TGSL Tuning & troubleshooting

[simonbaker]

File attached. There is no schematic, as it was quicker to just write the netlist by hand. Open the GEB.cir file in LTSPICE and run the simulation as normal. Right mouse click in netlist and select "Visible Traces" to see TX, TX_DELAY and SAMPLE1 in the plot pane.

Ok. I'm not good with netlists so it will take some time to make an equivalent schema. I'll try.

This is good. We can figure out our different assumptions I think by comparing Spice models.

Regards,

-SB

[Qiaozhi]

Ok. I'm not good with netlists so it will take some time to make an equivalent schema. I'll try.

This is good. We can figure out our different assumptions I think by comparing Spice models.

Regards,

-SB

You don't need a schematic. Just open the netlist in LTSPICE, and run the simulation as described above.

[simonbaker]

You don't need a schematic. Just open the netlist in LTSPICE, and run the simulation as described above.

Yes, it runs fine. I'm just not sure what it's simulating without seeing the circuit as a diagram (I like pictures ) -- I should be able to whip one up fairly easily, netlist pretty small.


Regards,

-SB

[Qiaozhi]

Yes, it runs fine. I'm just not sure what it's simulating without seeing the circuit as a diagram (I like pictures ) -- I should be able to whip one up fairly easily, netlist pretty small.


Regards,

-SB

The main netlist is as follows, with comments:

C10 TX_DELAY TX 100p ..... C10 100pF
R22 TX_DELAY 0 100k ..... R22 100k GEB trimmer
V1 P8V 0 8 ..... Positive 8V supply
V2 N5V 0 -5 ..... Negative 5V supply
V3 TX 0 SINE(0 6 14.5k) ..... 14.5kHz output from TX osc 12Vpp amplitude
XU1 0 TX_DELAY P8V N5V SAMPLE1 LM393 ..... GEB comparator LM393
RPULLUP SAMPLE1 P8V 3k ..... 3k pull-up resistor on LM393 output
.tran 0 200u 0 200n * Transient analysis from 0 to 200us with 200ns min step

The rest of the stuff is the LM393 subcircuit.
TX is the 14.5kHz sinewave source.
TX_DELAY is the signal at the inverting input of the LM393.
SAMPLE1 is the output of the LM393, which is also the GEB sample pulse.

[simonbaker]

Spice Wars

The main netlist is as follows, with comments:

C10 TX_DELAY TX 100p ..... C10 100pF
R22 TX_DELAY 0 100k ..... R22 100k GEB trimmer
V1 P8V 0 8 ..... Positive 8V supply
V2 N5V 0 -5 ..... Negative 5V supply
V3 TX 0 SINE(0 6 14.5k) ..... 14.5kHz output from TX osc 12Vpp amplitude
XU1 0 TX_DELAY P8V N5V SAMPLE1 LM393 ..... GEB comparator LM393
RPULLUP SAMPLE1 P8V 3k ..... 3k pull-up resistor on LM393 output
.tran 0 200u 0 200n * Transient analysis from 0 to 200us with 200ns min step
The rest of the stuff is the LM393 subcircuit.
TX is the 14.5kHz sinewave source.
TX_DELAY is the signal at the inverting input of the LM393.
SAMPLE1 is the output of the LM393, which is also the GEB sample pulse.

Hi Qiaozhi:

I drew a picture from your netlist and I see you simulated the GB phase shifter.
That's not really relevant to my point about ground balancing, which is that the GB pot position is not influenced by the null signal phase.

To analyze ground balancing you need to simulate the Synchronous Detector -- because the way the SD works is central to my point. I have attached another LTSpice sim that is a "minimal" model of the Synchronous Detector, and you can use it to see that ground balancing the detector is not sensitive to the null signal phase.

I now think maybe we have different assumptions on the physics of coils or the operation of the Synchronous Detector.

Here are my assumptions:

1. The RX coil obeys the superposition principle of electromagnetic fields. The "RX signal" is the sum of the target signal and the null signal.

2. Moving the RX coil slightly relative the TX coil (during nulling) does not affect the phase of the target signal. It does however greatly affect the phase of the null signal.

3. Ground balancing means: adjusting the phase of the sync pulse so that the ferrite target "sweep pulse" is minimal at the output of the Synchronous Detector. The "sweep pulse" is the pulse at the output of the SD when we sweep the coil over the target.


Attached is my LTSpice circuit which is the simplest representation of a Synchronous Detector I could make.

