Tweet
IB loops are typically factory nulled to -60dB or so; that is, VRX = 1mV per every 1V VTX drive. Usually this is done in two steps: coarsely using coil positioning, and finely using a single "fine-tune" loose turn of the RX coil.
It is not uncommon for loops to drift over time. With epoxy-potted loops, the epoxy can continue to cure for days or weeks after initial potting and cause movement of the coils. Coils placed in and out of severe environments (like the trunk of a car) can also cause drift. I have seen coil drift so bad that the coil null alone causes the detector to overload.
At White's I researched a number of ways to implement electronic nulling. My preferred goal was to do it all in the loop itself; the loop would have a TX coil, RX coil, bucking coil (maybe), the RX preamp, and a micro. The micro would quadrature-sample the RX signal and adjust a feedforward signal from the TX. (I'm trying to recall this as I type; I don't have my old notes, and I haven't thought about this in 10 years or more.) One idea is to do this only on command, either from a calibration box in production, or from the detector itself. That is, the coil would be nulled, the setting saved in non-volatile memory, and not adjusted again unless needed. I'll call this static e-nulling.
The other is to adjust the null on a continuous basis, even during use. The attraction of this is that the signal from ground mineralization (which has a phase close to that of a mis-nulled coil) can also be nulled, which could dramatically improve headroom. Let's call this motion e-nulling. The algorithm for implementing motion e-nulling will look a whole lot like something you would do for automatic ground tracking, and would require a target inhibit function.
As a starting point for discussion, here are two patents for e-nulling. US9989663.pdf is from John Earle at White's and uses 2 RX coils, one slightly undernulled and one slightly overnulled. A digital pot is then adjusted to select the best null. One thing I don't like about this approach is that it requires 2 complete RX windings, and depending on the inductance this can add some weight to the coil. Another issue is the bucking coil is wound onto one of the coils but not the other. This can create an R-null difference that is difficult to deal with. Another patent is US5729143.pdf from Zircon. They use PWM signals to create a correction signal with the proper magnitude & phase. This is a little closer to what I was working on, in that correction can be done without adding new coil windings. But I recall I was doing something with the present TX/RX signals, not creating a new signal. I'll have to re-think it out.
So the challenge here is this: think of ways to implement e-nulling. Remember that you want a good null before the signal even enters the RX preamp so you can get a decent first stage gain without risk of overload. Also think about how a multifrequency coil might be e-nulled.
It is not uncommon for loops to drift over time. With epoxy-potted loops, the epoxy can continue to cure for days or weeks after initial potting and cause movement of the coils. Coils placed in and out of severe environments (like the trunk of a car) can also cause drift. I have seen coil drift so bad that the coil null alone causes the detector to overload.
At White's I researched a number of ways to implement electronic nulling. My preferred goal was to do it all in the loop itself; the loop would have a TX coil, RX coil, bucking coil (maybe), the RX preamp, and a micro. The micro would quadrature-sample the RX signal and adjust a feedforward signal from the TX. (I'm trying to recall this as I type; I don't have my old notes, and I haven't thought about this in 10 years or more.) One idea is to do this only on command, either from a calibration box in production, or from the detector itself. That is, the coil would be nulled, the setting saved in non-volatile memory, and not adjusted again unless needed. I'll call this static e-nulling.
The other is to adjust the null on a continuous basis, even during use. The attraction of this is that the signal from ground mineralization (which has a phase close to that of a mis-nulled coil) can also be nulled, which could dramatically improve headroom. Let's call this motion e-nulling. The algorithm for implementing motion e-nulling will look a whole lot like something you would do for automatic ground tracking, and would require a target inhibit function.
As a starting point for discussion, here are two patents for e-nulling. US9989663.pdf is from John Earle at White's and uses 2 RX coils, one slightly undernulled and one slightly overnulled. A digital pot is then adjusted to select the best null. One thing I don't like about this approach is that it requires 2 complete RX windings, and depending on the inductance this can add some weight to the coil. Another issue is the bucking coil is wound onto one of the coils but not the other. This can create an R-null difference that is difficult to deal with. Another patent is US5729143.pdf from Zircon. They use PWM signals to create a correction signal with the proper magnitude & phase. This is a little closer to what I was working on, in that correction can be done without adding new coil windings. But I recall I was doing something with the present TX/RX signals, not creating a new signal. I'll have to re-think it out.
So the challenge here is this: think of ways to implement e-nulling. Remember that you want a good null before the signal even enters the RX preamp so you can get a decent first stage gain without risk of overload. Also think about how a multifrequency coil might be e-nulled.
Comment