OP AMP CIRCUIT DESIGN QUESTION?

[GOHO]

Is this the pre amp suggested?

http://docs.google.com/viewer?url=http://www.mouser.com/ds/2/302/BF862-840156.pdf

Any examples on how that would get implemented into the schmatic above?

Thx!

[bobaruni]

IMO, as a minimum, I would take 1 vector at the very beginning of the decay (1 vector being 2 consecutive samples, a tail and a head) for the very small metallic component, another vector a small time after this as the first vector will be swamped or saturated by large metallic components. another vector in the middle of the decay and one towards the end for the ferrous and magnetic components.
You can also take another vector (or at least a single sample) in a very quiet period for ground noise and magnetic field cancellation but as Kleinstein has stated, you can probably get away without the last vector.
That makes it a total of 6 to 8 samples per decay pulse.

I think even a Teensy 3.0 can do it in terms of processing power and ADC speed but do not expect the full 16 bits of useful data until you use a separate ADC.
If you want a faster ADC (80 MSPS) but only 12 bits, take a look at LPC4370, with some good analog front end variable gain amplifier VGA it could also work.
A bit hard to use the LPC as it's BGA so maybe  take a look at a board called LPC-Link 2 (can be had for under $30) but not so easy to implement flash based boot on this board as it runs user firmware in  RAM but there might be a way to inject your code into flash?
Teensy would be my first try if you have one laying around and if it's not good enough, add an external ADC or go the LPC with a VGA.

I'm not sure about the Teensy 3.6, I don't think this can do 16 bit conversions?

[GOHO]

  Thanks For the advice, I will give it a shot with my new Teensy 3.6...

  But now im back to my original problem of keeping the output of the opAMP to 3.3V MAX..    Should I keep the circuit positive gnd or switch to P channel mosfet and flip the gnd to neg? is one way more sensitive than the other?



 

[Kleinstein]

N-Channel MOSFETs are usually better in having less capacitance and more types to choose from. So there is a good reason to use a kind of positive ground, or negative supply to drive the coil. For keeping the signal within the 3.3 V voltage range, one option would be to have the last amplifier stage powered by those 3.3 V and limit the signal before that stage with diodes. This could be something like a times 50 from a FET+BJT based stage, followed by a times 10 with an OP powered from 3.3 V.

The linked BF862 should be good choice. However I would not use the inverting amplifier, as this usually has higher noise.

For taking the data something like 10 points in the decay region would be probably enough. For the final "DC" values a few more point from the 100-200 µs window. So something like a 500 kHz sampling rate would be fast enough. My choice would be a something like 1 MSPS.  A slightly faster sampling could compensate for lower resolution - I would expect 12 Bits to be good enough. With switchable gain, even 8 Bits could be good enough.

[bobaruni]

And something not mentioned is that BJT OPAs will be more susceptible to interference from AM radio stations causing random signals in the passband giving false signals through modulation and rectification effects in the OPA.
A fet input stage will be less susceptible to this phenomenon but even then some care must be taken (LPF) to make sure you don't have too much RF pick up and amplification above 500Khz.

Also, another trick to improve sensitivity to small metal objects is to increase the resonant frequency of the coil by reducing the capacitance on the coil by place a high speed high current low cap diode in series with the MOSFET and the coil.
Then the MOSFET's capacitance has much less effect on the tuned frequency thereby allowing you to sample earlier on in the decay.
But with only 3.3V supply, this is probably not feasible.
Also remember to drive the MOSFET gate with a low impedance source as the speed or slew of the off transition will help with sensitivity by sampling as early as possible not to mention the reduction in heating and loss by making sure the gate is driven hard enough to ensure minimum On resistance at high frequencies.

What is your battery voltage? Do you have more than 3.3V available at high current?

 

[StillTrying]

I can't see any any schematic of the pulse amp in this thread, - if there is one.

But, going by the scope traces in reply#14,
https://www.eevblog.com/forum/projects/op-amp-circuit-design-question/msg1141449/#msg1141449

and taking many things into account, I can't figure why the amp detects BOTH the energizing voltage AND the decay voltage as BOTH positive going. What (possibly obvious) thing am I missing?

[GOHO]

There is no pulse amp, just a 555 that triggers the mosfet.

[bobaruni]

and taking many things into account, I can't figure why the amp detects BOTH the energizing voltage AND the decay voltage as BOTH positive going. What (possibly obvious) thing am I missing?

Yes, good observation, they decay signal has overshot and is the wrong polarity.
Either the coil is not critically damped (Q = 0.5) causing overshoot or the Op amp stages are too slow or maybe some other problem with OPA?
Try reducing the value of R1 to see if the decay will invert then try increasing the BW of the gain stages.

[GOHO]

 That is the normal signal from that circuit. I have looked at a dozen or more different detectors that use this circuit and they all have the same exact signiture on the scope. Even a Brand new one right from the factory.

[GOHO]

The orignal circuit uses 9.8v-10.3v nimh but i am replacing with an 11.1v lipo 20c 5400mah.

