Ground Balance and EMF

[CAS]

I have always wondered why a separate EMF sample is taken in PI timing circuits when the EMF component should be present in the GB sample. Why not just subtract the GB sample from the main sample? Why subtract a separate EMF sample?

And before anyone says it - yes I have ITMD and have read it a few times (although maybe I missed that paragraph)

[Qiaozhi]

I have always wondered why a separate EMF sample is taken in PI timing circuits when the EMF component should be present in the GB sample. Why not just subtract the GB sample from the main sample? Why subtract a separate EMF sample?

And before anyone says it - yes I have ITMD and have read it a few times (although maybe I missed that paragraph)

Since the GB sample occurs a finite period of time after the main sample, it will be somewhat weaker in amplitude. Therefore the signal from the GB sample needs to be amplified so that it has the same amplitude as the signal from the main sample. Any target with a particular decay will then be eliminated when the amplified GB signal is subtracted from the main signal. Since both samples (main and GB) contain a contribution from the Earth Field ... you can now probably spot the problem(?). That is: when the GB signal is amplified, the EF is also amplified. So a simple subtraction process will not remove the EF.

Mathematically, with S1 = main sample, S2 = EF sample, and S3 = ground sample:

X = A1(S1-S3) - A2(S2-S3)

which subtracts the ground sample from the main and EF samples before finding the difference.
A2 is the ground balance control gain, Adjusting this will eliminate ground.

Alternatively, you can also use this equation:

X = A1(S1) + A3(S3) - A2(S2)

where A2 = A1 + A3

[CAS]

Ah of course. Thanks for the explanation. I have been working on a time base sampling method where multiple samples 1- 4 uS apart are taken over a short period (between 25 - 60uS) and compared directly with each of the previous samples that are stored in memory. In theory the GB and EMF components should be the same as the previous pulse sample. There is no need for separate GB or EMF sampling. Getting the timing right is the critical part.

[Qiaozhi]

Ah of course. Thanks for the explanation. I have been working on a time base sampling method where multiple samples 1- 4 uS apart are taken over a short period (between 25 - 60uS) and compared directly with each of the previous samples that are stored in memory. In theory the GB and EMF components should be the same as the previous pulse sample. There is no need for separate GB or EMF sampling. Getting the timing right is the critical part.

The contribution from the ground will decay with each successive sample, but the EF will remain the same. So you will still need to take a late sample to determine the amplitude of the EF without any target signal. Instead of using a gain stage to boost the GB signal, you can increase the sample pulse width, which is another way of increasing the amplitude.

[CAS]

I dont think I explained my idea very well. I have done a rough sketch to show you what I am doing



I record for example 3 data points A, B, C then on next sample record A1, B1, C1 - I then compare A to A1, B to B1 etc and use the result to see if there are any changes. The EMF and GB in each sample ie A vs A1 should be roughly the same.

[Qiaozhi]

That makes it a lot clearer.
Your idea is based on the assumption that only a real target will produce a change between associated data points (i.e. A-A1, B-B1, and C-C1), whereas the ground and EF signals will be eliminated since they remain constant.

I suspect (in reality) that this will not be the case. Firstly, the ground signal will only remain constant if the ground is homogeneous, and the coil remains parallel to the surface of the ground. In highly mineralized ground, the soil matrix is constantly changing, and there will be little difference between the ground signal and a real target. However, this approach might help to weaken the Earth Field response, even if it does not completely eliminate it.

You can get perfect EF elimination using bipolar pulsing, which can even ignore a large whiteboard magnet. That's leaves only the GB signal to deal with. You could try a combination of taking the difference (which would need to be an addition with bipolar pulsing) between successive samples, plus use the second sample for ground balancing. In that case, you might only need to take two samples per cycle.

As you seem to be testing this with real hardware ... how is your idea working in practice?

[CAS]

Its all coming together nicely. At the moment I am only using a Chipkit UNO as my controller as its quick and easy to program but am going to change that soon to something along the lines of a Nucleo board combined with a 24bit ADC. I have made the whole test rig modular so I can change bits that dont work as expected.

As for the GB not remaining exactly constant, I am incorporating an adjustable 'error' factor that can be used to compensate for things like that. I am hoping to have a full working prototype to take out to the property in Bendigo (lots of hot rocks etc) at Easter.

[CAS]

I was just thinking its a pretty simple thing to take later GB samples and subtract them from their corresponding data points before I do the data point comparisons. Its just another few lines of code.

The other thing this design incorporates is a variable constant coil current. The circuit all works but not sure of the real world applications yet but will be interesting to see what effect it has in the field.

While I use LTSpice and lots of other calculations to develop ideas, I much prefer to do physical real world testing where there a few more variables in play.

[Qiaozhi]

The combination of modular construction and software provides great flexibility for your idea.

As you say, LTSpice is extremely useful for developing ideas. It also has the advantage that you can "blow up" the circuit as many times as you like, without costing any money.
But ..... as Carl once said (concerning a particular detector):
"A day in the field is worth a month in the lab. Get it off the bench ASAP, you will quickly see whether that great idea is worth a crap."

A [slightly] more famous person also said:
"One good test is worth a thousand expert opinions." - Wernher von Braun

Let us know how the testing goes.

[Davor]

There is yet another good point for employing the EF sample: it turns the gain circuitry into a chopper stabilised system, which in turn drastically improves behaviour under 10Hz, and makes non-motion operation meaningful.