You really have to play with the simulator to show my point. Here is the basic idea: There are three main signals: the target signal (Vtargsweep), the null signal (Vnull), and the sync signal (Vsync). Then there is the output of the Synchronous Detector called Vdetect. Note that the "RX Signal", Vrx, is the output of a summing op amp which represents the RX coil summing the signals.

There is also a final output called Vout, which is basically the same as Vdetect but with the DC bias removed by a capacitor. This gives the flavor of passing Vdetect through the bandpass filters.


I use a "target sweep pulse" to switch the target signal in and out to simulate waving a ferrite target in front of the coil (or pumping the coil up and down over the ground).

---------------
Demonstration:


If you adjust Vsync phase equal to Vtargsweep phase, then you see the target sweep pulse Vdetect of about a .6 mV pp. This case is not ground balanced. You can change the phase of Vnull and it does not change the range of the target pulse -- it only changes the bias voltage of Vdetect.

If you adjust Vsync phase 90 deg off from the Vtargsweep phase (centered on the zero crossing of that component), then the output pulse hardly appears at all at Vdetect. This case is ground balanced. Again, you can change the phase of Vnull and it only changes the DC offset voltage of Vdetect, not the target pulse range.

Conclusion:

Ground balancing means setting the sync pulse (Vsync) phase 90 deg from the ferrite target signal (Vtargsweep) phase. Note that this does not refer to the "RX Signal" (Vrx) phase.

You never see Vtargsweep on your oscilloscope -- you always see Vrx, which is the sum of Vnull and Vtargsweep -- but its there and has a very real phase that is not related to the null phase.


----------
Examples:

It is difficult to show a few pictures and make this clear. Below are four images that show four cases. I can provide additional examples showing the relative phases for the input and sync signals if necessary. But you can use my LTSpice circuit to do it yourself.

Pic1: Vtargsweep phase = 0
Vnull phase = 0
VSync = 0

Notice that target sweep pulse (Vdetect) range is about .6 mV pp with a DC offset near -6 mV. This shows the ferrite signal is coming through strong (unbalanced detector), and the null signal is making a healthy bias on the output of the SD.



Pic2: Vtargsweep phase = 0
Vnull phase = 90
VSync = 0


Notice that target sweep pulse (Vdetect) range is still about .6 mV pp with a DC offset near 0 mV. The ferrite signal has not changed, but the bias due to the null signal has disappeared because the null signal is 90 deg from the sync signal.



Pic3: Vtargsweep phase = 0
Vnull phase = 0
VSync = 90

Notice that target sweep pulse (Vdetect) range drops to a neglible amount with a DC offset near 0 mV. This shows successful ground balancing where Vtargsweep and Vsync are 90 deg out of phase. The null signal is also being "removed" because it is the same phase as Vdetect and does not contribute much DC offset.



Pic4: Vtargsweep phase = 0
Vnull phase = 90
VSync = 90

Notice that target sweep pulse (Vdetect) range is again negligible, but the DC offset is around -6 mV due to the null signal being same phase as the sync signal. This also shows successful ground balancing where Vtargsweep and Vsync are 90 deg out of phase, again showing that the null signal phase does not influence the ground balancing.

This model represents my understanding of the ground balancing and relationship to target and null signal components of the received RX signal.

-SB

[simonbaker]

Continued...

This pic shows the schematic of the minimal Synchronous Detector used in the LTSpice file attached in a previous message (and here).

-SB

[Qiaozhi]

Hi Qiaozhi:

I drew a picture from your netlist and I see you simulated the GB phase shifter.
That's not really relevant to my point about ground balancing, which is that the GB pot position is not influenced by the null signal phase.

Now I can see you've been disagreeing with something completely different to my initial statement.

Of course, the GEB trimmer position is not influenced by the null position. You can see this clearly in my simulation. The GEB sample pulse is generated by delaying the TX signal with a simple RC network and then using a comparator to create the square wave pulse. I have never said this sample pulse is affected by coil nulling. What I did say is that the correct position for coil nulling is dependent on this circuit. In other words, the GEB sample pulse only has an adjustment range of 45 degrees, which tells you that the coils must be nulled with an approximate 20 degree initial phase-shift in order for the GEB trimmer to be positioned in the middle of its range. Also, since the DISC sample pulse has an additional 90 degree phase-shift, it is important to get this nulling correct, otherwise the minimum level of discimination will not occur when the DISC control is turned fully anti-clockwise.

Here are my assumptions:

1. The RX coil obeys the superposition principle of electromagnetic fields. The "RX signal" is the sum of the target signal and the null signal.

