Nulling / Ground Balance

[simonbaker]

A lot of TGSL Tuning & Troubleshooting thread was hijacked by me and Qiaozhi trying to get in agreement on nulling/ground balancing, so I'm starting separate thread for more nulling/GB stuff.

Some initial practical observations.

The coils' null position seems to have two semi-stable positions on either side of the minimum where small coil adjustments don't affect the zero crossing too much. Then there is the wishy-washy area in the middle where slight adjustments seem to move the phase around a lot.

Also, when comparing to the GB sync pulse, both sides seem to make a fairly convenient zero crossing to center the sync pulse on, but one is rising and the other descending.

I find it very hard to make subtle shifts in the null phase, so from a practical standpoint, it seems I'm stuck with one of the two quasi-stable choices.

To be continued...

-SB

[simonbaker]

target models

Realizing I have been assuming the wrong model for ferrite/ground, I would like to examine the physics assumptions for targets in general, in case there are other important facts that can be brought out.

Some beginning questions are:

1. Are there targets that only modulate RX signal amplitude without affecting phase?

(assumed answer: yes, ideal ground/ferrite).

2. Are there targets that transmit a signal that has a characteristic phase shift from the TX signal?

(assumed answer: yes, conductive targets such as metals: copper, aluminum, gold, silver, nickel. Not sure about iron and steel.)

3. Are there targets that create a characteristic phase shift in the RX signal, regardless of where the null phase is. In other words, rather than injecting a new signal with phase offset from the TX signal, they shift the RX signal a constant amount in a certain direction, which does not depend on the absolute RX null signal phase.

4. Similar to 3, but direction related to null phase. Are there targets which can shift the RX signal a constant amount with a sign determined by the null signal phase relationship to the sync pulse. In other words, if the RX null signal makes a positive offset at the output of the SD, the target shifts it X degrees to make the voltage more positive. If the null signal is negative at the output of the SD, the target shifts it X degrees to make the voltage more negative. But the amount of phase shift, like (3), does not depend on the absolute null signal phase.

5. Are there any other types of response that can be imagined, other than a combination of the above four?


It is reasonable to expect some targets might be a combination of the above models. A target with some ferrite and some metal for example I would expect would act in a predictable combinatorial way, unless something more subtle goes on.

There is some mention about ferrite not being ideal but causing some additional phase shift. I would like to examine that and understand the nature of the additional shift, if that is actually true.

-SB

[simonbaker]

ideal ground vs. electronics

I am trying to do some very careful tests to demonstrate the use of ferrite targets to ground balance the TGSL. My original tests seemed to show that the null signal phase did not affect the pot setting, which does not jive with the effect ferrite should produce. I am also learning how much time and care must be spent setting up a legitimate experiment that isolates the true effect from all the "red herrings". For that reason I would qualify any result with "it seems..." for experiments coming from my workbench.

I now think that my initial findings that null phase does not affect ground balance may have been a coincidence caused by the fact that the two stable null phases on either side of the minimum point have zero crossings that occur at a similar phase offset from the TX signal, although they cross in "opposite directions" (at least for the coils I have been using). Since the rule is to center the GB sync pulse over the zero crossing, it is not surprising that the pot setting is similar even though the null phases are radically different.

I'm finding something else that I would like to understand.

For ideal circuitry and ideal ferrite, we should center the GB sync pulse over the null signal zero crossing. We can tell we are there by checking the voltage at the output of the Synchronous Detector (SD) on capacitor C12 -- it should be zero.

Then when we wave ferrite, there should be no pulse in the GB channel and no beep in the audio.

However, I find that I must turn the GB pot to change sync pulse about 5 to 10 degrees more to stop the audio beep.

My question is: is this due to a non-ideal nature of ferrite, or is it due to non-ideal electronics?

Possible non-ideal electronics causes could be:

1. Synchronous detector has a threshold that causes different amounts of positive and negative RX signal to accumulate at capacitor C12. Ferrite changes the amplitude of the RX signal, which requires a slightly different balance point to make the voltage at C12 zero.

2. The RX signal is not symmetric. This is visibly true for my coils near the minimum null.

3. Sync pulse "dwell" is not 180 degrees. In other words, postive pulse is longer or shorter than the negative pulse. I'm not sure this would matter.

