Nulling / Ground Balance

[Jerry]

Good points Simon. I have not been able to get back to it so far today but maybe this evening I can.

For the pictures shown, Disc is at minimum (Full CCW) and the GB pot is about mid rotation, just at the point where ferrite is rejected. I Jumpered J2 with a 12 inch clip lead but a direct short is no problem.

One thing I was going to do is show a known frequency sine wave on the GB channel to compare the approximate frequency of the noise signal. Will do it both ways.

I think these pictures are telling us that somethings is getting in from other parts of the TGSL circuit, and/or the filters are ringing. It is periodic and not random. Probable suspect is the 4024.

Will report more when I have something.

Jerry

[simonbaker]

Good points Simon. I have not been able to get back to it so far today but maybe this evening I can.

For the pictures shown, Disc is at minimum (Full CCW) and the GB pot is about mid rotation, just at the point where ferrite is rejected. I Jumpered J2 with a 12 inch clip lead but a direct short is no problem.

One thing I was going to do is show a known frequency sine wave on the GB channel to compare the approximate frequency of the noise signal. Will do it both ways.

I think these pictures are telling us that somethings is getting in from other parts of the TGSL circuit, and/or the filters are ringing. It is periodic and not random. Probable suspect is the 4024.

Will report more when I have something.

Jerry

Hi Jerry:

Yes, I'm curious about the periodic nature. I find often the noise is periodic for a while, then gets more random, then back to periodic, etc. The filter section will enhance noise at its center frequency and make it look more periodic - like you say, ringing. But I don't think I saw much periodic noise when I shorted the RX coil.

All the TGSL signals (including 4024) are driven by the oscillator and should be stationary with respect to the SD and should not create any periodic demodulated signals. And its hard to believe the 4024 makes a low enough signal for what we're seeing. Well, 14.5 kHz / 2^7 (7 bit counter) gives 113 Hz, so maybe possible.

I'll be interested in how the waveforms look with very short jumpers and other experiments I mentioned.

Regards,

-SB

P.S. I'm thinking how nice USB oscilloscopes would be for capturing this kind of data.

[Jerry]

Well here is what I got this evening. The camera drove me crazy for a while trying to set the shutter speed, but I finally got it.

The scope was set for .1 sec/div and 10 mv/div for both channels. I have no idea on how accurate those numbers are, so I measured the Disc channel with my DVM and it varied from .2 to 1.0 mv. The scope will be showing peaks the the DVM will not respond to but it is not a large signal.

The camera shutter was set at 2 sec.

The second picture shows an alligator clip jumpering J2.

The exposure starts towards the right of the screen and wraps around. I was expecting something a little different from what I had yesterday because of the longer exposure. I didnt have time to put the signal generator on channel two for reference but now I am not sure what that would prove.

I thought I had something yesterday, now I am not sure.

Jerry

[simonbaker]

Well here is what I got this evening. The camera drove me crazy for a while trying to set the shutter speed, but I finally got it.

The scope was set for .1 sec/div and 10 mv/div for both channels. I have no idea on how accurate those numbers are, so I measured the Disc channel with my DVM and it varied from .2 to 1.0 mv. The scope will be showing peaks the the DVM will not respond to but it is not a large signal.

The camera shutter was set at 2 sec.

The second picture shows an alligator clip jumpering J2.

The exposure starts towards the right of the screen and wraps around. I was expecting something a little different from what I had yesterday because of the longer exposure. I didnt have time to put the signal generator on channel two for reference but now I am not sure what that would prove.

I thought I had something yesterday, now I am not sure.

Jerry

Hi Jerry:

Thanks for keeping at it. That looks more familiar to me as noise with no coil. Hang on to it, it is a good reference for comparing our basic "PCB noise" levels. Unscientifically I would call that a "15 mV pp" range noise.

The two signals seem quite well correlated if you look closely. I'll have to think about what that means. It depends somewhat on the DISC and GB pot settings. It might be interesting to increase your sweep speed one notch and then play with the pots and observe how the correlation changes.

I was thinking - a good way to test for noise from the 4024 chip would be to trigger your scope on one of the 4024 output pins, for example Q7 (the lowest frequency). Then any "noise" signal from it should be very locked and stationary, no drift left or right.

Final thought -- on your previous ferrite tests (with the whirly-bird), I saw one trace that seemed the inversion of the other. I'm thinking now that perhaps I would expect to see both traces the inversion if the null phase is flipped from one side to the other. However, this may depend on the exact phase of the nulls you picked and the sync pulse phases, so would need more data to fully analyze it.

Your pics are very good, that was a success.

Regards,

-SB

P.S. I generally don't believe multimeter AC readings much because they usually have very limited bandwith for measuring -- at least mine do.

