A Ground Balance Circuit (build and field tested)

[green]

I'm still trying to understand how the GB circuit works with the MPP integrator. I can't find the spice model of the MPP integrator Qiaozhi posted awhile back. Including a try at a spice model. Target sample(.1 volt, 10 to 20usec). GB sample(.1 volt, 25 to 125usec), no EF samples. To get the target sample and GB sample to cancel I had to make both signals equal value. Normally the GB sample would be maybe 1/10 the target sample amplitude and sampled 10X longer. Maybe someone could explain where I'm going wrong. Thanks

[Digger429]

My simple naive understanding of it is all about area under the curve.



If the area under the main sample part (pink) is the same as the area under the GB sample part (blue) then they contain the same energy and will charge the two capacitors in the integrator circuit ( C13/C14 ) an equal amount... and the output of the opamp will be zero.

The shape of the decay curve will vary depending on the soil type, and any combination of main sample time and GB sample time will work to cancel soil as long as the areas under the two parts are equal.

[green]

My simple naive understanding of it is all about area under the curve.



If the area under the main sample part (pink) is the same as the area under the GB sample part (blue) then they contain the same energy and will charge the two capacitors in the integrator circuit ( C13/C14 ) an equal amount... and the output of the opamp will be zero.

The shape of the decay curve will vary depending on the soil type, and any combination of main sample time and GB sample time will work to cancel soil as long as the areas under the two parts are equal.

I like your added circuit for GB. I don't understand how it works with the MPP integrator. The spice model reply #16 cancels with the GB area 10 times the target area. I'm missing something or doing something wrong. Seems like someone on here could help me see why it works.

[green]

I like your added circuit for GB. I don't understand how it works with the MPP integrator. The spice model reply #16 cancels with the GB area 10 times the target area. I'm missing something or doing something wrong. Seems like someone on here could help me see why it works.

Another try at a spice simulation why I think he MPP integrator won't work. Looking at integrator out, same input, 20usec sample time and 150usec sample time. Seems like it wouldn't cancel ground when sweeping the detector with different response times.

Looking at the simulations I see the first has a 10usec sample instead of 20usec. Play with spice reply#16

[green]

Another try at a spice simulation why I think he MPP integrator won't work. Looking at integrator out, same input, 20usec sample time and 150usec sample time. Seems like it wouldn't cancel ground when sweeping the detector with different response times.

Looking at the simulations I see the first has a 10usec sample instead of 20usec. Play with spice reply#16

Playing with spice reply#16 I see at least one mistake I made. Should have made the EF sample time same as target sample time and increased EF delay time to 500usec. Still a small difference in response when changing sample times, maybe not enough to matter. Maybe more errors?

[Digger429]

The output you've got for the integrator is a pulse. I've always understood it to be a constant voltage, one that will not change if you dont move the sample (or wave the coil.)

The sine wave you've got set up on the left, V3, is that mean to simulate the output from the preamp, such as seen at TP3 on the circuit? I'm not sure how that works because on the real circuit, the input (left side) of the resistors R24/R25 are joined together and so fed in phase, but your spice model not doing that. Dont you need your model R1 and R2 to be connected? Look like you're only feeding your R1 with a signal.

In any case, I just opened up my machine and inspected test point TP4 on the oscilloscope.
With GB off, TP4 is half a volt or so, and rises to >4 volts if I hold the coin near the coil.
With GB on and balanced to cancel the coin, TP4 is half a volt or so, and stays at that value if I hold the coin near the coil.

[green]

The output you've got for the integrator is a pulse. I've always understood it to be a constant voltage, one that will not change if you dont move the sample (or wave the coil.)

The sine wave you've got set up on the left, V3, is that mean to simulate the output from the preamp, such as seen at TP3 on the circuit? I'm not sure how that works because on the real circuit, the input (left side) of the resistors R24/R25 are joined together and so fed in phase, but your spice model not doing that. Dont you need your model R1 and R2 to be connected? Look like you're only feeding your R1 with a signal.

In any case, I just opened up my machine and inspected test point TP4 on the oscilloscope.
With GB off, TP4 is half a volt or so, and rises to >4 volts if I hold the coin near the coil.
With GB on and balanced to cancel the coin, TP4 is half a volt or so, and stays at that value if I hold the coin near the coil.

I'm not saying it won't work with the MPP integrator, just saying I think it won't work. I might have to build a MPP integrator to try with my circuit but first I would like to get a spice model working to learn something. Left some things out with the first spice model that might have caused a problem. Tried another model including target, ground, target EF and ground EF samples. Probably still something wrong, still needs equal amplitude target and ground signals even though the ground sample is 5 times longer to cancel. Maybe someone can see my error.

Found one error. Looks closer to working, think there might be more. Added another spice model

The half sine was an attempt at simulating coil swinging over a target.

[green]

Found another mistake. Looks a lot closer. Now all I have to do is understand where my thinking was wrong.