2. Moving the RX coil slightly relative the TX coil (during nulling) does not affect the phase of the target signal. It does however greatly affect the phase of the null signal.

3. Ground balancing means: adjusting the phase of the sync pulse so that the ferrite target "sweep pulse" is minimal at the output of the Synchronous Detector. The "sweep pulse" is the pulse at the output of the SD when we sweep the coil over the target.

We already know these to be true, so I'm not sure why you keep repeating this. So far I have only been discussing the coil nulling, as this is the problem area that concerns most coil builders, and is the least understood.

This also shows successful ground balancing where Vtargsweep and Vsync are 90 deg out of phase, again showing that the null signal phase does not influence the ground balancing.

There has never been any question about the null signal phase influencing the ground balancing. As I said in an earlier post, you have completely misinterpreted my original statement. The original discussions were about the correct way to balance the coils. Some people think they should be nulled to obtain the minimum voltage. My point is that this is incorrect, as you will not be able to ground balance the detector because the GEB sample pulse cannot be positioned over the zero-crossing of the residual RX signal. It is more important to have the correct initial phase-shift between the TX and RX coils to allow maximum adjustment with the GEB trimmer and to get the DISC control aligned correctly. Also, just to make this clear, the positioning of the GEB sample pulse over the zero-crossing of the residual RX signal is not the final ground balance position, as real ground does create a slight phase-shift. That's why the final adjustment is done with a ferrite slug or by bobbing the coil up and down outside.

The main reason I created the LTSPICE simulation was to try and break the deadlock. Now I can see you thought that I thought I was saying something completely different to what you thought I thought .... or something like that.

Anyway, the bottom line is that we seem to be in agreement.

So my question to you is now - what is the initial phase-shift of your coils, and what is the residual voltage after nulling?

[simonbaker]

We're getting closer... I think. But again, maybe you think I'm saying something a little different than intended. For example:

Now I can see you've been disagreeing with something completely different to my initial statement.

Of course, the GEB trimmer position is not influenced by the null position. You can see this clearly in my simulation. The GEB sample pulse is generated by delaying the TX signal with a simple RC network and then using a comparator to create the square wave pulse.

My point is not that "the null signal does not affect the sync pulse". We agree on that; the sync pulse phase is a controlled offset from the TX signal only. Clearly shown by your spice model.

I'm saying the the null signal also has no influence on where you position the GB pot when you ground balance the detector. So you can change the null phase all you want and you will always end up with the same pot setting to achieve ground balance. I'm not sure if you agree with that.

So my question to you is now - what is the initial phase-shift of your coils, and what is the residual voltage after nulling?

[/quote]

If by residual voltage you mean the DC value on capacitor C12 or C15, it will depend on the null amplitude and phase. I generally choose a null phase so that the residual voltage stays above -.5 volts over the range of the pots so not to forward-bias the JFets.

So in effect I do something similar to you -- I choose a null to keep the residual voltage in a range greater than -.5 volts. But it is purely a voltage range decision, not based on running out of phase range when ground-balancing the detector. In other words, if I pick a "bad" null phase, I can still theoretically balance the detector phase-wise with the same pot setting as any other null phase, but I will have to forward-bias my JFet to do it, which might work but is bad design and creates possible problems.

Using this criteria, I have freedom to choose a null phase which makes quite positive residual voltages if necessary to accommodate a coil that perhaps has some unusual small coupling that makes the null phase non-standard. Basically, my SD simulation shows that the choice of sync pulse phase for ground balance is determined by the ferrite target signal phase (not the RX signal phase, which is the sum of the ferrite and null signals) and does not depend on the residual voltage either.

I appreciate your effort to grind this out; it may be largely terminology, or not, we're still closing on it. It should help make our various advice and statements seem less conflicting to others in the long run. Or not.

Regards,

-SB

[Qiaozhi]

My point is not that "the null signal does not affect the sync pulse". We agree on that; the sync pulse phase is a controlled offset from the TX signal only. Clearly shown by your spice model.

Good ... so we have that one out of the way.

I'm saying the the null signal also has no influence on where you position the GB pot when you ground balance the detector. So you can change the null phase all you want and you will always end up with the same pot setting to achieve ground balance. I'm not sure if you agree with that.

No, I definitely do not agree with that. And I think your friend Don does not either. Remember this quote?

I learn something every day.. Maybe I have half a chance of making two coils where the GEB trimmer does not have to be readjusted when I swap coils!

Don

If the "null signal also has no influence on where you position the GB pot when you ground balance the detector" then why does Don have to readjust the GEB trimmer when he swaps coils?