[Jerry]

Hi Simon:

The biggest thing to get used to with that last picture is we are looking at a sub-audible noise. The entire length of the trace shown is two seconds. Because of the digital camera delay in starting the exposure, it is really hard to judge when to press the trigger button. This particular one starts one the right and wrapps around.

Will keep experimenting.

Jerry

[simonbaker]

Hi Simon:

The biggest thing to get used to with that last picture is we are looking at a sub-audible noise. The entire length of the trace shown is two seconds. Because of the digital camera delay in starting the exposure, it is really hard to judge when to press the trigger button. This particular one starts one the right and wrapps around.

Will keep experimenting.

Jerry

Yes, really low freq.

The pics are nice, wrap around is fine if there isn't too big a gap during flyback.

I think one notch faster sweep would be good for some tests, like the ferrite-go-round.

I realize it is quite a lot of effort to document these tests, so appreciate whatever you can do, and if reasonably entertaining for you.

Would video work and be easier?

Regards,

-SB

I like your Al box, looks ready for Afganistan...

[Jerry]

Hi Simon:

Here is the next installment I am starting to get the hang of capturing these traces. What works real good is a darkened room, camera tripod and a programable shutter speed. I sped up the sweep just a bit to show the sine wave and used a 1.6 second shutter speed.

Picture #1 shows a 16 hz sine wave with the GB channel on top.
Picture #2 is the noise on both channels with the input jumper removed.
Picture #3 is a 2 Euro + Nickel at one end the stick and a nickel on the other, turning at 28 cm. The GB channel is still on top so the coin signal shows up on bottom trace. I switched the scope so GB will appear on the bottom for the rest of the pictures.
Picture #4 is with the ferrite stick turning at 14 cm. GB is 50% rotation and still getting beeps.
Picture #5 is the same but with GB fully CCW.
Picture #6 shows the ferrite tuned out so speaker is silent.

The signal voltage on the ferrite at 14 cm was really strong so I turned the scope back to .5 volts/division.

I hope this is getting closer that what you were looking for.

Jerry

[simonbaker]

Hi Simon:

Here is the next installment I am starting to get the hang of capturing these traces. What works real good is a darkened room, camera tripod and a programable shutter speed. I sped up the sweep just a bit to show the sine wave and used a 1.6 second shutter speed.

Picture #1 shows a 16 hz sine wave with the GB channel on top.
Picture #2 is the noise on both channels with the input jumper removed.
Picture #3 is a 2 Euro + Nickel at one end the stick and a nickel on the other, turning at 28 cm. The GB channel is still on top so the coin signal shows up on bottom trace. I switched the scope so GB will appear on the bottom for the rest of the pictures.
Picture #4 is with the ferrite stick turning at 14 cm. GB is 50% rotation and still getting beeps.
Picture #5 is the same but with GB fully CCW.
Picture #6 shows the ferrite tuned out so speaker is silent.

The signal voltage on the ferrite at 14 cm was really strong so I turned the scope back to .5 volts/division.

I hope this is getting closer that what you were looking for.

Jerry

Very nice photos, you've got it down. I would be interested in the vertical scale in each if you remember.

Is the first photo GB with RX coil shorted?

It looks like maybe double the sweep speed would show your rotating target signals a little better, but advantages to both speeds. The more data the better.

It looks like in photo 4 if you turned the GB pot just a hair more it should quiet the ferrite. But ferrite is not always consistent.

Photo 5 seems to show how the null signal is "in phase" for one sync pulse and "anti-phase" for the other -- roughly.

Given your acquired skills, here is what I'm curious to see:

A) Using un-potted coils, null them to one side of mininum, so fairly stable phase.

With GB and DISC pots at minimum (CCW?),

A.1. Take a photo showing null signal (LF353 pin 7) as lower trace, TX coil ungrounded lead as upper trace. Adjust the vertical resolutions so both signals comparable magnitude. Position them to make it fairly easy to see phase offset.

A.2 Then do the ferrite-go-round test showing the outputs of LM308 chips as you did relative to eachother. Maybe twice the sweep speed you used. Note the vertical scale.

B) null the coils to the other side of the minimum, about same signal magnitude as before.

Take same pics as A.1 and A.2.

I can't say this is earth-shattering data, so don't feel pressure to do it. There are so many interesting signals to take and you already got some valuable noise data.

However, it should help me confirm my understanding of the effect of ferrite and get some ideas about how our choice of null might affect practical use in the field.

Depending on how those photos look, we might want to repeat the test with GB in middle and max positions.

Wish I could help more! I'll keep trying to find ways to experiment in noisy environment, or reduce the noise.

