Deus II: About the oscillogram

[Carl-NC]

I am sure of one thing - that many of the members of this forum have understood everything from the oscillograms . They saw absolutely nothing new to intrigue them and the topic is boring for them .Let's "undress" the the young French lady named ,,Deus II ( https://www.beertourism.com/blogs/be...t-des-flandres ) little by little ... - It certainly is H-bridge circuit. Hardly is DRV8872 - from pdf datasheet - tON(1) Turn-on time VM > VUVLO with IN1 or IN2 high - 40 to 50μs - it is too slow . And requires a voltage above 6.5 volts . Must operate stably in the range of 3.3 -4.2 volts . А joke - because the French are unbelievable stingy, tighter , scrooge , miser - much more even than the inhabitants of the city Aberdeen , Scotland -https://en.wikipedia.org/wiki/Aberdeen . In some modes of operation they stop at certain times from the oscillograms almost all and leave the young Frenchwoman to starve

I'm certain that the Deus2 is using an H-bridge for the TX. I only mentioned the DRV8872 as a particular chip I've used, there are lots of others (popular with motor drives). I think the 40-50us number is for coming out of sleep mode, not for general switching. I've used it to a few 10's kHz just fine.

[Skippy]

I have successfully recreated the three-level waveform, and fed it into a spectrum analysis package.

It was a sort of success, but the spec analysis program caused some hair loss, it wouldn't give repeatable data, seeming to remember old settings and crap like that.
Anyhow, it spat out some useful stuff, that matched my earlier attempts at modelling it. I tried it on a plain square wave first ( like a CZ ), te ensure it matched what is easily calculated with normal maths. Then tried it on the D2 waveform.
For the 'CZ' square wave, +5V / -5V levels:
Fundamental sine amplitude = 6.365 V
Third harmonic amplitude = 2.122 V , exactly 33.3% of the fundamental, as maths shows.

For the Deus2, 4.7k/14.2k signal, +5V / 0V / -5V levels:
Fundamental sine amplitude = 3.25 V
Third harmonic amplitude = 3.888 V; this is 120% of the fundamental.
( I'm not 100% sure on these actual amplitudes, but the 120% figure is correct, even if the individual voltage are not )

So ... third harmonic level is more than 3 times greater with that fancy 3-level waveform, compared to the Fisher-CZ type of wave. That's got to help the noise performance.
I'll post up some screengrabs tomorrow.
-----
Regarding the H-bridge design:
I think you have overlooked one thing. The three-level waveform used at 4.7k / 14.2k can only be created if the 'zero volt' actually applies zero volts to the TX coil. That is to say, it's not open-circuit during these times, it is actively driven to Zero volts. So some additional switches to ground would seem necessary.
Maybe XP are using a linear drive circuit ... a power op-amp, effectively. Possibly based around a regular opamp, with discrete or integrated power stage added. That way, it's also possible to do linear-current correction, by driving the power opamp with a slope-corrected signal ( eg. from a DAC )

[Riss]

I have successfully recreated the three-level waveform, and fed it into a spectrum analysis package.

It was a sort of success, but the spec analysis program caused some hair loss, it wouldn't give repeatable data, seeming to remember old settings and crap like that.
Anyhow, it spat out some useful stuff, that matched my earlier attempts at modelling it. I tried it on a plain square wave first ( like a CZ ), te ensure it matched what is easily calculated with normal maths. Then tried it on the D2 waveform.
For the 'CZ' square wave, +5V / -5V levels:
Fundamental sine amplitude = 6.365 V
Third harmonic amplitude = 2.122 V , exactly 33.3% of the fundamental, as maths shows.

For the Deus2, 4.7k/14.2k signal, +5V / 0V / -5V levels:
Fundamental sine amplitude = 3.25 V
Third harmonic amplitude = 3.888 V; this is 120% of the fundamental.
( I'm not 100% sure on these actual amplitudes, but the 120% figure is correct, even if the individual voltage are not )

So ... third harmonic level is more than 3 times greater with that fancy 3-level waveform, compared to the Fisher-CZ type of wave. That's got to help the noise performance.
I'll post up some screengrabs tomorrow.
-----
Regarding the H-bridge design:
I think you have overlooked one thing. The three-level waveform used at 4.7k / 14.2k can only be created if the 'zero volt' actually applies zero volts to the TX coil. That is to say, it's not open-circuit during these times, it is actively driven to Zero volts. So some additional switches to ground would seem necessary.
Maybe XP are using a linear drive circuit ... a power op-amp, effectively. Possibly based around a regular opamp, with discrete or integrated power stage added. That way, it's also possible to do linear-current correction, by driving the power opamp with a slope-corrected signal ( eg. from a DAC )

