BIPOLAR ALTERNATIVE TO H BRIDGE - GENERATING SINE WAVEFORMS - HALF / SINGLE & FULL

[Tinkerer]

Here is a simulation of Moodz's excellent Square current wave TX.
Next we will look at the details and the induced currents in the targets.

If anybody has questions about the simulation, please ask.

[Tinkerer]

I'm getting confused over which system you're describing. Here are the TX current & target EMF for CCPI:



The target EMF comprises short impulses that will depend on how the kick-start is done but will usually be close to half-sine voltages. Because they are rapid the target eddies have little time to build up so the eddy responses are likely close to a t*e^-t/τ. A low TC target will get a higher initial eddy current but decays faster, while a high TC target will decay slower but with a much lower peak.



The half sine TX and target EMF look like this:



To imagine the target responses start with a continuous sinusoidal response and simply truncate it. Here is a set of plots for some targets at 10kHz. The gray dashed lines exactly frame a half-sine of the current, and the plots to the right are all the responses cropped to these lines.



But this does not account for the exponential turn-on (or turn-off) inertia caused by the tau of the target. Suppose we take the US nickel; its tau is ~10us which means that its 5*tau settling is about the same as the half-sine pulse width of 50us, assuming a half-sine "frequency" of 10kHz. The total response looks like this (ideal truncated in gray):



It is still very much a VLF-ish response during the TX time (from which you can literally extract a target phase), and you also get a separate PI response during the off time.

Both approaches give very interesting but very different results.

Sorry for the delay, I was travelling.
My explanations are probably not very clear, but I think the subject is important enough to be worthy of detailed description and full understanding for anybody wanting to be involved in metal detector design. In my case, I am mostly interested in bipolar, square wave, continuous current PI. Therefore I will make another try, including simulations, to explain how I think this type of TX works and how it affects the targets and the receive circuit.
Probably the best way is a discussion. This will help to clarify and if needed to point out any possible mistakes in my thinking.

[KingJL]

Bom for the Moodz_TX(HS):
TX_BOM.pdf

[KingJL]

I just upgraded the schematic and gerbers to make it easier to accommodate Half-Sine operation. My original board was mainly focused on CC operation. Added the half sine coil capacitor and damping network to the board along with 2 jumpers to allow switching between HS or CC mode. For HS operation shunt J104 and remove shunt from J!05... For CC operation, remove shunt from J104 and shunt J105.

I just ordered a set of 5 boards from JLCPCB for $3.33 (would have been $2.00, but I opted for lead-free HASL) plus a $3.93 shipping charge.


TX.pdf
Moodz_TX(HS)_Gerbers(1-13-2023).zip

[Carl-NC]

And how is that different from the bipolar square wave current shown in "CC bipolar pulses"

I'm getting confused over which system you're describing. Here are the TX current & target EMF for CCPI:



The target EMF comprises short impulses that will depend on how the kick-start is done but will usually be close to half-sine voltages. Because they are rapid the target eddies have little time to build up so the eddy responses are likely close to a t*e^-t/τ. A low TC target will get a higher initial eddy current but decays faster, while a high TC target will decay slower but with a much lower peak.



The half sine TX and target EMF look like this:



To imagine the target responses start with a continuous sinusoidal response and simply truncate it. Here is a set of plots for some targets at 10kHz. The gray dashed lines exactly frame a half-sine of the current, and the plots to the right are all the responses cropped to these lines.



But this does not account for the exponential turn-on (or turn-off) inertia caused by the tau of the target. Suppose we take the US nickel; its tau is ~10us which means that its 5*tau settling is about the same as the half-sine pulse width of 50us, assuming a half-sine "frequency" of 10kHz. The total response looks like this (ideal truncated in gray):



It is still very much a VLF-ish response during the TX time (from which you can literally extract a target phase), and you also get a separate PI response during the off time.

Both approaches give very interesting but very different results.

[KingJL]

... Moodz Alt2H-Bridge... posting the schematic and gerbers...

Just checked the JLCPCB site... you can get 5 boards for $2.00 + shipping!

