ARMD (ARMRADIO based Metal Detector) VLF IB PROJECT

[ArchibaldSTM]

If you need a sine, you can do it as in Stalker-IB

[moodz]

Explain then why Carl burned his set at the first attempt?


TC4428 I can not obtain, as I mentioned several times, nor the other lookalikes.
Besides I like to try with "simple" components. that can be obtained anywhere.
Is there a problem with the schematic?
That's why I posted it; to hear other opinions and remarks.
But nothing like ".... use this or that driver chip..."

OK you wanted to know whats wrong with your schematic ( AKA CCPI )

So I redrew the ARMD TX configuration so you can understand it .

There is "shoot thru" if we were designing a mains switcher this would be a major problem but at 2 volts DC ... meh !!
The shoot through on my circuit is about 75 - 90 nanoseconds on every transition ... but since the transitions are every 60 to 120 microseconds I did not worry about them.

So your schematic does nothing really to fix the shoot through .. playing with the transition timings is messy and could degrade circuit performance.

The solution ( a solution ) that I like ... is to put a current snubber ( not voltage snubber ) in line with the DC power supply. This prevents the shoot through from occurring.

I just used some random values for the snubber ... you can calculate an optimal value I will leave it as a challenge for you.
The pics below show the improvementd D3 R2 L8 below. The red trace is a 20 amp ST. The green trace is the ST eliminated.

[ivconic]

Thank you!
I didn't express myself clearly, I think.
When I took to consider such a TX; I didn't mean to just follow the development of one idea, for example yours.
TX itself is already interesting to me for further experimentation.
As such, for tests with VLF-I/B and not only on PI.
This implies different supply voltages, different configurations of the rest of the circuit, etc.
Hence the desire to do everything independent of the MCU, analog, with discrete circuits.
Maybe I will power such a TX with 5v, 10v, 12v. Maybe I will generate pulses with NE555, maybe with MCU, maybe with some cmos oscillator.
It's all in the game.
The idea is to make it a universal circuit and that it can work equally well in a wider frequency range.
I don't know how it will go further, I'm not thinking about it, right now I'm just thinking about how to make TX resilient and immune to all possible contingencies.
Long ago, many years ago, I noticed that a similar approach is good. First time with the Fisher CZ5.
There is a half bridge. Square signal. In practice, the CZ5 has proven to have very good discrimination against deep unwanted targets.
I noticed that at much greater depths it clearly discriminates the famous "horse shoe", which other detectors could not.
I am convinced that such quality of work first starts with a precise and robust TX stage.
ArchibaldSTM has already proposed a solution as on the Quasar ARM detector. Pretty similar to the CZ5.
Moodz, you think rather narrowly, within the framework of your idea that you are following.
Okay, that's fine. But that's not what I'm interested in either.
Except the similarity in the TX stage; everything else that interests me will have no points of contact with what you propose here.
Snubber... ha! And when I suggested that on the AMX thread; there were no reactions. No one paid attention to it.
I know very well what a snubber is and how to apply it.
I have been using a similar method on a project that has been commercial for 6-7 years.
Motor control with full bridge configuration and currents of 14-30A.
And the whole project was done in a similar way, which I am still following.
Discreet electronics, nothing left to chance, everything fully secured.
There may be more components, but the point is that it works safely and without excess.
Unlike the joke and play that goes on here; I do things that apply in the serious world where people's lives depend on such electronics.
Specifically, it is about winch control on forestry tractors that pull up to 8 tons of logs at once.
The slightest glitch and a worker in the forest is cut in half in a split second.
There is no joke and no white smoke in that world.
I respect your ideas, I'm waiting to see what you will make of them realistically and I'm rooting for you!

[moodz]

Well I am glad you are finding it useful for more than metal detectors.

So we have fixed the ShootThru. The next fix to add is the issue of current burning up the circuit if there are no TX clocks on the Mosfet gates.
I have added a fix for that also. ( C8 D2 R5 C2 D1 R4 ) turn off the ground mosfets if there is no clock.

See below. Any issues let me know.

[ivconic]

The full bridge configuration I used so far is working quite alright for the purpose it is aimed.
But this is interesting enough to try it too.
I like it beacuse of the wider range of supply I can use.
...
Now you talking!
Will check it of course.

[moodz]

So the ARMD TX format is now tested and mainly in its final form ready for part selection and PCB layout.

We will now address the RX section. I always like PI detectors because of the TX power. Not so the VLF detectors .. they have wimpy milliamp class transmit signals and I think it is because they are CW continuous wave transmitters and the RXs cannot be highly gained due to the targets causing IB unbalance and overloading the RX amplifier.

For the first iteration of ARMD I am going to use an RX front end that will do the following.

