Is this a major advance in PI sensitivity ??

[mickstv]

coil capacitance big problem cousing the noise

http://uploadpic.org/v.php?img=A0cMhpLrEq

I've been building a modified version of this frontend design and haven't noticed any problem with noise on either input or the output of the diff amp, the diff amp is running at about 500 and thats with an LCD and computer within 2 metres of the test bench. The coil I'm using at the moment is a spider coil made out of a twisted pair removed from a length of cat5.


I've built one sample circuit so far and it has no post sample amplification yet, whilst monitoring the output of the sample circuit with a cro set to 100mv per div i can detect the presence of a 0.4 gram nugget at about 10cm in air which I think is pretty good.




MIck

[moodz]

coil capacitance big problem cousing the noise

http://uploadpic.org/v.php?img=A0cMhpLrEq

If the coil is balanced the capacitance will be the same on each coil and noise will be balanced and cancel ... the amount of capacitance is relatively immaterial. The high frequency noise is coupled via gate leakage from the TX mosfet on one side of the coil.

moodz

[ashkelon]

coil capacitance big problem cousing the noise

http://uploadpic.org/v.php?img=A0cMhpLrEq

[samu]

Metal isolation?

Hi,

I´ve mounted the circuit and it works pretty well. Now, I´m trying to program a good software for the pic.

What I´m planning to do, is to use a large coil installed in a large metal box. I tried it and the result was logical, the circuit detects the box. I can compensate the box signal, but then the sensibility decreases a lot, I guess due to a kind of saturation and makes the MD not useful.

I don´t know if there is a way to magnetically isolate the metal box from the coil so the eddy currents are not induced or avoid. I tried to connect the box to ground (big fail), maybe there is a material to do this, or a way to stop eddy currents.

I think it would be nice if we can shield the coil in every direction but the one in which metals are expected.

Thanks to everyone.

[moodz]

Gday Tinkerer ... thanks for your reply .. in a professional design I would use a separate drive for the bias generator as the present circuit configuration is highly dependant on a fast pulse repetition rate ( eg > 4 Khz ) to maintain low ripple in the bias generator outputs.
C6 an C7 are actually 10 uf ... I was surprised too that the impact on gate drive timing was so minimal ... the caps dont really have time to discharge much between pulses ( approx 70 mv across 200us ). The diff amp only draws around 10 ma from the supply.

I do hope to get onto descrimination on the FPGA project .... I just need to finalise the dspic code for the Model T demonstrator ( which wont have descrimination )

Regards,

moodz.

[Tinkerer]

moodz,

thanks for the link. That seems to be just what I was looking for.

About C6 and C7, I dont't see any values on your schematics, just guessed at a few tens of nF and looked up the datasheet. The TC4421 indeed has lots of power and you have paralleled the 2 outputs. 5ns is nothing. Good show!!!!

I found that in a situation like you show on the scope, it helps to add a small feedback capacitor. It reduces the band width and the signal amplitude, but it also integrates the signal and gives more time to sample it.

For the discrimination you must look for THE PIVOT. This is what gives the information about the characteristics, FE, foil and volume of the targets.

All the best

Tinkerer

[moodz]

Driving the capacitors C6 +C7 with the TC4421, slows down the switch OFF speed considerably. This has a lot of influence on the Flyback and damping time.

I have been using a differential input preamp for some time now and am very satisfied with the results. However, it is not only differential, it also gives different results. It did take me quite a bit of experimenting with different input impedance, damping, and gains to find the mode that I am using now.

I don't think I will ever go back to a single ended preamp.

There are still many more things to try, for example I am looking for a good differential preamp that has not only differential input, but also differential output.

Can anybody recommend a low noise one?

Tinkerer

I have to respectfully disagree with you tinkerer .... I did originally drive the bias generator off a separate 4421 dedicated for that purpose however in order to reduce the component count I measured the impact on the fall time to the gate and it is approximately 5 ns ..... given the modest design goals of the Model T .... 5 ns is neither here nor there. The 4421 is a surprising beast in terms of drive capability .... if it had a higher voltage rating it would make an excellent coil driver in its own right.

Here is a nice diff amp for you.

http://www.thatcorp.com/1570_Low-Noise_Differential_Audio_Preamplifier_ICs.shtml

regards moodz,

[WM6]

Driving the capacitors C6 +C7 with the TC4421, slows down the switch OFF speed considerably. This has a lot of influence on the Flyback and damping time.

