Minelab GPZ 7000

[Qiaozhi]

With this method, a bad coil is still highly usable, see simulation below:

L2 = Tx coil, 300uF and 2.2nF give a transient of 1.7us to 1A, which is not bad. R3 damps the ringing a bit.

This method clearly minimises the effort required for fast coil construction. Have you actually built this, or is it just a simulation so far?

Also, I note from post #82, that there is some ringing present.

[Teleno]

This method clearly minimises the effort required for fast coil construction. Have you actually built this, or is it just a simulation so far?

Also, I note from post #82, that there is some ringing present.

The GPZ 7000 has the same ringing, and even worse. See the scope capture by Kev in post #39 http://www.geotech1.com/forums/attac...9&d=1444074513

The ringing is due to the diode D3 being in high impedance for a brief moment during reverse recovery.

It can be avoided by replacing the diode with a MOSFET turned on at the right time with its source connected to a second voltage source Vdd2 higher than Vdd, removing M2 and replacing it with Vdd.

So far it's only a concept, haven't built it yet.

[Mechanic]

The GPZ 7000 has the same ringing, and even worse. See the scope capture by Kev in post #39 http://www.geotech1.com/forums/attac...9&d=1444074513

The ringing is due to the diode D3 being in high impedance for a brief moment during reverse recovery.

It can be avoided by replacing the diode with a MOSFET turned on at the right time with its source connected to a second voltage source Vdd2 higher than Vdd, removing M2 and replacing it with Vdd.

So far it's only a concept, haven't built it yet.

The ringing in Kev's scope shots is due to using an inductor as a pickup without any damping of the inductor.

There is a video somewhere where Bruce Candy shows the output from the Z using a magnetic sensor and there is no ringing..

Cheers Mick

[Teleno]

The ringing in Kev's scope shots is due to using an inductor as a pickup without any damping of the inductor.

There is a video somewhere where Bruce Candy shows the output from the Z using a magnetic sensor and there is no ringing..

Cheers Mick

For the looks of it, the GPZ 7000 is as good as a PI can get. The transient is extremely fast and the use of the rising edge allows unattenuated detection of all time constants (including the long ones) and a perfect 1/t ground response. Contrary to intuition (for those not versed), it accomplishes all of this with very short pulses (<5us) .

[Qiaozhi]

For the looks of it, the GPZ 7000 is as good as a PI can get. The transient is extremely fast and the use of the rising edge allows unattenuated detection of all time constants (including the long ones) and a perfect 1/t ground response. Contrary to intuition (for those not versed), it accomplishes all of this with very short pulses (<5us) .

As you didn't post the LTSPICE files, I recreated the simulation myself, and get the same results. When I probed the junction of D2/D3/L2, I discovered something interesting ... a truncated sine-wave! Which reminded me of this patent from White's Electronics ->

[Teleno]

As you didn't post the LTSPICE files, I recreated the simulation myself, and get the same results. When I probed the junction of D2/D3/L2, I discovered something interesting ... a truncated sine-wave! Which reminded me of this patent from White's Electronics ->

Related to that, check out the current in D1 or D2 ... it's practically a constant current equal to I/2 (I being the coil current). L2 acts as a constant current sink. This is the moodz patent AU2013101058 where the energy of a coil is completely discharged without overshoot, decay or ringing.

[Qiaozhi]

Related to that, check out the current in D1 or D2 ... it's practically a constant current equal to I/2 (I being the coil current). L2 acts as a constant current sink.

You also need to check out the current flowing through V1 (the 12V DC source). It's a battery hog.

Apart from the battery consumption problem, your main challenge will be getting this design to transmit a continuous sequence of pulses, because (at some point) you will need to dump the current from L2 either before or during polarity switching.

[Teleno]

You also need to check out the current flowing through V1 (the 12V DC source). It's a battery hog.

Apart from the battery consumption problem, your main challenge will be getting this design to transmit a continuous sequence of pulses, because (at some point) you will need to dump the current from L2 either before or during polarity switching.

Done this already. After 5us L2 discharges into a constant current sink, the moodz way. It gives a symmetric pulse repeatable as often as wanted.

