Another foray into Multi-loop Coil Antenna design

[mgburr]

I'm still working through some of the transmit refinement for our antenna design. I've had a chance to actually see what is being measured by using a Tektronix Spectrum Analyzer. Definitely an eye opener. Now comes another interesting question about our antenna system. If you check my original thread,

http://www.eng-tips.com/viewthread.cfm?qid=231049

, I drew up a representation of the receive antenna(similar to a Magnetic Read Head) design. Currently we are using a static 5V DC derived from 24V and a voltage divider, to monitor a break in the wiring. We are blocking the DC from the rest of the receiver section with polarized tantalum caps to reduce issues with specific frequencies becomming unstable. What my question comes down to is this.

Does any one know, or had experience with, the affect of leaving a static DC potential on a large inductor and then bombarded it with a bipolar pulse from another antenna. I've noticed that with the antenna monitoring circuit enabled in the system, I can see a fairly vigorous transition on the supply system when our LCD backlight extinguishes, along with the output of the post processing stage amplifiers. When the monitoring is disabled, I no longer see any transitions in the power supplies, or the post processing during the backlight transitions. Because we work with such low frequencies, and the method to detect the metal is treated as a slower(5-12Hz) sine wave. I'm unsure of a viable way to shunt the current spikes to ground while still enabling the monitoring and detection signals to properly pass. Realistically I'd like some ideas as to what effect the power supplies can be seing from the static poential resting on the antenna when it's being bombarded with a frequency, and if it's possible to filter that response out.

 

[VEBill]

Extremely generalized answer: So long as a system stays linear, then each frequency (including 0 Hz, i.e. DC) will remain at their own frequency and not mix. In other words, they're easy to keep separate in the frequency domain.

For example, if you ran DC through a coil with a ferrite core, it might drive the core to a point where the signal of interest pushes the core into a non-linear region.

Based on the description of your system, it seems obvious that a modest DC current through the core is likely to have no impact.

 

[mgburr]

VE1BILL, thanks for the quick look. And yes I shouldn't see any affect with the dc, however reality does intrude. What I'm curious about is with the minor ripple from the SMPS sources sitting atop the reference voltage. Could these be setting up a small magnetic field that when current is surged on the 24v system, it causes a small signal throughout the processors that gets amplified enough to finally seen with an O-scope? I've done som physical testing, and seen that by removing the monitoring voltage from the receiver antenna, the output signal no longer jumps with the LED transitions. But when it is put back in, shows a minor transition when the backlight transitions. It also shows a small bump when pressing keys that are using 5v pull-ups and supply a ground indication back to the Microcontroller. I have to amplify the output from the detection circuit by approximately 300 times to visualize the transitions.

 

[mgburr]

VE1BLL, sorry about the typo, responded on my phone and it decided it wanted BILL instead of BLL.

 

[VEBill]

Another very generalized answer: when dealing with very small, weak signals, then you need to avoid introducing noise - specially if the noise gets into the same frequency range as the signal you're trying to detect. Power supplies should be as clean as you can make them. There's no reason to allow self-made noise to propagate around the system.

Usually when a system is spread out over a large area, from conveyor belt to user interface, the subsystems could be distributed in a manner such the the user interface (in the office?) would be very far away from the signal processing subsystem (adjacent to the coils). Inherently designed so that all the signal processing circuits have a nice quiet shielded chassis all to themselves. It seems like this system isn't designed in that manner.

 

[mgburr]

The electronics are in enclosures, and depending on the application, may be as close as 15ft. We do see external noise from variable speed motors, CB's etc and try to mitigate those as much as possible. When we setup the transmit/receive pair the pulsed field is nulled. What we are actually detecting is the change in the static nulled magnetic field which produces offsets when a piece of metal comes through the aperture. Which is usually a very small deviation, hence the massive gain through various stages. We do use a time sliced sampling process that reduces the noise received, but this doesn't address the issue we are seeing with the LED backlight and switch actions, and why it dissapears when we remove the monitoring voltage from the receive antenna.

 

[VEBill]

These sound like 'the usual' E3 (a.k.a. EMI/EMC) problems.

The usual E3 problems are related to shielding, "grounding" (not necesarily involving the planet's dirt layer), signal return paths, loop area, unintended coupling, wiring practices, etc.

In the military avionics world we have E3 specialists that are involved from start to finish. They tell (or remind) the system and box designers what to do with respect to such details as (for example) proving 360° bulk shield termination around each mil spec connector (details!).

In your case, based on very limited information, in sounds like you have noise getting on the power supply.

The problem is, the coupling mechanism might be so obscure that you will never notice it - and therefore you will never 'in a million years' describe it in your posts. In other words, we can't see it from here. It's one if the inherent limitations of this medium (must keep it in mind).

Advice: start thinking in terms of weak signal communications systems. Such receivers typically have plenty of shielded modules to keep noise away from the low amplitude front end circuits. Try to isolate the noisy user interface from the signal processing circuits. Maybe use an isolated battery to power the signal side as a temporary test to confirm or exclude coupling paths.

Ultimately, you may have to read up about E3 design practices, or hire a subject matter expert. The short version of E3 is that the control methods are the exact opposite of what's done to intentionally radiate radio signals.

 

[mgburr]

Sounds like NAV/MAR, don't remember Army paying that much attention to things as such. I worked Navy Helicopters for 20 years. I used to have my ASEMICAP login to their resources. The biggest problem with what I'm batteling is, I didn't design it. I'm trying to debug someone elses H/W design and make improvements. There is no doccumentaiton as to "why" they did things the way they did. Just trying to keep the system stable enough so I can re-design the whole thing. I can almost guarantee that I've got a parasitic inductance or capacitance that is contributing to it, just not the resources to be able to track it down efficiently. Most of the grounding and noise reduction issues I've seen are easily traceable. This however isn't duplicated easily. Some antennas work fine, others are sensitive to the fluctuations caused by LED switching. If you were somewhere close I'd even invite you over to see what I'm fighting. But, I definitely value the different set of ears to think about the possible issues.

 

[mgburr]

Just for an update. I changed the values of the voltage divider for the monitoring circuit. Decreased the current by a factor of about four. My monitored voltage surge on the signal conditioner output caused by LED jump dropped from 200mvpp, to 20mvpp. And duplicated it on two other units just to make sure. So it looks like the varying coil size is contributing to the issue, along with the current provided via the monitoring voltage.