Maybe we need to experiment with different lower value of C6 + C7.

[Tinkerer]

Driving the capacitors C6 +C7 with the TC4421, slows down the switch OFF speed considerably. This has a lot of influence on the Flyback and damping time.

I have been using a differential input preamp for some time now and am very satisfied with the results. However, it is not only differential, it also gives different results. It did take me quite a bit of experimenting with different input impedance, damping, and gains to find the mode that I am using now.

I don't think I will ever go back to a single ended preamp.

There are still many more things to try, for example I am looking for a good differential preamp that has not only differential input, but also differential output.

Can anybody recommend a low noise one?

Tinkerer

[moodz]

The change of your balance resistors may give less noise, but they extend the flyback /decay appreciably. During the time that the flyback voltage is above the diode threshold, 1000 ohms is put in parallel with the damping resistor causing it to be significantly overdamped. I have been using 2.2K. Noise may be higher, but it still below any valid target signal. The noise with a 2.2k and a THAT1500(or INA217) is still below the typical single ended PI front-end. I guess the trade-off you accept is what suits your design goal.

There have been several enlightening discussions on this subject by bbsailor and others.

Regards,
J. L. King

Actually my design goal with this version of the circuit is to investigate the target signal at the base of the flyback. I have examined the input signal to the diff amp on a CRO and it is quite different from normal target decays. I am not sure how it is produced as simulations cannot reproduce it ...

regards,

moodz

[J_Player]

The change of your balance resistors may give less noise, but they extend the flyback /decay appreciably. During the time that the flyback voltage is above the diode threshold, 1000 ohms is put in parallel with the damping resistor causing it to be significantly overdamped. I have been using 2.2K. Noise may be higher, but it still below any valid target signal. The noise with a 2.2k and a THAT1500(or INA217) is still below the typical single ended PI front-end. I guess the trade-off you accept is what suits your design goal.

There have been several enlightening discussions on this subject by bbsailor and others.

Regards,
J. L. King

Hi KingJL,
I have been wondering about what you mean when you say "design goal".
The majority of PI detector users are looking to find metal items that may weigh between 1 gram and 1 ounce.
But there are other PI users who are looking to find larger buried objects, such as large meteorites or large caches of metal treasures.
When you say "design goal", I am thinking you may be focused on the more common goal to find the smaller targets. Am I correct?

Best wishes,
J_P

[WM6]

The noise with a 2.2k and a THAT1500(or INA217) is still below the typical single ended PI front-end.

I ordered some parts (TC4421, INA217) over eBay. Two weeks of delivery in best case.

[KingJL]

The input balance resistors have been dropped to 500 ohms.

The change of your balance resistors may give less noise, but they extend the flyback /decay appreciably. During the time that the flyback voltage is above the diode threshold, 1000 ohms is put in parallel with the damping resistor causing it to be significantly overdamped. I have been using 2.2K. Noise may be higher, but it still below any valid target signal. The noise with a 2.2k and a THAT1500(or INA217) is still below the typical single ended PI front-end. I guess the trade-off you accept is what suits your design goal.

There have been several enlightening discussions on this subject by bbsailor and others.

Regards,
J. L. King

[WM6]

You could try a fast diffamp based on an op-amp instead of the THAT1510 ... I will look into it.

THAT1510 is not obtainable here.

I found by google this:

The THAT 1510 is a high performance audio preamplifier suitable for use in microphone preamp application. It is pin compatible with the now discontinued, Analog Devices SSM2019 and SSM2017 IC chips, and the Texas Instruments INA217 and INA163 mic preamp chips.

The SSL 9K probably has a very simular topology and Standard NE5532 opamps too.

[moodz]

I have measurement equipment, but those parts (as THAT1510, TC4421, FQPIN60, SF18G) are totaly unknown on my components stock list. I have to see first where to purchase those parts - if possible.

What are adequate FQPIN60 replacement? From memory I have IRFs 740, 640, 840 in my lab stock.

I did some tests ...
IRF740 is fine .... I just use the FQPIN60 because it has 10 ohm on resistance to limit power.
A 1N4148 works OK for the damping diode ... it can handle 4 amp spikes. The SF is more rugged.
A 1n4004 works fine on the Mosfet switch.
You could try a fast diffamp based on an op-amp instead of the THAT1510 ... I will look into it.
The TC4421 could be replaced with discrete component gate driver.

moodz.