Sensitivity Issues - part reply to Walter Marton.

[Anonymous]

Hi Walter,
I

 

[Anonymous]

Hi Eric, thanks for the response,
for my understanding, the power induced into an target is proportional to the current change rate (and therefore to the back emv) but not to the absolute current value. Of course, the longer the pulse (same emv), the bigger the start current, if it should end at zero. At 400V, 300uH and 1A it would last 0,75us. Do you think, this is long enough for smal targets like coins? Using 600V-MOSFET would reduce it to 0,5us, but with 50% more peak power ... for the same power input! Farnell is offering "Low Gate Charge" N-MOSFETs rated for 400V to 900V, even a Low Gate Charge version of the IRF740. Is there a benefit of higher voltage (peak Power) vs. Pulse length? I mean, if I double the start current value and the back emv, Pulse length will remain the same - I believe to get twice the signal for the cost of twice the power. What about doubling back emv/half pulse length? (measuring at the same delay after mosfet comes out of avalanche) After the current reaches near zero, the voltage starts to fall down. In case of dumping resistor you have a long expo decay. If no resistor, the inductor current is going down much quicker, but at the time the voltage reaches zero, current has allready negative value and so it starts ringing ... Now lets switch (with another mosfet) the damping resistor on with some delay and vary the delay - the response changes from expo decay over much quicker zero setting to overshoot and ringing. If resistor switching is done precisely enough (10-20ns resolution), you can save some time. Can you comment my idea please?
Can you figure out, please what factor of noise cancelation can be achieved with a differntial noise cancelling coil, e.g. as posted on the forum? What are the reasons not to use it in every detector (except of larger costs)?
regards Walter

 

[Anonymous]

Hi Walter,
I shall present all the equations for signal voltage etc. in the book I have just started to write, but ultra-fast switching of the TX current is of no benefit unless you are wanting to detect very tiny bits of metal. There is nothing in the equation that represents the rate of current switch off. The only proviso is that the rate is at least five times faster than the object time constant. If this is met, then the shape or rate of switch off is immaterial. A thin gold ring might have a TC of 25uS which means that the total decay time would be about 250us. Provided the TX field is removed within 5uS, then it is as if it were removed instantaneously and no more signal would be obtained even if you switched off in 0.5uS. What does go up, of course, is the back emf from the coil and this might start to put a strain on various components.
Try some experiments with fast switching and switched damping resistor. I have found that you never really know until you try it. Working things out in theory is all very well, but there are so many unknown factors which can help, or hinder, the final outcome. One of the problems is that the TX is switching amps and the receiver is looking for microvolts. Adding devices to the TX or RX front end often gives rise to low level effects that cannot be forseen from the equations. I have had just this problem within the last 24 hours. Added a little circuit to the receiver front end to perform a certain function and found that the differential integrator no longer performed as it should. After much head scratching and a bad night it all came down to a very low level diode non-linearity that you could barely see on a scope.
Noise cancelling coils are very effective at powerline frequencies and moderately effective at low r.f. below 0.5MHz. Above this, ordinary bundle coils aren

 

[Anonymous]

Hi Eric, thank you - with your help I made a real jump in understanding PIs. But the questions in my head are going on - more and more. So the signal of the thin gold ring having a TC of 25us would be 37% of his initial value after 25us. We could double this value (to 74%) if we measure at 30% of the TC, meaning 7,5us. And I suppose - please tell me, if you know - the TC of bronze (brass) coins is smaler because of the high resistivity - perhaps 10us or even less - so it would make sense to mesure even after 3us! On the other side, a TC of 10us corresponds to LP Filter at 15,9kHz (1/(2*Pi*TC)), so why using such a high bandwith like 1MHz with all the noise burried in? If it is only to avoid saturation from the foregoing puls, the a RX-TX coil separation (DD-coil) realy makes sense. And if it is possible to get the bandwith significantly below 0.5MHz, then noise canceling makes also much more sense!
regards Walter