New advanced PI project

[Tinkerer]

The SQUAREWAVE without the Flybacks.

[eclipse]

Is a sub circuit required to cover for resistive losses /kick circuit in Deemons design idea / also used to adjust square wave slope

[Tinkerer]

Is a sub circuit required to cover for resistive losses /kick circuit in Deemons design idea / also used to adjust square wave slope

Our coil current is software controlled.

[willybayot]

I have compared the power taken from the battery powering the transmit circuit on the simulations.
Between the two types of circuit, I get a ratio of almost 10/1 in Watts.
The traditional PI takes 4.3W and the Square Wave type takes 0.65W delivering 5X more pulses per sec in bipolar mode.(200 microsec period vs 1000 microsec period)

[eclipse]

Our coil current is software controlled.

Yup sounds good.
Shouldn't be too hard to implement and it will work better because coil parameters can change, and automatic monitoring and re-adjustments can solve lots of problems and improve efficiency.

[Monolith]

How complicated or difficult is the demodulation?

[eclipse]

It's not too complex. You just invert every 2nd TX cycle and sum them /no EFE subtraction/ so it's actually much better than unipolar.

[green]

Here is the SQUARE WAVE simulation.

SQUARE_WAVE.rar (reply #44)

Ran the simulation. Tx current not flat topped, slope about 30A/sec. What parameters are measured and controlled to get Tx to flat top?

[Tinkerer]

SQUARE_WAVE.rar (reply #44)

Ran the simulation. Tx current not flat topped, slope about 30A/sec. What parameters are measured and controlled to get Tx to flat top?

Thank you for the feedback, green.

You are correct.
It is far from a perfect square wave. The verticals are actually exponentials. The "Flattop is sloped about 3mA over the 100us of target receive time.
Unfortunately, there is no perfect electric world.

The slope shown is due to the ohmic losses.
What can we do to improve? In general, reduce all the resistances in the circuit.

The high voltage Mosfets used in the simulation have a relative high resistance. We use better ones in the real circuit.

I also found that using genuine Litz wire for the TX coil makes quite a difference in reducing the impedance.
Fine multi strand, tinned wire is the next best.
But for rough tests I just use magnet wire. Spacing the coil loops with fishing line reduces the interwire capacitance.

After reducing the total current path resistance / impedance as far as possible we still end up with a slight slope.

How much slope does matter?

How much eddy currents in the soil or target does it need to be detected?

The "Flattop slope" will affect a target with an TC of 5 times the slope time most. With a slope length / time of 100us that would correspond to a target TC of 20us. Let's look at what the simulation says.

To eliminate the slope completely we need to replace the ohmic losses at every cycle. DEEMON's original design already included a circuit for compensating the losses. If we do not like his compensation circuit, we can invent our own. In our prototype we control the current by micro-processor. For myself, this means that I dial-in the amount of compensation I want in a parameter file.

How does the slope manifest itself in the RX? basically it shows up as an offset.

[Tinkerer]

It's not too complex. You just invert every 2nd TX cycle and sum them /no EFE subtraction/ so it's actually much better than unipolar.

So, let's talk about some of the improvements the SQUARE CURRENT WAVE offers over the traditional PI TX.

1) We can use a much higher repetition rate. A higher repetition rate means we can integrate so many more cycles. This improves the S/N and the signal amplitude.
2) Considerable les power consumption for similar results.
3) Much less limitations as to the coil diameters / amount of coil turns.
4) Automatic Earth Field compensation.

Any questions? anybody?

Anybody disagree?

[eclipse]

Here's a relevant interesting topic to read:Optimizing-Target-Responses

Have you noticed increase response by hot rocks with Square Wave PI vs regular PI?

I've performed brief tests on few hot rocks I have and the response was quite noticeable /subjectively speaking/

Mixing several TX periods will work nice for this type of PI: 100us + 300uS because narrow tx without spacing may hide some target Reponses with long TC so this can be experimented, however on longer TXs you'd need to have that current control circuit.

[Tinkerer]

Here's a relevant interesting topic to read:Optimizing-Target-Responses

Have you noticed increase response by hot rocks with Square Wave PI vs regular PI?

I've performed brief tests on few hot rocks I have and the response was quite noticeable /subjectively speaking/

Mixing several TX periods will work nice for this type of PI: 100us + 300uS because narrow tx without spacing may hide some target Reponses with long TC so this can be experimented, however on longer TXs you'd need to have that current control circuit.

Thank you for the link, eclipse.
These are most excellent descriptions.

It is now a matter of choosing the design parameters for the desired purpose of the detector.

A receive time of 50us will be perfect for a targets up to a TC of 10us and quite good for a target of 20us.
A receive time of 500us for respective targets.

A detector could have several modes, like 50us, 200us, 500us.

Or, we could design a detector with CHIRP mode. Cycling consecutively 20us, 50us, 100us, 200us, 500us, 20us, 50us, 100us, 200us, 500us, or any repetition rate we need to best "illuminate" a great variety of targets.
It is now only a matter of a software controlled timing schedule.

Yes, the current control needs to be software controlled too.

[ivconic]

Tony, I am not getting any younger!

[Carl-NC]

For nugget hunting I would suggest something like 4 pulses @ 25us and 1 pulse @ 100us, separately demodulated. Subtract the channels for GB. This will give you a target hole so an alternative is to use 4 channels and GB each pulse width individually. This should give you 2 different target holes; the target hole for the 25us channel is filled by the 100us channel, the 100us channel hole is not filled but is up in a high conductive range where there are few nuggets. You could also triple-sample the 100us channel and eliminate its target hole.

[green]

I've been trying to figure how to measure change in coil current during the target sample and ground sample periods. Any thoughts?

https://www.geotech1.com/forums/atta...3&d=1655214071

easy with the simulation, don't know how with real circuit.