Vallon VMH3CS Mine Detector

[Qiaozhi]

I had assumed that the HV kick was from capturing the HV flyback, but the TX voltage waveforms seem to refute this... capturing the HV flyback reduces the voltage significantly from the 250v that is indicated by the TX coil waveforms. Also capturing the flyback of 1 amp coli current will only kick start to about 750 ma. I suspect there is an addition HV being generated to kickstart at least 1 MOSFET in each TX cycle to 1A with the other MOSFET working off LV with total resistance set to limit current to 1A.

I think we're both arriving at the same conclusion. The Vallon does not appear to be attempting to recover energy from one flyback to drive the other, otherwise the flyback voltage would be greatly reduced. Also, the waveforms do not look consistent with the energy recovery process. Perhaps there's two poly-phase boost converters onboard, one for each polarity. Hence the 4 toroidal transformers. Using a higher voltage to stimulate the coil allows the energy to build up in a much shorter time period. If they were using an induction-balanced coil, I would hazard a guess that sampling was being done during TX turn-on, since it would allow for earlier sampling to occur. However, Eric says that this is not the case, as the coil is a mono. Therefore sampling must be occurring during TX-off, which is contrary to the use of a 1.5mH inductance coil.

I am trying to deduce whether the complex bridge arrangement is leaving the coil disconnected from the damp for a portion of the flyback.

Since the Vallon is designed to detect minimum metal mines, and hence low conductivity targets, the only possible way to sample early enough is to start the flyback with the damping resistor disconnected. If it's then connected at the point where flyback is at maximum (or at least close to it) the decay time can be massively reduced, allowing earlier sampling.

The coil seems to have a floating ground as there is no connection to the yellow/green wire adjacent to the supply wires.

As the Vallon uses bipolar pulsing, I would expect that to be the case.
Interestingly, the bipolar pulses are grouped relatively close together, followed by a longer period. This is either to allow for some complex signal processing to happen and/or sampling only occurs after the positive pulse. The negative pulse would then only be there to ensure the ground matrix and any targets do not become magnetized in the process. If it's the latter, any Earth field elimination would require a later sample as per a mono-polar transmitter.
Personally I imagine it's to allow sufficient time to do all the calculations, with the EF being eliminated by bipolar sampling.

Just a few thoughts.

[Ferric Toes]

Eric,
On the torroid side of the board, there are some black cylindrical components... 6 at the top right just to the left and down from the yellow and orange wires and 4 more in the middle of the four MOSFETs. I believe these are diodes in a DO213 package... I have only seen this package used in schottky diodes. Can you discern any identifying markings? I believe the 8 MOSFETS comprise a complex bridge arrangement (still working on it). Is there a damping resistor in the distribution box where the yellow and orange wires are connected to the coax or is it somewhere on the pictured board. I am trying to deduce whether the complex bridge arrangement is leaving the coil disconnected from the damp for a portion of the flyback.

Hi Jim,
They are diodes, and after a hard job scraping off a bit of the conformal coating to connect to a test meter, the one to the left of the Mosfet at the top right of the board has a forward voltage of 0.488V. There is another size similar diode to the left of the blue and red wires and this has the same forward voltage. It is reverse connected across the supply wires and may be part of a reverse battery protection arrangement, the other part of which may be on the distribution board. If you put the batteries in wrong they do not connect anyway. The only marking I can see on the ones associated with the Mosfets is the figure 7. On the one by the supply wires it has 10 as well as a 7. I do not know what these numbers relate to. Also .488 Vf seems a bit high for a Schottky. There are 14 of the no.7 diodes on the underside of the Mosfet board, 12 of which seem to be associated with the toroidal transformers. Also a whole heap of other diodes.

No sign of any damping resistor anywhere. A testmeter across the yellow and orange wires reads 1.2Meg. The coil seems to have a floating ground as there is no connection to the yellow/green wire adjacent to the supply wires. This Y/G wire connects to the negative blue wire and the shielded box which houses the electronics.

Eric.

[KingJL]

Eric,
On the torroid side of the board, there are some black cylindrical components... 6 at the top right just to the left and down from the yellow and orange wires and 4 more in the middle of the four MOSFETs. I believe these are diodes in a DO213 package... I have only seen this package used in schottky diodes. Can you discern any identifying markings? I believe the 8 MOSFETS comprise a complex bridge arrangement (still working on it). Is there a damping resistor in the distribution box where the yellow and orange wires are connected to the coax or is it somewhere on the pictured board. I am trying to deduce whether the complex bridge arrangement is leaving the coil disconnected from the damp for a portion of the flyback.

[Ferric Toes]

I am assuming that the yellow and orange wires at the top right of the board with the torroids are the wires that go to the coil. Or maybe I shouldn't assume!

Yes that is correct. The red and blue wires are the supply volts. That connector plugs into a third pcb which is basically a distribution board with cable runs to the control switch, display/push button, and data/headphone connector. There appears to be a small switch mode psu on that board also. Maybe for the LED display.

On that third board there is a second 8 pin connector where the raw battery supply comes in (red and blue wires also) and the coax cable to the coil goes out from that. The coax links by pcb tracks to the orange/yellow wires, with the orange being the coax braid.

Eric.

[KingJL]

I am assuming that the yellow and orange wires at the top right of the board with the torroids are the wires that go to the coil. Or maybe I shouldn't assume!

[KingJL]

What do you make of the 8 Mosfets and the ring core transformers? I surmise that the Mosfet's gate drive is via the transformers.