[bobaruni]

That is the normal signal from that circuit. I have looked at a dozen or more different detectors that use this circuit and they all have the same exact signiture on the scope. Even a Brand new one right from the factory.

Ahhhm, OK well then I'm confident that with the information here that you will better the dozen or more detectors out there as they are not damping the coil enough to achieve excellent sensitivity.
I have run some simulation and by setting the coil damping resistor to 390R, the result is underdamped, by setting it to 330R, the result is damped well enough to be the correct polarity.

See attached LTspice simulation and output image, the green trace is the desired polarity using 330R, the blue trace is underdamped using 390R.

[GOHO]

Thats interesting... ill test it out.

The search coil resistance is 1.6R and inductance is 27uH.

[bobaruni]

27uH is very very low, 300uH is the norm for high end PI detectors, at least of the Minelab variety and starting price is about 3K on those.
Do you know the coil capacitance and cable capacitance or at least the search head resonant freq?
Plug those values in the sim and have a fiddle.
With such a fast coil, you will be chasing very small shallow nuggets, down here we are chasing very large but deep nuggets in heavily mineralised soil and not many prospectors use discrimination when looking for gold.
What are you going to use it on, coins/relics/gold?

[GOHO]

when i replace the 470R damping resistor with a 330R this is what i get... i loose the induce current signature.

  This detector is a coin and relic detector but i would like to increase the sensitivity to small gold items.

[bobaruni]

What you are now getting is a good starting point, just add your MCU, I wouldn't worry about the FET OPA as your coil is very low impedance if it is really 27uH.
To get better sensitivity to small items, you need to sample as early as possible in the decay period.
To be able to sample earlier, you need to reduce the capacitance on the coil, try adding a high voltage low capacitance high current diode in series with the MOSFET like in the sim.
Your MOSFET has 170pF of output capacitance, with a series connected diode, this could drop to about 20pF increasing the resonant freq substantially.
Also, the MOSFET gate should be driven really hard to speed up recovery, definitely more than an MCU I/O port can do but should still have a gate stopper resistor.
Use a pair of BC327 / BC337 in an unbiased push pull voltage follower arrangement or a proper MOSFET driver IC like the TC4427A family.

[GOHO]

I agree that in order to detect small gold i need to sample higher up the curve but this circuit wont allow that . I have been successful replacing the mulitvibrators with a micro controller and i used to reprogram the sampling locations but once i got on the curve i couldn't balance out the detector, im sure it could be modified but i was unable to get it to work better than original..., well, maybe slightly better..

  Here are the original factory sample gates, as you can see they are way down the line, to far to detect small gold but when i have it on my scope, i can detect small gold items by watching the curve move. 

[GOHO]

sorry, i miss quoted the inductance... its .27mH not uH!

[bobaruni]

0.27mH or 270uH will work very well.
I wouldn't expect to be able to balance it with modified timings.
That's the beauty of doing it in an MCU, you can experiment with different ways of getting it sensitive and balanced.

[Kleinstein]

Right after the current is turned off, there is a background from search coil itself too. If well done this background should decay rather fast and it should stay constant over time. Still it will limit the time on how close one should go to the pulse. To close to the start there will be a lot of background (that can also drift). So the starting point is compromise between the signal (that is highest at the start) and the background that is also highest there. But hopefully the background decays faster than the signal.
 
The old sampling gates look rather late, but this is likely without an object in range. So the initial pulse is likely just background you don't want anyway. Still strange that the second background sampling window is also so short. More time for the background might give slightly better SNR.

The advantage with a µC is that you get not just those 2 sampling windows, but can look at the whole curve. The logical probe background to zero out the detector is the decay curve you get in a clean environment. It is after subtracting that background one can test methods to get a more simple (e.g. single number) value from the curve.

[GOHO]

I have a question about a brd design for a detector. Would it cause noise to have a GND plane?, For the schematic i attached above would it affect the the signal noise if the +9V/gnd was a plane on the circuit brd top and bottom layer? Is it better to design a detector brd without power or gnd planes?

[Kleinstein]

The electronics board would be detected by the circuit as well. So the usual ways is to have the electronics and battery just at the other end of the pole to get the highest possible distance. A GND plane would give same relatively slow background, just like a piece of copper so it might not be such a good idea. It could be OK if it is really small. Also make sure current loops are small.

A thing to avoid is a metal enclosure with contacts that are variable - like aluminum with not so perfect screws. The background from the electronics part itself is not that bad, as long as it stays constant. The worst thing is a background that changes with movement, like wires flapping around or a flimsy battery holder.

[bson]

So the NE5534 configured at a gain of 100 would only have a bandwidth of about 100 kHz.

Well, it still produces output, only it's attenuated past -3dB, so instead of +40dB you get +37dB.  At 1MHz you get a -30dB signal out of it, which is only +10dB.  This might be perfectly acceptable for waveform recognition, as long as the loss is part of the waveform pattern or template.

But I agree, why go to such great lengths - working with distorted waveforms - just to use such a crappy old limited vintage opamp?  There are much better devices around today.