Regards,

-SB

[Jerry]

Hi Simon:

Here are the results from the latest test:

#1. Inside of null with .7 VDC at U101 pin 7.
#2. Ferrite turning at 14 cm. Vertical deflection is .5 volt/div.
#3. Outside of null with .74 VDC at U101 pin 7.
#4. Ferrite turning as in #2.

One thing that stood out on this test is how fussy it was to get a specific null point and steady voltage compared to measureing rx at J2-1. This makes sense because the amp has a gain of about 47 which makes it like trying to hand hold a 100 power telescope. Very hard to hold it still.

I did repeat the test by using U101 input I can post it if you want.

Jerry

I re-checked my notes and I have the ferrite pictures in the wrong order. #1 and #4 should be paired and #2 and #3 should be paired. I tried to save some time and took the pictures out of order. I sure wish the camera would show me a number to write down. This agrees with the second test I made in correct order. Sorry about that.

[Jerry]

Looking that last post over again and it does not make sense. I will post the second test where I am more sure of what is going on.

I set the Null voltage at 12.1 MV on both the inside and outside null points

The order is as before:
1. Null inside
2. Ferrite turning
3. Null Outside
4. Ferrite turning.

The null pictures look much different.

Jerry

[simonbaker]

Looking that last post over again and it does not make sense. I will post the second test where I am more sure of what is going on.

I set the Null voltage at 12.1 MV on both the inside and outside null points

The order is as before:
1. Null inside
2. Ferrite turning
3. Null Outside
4. Ferrite turning.

The null pictures look much different.

Jerry

These are outstanding tests, thanks!

In your first post, it looks like maybe you didn't get different sides of the null point, not sure.

Second set looks more understandable.

When using output of LF353, you probably weren't shoving the null past as far because signal looked so large. If you reduce gain of scope, maybe would feel the same.

Is the top trace in the ferrite test the GB channel?

Are both GB and DISC pots fully CCW?

Anyway, it fooled me again. Both nulls show the GB and DISC channel response moving in the same direction. I will have to draw myself pictures to understand why.

I guess the difference between these two null points requires a dynamic test moving the pots, then we should spot different behavior from the two null points at some settings.

But that is getting too complicated and frankly I'm not sure what I would make of the results. But it seems some people say one null is preferred over the other and I'd like to collect data that shows why.

You have a great setup for testing. Perhaps you could just do some quick tests where you turn the pots and see if you notice any significant difference in the ferrite (or Euro) responses comparing the two null choices. Personally I'm looking for cases where the GB and DISC channels coordinate their motions differently depending on which null is chosen.

Thanks for your hard work!

Regards,

-SB

[Jerry]

Hi Simon:

The bottom trace is the GB channel and Disc and GB controls are fully CCW (Minimum)

I was looking back at two of the pictures I took yesterday and think your idea for a dynamic test will be intesting. I have a ten turn, 100k pot with a three digit scale to use for a GB control. Should be able to get really good precision and repeatability for that transition between fully CCW GB and the point where ferrite is silenced completely.

I think I will start at the point where ferrite is completely silences and work towards the ccw position.

We may take the camper out again this weekend so will probably be next week before I get to it.

Jerry

[Qiaozhi]

But it seems some people say one null is preferred over the other and I'd like to collect data that shows why.

There are two good reasons why you should select the inner null point:

1. The crossover area between the coils is larger and therefore provides a greater detection area.

2. This position provides a phase-shift on the RX signal, relative to the TX, that is most compatible with the Tesoro design.

[Jerry]

Two very good points. I have been using the inside of null setting. One thing I have found is there seems to be a fairly broad range of phase shifts that work well on my detector. Twenty degrees works very well but there is 10-12 mv residual null voltage. Forty and Fifty degrees seem to work about as well and the residual voltages are much less.

When I made that phase vs null chart a few weeks ago, I was looking to see if I could find a "sweet Spot" to set the coil overlap. So far that is not determined. I easily get 34-35 cm air test on the 20 cm coil I am using within that range of phase shifts.

I am having a lot of fun testing while I am waiting to get my vacume forming setup finished.

Jerry

[simonbaker]

There are two good reasons why you should select the inner null point:

1. The crossover area between the coils is larger and therefore provides a greater detection area.

I like that reason too.

I used to worry about bias on SD caps C12, C15 that might forward bias the JFets and ruin their operation. But seems typical null voltages don't stray there... something to watch though.

2. This position provides a phase-shift on the RX signal, relative to the TX, that is most compatible with the Tesoro design.

That is where the question lies...

Our discussion of null phase and ground balance seems to give some rationale for choice, but now I want to know how much we can control the "magnetically-coupled" null phase component.

If most of the phase latitude is available because of a large capacitive coupled component, then practically we may have only two stable phase choices, one on either side of the minimum... unless you want to tinker with conductive targets taped to your coil to shift the null phase where you want it, which is not out of the question.

-SB



-SB