Eric

I am still studying that... It seems the arrangement is different between the MOSFETs on the bottom vs the top. The 4 on the bottom of the board (the side without the torroids) are arranged in pairs with the drains connected together and sources tied together thru resistors to the same point for each pair (can't tell where the gates are connected), whereas the 4 on the top are arranged in pairs with the source connected to the drain and gates tied together for each pair. The torroid transformer windings appear to use awfully heavy gauge wire for use a pulse transformrs. There needs to be a means of generating the HV for driving the leading edge of the TX for ~10us. I had assumed that the HV kick was from capturing the HV flyback, but the TX voltage waveforms seem to refute this... capturing the HV flyback reduces the voltage significantly from the 250v that is indicated by the TX coil waveforms. Also capturing the flyback of 1 amp coli current will only kick start to about 750 ma. I suspect there is an addition HV being generated to kickstart at least 1 MOSFET in each TX cycle to 1A with the other MOSFET working off LV with total resistance set to limit current to 1A. I agree that the gates are most likely transformer driven. My experiments with driving MOSFET gates with pulse transformers suggests a 2:1 ratio with a minimum 500uH in the primary (preferably 1mH).

It keeps getting curiouser and curiouser!!!!

[Ferric Toes]

The Vallon coil is 11.5in x 6in and my coil is 10.5in circular. Measurements are take from centre to centre of the winding bundle. My winding should be 11in diameter but the winding jig was a bit too tight so I lost 1/2in. However, there was little noticeable difference in performance. Vallon make a 12in circular coil but quoted me 573 euros. The battery cap is 35 euros, so you can imagine what a new VMH3CS would cost. It is a shame that other manufacturer's coils don't work as I have quite a selection. I am going to wind up a 15in coil at some stage, as that should show a noticeable improvement.

What do you make of the 8 Mosfets and the ring core transformers? I surmise that the Mosfet's gate drive is via the transformers.

Eric

[KingJL]

Eric,
How did your field testing with the Vallon w/your home made round mono work out? Was there a noticeable difference with it vs the original vallon coil?

[Ferric Toes]

These are the boards referred to in 443. The boards are plugged together and the first picture shows what you would see on top and bottom side when you turn it over in the plugged state. The second picture is what is inside (middle layers) when unplugged. The furry component is an inductor which is later versions is slightly smaller and mounted on the recessed end of that board.



The four IRFR320 mosfets are at the top of the board with the OPA627 preamp on the right near the yellow/orange twisted lead.



Four more IRFR320 mosfets on the underside and the 4532 P/N cmos gate almost under the OPA627.

Eric.

[Ferric Toes]

OK, I see now there is much work on this topic ! But I don't understand why Eric was surprised by the voltage waveforms I have measured across the coil ...

Philippe

Hi Philippe,

Your waveforms are quite correct, but we approached the problem differently. I look at the unloaded TX waveform first to check the shape, amplitude, width and the frequency. I then look at the inductively loaded current waveform across a 0.1R resistor to check the turn on time and turn off time, as it is this that generates the necessary magnetic field and generates eddy currents in the target. At some point I also measure the flyback voltage but didn't examine the detail of its shape. The difference in turn off and turn on was new to me, but it make sense now we are certain that the turn on is assisted by the stored 250V flyback from the previous alternate pulse. When I use the x10 probe to give 20V/cm vertical scale, I can then see the minus/plus 6V alternate TX pulse along with 80V of the high voltage spike either way. By reducing the timebase to 10uS/cm I can clearly see the differences in shape of the two spikes. That was something I had not done before.

Eric.

[Ferric Toes]

I have 2006 VMH3CS where it is easier to separate the two circuit boads. In the TX there are 8 IRF320 mosfets; four on the topside and four on the underside positioned just slightly offset to one another. It is not possible to see how they are interconnected as it is a multilayer board. It looks like three layers, as up against the light there is an internal ground plane layer that does not appear on the top or bottom. I must say that this is the most complex detector electronics that I have ever seen. The OPA 627A, which I reported as possibly being the RX preamp, has another IC on the underside with the number 4532A on it, which are N and P channel mosfets.

Eric.

[Phiphi]

In #342, KingJL posted a simulation with real components. There's an updated version in #355.
I also simulated something similar using ideal components in #362, and Teleno offered a monopolar version in #364.
KingJL's simulation looks the closest to the actual results.

OK, I see now there is much work on this topic ! But I don't understand why Eric was surprised by the voltage waveforms I have measured across the coil ...

Philippe

[Qiaozhi]

In #342, KingJL posted a simulation with real components. There's an updated version in #355.
I also simulated something similar using ideal components in #362, and Teleno offered a monopolar version in #364.
KingJL's simulation looks the closest to the actual results.

[Phiphi]

I have made a simulation with PSPice trying to get the same current and voltage coil waveforms than those measured by Eric for the first one and by me for the second one.
Here is the schematic :



That's the classic one but the Mosfet source is polarized with a signal going from -300V to -6V in 8 us instead of a constant voltage.
The result is shown below



On top is the current in the coil increasing quickly to about 850mA in 8us with a parabolic shape followed by a linear increase to about 900mA in 40 us.
On bottom is the voltage across the coil with the first part with a sawtooth shape and the second one with the classic back-emf.

So now we have to imagine how to realize practically the source polarization.

[Phiphi]

Eric

The image in your post 53 is exactly as the description I made in my previous post : an increasing current with decreasing slope during about 8 us up to 840 mA followed by a linear increase up to 1A for the last 40us.

Have you done the measurement I made by just setting a scope probe on the coil pins ? You should have obtained the same results than mine, I guess.

Philippe