1. Deal with a high dynamic range input signal ( eg 1000 : 1 ) or better.
2. Recover Accurate phase information.
3. Recover Accurate amplitude information.
4. Very low power.
5. Handles noisy inputs.

and

6. Made from parts that even Ivica can find.

Here it is below ... as tested.

Its a synchronous oscillator that locks on signals at better than 1000:1 input range and recovers phase and a log amplitude response ( ie does not overload ).
The traces below tell the story .... with no input you get low amplitude random phases at the output.
With valid input you get lock and log amplitude output from 1 millivolt to 1 volt.

[profesor]

So the ARMD TX format is now tested and mainly in its final form ready for part selection and PCB layout.

We will now address the RX section. I always like PI detectors because of the TX power. Not so the VLF detectors .. they have wimpy milliamp class transmit signals and I think it is because they are CW continuous wave transmitters and the RXs cannot be highly gained due to the targets causing IB unbalance and overloading the RX amplifier.

For the first iteration of ARMD I am going to use an RX front end that will do the following.

1. Deal with a high dynamic range input signal ( eg 1000 : 1 ) or better.
2. Recover Accurate phase information.
3. Recover Accurate amplitude information.
4. Very low power.
5. Handles noisy inputs.

and

6. Made from parts that even Ivica can find.

Here it is below ... as tested.

Its a synchronous oscillator that locks on signals at better than 1000:1 input range and recovers phase and a log amplitude response ( ie does not overload ).
The traces below tell the story .... with no input you get low amplitude random phases at the output.
With valid input you get lock and log amplitude output from 1 millivolt to 1 volt.

Finally, the approach I was waiting for. Can this oscillator be used with a classical pulse signal?

[moodz]

Yes it works on the principle that two loosely coupled oscillators will synchronise ... we use one oscillator that is "weak" and wants to be driven by a periodic signal ( even buried in noise ) that is close to its natural beat frequency.
This principle will also work with harmonics.
eg two clocks on the same wooden beam will synchronise ( eventually ).

[moodz]

Here is the performance of the circuit pulling a 20 millivolt peak to peak signal buried in 1 volt peak to peak white noise. No problem and no phase slip.

The green trace is the recovered waveform.
The blue trace is the sinewave at the input before noise adds.
The red trace is the 1 volt pp white noise and 20 mv pp signal added at input to cct.

CHALLENGE QUESTION : This circuit seems almost too good to be true ... What is the the downside ( there is one) ???

[moodz]

...the same circuit locking in on a 1 millivolt signal in 1 volt pp of noise ... the wanted signal is below the noise floor of the white noise.

[Gunghouk]

Does the phase shift matter in the recovered signal?

[Teleno]

Here is the performance of the circuit pulling a 20 millivolt peak to peak signal buried in 1 volt peak to peak white noise. No problem and no phase slip.

The green trace is the recovered waveform.
The blue trace is the sinewave at the input before noise adds.
The red trace is the 1 volt pp white noise and 20 mv pp signal added at input to cct.

CHALLENGE QUESTION : This circuit seems almost too good to be true ... What is the the downside ( there is one) ???

Interesting. What I see at first glance:

The circuit is basically a Colpitts oscillator where the emitter resistor has been replaced by a current source. I wonder if it still oscillates when there's no input signal.

The biasing of Q1/Q2 is very sensitive to beta and temperature variations. You can use the LTSpice command .step temp list 0 25 40 to see what happens.

[moodz]

Does the phase shift matter in the recovered signal?

The recovery is "in phase" ... any shift should be constant.

One of the traces posted previous was unlocked at 100 uV ( but sort of locked inverted )... any phase shift is fixed and way more than adequate for metal detector purposes.

input output shown below ..

[moodz]

Interesting. What I see at first glance:

The circuit is basically a Colpitts oscillator where the emitter resistor has been replaced by a current source. I wonder if it still oscillates when there's no input signal.

The biasing of Q1/Q2 is very sensitive to beta and temperature variations. You can use the LTSpice command .step temp list 0 25 40 to see what happens.

Thats a good point about the temperature dependance however it is not the answer to the challenge question above.

I make it a point now not to publish "theoretical" ltspice simulations on Geotech unless I have checked "real world on the bench circuit" as well.
The circuit does oscillate without input and it does not demonstrate any temperature variation dependancy I know of from around 10 to 35 celcius ( not expecting it to ).

Below is the actual CRO capture of the "real circuit" locked on a noisy input. That circuit is running at around 40 Khz so it has some different component values from the ltspice sim above.

[moodz]

I had better credit my inspiration for this work ....

https://www.semanticscholar.org/pape...817d5b996409d9

Vasil Uzunoglu ... theres a patent as well. ( expired ).

Vasil did all of his work with RF.

I have reconfigured it so the LC circuit is grounded and produces a nice output for the ADC on the ARMD DSP processor.