Warbling

[Anonymous]

Speaking of technical terms, I caught the term Warbling used, which I think means the low level low frequency "noise" caused by such things as external signals being demodulated by the double correlated sampling receiver commonly used in pulse induction detectors.
External controls are provided on some detectors to "tweak" the transmit frequency (transmit period or receive period) which also changes the two sample spacing, I guess. On pulse induction detectors I have built, I have had the ability to adjust transmit on off periods, delays to first samples (two channel), delay between samples, sample widths. Basically I could adust everything in an analog format. Tweaking all these could have an interesting effect and could find some real sweet spots where the "noise" (unwanted signal) would go way down and I could increase total gain alot and sensitivity would go way up.
Of course the external signal is "beating" with the demodulator and all sorts of beat type waveforms are generated of various amplitudes while changing timing. I see this Warbling (if this is what this is) as the greatest limiting factor to a really great pulse induction detector.
My question is this, Has anyone ever really gotten their arms around this effect, to describe it mathmatically might help, but are there any clever ways to reduce it? Extreme many pole low pass filtering makes it go down but makes the response of the detector slow also. I am also not sure how anyone gets away with not being able to adjust at least one parameter to reduce this unless they just limit the sensitivity of their detector and make it slow, so the operator won't have to mess with it (his loss).
JC

 

[Anonymous]

Hi JC,
A lot depends on where you are. In the Oxford area, the biggest interfering signal is from a high power 200kHz station at Rugby, which is probably 40-50 miles away. This causes severe interference if your TX frequency and sample timing are set wrong. In Hawaii and the East coast of England, there are submarine communication systems operating around 20kHz which can cause bad problems. For many years I have had a variable TX frequency control on most of my detectors. This only gives a shift of a few pulses per sec. but it does minimise most interference situations. As you have found from your PI where you can vary everything, the interference situation is complex. However, it is possible to make a PI that is almost interference free, as I have done for some industrial installations. The noise performance is then that of the first amplifier in the receiver chain. The best thing to do is to stop the noise entering the system in the first place and that means attention to the search coil, its configuration and shielding. After that there are various filter techniques that can be applied so that only the wanted signal passes though the receiver chain. The biggest industrial problem is spikes and transients, rather than continuous modulated or unmodulated carriers. Spikes can be filtered out by various methods, leaving an almost noise free dc signal that can drive a meter, relay or audio device. Additionally, increasing the transmitter power improves the signal to noise performance and there is great scope for doing this in a PI. The downside, of course, for a portable detector, is the increase in the size and capacity of the battery pack that is needed. Industrial units are often mains powered so this is of little consequence. One environment where there is little noise, except if your are near to a sub. comms. system, is underwater in the sea. This acts as an effective shield for most rf frequencies.
Eric.

 

[Anonymous]

Eric thanks for the response.
This is on a monoloop, and I could play around with the figure eight, but the detectors are working fine now, running 3 Amps on transmit and carrying a 4 pound battery. Lots of low frequency filtering after the integrator takes care of most of it, but can slow searching down if I go too low. I am also running direct coupled from the front end to the rear. Just getting greedy and wondering if I can get more. If I can get something to work better, I will probably turn down the transmit and get a little battery.
Nice to be able to tweak something to find a quite spot for best performance.
I going to work on filtering after the front end amp. Not sure I really need all the bandwidth I have now.
One reason I like to build my own detectors, can't stand a AC coupled detector whether pulse or vlf.
I know the down sides, but I love the direct coupling no sweeping required, I can even jab em in between rocks and bushes (prospecting). Need low offset drift amps and still adjust once and a while, not something you would sell to the general public. But perfect for me.
I just see this warbling as a limiting performance factor in pulse induction metal detectors.
Thanks again
JC

 

[Anonymous]

Hi JC,
Interesting that you use a fully dc coupled system. Some of my earlier PI

 

[Anonymous]

Hi Eric
Yes I get some warm up drift and some drift throughout the day. I use Burr-Brown OPA4137 low power quad op amp for everything past the front end amp. Some of these modern amps don't drift near as bad as the older ones used to, even with large value resistors. The front end is set up to use lm709, ne55324, lm318, or other favorite flavor. Max352 (pricey) analog switches.
For ground balance if you can call it that, I am running two channels of diff integrator and about 6 pole low pass at 0.7 Hz to two meters. The two channels have different sample timing (say 15 us and 25 us). I have to use my brain from this point to notice how the meters move relative to each other and try to figure out if it is ground making them go up or a target.
I have used IRF740s for alot of things and its the only Fet I have tried. Haven't given it much thought but you make a good point. Using your suggestion of the fast blocking diode to block the capacitance of the fet. Lots of current through the diode and fet and these junctions are cooling down during the receive period, may have something to do with it.
Pretty much dirt simple, could blend the outputs of the two channels in some fashion I guess, but would have to get the gains just right or probably goof things up. I fact I don't think I can mix these two channels linearly, probably some log function in there some where, but don't know.
I am open for ideas for ground balancing as I am not sure how to do this well. Works in mineralization ok but not in iron. Iron is very strong in one area I search and it is a problem.
AC coupling after the integrator on vlf detectors would let you add another gain of 10 amplifier onto the detector (more sensitive). The same thing without AC coupling will drift around like crazy, but if you don't mind adjusting offset once in a while and use better amps then what comes in commerical detectors, it sure is nice to always be dc coupled after the integrator. Not so bad with meters where you only care about movment of the needle rather than absolute position, thresholds and beeps gets messier where you would have to follow the drift. Downside is you have to keep looking at the meters while you search. Upside is don't have to sweep back and forth ten times to try and figure out what is down there. Especially in areas where you can't sweep at all.
I know the AC coupling time constant was used (still is?) as a form of ground balancing/rejection since the target "should" change sharper than the ground will, and therefore be passed better. I lose that feature.
Drift with the pulse is not so bad right now (maybe don't have enough gain) and sensitivity is a bit more than I read on the forums, but my TX current is probably higher than the commerical units too.
As far as the warbling is concerned some things bother me that makes me think this is not all external. Certainly I can tell if the television is turned on or off from across the room, but yes the warbling increases if the transmitter is on vs off, and the other thing that bothers me is that adjusting the transmitter periods slightly seems to have a greater effect than just adjusting sample spacing slightly. If it is truly demodulating external noise this shouldn't make a difference. This "noise" is normally about 10 mv to 100 mv on the output of my detector. Not so bad but love to add another gain of 10 to the detector. Certainly easier than bumping the TX current by a factor of ten.
Turn the transmitter off and disconnect the coil and the electronics is quite as a church mouse. Still investigating this beast.
Thanks again,
JC