If due to non-ideal ferrite, is it due to the equivalent of a conductive target component, or some other effect. If due to a conductive target component, then theoretically setting the phase of the null equal to the phase of the conductive target component should eliminate the phase shift. However, that phase may not be in the range of the GB sync pulse and not easy to test the theory. Rigging up a wider GB sync pulse range could answer that question.

-SB

[simonbaker]

TGSL ground balance technique

It's been stated that the principle of ground balancing is to center the GB sync pulse over the RX signal zero crossing. This makes the GB channel insensitive to ferrite/ground material (which only affects the RX signal amplitude without affecting the phase).

However, we can ask a more general question as to the purpose of ground balancing and some considerations for the TGSL design in particular. A more general criteria for ground balancing may be to "minimize the interference of the ground with target detection".

For the TGSL, it turns out that any effect (conductive target or ferrite/ground) that makes the DISC channel pulse and GB channel pulse move in opposite directions will not make a beep.

So does this open a broader door for ground balancing? Rather than strictly centering the GB sync pulse over the RX signal zero crossing, we could set the null signal phase and the GB sync pulse so that the GB channel and DISC channel move in opposite directions in response to ferrite/ground, and so make a detector that does not "respond" to ferrite/ground.

We have to be careful though. The DISC pot is turned over about a 90 degree sync pulse range during normal operation. Depending on where we set the null signal phase, we could have a situation where a change in the null signal amplitude causes a positive pulse at one end of the DISC pot scale and a negative pulse at the other. This effect may depend on which side of the null minimum we choose for the null signal phase.

But anyway, we have some latitude. So the question is, can we do better than centering the GB sync pulse over the null signal zero crossing? Can choose a configuration where we truly minimize the interference of the ground with good targets?

For example, if our normal ground balancing causes us to choose a null phase that falls near the phase of silver targets, then even though we make the GB channel insensitive to ferrite/ground, the DISC channel will be very sensitive to ground, and possibly the ground effect will oppose a silver target signal (it could reinforce it too). We have not achieved a detector that is insensitive to ground.

I don't know if such a configuration is possible or likely, but it is worth analyzing to see what we want to do. We may want to split the difference and allow a moderate sensitivity in both the GB and DISC channels as long as ferrite/ground moves the signal in opposite directions.

Also it may be advantageous to choose a null signal that moves the GB sync pulse as far away as possible from good targets, so that the GB channel is more sensitive to good targets (Qiaozhi suggested this might make a more noisy detector, I'm not sure why).

Maybe worth studying these considerations.

-SB

[Altra]

Hi Simon,

Take a look at this article it may have a few tidbits you can use.

http://jb-ms.com/Baron/gb.htm

Regards
Mark

[simonbaker]

Hi Simon,

Take a look at this article it may have a few tidbits you can use.

http://jb-ms.com/Baron/gb.htm

Regards
Mark

Thanks! I have been looking at that, his terminology is a little misleading in places, but I'll continue to look at it.

-SB

[simonbaker]

nulling with ferrite blob

I recently noticed something.

Waving a ferrite blob across our DD coils actually "unbalances" the coil dynamically. It can even temporarily reverse the polarity of the null as it passes across the coil overlap.

This seems potentially quite different than pumping a coil up and down over the ground. Especially if the ground has a fairly even distribution of ferrite.

I also observed this. If you choose a null on the side with "less overlap", passing ferrite over the overlap can not change the polarity of the null, rather it increases the null signal.

If you choose the side with more overlap, passing ferrite over the overlap reduces the null signal and can actually change the polarity of the null, eventually increasing it with the opposite polarity.

Don't know if this has any importance.

However, I'm looking again at the business of nulling and ground balance and I think I will have some new conclusions (eventually), a little different than my last ones.

-SB

[simonbaker]

I recently noticed something.

I also observed this. If you choose a null on the side with "less overlap", passing ferrite over the overlap can not change the polarity of the null, rather it increases the null signal.

If you choose the side with more overlap, passing ferrite to either side of the overlap reduces the null signal and can actually change the polarity of the null, eventually increasing it with the opposite polarity.

