Best way to pick up 8.2 MHz EAS signal

Hi!

I am working on an EAS (Electronic Article Surveillance) design. The
objective is to design an active tag that a) beeps (screams) in the
presence of the 8.2 MHz field generated by the surveillance antennas
_and_ b) triggers the conventional alarm mechanism of the system.

To detect the RF field, we have built an LC tank where the inductor is
made of some turns on a FR4 PCB (half on each layer) spiraled around a
ferrite bar inserted through the PCB. This also triggers the basic alarm
of the system.

The question: Which is the best way to increase sensitivity of this
setup? [Sensitivity is ok by now, but we may expect that in the future
someone expects "more"]

* We can increase the number of turns and decrease line thickness ("y"
dimension is limited). This will increase inductance by a factor, but
will increase resistance due to the increased length and the decreased
cross section. Will there be net gain? Skin depth is roughly 20um at
this frequency and so does not affect on 35um copper.

* Widening the loop in the "x" direction keeping the same ferrite? This
does not make sense, since the effect of the ferrite gets more "diluted".

Additional constraints: power consumption has to be kept below 10 uA
(3V) when not beeping.

Any suggestions/pointers are welcome.

Pere

 

1. The ultimate limit of the sensitivity is set by the ambient RF noise.
However it is unlikely that you are approaching this limit with the
setup that you described; it seems to be self noise limited.

2. The amount of energy captured by the antenna is coarsely proportional
to the volume of the ferrite. Without increasing the size of the ferrite
in either dimension, you can only optimize the parameters of the
winding, the input amplifier and the matching between the antenna and
the amplifier for the best noise performance. Depending on the current
situation, that optimization may or may not provide for the substantial
improvement.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

Thanks! This seems an useful rule of thumb indeed. Is there any
reference on this so I can try to grasp the reasoning behind this?

In the current (very crude) configuration the tank is directly connected
to the BE of an unbiased BFR92. Field strength is now sufficient to turn
it on. Inductor reactance is aprox. 600R (12uH) and the inductor Q is
approximately 120. Any guess on the input impedance of the unbiased
transistor, so that I could try a better match?

Pere

 

Besides antenna/ferrite mods, what is the noise and impedance matching
situation of your current front end? If it falls into the category of
"not so great" you might want to consider a switched pre-amplifier.
Switched in order to make the 10uA because you can't really get a good
noise figure at quiescent currents this low.

The feature you need is nowadays called "wake-on-radio" or WOR. It turns
on, sees if RF is there, and if not goes back to sleep.

 

The current configuration could not even be called a "front end" but is
a direct connection of the tank to the base of an unbiased BFR92! I
could design a superregenerative receiver as the front-end (however,
going down to the few uA that are available would be hard) but first I
would like to exploit the passive possibilities that might be available.

If, as Vladimir says, there is little optimization possible without
increasing ferrite volume (which is not possible due to size
constraints), perhaps there is a matching circuit that might do the job.
I have no idea what the unbiased BE junction of the BFR92 looks like
(nothing in datasheet), however I suspect it will not be easy to match
to the tank. If it's smaller than the tank resistance at resonance, a
small series capacitor might do the job, however there should also be a
path for dc... Any comments?

Pere

 

bias up the base of the BRF92 to about 0.5V DC so it is closer to
threshold...then it will take less RF to turn it on...

or add an impedance matching network to step up the RF voltage applied
to the base...the base is a Hi Z when off...

or both

Mark

 

Yep. Without any bias it'll be deaf like a door knob. But DC-regulate
that bias, i.e. via an emitter resistor. Else it'll drift all over the
place with temperature.

I'd say "and" add such a network. Sans bias there isn't much hope here.

 

IME it really never made much of a difference whether a FET or BJT was
used. Not at those frequencies. For really, really short antenna stubs
it might though. But it's got to be biased, else you almost need a
nearby lightning strike to see some signal.