Actually, there is no need to have a front end that is flat down to dc. Some of the industrial PI detectors that I designed in the 1990’s had a bandpass characteristic that rolled off rapidly below 200Hz, This was done by means of an active filter in the feedback loop. The purpose here was to give more attenuation to both 50Hz and 60Hz line frequencies, the levels of which can be high in an industrial environment. The active filter involved a couple of capacitors which could be described as being in the signal chain. Generally, industrial detectors are just required to sense the presence or absence of metal in a product and the decay characteristic is irrelevant.
However the requirements of gold nugget detectors are somewhat more stringent, especially if effective ground balance and/or discrimination is required. Whatever circuit elements are incorporated, IMHO they should not distort the received waveform, particularly so for accurate ground balance at all coil heights i.e. large changes in signal amplitude with a mono coil.
I find that with conventional dual sampling and analogue integration I can achieve all the earth’s field cancellation I need when using a 25in diameter coil and also be completely insensitive to maghemite “hot” rocks that have a level of remanent magnetism in addition to the viscous signal. At this point I will say that my preamp and integrator is quite different to what you see in popular PI schematics. Recently, I experimented with a dual differential 25in coil which would give additional high levels of earth’s field attenuation, but was primarily to cancel low frequency em noise. As far as earth’s field went the improvement was not noticeable because the main noise source is still the front end amplifier. Strictly speaking, not the amplifier but the 1K input resistor, and I know what to do about that.
I used to wave ferrite magnets over the coil to check earth’s field cancellation, but now use a low frequency signal generator to inject a 5Hz sine wave into the RX input. This way you can control the amplitude and set it such that the signal swing is well within the linear range of the amplifier/integrator. Ferrite magnets close to the coil can induce a superimposed magnetostrictive ringing on the RX waveform, but when only a few inches away the 1K noise predominates.
I am old school electronics but it is good to see new ideas, such as those which Moodz is developing, being investigated and applied in PI technology. Let’s encourage him and others to persevere with their work, particularly in reducing noise by means of novel front ends, new coil configurations, dsp, etc. After all, the performance of any detector is basically a S/N problem.
Eric.
However the requirements of gold nugget detectors are somewhat more stringent, especially if effective ground balance and/or discrimination is required. Whatever circuit elements are incorporated, IMHO they should not distort the received waveform, particularly so for accurate ground balance at all coil heights i.e. large changes in signal amplitude with a mono coil.
I find that with conventional dual sampling and analogue integration I can achieve all the earth’s field cancellation I need when using a 25in diameter coil and also be completely insensitive to maghemite “hot” rocks that have a level of remanent magnetism in addition to the viscous signal. At this point I will say that my preamp and integrator is quite different to what you see in popular PI schematics. Recently, I experimented with a dual differential 25in coil which would give additional high levels of earth’s field attenuation, but was primarily to cancel low frequency em noise. As far as earth’s field went the improvement was not noticeable because the main noise source is still the front end amplifier. Strictly speaking, not the amplifier but the 1K input resistor, and I know what to do about that.
I used to wave ferrite magnets over the coil to check earth’s field cancellation, but now use a low frequency signal generator to inject a 5Hz sine wave into the RX input. This way you can control the amplitude and set it such that the signal swing is well within the linear range of the amplifier/integrator. Ferrite magnets close to the coil can induce a superimposed magnetostrictive ringing on the RX waveform, but when only a few inches away the 1K noise predominates.
I am old school electronics but it is good to see new ideas, such as those which Moodz is developing, being investigated and applied in PI technology. Let’s encourage him and others to persevere with their work, particularly in reducing noise by means of novel front ends, new coil configurations, dsp, etc. After all, the performance of any detector is basically a S/N problem.
Eric.
Comment