-SB

Aggh - totally screwed those statements up.

This is correct according to my last tests:

If you null with more overlap of the coils, ferrite on either side of the overlap drives the null signal towards the opposite polarity. Ferrite directly over the overlap amplifies the null signal.

The opposite is true if you choose null with less overlap.

These exact results may depend on the size of the ferrite. Ferrite must be small enough to fit inside overlap region, and big enough to reverse the null signal polarity.

-SB

[simonbaker]

proposed test

Here is a test I'd like to see to satisfy my curiosity. I can't do it well due to EMI noise on my bench. Anyone with a low-noise basement and a rotating target presenter would be ideally suited.

Basically, here's the test:

I'd like to take some video of oscilloscope showing the outputs of the LM308s as dual traces, adjusted close together but not overlapping. Horizontal should be .1 seconds / div. Vertical should be .5 V or 1 V per division or whatever makes best trace. Audio of the speaker should be recorded also.

The target is smallish piece of ferrite, fairly close to coils. The coils should be un-potted so we can adjust the overlap for nulling.

Two videos will be taken, one for each nulling of the coils.

The first video will have the null on side with more overlap. Adjust so RX null signal at output of LF353 is 1 to 2 volts.

The second video is similar except null on side with less overlap.

Turn brightness of scope very high so trace persists fairly long.

I'm interested in shape of output waveforms. Does inverting the null signal essentially "invert" the waveforms at the LM308 outputs?

It is pretty hard to assess this, it requires very consistent sweeping of the target. It helps if the target sweep can be synchronized with the scope sweep so the waveform pulse is centered, but anything is OK if enough video is captured. Probably want to capture 20 to 30 seconds for each video.

For fun, same tests with a Euro target can't hurt either. Waveforms shouldn't change regardless of the null, unlike the ferrite.

-SB

[Dennis the Mennis]

Here is a test I'd like to see to satisfy my curiosity. I can't do it well due to EMI noise on my bench. Anyone with a low-noise basement and a rotating target presenter would be ideally suited.

Basically, here's the test:

I'd like to take some video of oscilloscope showing the outputs of the LM308s as dual traces, adjusted close together but not overlapping. Horizontal should be .1 seconds / div. Vertical should be .5 V or 1 V per division or whatever makes best trace. Audio of the speaker should be recorded also.

The target is smallish piece of ferrite, fairly close to coils. The coils should be un-potted so we can adjust the overlap for nulling.

Two videos will be taken, one for each nulling of the coils.

The first video will have the null on side with more overlap. Adjust so RX null signal at output of LF353 is 1 to 2 volts.

The second video is similar except null on side with less overlap.

Turn brightness of scope very high so trace persists fairly long.

I'm interested in shape of output waveforms. Does inverting the null signal essentially "invert" the waveforms at the LM308 outputs?

It is pretty hard to assess this, it requires very consistent sweeping of the target. It helps if the target sweep can be synchronized with the scope sweep so the waveform pulse is centered, but anything is OK if enough video is captured. Probably want to capture 20 to 30 seconds for each video.

For fun, same tests with a Euro target can't hurt either. Waveforms shouldn't change regardless of the null, unlike the ferrite.

-SB

Hi Simon ,

I've did some testing as described above ( only I've used a ferrite rod instead of the prefered smallish piece of ferrite , missed that part ...
) on my TGS (not L) .
As far as I can see , and that is pretty tough on my ancient scope , the signals and waveforms on the outputs of the LM308's do not change for both nullings .
This is the same for the test with the euro coin : the waveforms do not change .
I've tried to capture the waveforms but I think the shutter speed of my digital camera is not good enough : the video doesn't make any sense.

I will try again with a smaller piece of ferrite if time's there.

regards


Dennis the Mennis

[simonbaker]

Hi Simon ,

I've did some testing as described above ( only I've used a ferrite rod instead of the prefered smallish piece of ferrite , missed that part ...
) on my TGS (not L) .
As far as I can see , and that is pretty tough on my ancient scope , the signals and waveforms on the outputs of the LM308's do not change for both nullings .
This is the same for the test with the euro coin : the waveforms do not change .
I've tried to capture the waveforms but I think the shutter speed of my digital camera is not good enough : the video doesn't make any sense.

