Opinion needed for optimal PI coil for gold hunting

[homefire]

Thank You!

[bbsailor]

The GBW (Gain Bandwidth) of an opamp is the bandwidth at unity gain (gain of 1). Strictly speaking, this is called the Gain Bandwidth Product, or small-signal bandwidth. For example, if your opamp has a GBW of 4MHz, then it will have a BW (bandwidth) of 400kHz when configured for a gain of 10. You simply divide the GBW by the gain. As a second example, the NE5534 has a GBW of 10MHz. If configured for a gain of 1000, the bandwidth of the circuit becomes 10kHz. By splitting the pre-amp into two stages (each with a gain of 33) you can boost the usable bandwidth to 303kHz.

Tau is the time required for the voltage to rise from zero to 63.2% of its final value.

Qiaozhi and others interested in this topic,

See this Eric Foster Post. http://www.findmall.com/read.php?34,...465#msg-129465

The coil discharge time constant is, to a large extent, controlled by the value of the damping resistor. Eric states that you want the coil discharge time constant (TC) to be at least 5 times faster than the target TC. The value of the damping resistor is controlled by the total capacitance in the TX circuit including, MOSFET COSS, coil, shield, and coax.

Higher values of damping resistor allow for smaller targets with lower TCs to be detected.

bbsailor

[Qiaozhi]

Qiaozhi and others interested in this topic,

See this Eric Foster Post. http://www.findmall.com/read.php?34,...465#msg-129465

The coil discharge time constant is, to a large extent, controlled by the value of the damping resistor. Eric states that you want the coil discharge time constant (TC) to be at least 5 times faster than the target TC. The value of the damping resistor is controlled by the total capacitance in the TX circuit including, MOSFET COSS, coil, shield, and coax.

Higher values of damping resistor allow for smaller targets with lower TCs to be detected.

bbsailor

Do you have the Fisher patent mentioned by Dave Johnson (2 back from Eric's post)?

[bbsailor]

Do you have the Fisher patent mentioned by Dave Johnson (2 back from Eric's post)?

Qiaozhi,

Sorry, but no.

There seems to be two ways to increase the sensitivity of PI machines for specific targets.

1. Brute Force: Pump a long, (few hundred microsecond) high current pulse into a coil and try to sample as soon as the flyback signal is damped and the preamp comes out of saturation. However, early sampling and high current pulses are variables working in opposite directions. High current pulses make for higher flyback voltages and these higher flyback voltages take longer to damp.

2. High Frequency: Stimulating a few thousand pulses into a coil with shorter TX pulses results in lower flyback voltages, damping with higher value resistors, earlier sampling and the integration of a few thousand samples to increase the target sensitivity (kind of like how a lock-in amplifier works to extract weak signals from noise).

Each method can be built to be optimum for various type targets but not at the same time. Large targets, by their nature, need long pulses to penetrate their mass and return their strongest signal at good depths. Therefore, large targets need slower and longer pulses. Small targets, like the 1mm links on a gold chain, need many small pulses and the ability to sample at the earliest possible time or 10 microseconds (for larger gold links) or even down to 8 microseconds for these very tiny 1mm gold links. Unfortunately, you would need to scan the same area at least twice with a different machines set and optimized, one PI machine for large targets and another PI machine or settings for small targets.

bbsailor

[Carl-NC]

http://www.geotech1.com/pages/metdet/patents/US4868504.pdf

[Qiaozhi]

http://www.geotech1.com/pages/metdet/patents/US4868504.pdf

Thanks. I should have known to look there before asking.

[Tec]

Higher bandwidth fast response

The SMPI uses a single-stage preamp with a gain of 1000. The GBW is ~10MHz, so that means the BW=10kHz or a tau of 100us. That's too slow for a fast-response gold detector. Split the preamp into two stages, each with a gain of 33, and now the tau is 3.3us. The noise increase is minimal, and simply switching the first stage opamp to one with a lower noise will beat the single-stage NE5534.

On this topic, something doesn't feel right..
Is the 'bandwidth' you are calculating for a particular gain an indication of a point -3dB down from the maximum gain of the amplifier ?

Is the actual gain at GBW 1 always, regardless of the gain of the amplifier.
i.e '1' would be -3dB of a gain of 2
but 1 would be much less in dB compared to a gain of 10 or a 100.
But my point is, if this is the case the gain is still 1 at this high frequency regardless of the amplifier configuration.
If the 'bandwidth' is the -3dB point of the amplifier, or some other figure relative to the gain of the amplifier.

So this high frequency is going to experience an amplification of 1 if you put it through two amplifiers with a gain of 33, or one amplifier with a gain of 1000.
You are not improving the actual amount of amplification the fast 'gold' signal is getting.

Just lowering the gain of the lower frequencies to make the bandwidth look better..

So, the fast signal goes through amplifier 1 and gets amplified by 1 (nothing), and then goes through amplifier 2 and gets amplified by 1 (nothing), making a grand total of 1.

I can;t feel or understand how you improve the actual speed of the op-amp, by jiggering around with the gain calculations.
It would have no effect on the actual amplification of these high frequencies..

[Carl-NC]

It is true that if you apply a 10MHz signal to an opamp with GBW=10MHz, the effective gain is 1 regardless of what you set the LF gain for. But we usually choose an opamp so the GBW is much higher than the signal of interest. That is the case with PI preamps... we're generally looking at taus of a few µs, not 0.1µs. So for our signal BW, running 2 stages with lower gain results in a wider effective BW. It also improves the large signal response, which is not modeled by the small-signal GBW curve.

[Dr Vel]

Qiaozhi and others interested in this topic,

See this Eric Foster Post. http://www.findmall.com/read.php?34,...465#msg-129465

bbsailor

Interesting read. Has anyone looked into devices with a lower reverse transfer capacitance? For example the IRF520 is used for power amplification on 10 and 11 meters, where low capacitance is a mandatory parameter. The Ranger RCI 2970 DX uses 8 of these for 200 watts RF power from 26 to 32 MHZ. Without looking up the specs I do know the Drain-Source resistance is very low at around 4 volts on the gate so I do not see peak power in the application here being much of an issue. Just seems to make logical sense that the best numbers for 30 MHZ power will also be superior for the rapid turn off needed in a PI MD.

Or am I missing something or other?

[moodz]

Real mensch use these ....

SJEP170R550 SIC JFET 1700 volts 550 mohm T0-247 8 AMPS

very fast switching.

**** 20 pf output capacitance ****





from www.semisouth.com

your own solid state valve.

[Carl-NC]

$25... ouch.

[Davor]

Or am I missing something or other?

Yu are not missing much, except that irf520 is a 100V device and nowadays PI suffer from somewhat larger flyback. Besides, varactor - type switchers tend to be more efficient than the pure resistive ones due to the beneficial capacity change during switching. Hence, a slower and more capacity loaded device can be a better choice in a somewhat slower switching case.
We are talking of di/dt here, and it directly leads to voltage. In case stray capacitances are in range with the MOSFET's, and the mosfet being a varactor in its spare time, it is a win-win for slower devices that can withstand a bit more voltage kick, and that have a capacity in match with the coil.

[Davor]

$25... ouch.

But you only need two per channel to make a perfect A class audio amplifier. OK, and a lot of juice.

[moodz]

But you only need two per channel to make a perfect A class audio amplifier. OK, and a lot of juice.

I have a quantity of them ... I will do some tests and post the results ...been too busy with coding to get on to it.

moodz.

[Aziz]

SiC JFets will be the next generation of PI's! (High voltage, high current, high temp)
They (SiC JFets) start to flood the market now. Although not cheap yet.

Aziz