Skippy , about 5 years ago a colleague of mine told me that a Polish detector Rutus works with such a waveform . Links - https://www.rutus.com.pl/en and https://www.rutus.com.pl/en/page/26/about . Take an H-bridge driver , for example the good old HIP4082 - https://www.renesas.com/us/en/docume...4082-datasheet . Give him (HIP4082 ) a 10 volt power supply .Connect only the drains of the two above N -channel VMOS on the H-bridge to the plus of the battery 4V via a Schottky diode( at the common point of the drains and the Schottky diode - there is nothing else ! - there is no electrolytic capacitor with a large capacity - it is connected elsewhere ) and run the computer simulation of the current through the coil . In time intervals where the voltage level is 0 volts - exactly then French engineers disconnect the diode anode from the battery with a key ( for example P - channel VMOS ) (it is at these moments that the French girl is left to starve ) and see what the computer simulation will show for the current through the coil . The coil - (a French girl ) is not stupid , she is frugal , provident - he knows that the French do not like to give , and may suddenly stop feeding her . Exactly because of that she (coil ) has accumulated energy .Skippy , you have mathematical thinking . I hate math , but at the same time I have a lot of respect for this science and I admire people with mathematical thinking . In fact, things are very simple and easy to understanding . I like to watch oscillograms . I like the oscillograms of Deus II , although there is nothing new in them . To leave mathematical simulations and sinusoidal interpretations . Here things are most clearly seen , in this photo - https://md-arena.com/multichastota-x...comment-133217 - photo with red ,,play " . Button is over
,,E.T. on 16.01.2022 в 08:48 " - .On this oscillogram things are very clear (what exactly happens to the current through the coil ) and we don't need to measure the current through the coil - according to the voltage oscillogram we can draw the oscillogram of the current through the coil. Also and the reactions of the ferrites and the targets . XP use a modern bridge that can operate with 3 volts for example. so nothing new under the sun ...

[Carl-NC]

I find the exact waveform for the D2 to be a little curious. A simpler voltage waveform results in the following current waveform:



But the D2 waveform has those extra "pauses" which produce an odd glitch in the current slew:



I'm not sure what they are for as they produce asymmetrical slews in the current. The induced EMF in the target has the same (but negated) waveform as the TX coil voltage so the eddy current response will also have a "hitch" in the response at those short pauses. Right now I'm not getting it.

As far as how to create these waveforms, you (Skippy) are right, an H-bridge will need a couple of extra ground switches. Which also makes me wonder about current recycling. In a normal H-bridge TX circuit a big cap is placed across the H-bridge supply and the coil current is recycled in a manner that has almost the same efficiency as an LC sinusoidal oscillator. In the D2 scenario I'm not sure how that plays out, but I assume it does because power efficiency is a key element of the D2 design.

[ivconic]

I see I would lose the bet.
But still ... let's wait for someone to break the Deus 2 coil and take clear shots of the pcb ...

[ivconic]

Back in a days i had Cscope 33XD.
I wanted to boost its capabilities by adding strong standalone TX, parted 15-30 meters away from the searching area.
So I made this circuit and it gave ~200vpp at the output, basic clock frequency.
Results were good, 33XD worked much better with such booster on.

...
Now... with the drivers you mentioned here guys; i see the rough picture of how it can be done.
Actually the current can be fine controlled by the pwm coming from mcu. (I'm definitely losing the bet now!)
Question:
is there any capacitor in series with the coil?
So many different frequencies will make it hard to imagine.
Except if they again used relays to switch bettween caps.
Let's ask somebody who already got Deus 2, to place the ear on coil surface and listen carefully for the "clicks"!

[ivconic]

...As far as how to create these waveforms, you (Skippy) are right, an H-bridge will need a couple of extra ground switches. Which also makes me wonder about current recycling. In a normal H-bridge TX circuit a big cap is placed across the H-bridge supply and the coil current is recycled in a manner that has almost the same efficiency as an LC sinusoidal oscillator. In the D2 scenario I'm not sure how that plays out, but I assume it does because power efficiency is a key element of the D2 design...

If the PWM coming from mcu is sinus; it doesn't really matter the question: half or full h bridge.
In half-bridge arrangement you don't need extra ground switches.
No risk from possible "limbo" state on the bridge, latching etc. (mcu can always go crazy for not so obvious reason)
That's why i voted for it.
The "dang" is; it need fat caps on the other side.
My bet is still having chances...

[Skippy]

Quote Carl:"they produce asymmetrical slews in the current"
Shouldn't they produce 'flat spots' , where the current continues at whatever value it was prior to zero volts being applied? ( Obviously the coil isn't perfect, it has L and R, the current will actually decay, no superconducting coil wire here ... )

I = I₀ + integral of ( dI/dt )

dI/dt = - V / L ; if V is fixed, I is the linear ramping slope up / down. If V = 0, no ramping, just I₀

In my head I see rising current slopes, descending current slopes at the same dI/dt rate, and flat current spots. I haven't Spiced it.