The consumption problem is due to the pulse in L2 being infinite in duration. In real life it would be 5us width with a repetition rate of 100Hz, which gives a duty cycle of 0.05%.

Quick calculation:

L1 ramps from 0 to 2.9A in 50us. At 12V the average power is 17.4W
L2 stays at 1.6A for 5us. At 12V the average power is 19.2W
At a repetition rate of 100Hz (10,000 us period) the average power is: (50 * 17.4 + 5 * 19.2) / 10,000 = 0.0966; roughly 10mW

[Qiaozhi]

Done this already. After 5us L2 discharges into a constant current sink, the moodz way. It gives a symmetric pulse repeatable as often as wanted.

The consumption problem is due to the pulse in L2 being infinite in duration. In real life it would be 5us width with a repetition rate of 100Hz, which gives a duty cycle of 0.05%.

Quick calculation:

L1 ramps from 0 to 2.9A in 50us. At 12V the average power is 17.4W
L2 stays at 1.6A for 5us. At 12V the average power is 19.2W
At a repetition rate of 100Hz (10,000 us period) the average power is: (50 * 17.4 + 5 * 19.2) / 10,000 = 0.0966; roughly 10mW

I was trying to understand how you were planning to achieve this without critically damping the switch-off, but I now see that you are actually damping it with a CC sink.

Good luck with your experiments. It will be most interesting to see the results when you attempt to turn the simulation into a real working circuit. I suspect you may be releasing some magic smoke during the process.

[Teleno]

I was trying to understand how you were planning to achieve this without critically damping the switch-off, but I now see that you are actually damping it with a CC sink.

Good luck with your experiments. It will be most interesting to see the results when you attempt to turn the simulation into a real working circuit. I suspect you may be releasing some magic smoke during the process.

Well, thanks. When I do it (if I do) be sure I'll post the results here even if everything blows up in smoke.

Just for the fun of it and since you took the trouble of duplicating the simulation (which I appreciate), try connecting a 300 Ohm resistor (approx.) between the drain of M2 and Vdd. Voila! there's your critical damping! Then put some repetition rate of your liking into V1 and V2. There's your pulse train.

The switch-off waveform can be any as long as it decays to 0 between pulses (damped oscillation, exponential ...), it's immaterial because target sampling is done during the on-time. What's critical here is to obtain a fast switch-on and a stable on-current (< 1mA change).

Cheers!

[Teleno]

For the looks of it, the GPZ 7000 is as good as a PI can get.

After a clarification by Monolith regarding the Kev's scope shot, I have to withdraw the above statement.

If Kev used a coil in his measurement, then his scope shows the voltage induced by the flyback of the GPZ 7000. Since this is created by connecting the coil to a high-voltage source, the coil current rises as a linear ramp, therefore the induced voltage is a constant during the 4.5us of flyback (the step in the scope).

It really makes no technical sense to maintain a high coil current for long periods of time, which is what the 7000 does. What's worse, when you count the reverse periods its duty cycle is in the order of 90% or above. This detector is a power hog, just look at its heavy battery. It uses two switched power supplies (positive and negative) for the flyback besides the + and - battery supplies to hold the coil current. I wonder how they keep it from going up in smoke.

What do you get from all this? better response from long tau targets (big nuggets). Small nuggets remain out of reach (4.5us flyback is too long).

The same can be achieved with less energy waste by a short constant-current pulse because long taus respond in full after the rising edge of the coil current.

[greylourie]

User reports show it picks up those short tau type targets at improved distance than was the norm with previous models. Coil design makes pinpointing difficult. 72Wh battery lasts all day.

[Tinkerer]

This is the TX and target response of a traditional PI.
The TX ON is 50us. This is better for short Tau targets. For long Tau targets it would be better using much longer TX times.

It could be interesting to compare the TX and target response of a GPZ 7000

[crane]

[ATTACH]33996[/ATTACH]

This is the TX and target response of a traditional PI.
The TX ON is 50us. This is better for short Tau targets. For long Tau targets it would be better using much longer TX times.

It could be interesting to compare the TX and target response of a GPZ 7000

Perhaps you should read this patent!!

[Tinkerer]

Perhaps you should read this patent!!

Could you show us a simulation?