Hi Jeff,
Despite my bad english, I'll try to explain
Theoretical limits of sample delay
1. There is only a practical limit to shorten SD. That limit is the CF of small object (SO) or its time-constant TC. At SI, the Bode diagram of a small nonferrous metal object seems like HP filter. It is no use to excitate SO above its CF (or to take first sample earlyer than its TC value) because the amplitude of useful signal can't more increase with frequency. In reallity, the SO has series of TCs, inverse proportional to 1; 9; 25; 49 etc, but that isn't important for Bode diagram because cuttoff frequency is the some.
2. However, there are two practical limits to prolong SD. The limits are depending on 3 cuttoff frequencies (3 time-constants):
a) CF of large object LO (matrix, medium). Fortunately we can alternate that CF by means of TX coil diameter,
b) CF of unwanted (discriminated) object UO (which we can't alternate if its eddy diameter is relatively small) and
c) CF of SO (TC of wanted small object).
In theory, for best results excitating SI frequency (delay of first sample at PI) should provide quadrature (90 deg diference) between wanted and unwanted signal. In praxis, the necessary phase shift depends on wanted/unwanted signals ratio. At very low ratio, we should drive synchronous demodulator by interference signal passed by inertial (integrating) device. That is different than GB (or GEB - ground exclusion balance) and different than Corbin's eliminator. They both are using only a model of unwanted signal. See how acts a modern FM receiver. Its demodulator chip uses tank circuit to memorize large carrier signal, in order to detect small phase difference. I'll stop here, since that lead us in frequency domain to BFO with angle (frequency) demodulator, and in time domain - to coplanar PI gradiometer. There is such one shematics by R. & D. Crone in ETI 9/1989:
http://tthn.com/geotech/metdet/pi/twinloop.html
The circuit of Crone Bros has several advantages, which I have listed in a post to Eric's old forum. That's the right princip, but we should improve the sensor configuration.
Mike BG
Post Scriptum: Abbreviations and terminology:
CF - cutoff frequency CF=1/TC. Frequency of SI, at which object response is shifted 45 deg, and amplitude drop is 3 dB.
HP - high pass,
LO - large object. Conductive object which eddy current diameter is large than TX coil dia. By alternating TX coil dia, we can alternate LO TC or LO CF.
PI - pulse induction. Excitation by step function. Synonymes are TEM, "EM decay" and "EM relaxation". In excitation theory, termin "pulse" means something anoder. That is infinitely short and infinitely large function, which integral is the step function.
SD - sample delay. Synonymes are "pulse delay" and "sample moment",
SI - sine induction - excitation by trigonometric function. Synonymes are CW (continous wave) and VLF,
SO - small object. Object which eddy diameter is small in relation to TX coil diameter
TC - time constant TC=1/CF. Time of PI, at which response signal decay is e=2,718 times,
TX - transmit, excitating
UO - unwanted (discriminated) object.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
">