[Skippy]

Attached are: The plain square-wave that I used to evaluate the spectrum, eg. Fisher CZ tx drive.
And the two spectrum outputs, 'CZ' and Deus2'. The square wave has the familiar diminishing amplitude of the third and 5th harmonics ( 0.333 and 0.200 of the fundamental ).
The Deus2 has the third harmonic 20% stronger than the fundamental. The 5th, 7th harmonics were quite low, the higher orders had more strength.
The amplitudes shown on the spectrum are half what they actually are... problems with the 'ScopeDSP' software.
And a Desmos plot of y = 3.25 sin (wt) + 3.888 sin (3wt) ,
showing what the Deus waveform behaves like, considering just the useful parts of the 'square' signal.

[Skippy]

My 'notebook' sketch of coil current for the 3-level waveform, attached.
It looks right when considering the fundamental freq. It's 90 degrees away from the voltage, like current through an L is, doing a:
I = - cos (wt) waveshape.

[Carl-NC]

Shouldn't they produce 'flat spots' , where the current continues at whatever value it was prior to zero volts being applied? ( Obviously the coil isn't perfect, it has L and R, the current will actually decay, no superconducting coil wire here ... )

In the absence of resistance, yes. In a real coil you need a small DC voltage to "hold" the coil current:



CCPI (Minelab GPZ) works the same way.

[Carl-NC]

Forget the last one. Skippy, your waveform in #70 is probably right. Still requires some DC offsets at the held peaks. Also, this ceases to be MF, maybe it's one of the SF modes.

[ivconic]

You stole that from the Mincecraft ...

notty notty!

[Riss]

Here is the oscillogram The wife of the man who invented this oscillogram was born either in the city of Aberdeen, Scotland, or Gabrovo, Bulgaria. She cooks for him every other day (for example only on odd days ) to save money for food products and the electricity bill -the energy spent on cooking is halved . And because when you're hungry, you think much faster ( FMF ) -that's why he invented this current through the coil - by analogy with his eating mode . The inhabitants of both cities (Aberdeen and Gabrovo ) are incredibly stingy . They also have a very good sense of humor -they joke about their stinginess . And for this reason the two cities are twinned -maintain intensive communication . There is a humor festival in Gabrovo every year . I guess in Aberdeen too , but in Aberdeen festival it is not every year ? after all, money is spent on the festival -these are expenses from the city budget ? Here are some Gabrovo jokes :
"They say that people from Gabrovo cut the tails of cats so that the door closes faster after them and the room does not cool down." That is why the black cat with the cut tail has become a lasting symbol of the city of Gabrovo. They also say that they put taps on the eggs - to drain as much as they need - "a whole egg is a lot for one soup!" "? That they stop their watches at night so that their parts - gear wheels (tooth wheel ) don't wear out."
? - In order not to pay the chimney sweep, they let the cat through the chimney with a burning newspaper on the tail ".
- I can also tell why the TX coil works in mode ,,short circuit?? in the periods , when the current is kept constant . Then the coil is powered with a minimum of energy - just as much as is necessary to compensate for the losses from the active resistance of the coil - no more , no less energy . What are the advantages of this type of work . But I expect someone else to do it , because I will save :
- electricity bill to power my computer - the computer keyboard will wear out -my brain will expend too much energy to think how to translate it from bulgarian to English more understandable. To compensate for this (energy losses in my brain ) , I need to eat a little more - and these are costs? -I will lose a lot of time for writing (and time is money !!!)
Skippy , thank you so much for showing the oscillogram . I could show it too , but I judged , that it will soon be shown by someone else . And so I will save :
- a sheet of paper
- a little graphite on my pencil
- I must then sharpen the pencil again , and so the knife will wear out
- energy for the computer
- power from the smartphone battery and its flash
- the smartphone's chips will wear out
- but what terrified me the most was that the electrons in wires move forward-backward - like a saw . Which means - my cable connected to my computer and the internet - and it will wear out too !!!

[Carl-NC]

I might know why XP is using this waveform. A simple way to implement a SF detector using a digital TX drive is to just use a square wave, a la Fisher CZ. The current is then a triangle wave. At higher frequencies where inductive reactance dominates this works well. At low frequencies (maybe below 5-6kHz) the inductive reactance is dropping and coil resistance becomes a bigger factor. Instead of nice linear ramps the current waveform slopes become more exponential. The ohmic losses waste power.

Adding in some dead time (especially at the zero crossings) allows the use of faster slopes in the current ramping which improves efficiency. XP doesn't quite add 50% dead time but if it did it would make a 5kHz waveform have a similar efficiency of a 10kHz square wave. I suspect that they are channel-filtering the preamp signal and then doing a software IQ demod for traditional phase-domain results so all the weird artifacts go away.