[Tinkerer]

So during the first quarter TX sine wave negative eddy currents are induced in the target. Then, during the second quarter TX sinewave, the negative eddy currents in the target are cancelled (or subtracted) and then new positive eddy currents are induced in the target.
During the "listening period" of zero current, the remaining eddy currents decay and can be sampled.


And how is that different from the bipolar square wave current shown in "CC bipolar pulses" https://www.geotech1.com/forums/foru...046#post408046


During the first quarter TX sine wave POSITIVE eddy currents are induced in the target. The TX current changes very fast from +5A to 0A. The energy in the inductance of the coil is transferred into the capacitor.
At the time when the current is 0A, the capacitor is fully charged to the peak Flyback voltage.
The capacitor is then discharged very fast into the coil (inductance) in the opposite direction, again inducing POSITIVE eddy currents in the target. The eddy currents induced during the first quarter sine wave did not have any time to decay at the time the coil current was 0A, thus, the new eddy currents induced during the second quarter are ADDED to the first eddy currents.

The eddy currents are corresponding to a 10A TX pulse, from +5A to -5A.

During the "listening period" of 5A current, the eddy currents decay and can be sampled.

Why do we not see any half or quarter sine waves on the graph? This is because the frequency is so high that in the graph we see them only as a straight line.

[Dean Sarelius]

Hi moodz, Carl,
Replying to moodz statement Multifrequency is your best bet ( if your frontend can process it ) , in your opinion what would be the minimum specs for a good A-D front end. For example woud you prefer an A-D with a fast sample or hold or one with highest resolution.
In my experience the difference between a good 12 bit A-D and a 16bit A-D is marginal so long as the noise of the flyback is kept to a low level. Typically a 12bit A-D will be faster also.

[KingJL]

I have been playing with the Moodz Alt2H-Bridge for a while now. The PCB that I am using uses SMD components (except for the coil connector and a 100K KEMET potentiometer). If anyone is interested, I am posting the schematic and gerbers.

So far I have been focused on the CC operational mode... The PCB works well in this mode. I have not yet ventured to the half-sine mode of operation... have a few loose ends to tie up with the current effort first.


edit: I inadvertently posted an old version of the Gerbers... new version functionally the same... just cleaner!

[moodz]

Hi Moodz,
I am just looking into the different aspects of bi-polar switching before trying some different simualtions. Have you expermented with any dual frequency bi-polar designs as yet or do you think that a single frequency front end should suffice..?

... I would go with Carls advice ... Multifrequency is your best bet ( if your frontend can process it ).

[Carl-NC]

I did both truncated and non-truncated, it's just a matter of how the timing pulses are done. Yes, the drive voltage is proportional to the "frequency". Multifrequency also gives you more info on targets and ground, and in the half-sine it also gives you "multi-pulse" variation. There is lots of potential here.

[Dean Sarelius]

Hi Carl,
For the multifrequency half sine experiments were these truncated half sines and did you use different driving voltages for example a lower drive voltage for the lower frequency pulses. My understanding is that the only reason you would want to use multifrequncy is to null out salt water influences but are there any other advantages I guess..?

[KingJL]

The response during the TX half-sine on time is REALLY interesting!

[Carl-NC]

You can initially think of the target response as exactly the same as a continuous sinusoidal VLF response, only truncated at half-sine intervals. But this ignores the exponential start-up inertia of the eddies, so that gets superimposed on the steady-state response. The inertia exponential is most prevalent for high conductors, less so for low conductors.

Obviously with half-sine you want to run an IB coil and look at the response during TX as well as after. This gives you a true hybrid:


You do bring up an interesting point, as to whether the inertial part of the response might allow us to distinguish a single-domain target (like a coin) from a multi-domain target (like a pulltab). I never got that far with my research.

[Tinkerer]

So during the first quarter TX sine wave negative eddy currents are induced in the target. Then, during the second quarter TX sinewave, the negative eddy currents in the target are cancelled (or subtracted) and then new positive eddy currents are induced in the target.
During the "listening period" of zero current, the remaining eddy currents decay and can be sampled.

The advantage seems to be a different response for certain targets according the TC`s of the target, possibly enhancing or diminishing part of the response of targets that have 2 TC`s or multiple TC`s.