[green]

Found another mistake. Looks a lot closer. Now all I have to do is understand where my thinking was wrong.

I apologize for suggesting the MPP integrator might not work with GB. Didn't understand how it worked and the spice models I tried had errors.

[green]

There are many ways to skin a cat, but they all end up with the same result. This circuit is not revolutionary in that it achieves the goal that has already been discussed in several other threads, but it is perhaps interesting in that it uses circuitry that anybody who has built a Minipulse Plus has already seen and is familiar with, and does not require the cutting of tracks or any circuit modifications to the Minipulse Plus kit at all. Just 5 wires tapping onto the existing board (plus 2 more for power supply).

The general concept of PI ground balance circuits is that they take a Ground Balance (GB) sample very soon after the main sample, and then at a much later time also take an Earth Field (EF) sample of equivalent sample time to counteract any EF effects. Because there is already a EF sample built into the Minipulse Plus, people have discussed addition and subtraction calculations, often with microcontrollers to do the work of determining the correct length of the resulting EF sample.

My design doesnt do any of that, it leaves the standard Main and EF pulses alone, and just places new GB and EF samples in parallel at the appropriate times. The key to inserting samples in parallel is to use a transistor OR gate, as seen here:

http://hyperphysics.phy-astr.gsu.edu...angate.html#c2

and as seen on the circuit:

https://i.imgur.com/wdrVVoB.jpg

I'm not sure the type of FET is critical as long as it is similar to the J113. I had some MPF102s lying around, so I used those.

To generate the new GB and EF pulses, you need to tap a wire onto the existing test point TP7, and make a new copy of the kit circuit that creates the pulse train:

https://i.imgur.com/oYFuQwz.jpg

For the new circuit, pulses that went to the integrator at point C now go to D, and pulses that went to D now go to C... because the ground balance circuit works to cancel signals.

---

At top left there is a potentiometer on the circuit diagram, but in my final build I eliminated that and just used a 1K resistor with a 5.6pf capacitor for C8, so the ground balance sample is taken 3 uS after the main sample is done.

The circuit diagram mentions 192uS for the time between GB and EF, but thats just what I used so the timing of the later EF pulse would not impinge on the next transmit pulse. Make it as long as you can, depending on your sample pulse rate.

R11 (and C11) controls the width of the ground balance sample. It also acts effectively as a "strength" control in that a longer time means more ground balance effect. With a potentiometer, mine can vary between 10 - 150 uS. You will need to experiment a bit here to see what works for the soil you have locally.

I also put a dual gang potentiometer for the outputs that go to C and D to control the ground balance strength. With field testing I think I'll throw that out and just put a switch, because the potentiometer at R11 does the same job.

---

It is not a miracle worker, but it is better than nothing. Its the difference between giving up in frustration at that damned ground and going home, and being able to stay out and swing the machine some more. Field testing at sites around Ballarat (Victoria, Australia) lets me cancel the ground completely if I have a main pulse width of about 30uS and a GB sample width of about the same. For a sample of very very mineralised soil I took from Wedderburn, I need to wind the GB width dial all the way up to 150uS to cancel it.

As other people who have built GB circuits have noted, a strong GB balance circuit can also cancel out targets making a "hole" in response so they wont be detected at all, and some large targets can be "negative" in that they make the metal detector go quiet instead of beeping. So use the minimum amount of effect you can get away with.

I again apologize for side tracking your thread. I need GB to detect in my back yard so adding GB is important. I use a quart zip lock bag of dirt from the back yard and another from California to test canceling for my bench circuit. Most of the time I monitor integrator out with a digital voltmeter when adjusting for GB or testing for change with targets. You mention it's not a miracle worker, what problem do you have with the GB circuit? The dirt from California and my back yard cancel with the same settings so I was surprised you had such a big difference with your samples. I added an absolute value circuit to see targets both side of the hole but if your looking for smaller nuggets cancelling out larger targets might be an advantage.

[Tibuck19]

Well done.

That was on my "planned list" of future projects. Now it looks like we just need an add-on PCB to help all you nugget hunters.

I want one please lolll

[homefire]

I would be most pleased with a Simple Schematic and PCB Foil stencil .

[6666]

Most of the time I monitor integrator out with a digital voltmeter when adjusting for GB or testing for change with targets.

I was wondering how you monitored the changes.

[geoscash1]

R11 (and C11) controls the width of the ground balance sample. It also acts effectively as a "strength" control in that a longer time means more ground balance effect. With a potentiometer, mine can vary between 10 - 150 uS. You will need to experiment a bit here to see what works for the soil you have locally.

This sounds like a nice quick solution and would be great to have a visual, like with an on panel meter so you could see the ground cancelling effect at the ranges specified (10 to 150us).

[Tibuck19]

for us none technical guys, is the photos and help on how to make this work for our MMP ? should I make a new coil with small magnetic wire and shielding for this setup ?