voltage). Any temperature dependency of this too. Particularly both parts have high temperature behavior.
Alternating pulses will produce less heat on the parts too.

Balancing is quite trivial and can be made with different solutions.
In the coming days, I will order the mentioned inst. amps.


Hi porkluvr,

I meant driving altenate transmit pulses of course to get alternating magnetic pulses. One could also drive simultaneously to increase the coil current (parallelling the coils with same phase). A differential signal won't be there then of course.

Aziz

[moodz]

Thanks Pork ... I do try to avoid complexity if possible .... anyway here is the v6 schematic ... this is the actual circuit that produced the results above ... I have even added some bias for Carl .

NOTE : if the MOSFET goes into avalanche you will NOT get balance ... you will get something ... but not the nice pix previously posted ... adjust V+ to maximise flyback peak voltage to below MOSFET avalanche or get a higher rated MOSFET .... and where are the clamping diodes ? ... you tell me
... remember this aint no spice simulation.
There are only 8 turns in the coil ... whaaaat you say ... remember it is bifilar so it is the equivalent of a 16 turn coil. Save your money ... use plain 0.8mm dia magnet wire .. twist with electric drill to 1 turn per inch ... I think the turns provide air spacing anyway ... sure you can use more turns ... 22 turns gives me 8 us sampling .. which is also good .. this needs to be experimented with ... hey you ( the reader ) could do it and let us know !

Regards,

moodz

[moodz]

doh ... forgot to crop schematic after resizing ... sorry Carl.

[Aziz]

Hi Paul,

Well done mate.
When will you get rid of your CA3140? This mature chip is quite noisy.

Aziz

[moodz]

Hi Aziz ... thanks ... I have mainly been using the old 40 because it demonstrates that the differential method that you originally devised works better even with old amps and also because I dont feel guilty if I blow one up ... which I actually have not yet with this circuit despite the p/p voltages reaching over 1000 volts .

A further note on the schematic .... V+ for the mosfet drive was set to 6 volts but the amp supply was 12 volts otherwise the mosfet went into avalanche and the balance point shifts unpredicatably ... maybe someone could spice this ... dont know.

Regards,

paul.

[moodz]

Hi Aziz ... have you seen this diff amp ... it is called THAT1510 .... 7 Mhz bandwidth at gain = 100 and 1 nv root hz at 60 db gain.
Seems to be a bit faster than the SSM2019.
$A6.44 here in oz.

moodz.