I will try again with a smaller piece of ferrite if time's there.

regards


Dennis the Mennis

Thank you very much for trying the test! Hope you had a great holiday. Good to have you back here.

Well, that is very interesting. I'm not completely sure what to expect, but I thought the different nulls would show waveforms that moved in "opposite directions", if we could carefully compare the waveforms. I'll have to think about it some more too.

The smaller piece of ferrite might help.

It's difficult to see the waveform shape because it depends on how you scan the ferrite in front of the coil. That is why a machine like dfbowers "merry-go-round" for scanning the target would be useful.

Basically, targets often make a "W" shaped waveform at the output of the LM308, or "M" if inverted. That's kind of what I'm looking for.

Maybe we can capture it somehow.

By the way, set the GB pot and DISC pots to mininum for test.

Regards,

-SB

[Jerry]

Well I made an attempt at doing the ferrite test this afternoon. I slowed the sweep way down on the scope and took about 1 second exposures. I had to play with it quite a bit to get anything but I saw several interesting things.

In all cases the GB channel is the lower trace. I did take some both inside and outside of null, but I did not know what the frame number was so will have to repeat that part of the test.

I post the pictures to see if this is even close to what you are looking for.

Jerry

[simonbaker]

Well I made an attempt at doing the ferrite test this afternoon. I slowed the sweep way down on the scope and took about 1 second exposures. I had to play with it quite a bit to get anything but I saw several interesting things.

In all cases the GB channel is the lower trace. I did take some both inside and outside of null, but I did not know what the frame number was so will have to repeat that part of the test.

I post the pictures to see if this is even close to what you are looking for.

Jerry

Hey, very good! I see something I was looking for. Third photo, GB channel trace looks kind of inverted. I was expecting the GB channel on one side of null to look inverted to other side.

Are the last two photos for a metal target?

Really appreciate it. Excellent work. No way could I get that kind of consistency.

I'm trying to formulate an opinion about which side of null is preferable to null coils to, and why. I'm sure many people have opinions, would like to hear their reasoning.

Regards,

-SB

[Jerry]

Hi Simon:

The last two photo's were with a different scope sweep speed. I actually took those before the others. I slowed down a lot for the first three and probably should have left the last two out of my report.

I need to get better at recording stuff as I do it.

Here are two more pictures that have nothing to do with the test but are still interesting. Scope is connected as before but I shorted out the input terminals on J2. The pictures are a time delay of the circuit noise. I need to do this again and sort out the various frequencies shown.

It is too bad I cannot see this in real time. These pictures are using a pretty slow shutter speed. All you see on the scope visually is a dot jumping up and down as it sweeps from left to right. I will do this again tomorrow and write down all the settings for camera and scope.

Jerry

[simonbaker]

Hi Simon:

The last two photo's were with a different scope sweep speed. I actually took those before the others. I slowed down a lot for the first three and probably should have left the last two out of my report.

I need to get better at recording stuff as I do it.

Here are two more pictures that have nothing to do with the test but are still interesting. Scope is connected as before but I shorted out the input terminals on J2. The pictures are a time delay of the circuit noise. I need to do this again and sort out the various frequencies shown.

It is too bad I cannot see this in real time. These pictures are using a pretty slow shutter speed. All you see on the scope visually is a dot jumping up and down as it sweeps from left to right. I will do this again tomorrow and write down all the settings for camera and scope.

Jerry

That's interesting. I'd like to know what the approx noise voltage is with RX coil shorted at the PCB (you need very small jumper to short, otherwise, antenna).

Also relevant is where the DISC and GB pots are set, just for interpreting phase of signals.

It also is useful to make the two traces a little closer together vertically to compare phase.

It might be interesting to ground the non-inverting inputs of the LM358 chips to see how much noise contributed by the main filter section. I would expect very little noise at output of LM308.

I'm curious also about observations of noise at output of LM308 if the input pins of the LF358 are shorted with very small jumper, compared to shorting the RX coil. In other words, how much noise comes from the path from the RX coil entry point to the PCB to the LF358?

-SB