If by residual voltage you mean the DC value on capacitor C12 or C15, it will depend on the null amplitude and phase.

No. By residual voltage I mean the voltage across the RX coil once the coils have been balanced. A commercial Tesoro coil will generally have about 15mV residual voltage.

[simonbaker]

Good ... so we have that one out of the way.


No, I definitely do not agree with that. And I think your friend Don does not either. Remember this quote?


If the "null signal also has no influence on where you position the GB pot when you ground balance the detector" then why does Don have to readjust the GEB trimmer when he swaps coils?


No. By residual voltage I mean the voltage across the RX coil once the coils have been balanced. A commercial Tesoro coil will generally have about 15mV residual voltage.

Fantastic, we're on the same page, even though different opinion.

I would ask you and Don to carefully do the test again. I'm not sure Don actually did the test, rather just took your method for useful advice. Which I think it is, by the way. Nulling all your coils the same can't hurt.

And by coincidence, your technique tends to set the null phase very close to the ferrite target signal phase, which works out pretty well for the JFet biasing.

There are reasons you might need to ground balance different coils differently for reasons unrelated to the null phase. For example if the offset between RX resonance and TX frequency is different, you can get a different target signal phase for the same target.


So if you and Don (or anyone) decides to do the test, here is what it is:

1. Null your coils, just making sure the voltage on the C12, C15 capacitors are not below -.5 volts over the GB and DISC pot ranges. Pick any null signal phase that meets those requirements.

2. With DISC pot a minimum, wave a ferrite target at about 8 cm and turn the GB pot until audio signal just disappears. To be really accurate, look at the pulse at the output of the LM308. When the pulse goes flat, you have ground balance. Turn the pot more and the pulse will appear again (opposite polarity). Try to find that sweet spot where minimum.

Make note of pot position.

3. With GB pot at minimum, use a US Nickel and do same procedure with DISC pot and other LM308 output.

Make note of pot position.

4. Now choose another null phase, but still obey requirements of step 1. Repeat steps 2 and 3. The pot positions should end up virtually the same as in steps 2 and 3 according the results of my spice simulation and my assumptions. But always glad to acquire new assumptions.

Regards,

-SB

[dfbowers]

Fantastic, we're on the same page, even though different opinion.

I would ask you and Don to carefully do the test again. I'm not sure Don actually did the test, rather just took your method for useful advice. Which I think it is, by the way. Nulling all your coils the same can't hurt.

And by coincidence, your technique tends to set the null phase very close to the ferrite target signal phase, which works out pretty well for the JFet biasing.

There are reasons you might need to ground balance different coils differently for reasons unrelated to the null phase. For example if the offset between RX resonance and TX frequency is different, you can get a different target signal phase for the same target.


So if you and Don (or anyone) decides to do the test, here is what it is:

1. Null your coils, just making sure the voltage on the C12, C15 capacitors are not below -.5 volts over the GB and DISC pot ranges. Pick any null signal phase that meets those requirements.

2. With DISC pot a minimum, wave a ferrite target at about 8 cm and turn the GB pot until audio signal just disappears. To be really accurate, look at the pulse at the output of the LM308. When the pulse goes flat, you have ground balance. Turn the pot more and the pulse will appear again (opposite polarity). Try to find that sweet spot where minimum.

Make note of pot position.

3. With GB pot at minimum, use a US Nickel and do same procedure with DISC pot and other LM308 output.

Make note of pot position.

4. Now choose another null phase, but still obey requirements of step 1. Repeat steps 2 and 3. The pot positions should end up virtually the same as in steps 2 and 3 according the results of my spice simulation and my assumptions. But always glad to acquire new assumptions.

Regards,

-SB

Hey guys.. I'm out this week, so I have yet to verify what I said about readjusting the GEB trimmer when swapping coils. It may have more to do with slight differences in inductances but not sure yet.. I'm at the beach actually using the TGSL this week!!.. Nice debate though!

Which one of us is going to write the tuning & setup manual when we figure this all out?

Don

[simonbaker]

Hey guys.. I'm out this week, so I have yet to verify what I said about readjusting the GEB trimmer when swapping coils. It may have more to do with slight differences in inductances but not sure yet.. I'm at the beach actually using the TGSL this week!!.. Nice debate though!

Which one of us is going to write the tuning & setup manual when we figure this all out?

Don

The one enjoying the beach...

