Digital TGSL

[Qiaozhi]

Hi Max
Do you have Block diagram of TGSL?
I want to make it with only one demodulator that phase angle of square wave signal changes by a micro(phase scanner).
in my initial tests i saw some metals change only phase angel and others change amplitude!
i was confused. i'm searching what? phase shift or amplitude change?
why the resonat frequency of TX and RX must be different?
if i have Block Diagram, i can find answers.

Hi,
no... I haven't, sorry. But it's quite easy identify various parts... you have op amps used as phase shifters that control the gate of fets... fets are the sampling devices in TGS. That's about same stuff for disc and gnd parts... you have 2 phase shifter networks and 2 fets for that (TGSL).

After the fets there are the low pass filters... so are other blocks.

All tesoro coils for uMax shift to the left (I saw that on 9x8 and 8 round but think they are all the same about this) when a metal is near, no matter of kind of metal, are eddy currents that make it happens. The non-ferrous give rise in amplitude... and iron gives a fall in amplitude at rx, this is related to flux lines behaviour.

It's a bit complex issue: on geb channel you get amplitude variations that are driven actually by phase shift introduced by targets.
In the disc channel you get, instead, amplitude variations related to amplitude variations at rx coil.

It's so cause geb channel don't care about target composition but only that mineral response need to be ignored.

But you'll better ask Qiaozhi about subtle details, he knows about these designs strategies very, very good.

Kind regards,
Max

Can you explain in more detail what you are trying to achieve here?
Why do you only want to use one demodulator?


Your description is basically correct, but I'm not sure what wild_desert is trying to do. It doesn't make any sense at the moment. We need to wait for an answer.

Hi,
I think he wanna use one channel "demodulator" ... some sampling stuff connected to something atmel MCU.

But the problem is that usually there are two channels... one for geb and the other for disc path.

In case of amplitude stuff he'll use ADC; for phase issue instead he would use (probably) a comparator only.

But we know the problem is somehow a bit more complex...

Kind regards,
Max

Thanks to Max and Qiaozhi
Ex me for my delay(I have to make circuits for living and MD is my research).
You said: we must have 2 demodulator simultaneously. one for detecting phase change and other for amplitude change. And an or logic for detecting a target.
Is it true?
If true i say: micro can make precisely 2 squre wave needed for 2 demodulator and 2 low pass filter can make smooth dc voltage. next two 10bit ADC of AVR will sample dc voltages and micro can compare it with presaved numbers. any change in any channel means metal vicinity. Then micro can change phase clocks for discrimination purpose.
You see, we may have smaller circuit with more flexibility. Instead of potentiometers we have keyboard. Also auto GEB is a short procedure in our program. we can change many parameters by 4 keys + a basic LCD + menus.
I wait your answer
Kind regards

Hi,
it's like in digital VLFs: usually the phase of sampling gates is controlledby the MCU , often another MCU or the same do also the reading of values after adc stuff and make other computations, like for VID stuff, related to "phase" of targets.

If you try to generate the phase for sampling using the MCU you'll sure have easy work... just need to sync using an interrupt driven routine and use internal timer(s) to generate an opportune delay(s).

The adc reading, instead, seems much harder work if you'll use just simple sampling strategy... require hi-priced device to do so, the internal adc will fail on that if there aren't analog blocks before it, e.g. like just after sampling gates.

You could use another strategy: numerical integration.

Look for Robert Hoolko's project , it's a thread inside this forum: he made something similar and works, but then abandoned the project.

Kind regards,
Max

Hi Max
I don't want to do sampling on sine wave, although that is better but the ADC become expensive and busy the micro.

Let me show a block diagram:

Don't worry Ivconic, I like TGSL topics and want to be here
But Max and Ivconic
The 1024 discrete levels is better than offset errors of LM393. Is not true?

Now "magnetometer". I guess the wire path of OSC to Coil and ground effect on magnetic field can make some phase error. if we have magnetic field directly, we can overcome this errors. Is not true?

I think you are going to completely saturate you magnetometer with such strong fields so close.
Ivconic :
, regards,
Fred.

OK - I've copied everything over to this new thread.
Wild_Desert - please post here on this subject.

[wild_desert]

Thanks to Friends

Hi all friends
I understand your problems. I caught myself when searching here. But my experiences in MDs is not enough to decide to make a new topic!
I apologize you and thanks for your attentions.
I want to start below diagram for making schematic and wait your comments:

[Max]

Hi all friends
I understand your problems. I caught myself when searching here. But my experiences in MDs is not enough to decide to make a new topic!
I apologize you and thanks for your attentions.
I want to start below diagram for making schematic and wait your comments:

You don't need any magnetometer there.

Kind regards,
Max

[wild_desert]

Hi Max
My idea is: The magnetometer reads actual magnetic field. Is it wrong? why?

[Max]

Hi Max
My idea is: The magnetometer reads actual magnetic field. Is it wrong? why?

Hi,
the magnetometers are good to find ferromagnetic stuff... flux lines concentrate inside the ferromagnetic "core" and you'll see an increased reading when over the target.

