Technology Forum

Carl,

You are right, the gain of 212 could be lower and I could sample in even closer. I reduces my Preamp gain to the point where I could sample at 10us from the end of the coil pulse and get good target data, see attached scope picture and target samples. I stopped there simply because I met my objective for the project. The detector will detect my 1.5 grain gold flake test target at a bit over an inch and a half in air with my 8" coil and it will detect my 3 grain test target at over 3" with the same coil. That was the objective I had for this PI detector.

I guess my thinking must be all wet but it seems to me that other than a bit faster recovery speed, with a total gain of 1,000 in the preamp, and taking a target sample at the output of the preamp, an input signal to the preamp of .005V should drive the output of the preamp into the + and - 5V rails whether it has 1 stage or 2 stages to get the gain to 1,000. That means that you can not sample in any closer than at the .005V level on the decay curve. It just seems to me that if you sample after the first low gain stage you can sample in further on the decay curve because the curve is not amplified as much, then amplify the target sample to where it is usable. Right or wrong, that is my thinking. When I had my preamp gain at 1,000 I could not sample in any closer that about 18us. I could not detect the 1.5 grain gold flake test sample at all and only detect the 3 grain when I was almost touching the coil. When I reduced the gain to 212, I was able to sample in at 10us very nicely and could detect both the 1.5 grain gold sample and the 3 grain one as indicated.

Right now, if I ever build another PI I would probably lower the preamp gain to 150, take my samples at the output and then amplify the samples about 6+ times and then pass them on to an integrator. I should be able to sample into around 8us or less that way and still have an amplification of 1,000 going into the integrator.

God Bless!!

Smitty II

Sorry about that, I was looking at the wrong caps. I am not sure why the value is so high. I would have not used one that large.

God Bless!!

Smitty II

Carl,

I didn't mention this in my other post but I agree with you, the speed of the preamp increases when you lower the gain. Since an op-amps max bandwidth is at unity gain, the higher the gain the slower the op-amps response time is. With lower gain in the preamp you get two benefits, faster speed ( helps sample in closer) and lower amplification so you can sample in further before you reach a decay voltage that is high enough to drive the preamp into its supply voltage rails. I would guess that the lower gain limit would be at the point where you don't have enough signal to noise ratio.

Just thinking, thanks for your thoughts.

God Bless!!

Smitty II

Carl,

I have been thinking more about your split amplifier approach to taking the target sample in closer to the end of the coil pulse. Correct me if my thinking is wrong. If I use a single amplifier with a gain of 1,000 and measure it rise time, then using the same amplifier and change the gain to 33 and measure its rise time. If I put 2 of those amps with a gain of 33 in series so I have basically the same gain at the output, the rise time at the output should be the square root of the sum of the square of the individual rise times. That would tell me how much actual speed improvement I am getting over the single amplifier.

A spice simulation should show that, your feeling that I don't need to sample where I am sampling is a good point. I think I can make a comparison between the two approaches.

Thanks for your input.

God Bless!!

Smitty II

Yup, that's about right.

Preamp with Low Cutoff Frequency at PI
Error or Smart Design Solution?

How to calculate capacitance of C24?

A visual analysis in frequency domain shows what happens when the RF section of RX circuit has relative low cutoff frequency. The attached Bode diagrams represent magnitude characteristics (frequency responses) of both target properties creating signal in RX coil permeability and conductivity.

The Bode diagram of pure permeability has slope 6 dB/oct (or 20 dB/dec). That means the electromotive voltage induced in RX coil increases twice when TX frequency increases twice (or 10 times when TX frequency increases 10 times). Such transfer impedance has mutual inductance (separated TX and RX coils), self-inductance (monocoil) and targets having no conductivity (hot rock, pottery, meteorite).

The system (RX & target) can have flat frequency response if we design RX circuit having RF amplification with slope minus 6 dB/oct (or minus 20 dB/dec).
What means such preamplification in time domain? Each sample at pulse induction will have the same amplitude. A difference integrator will eliminate signal from magnetic ground, hot rocks, pottery and meteorites.

The Bode diagram of pure conductivity (half-space and nonferrous metal) has a cutoff frequency fc which separates the spectrum in two regions:
LF region where the magnitude characteristic of target increases with slope 12dB/oct (or 40 dB/dec) and
HF region where the slope is 6 dB/oct or 20 dB/dec.
The diagram is valid only for first order target having only one timeconstant (annular object).
http://www.geotech1.com/forums/showpost.php?p=103162&postcount=125

How will detect conductive targets a PI metal detector if its RX preamplification has slope minus 20 dB/dec?
The difference integrator will not detect the target when both samples appear in its HF region because the resultant slope of system response is zero. The target will be detected when the samples appear in the LF region of target (relative large sample delay). In this region the slope of system is 40 20 = 20 dB/dec.



Edited 1 time(s). Last edit at 10/13/2011 10:45PM by mikebg.

The power supply for this project is here



http://www.findmall.com/read.php?34,1537000

Share please PCB. I very much want to build this detector.

Its made on Vero board

For visual analyzis of the system (target & preamp), I posted spectral response of permeability and conductivity on October 13, 2011 as magnitude functions (magnitude vs. frequency).
Despite this is not an error, it is not enough for spectral description because a function in time domain should be described always by two functions in frequency domain. Here is the phase response of permeability and conductivity. Note that at cutoff fgequency fc, the conductivity makes phase lag 45 deg relative to current in TX coil.
Remains to show spectral response of a preamp having low cutoff frequency.



Edited 2 time(s). Last edit at 02/02/2012 09:46PM by mikebg.