-SB

[Qiaozhi]

Hi Simon,

The ground balancing procedure you are using is incorrect. Let me refer you to the Tesoro Bandido User Manual:

http://www.tesoro.com/info/manuals/older/bandidoiiumax/

I have purposely referred you to the Bandido manual, and not the Golden Sabre, as the TGS does not have an external GEB control, since the user is not expected to adjust this setting. It is preset at the factory. Note that the Quick Start Mode requires you to set the Ground Balance in the "middle" and the Discrimination at "minimum".

Also, read this ->
NOTE: The GROUND control does affect both operating modes but should be adjusted in All Metal Mode.

[simonbaker]

Hi Simon,

The ground balancing procedure you are using is incorrect. Let me refer you to the Tesoro Bandido User Manual:

http://www.tesoro.com/info/manuals/older/bandidoiiumax/

I have purposely referred you to the Bandido manual, and not the Golden Sabre, as the TGS does not have an external GEB control, since the user is not expected to adjust this setting. It is preset at the factory. Note that the Quick Start Mode requires you to set the Ground Balance in the "middle" and the Discrimination at "minimum".

Also, read this ->
NOTE: The GROUND control does affect both operating modes but should be adjusted in All Metal Mode.

Hi Qiaozhi:

I did a quick read and didn't find anything that struck me as inconsistent with my test procedure or my model for how ground balancing works.

Pumping up and down is analogous to waving a ferrite target. Finding the minimum audio change is analogous to seeing the output of the LM308 go flat.

They recommend All Metal Mode for dorks who leave their Disc control pot at max and block the signal ; in other words, to take the disc channel out of the equation. In my test procedure, if you look at the output of the LM308 for the channel of interest, the pot of the other channel is irrelevant.

Also, notice the two highlighted setup instructions:

• Threshold Control at 12:00.
• DISC LEVEL control set to MIN.
• SENSITIVITY control set to POW OFF.
• TUNE switch set to AUTO.
• MODE switch set to ALL METAL.
• GROUND control set to the middle of its range:

For the TGSL, if you are in ALL METAL mode, it doesn't matter what DISC LEVEL you set -- it is ignored. Do they know what they are doing?

Also, I certainly would not extend my statements to other detector designs -- I only am looking in particular at the TGSL front end. Your experience with other detectors may include some with a whole different type of processing that would have a different sensitivity to null signal phase.

I still don't believe the discrimination calibration or GB calibration (clearer terminology I hope) are sensitive to null phase, but if it really is so, then it will be interesting to incorporate that new information into my LTSpice model to see why so and how it works.

Regards,

-SB

[Qiaozhi]

Hi Qiaozhi:

I did a quick read and didn't find anything that struck me as inconsistent with my test procedure or my model for how ground balancing works.

Pumping up and down is analogous to waving a ferrite target. Finding the minimum audio change is analogous to seeing the output of the LM308 go flat.

They recommend All Metal Mode for dorks who leave their Disc control pot at max and block the signal ; in other words, to take the disc channel out of the equation. In my test procedure, if you look at the output of the LM308 for the channel of interest, the pot of the other channel is irrelevant.

Also, notice the two highlighted setup instructions:

• Threshold Control at 12:00.
• DISC LEVEL control set to MIN.
• SENSITIVITY control set to POW OFF.
• TUNE switch set to AUTO.
• MODE switch set to ALL METAL.
• GROUND control set to the middle of its range:

For the TGSL, if you are in ALL METAL mode, it doesn't matter what DISC LEVEL you set -- it is ignored. Do they know what they are doing?

Also, I certainly would not extend my statements to other detector designs -- I only am looking in particular at the TGSL front end. Your experience with other detectors may include some with a whole different type of processing that would have a different sensitivity to null signal phase.

I still don't believe the discrimination calibration or GB calibration (clearer terminology I hope) are sensitive to null phase, but if it really is so, then it will be interesting to incorporate that new information into my LTSpice model to see why so and how it works.

Regards,

-SB

The All Metal mode effectively disables the DISC channel. Therefore you must set up the ground balance for the ALL Metal Mode. It is possible that you are finding an alternative position for the pots where ferrite is being excluded, because both GEB and DISC channels are active during your tests. However, this would not be the correct ground balance position, and the ALL Metal mode would fail to exclude the ground effect, or even reject the ferrite.

Pumping up and down is analogous to waving a ferrite target. Finding the minimum audio change is analogous to seeing the output of the LM308 go flat.

To be more precise you should be monitoring the DC voltage on the output of the GEB sync demod (source of TR5).

Previously I asked:

So my question to you is now - what is the initial phase-shift of your coils, and what is the residual voltage after nulling?

I would still like to know the answer.