In MDs you usually look for non-ferrous stuff... valuable items... so why bothering about ferromagnetic materials ?? Iron ??

That's why you don't need it I think.

Kind regards,
Max

[wild_desert]

Thanks Max
Ok
I edited the block diagram. no other comment?

[Max]

Thanks Max
Ok
I edited the block diagram. no other comment?

Hi,
why using ADC just after low pass ?
Isn't better using an op amp in the middle ?

Kind regards,
Max

[Qiaozhi]

Hi,
why using ADC just after low pass ?
Isn't better using an op amp in the middle ?

Kind regards,
Max

I was just studying the new diagram and thinking the same thing. The signals after the LPFs have gone through a process of differentiation. You should try putting the RX signal from the pre-amp through a schmitt trigger, as you've done with the TX.

[wild_desert]

Oh, LPF is integrator not differentiator and TX is phase reference! Is it wrong?
We can use Active Filters to achive gain and filtering simultaneously.
Like below: Order=2, Gain=-100, Cutoff=33Hz

[Max]

Oh, LPF is integrator not differentiator and TX is phase reference! Is it wrong?
We can use Active Filters to achive gain and filtering simultaneously.
Like below: Order=2, Gain=-100, Cutoff=33Hz

Hi,
yes, you can. Usually in tesoro's designs the first is a passive low-pass filter... then others are, actually, active filters.

Kind regards,
Max

[Sean_Goddard]

Why not have a DAC to set a schmitt trigger threshold (like the "sensitivity" control on the Analogue GS)?

I think trying to interrupt the Atmel at 14KHz AND process the phase info is going to be very hard (not fast enough). What you CAN do is generate an interrupt when the schmitt trigger fires (target found) and process the phase info then (much easier as you are using the timers to generate the Tx signal).

You can bypass this schmitt when setting up ground phase measurements or add a slowly changing channel (FpassMax of around 0.5Hz) and use this inphase with Tx to monitor the ground phase signal (you can then do auto ground tracking and balancing in real time).

Your route is the way Tesoro went (doing signal detection in the digital domain) and as we've seen, it didn't work for them. The Euro Sabre sucked HARD, but the old SS uMax ran it ragger (all analog design). This hybrid methodology works very well for Whites BTW.

What about switched gain stages and sample and holds to stop signal droop on the LPF outputs?

Look for the Eagle Spectrum diagram (somewhere on this forum). That's a machine that got this type of design PERFECT.

Just a thought .

[wild_desert]

Why not have a DAC to set a schmitt trigger threshold (like the "sensitivity" control on the Analogue GS)?

I think trying to interrupt the Atmel at 14KHz AND process the phase info is going to be very hard (not fast enough). What you CAN do is generate an interrupt when the schmitt trigger fires (target found) and process the phase info then (much easier as you are using the timers to generate the Tx signal).

You can bypass this schmitt when setting up ground phase measurements or add a slowly changing channel (FpassMax of around 0.5Hz) and use this inphase with Tx to monitor the ground phase signal (you can then do auto ground tracking and balancing in real time).

Your route is the way Tesoro went (doing signal detection in the digital domain) and as we've seen, it didn't work for them. The Euro Sabre sucked HARD, but the old SS uMax ran it ragger (all analog design). This hybrid methodology works very well for Whites BTW.

What about switched gain stages and sample and holds to stop signal droop on the LPF outputs?

Look for the Eagle Spectrum diagram (somewhere on this forum). That's a machine that got this type of design PERFECT.

Just a thought .

Thanks a lot Sean_Goddard
I'm trying to remove analog parts as much as possible. A good program plus correct use of internal hardware of AVR. Also maintaining constant phase errors in the program is the key of precision and 16Mhz AVR may be fast enough.
Your suggested auto GB seems to be analog but we can save ADC values in registers and wait for changes. any slow change may be omitted in program and only uptate GB registers, instead of target alarm. this is my idea for auto GB but i need further remarks from friends.

If analog MDs have better history we can change it by effort.

Programmable Gain Amplifier(PGA) was used in diagram may be switched resistance. LPF outputs is DC and there is no drop on it. am i wrong?

I saw Eagle Spectrum after your post. It is new for me. Thanks. I want to study it in detail.

[KingJL]

I saw Eagle Spectrum after your post. It is new for me. Thanks. I want to study it in detail.

The schematics for the Eagle Spectrum are on the forum. Also you may want to look at U.S. patent 4868910. This is the patent for the Eagle (I think Eagle II) and provides detailed discussion of the circuitry and programs. Much of the hardware and much of the software are unchanged in the Eagle Spectrum, and XLT. Functionally everything is the same. The reading of the White's patent for their VDI (US 6879161) brings in the missing pieces for the Spectrum Eagle.

Like Sean said, Whites Electronics got this one right!

[aft_72005]

Hi wild_desert
Where are you? Are you have activates on your project?
Beat regards.

[wild_desert]

Hi aft
I'm here. but searching better methods. i think we have lack of block diagrams
regards