Circuits for magnetic pickups

[Darrell Norquay]

At 09:16 PM 8/2/96 -0700, Sandy wrote:

>Do I need to ground the - on the sensor at all? If I do, don't I lose any of
>the balanced input benefits that I'm trying to get? If I grounded the (-) at
>the opamp, I would guess that I would lose ALL of the benefits of the
>balanced inputs.

Ideally, you don't want to ground either side of the sensor unless the
sensor requires a return path to ground (ie a Hall effect sensor as opposed
to a magnetic reluctor type pickup).  Your assumption about losing the
common mode rejection is correct.

>Also I messed around with some input voltage limiting, and found that zeners
>won't work, as they are too slow. I think however, that the TVS's will work
>like a champ, with only a few ns to clamp.

I don't agree generally that Zeners are too slow, the are only a few percent
slower than tranzorb type devices.  Zeners should be more than adequate for
automotive use.  Make sure the voltage is high enough to prevent shunting
your signal, though.

>     
>     |----------------*---------()()()-------*---*------>>> OpAmp +
>     |                |                      |   /
>  [pickup]          [TVS]       Inductors    |   \ Gain   
>     |                |                      |   /
>     |----------------*---------()()()---*---|---*------>>> OpAmp -
>                                         |   |
>                                        --- --- 
>                                        --- --- Small Caps
>                                         |   |
>                                         --|--
>                                          GND         
>
>This is what I was thinking, with some hysteresis on the opamp side, should
>do the trick, but again this is just a guess. What do you think, and what
>problems will be caused by not have a GND in the pickup side?

This circuit should work well, with a couple of mods.  You don't really need
inductors, a couple of resistors will work nicely.  Move your tranzorb
device to the opamp side of the resistors, to provide some current limiting
if the device goes into conduction, this will prevent blowing out the
tranzorb (or Zener).  You may also want to use a pair of diodes to clamp the
input voltage to the supply rails.  Place one cathode to +V, anode to +
signal, and the other cathode to - signal, anode to ground.  This will
prevent high common mode voltages from killing the opamp.  The standard
noise filtering for differential lines usually includes a cap across the two
lines, in addition to the two you have shown from each line to ground.
Calculate the values of the caps from the f = 1/(RC) equation, and make the
frequency cutoff point at least twice your maximum sensor input frequency.
This will give you good noise filtering without affecting your signal.

The resistor you show as "gain" in the above will serve to attenuate the
signal in combination with the two series resistors, eliminate it if
possible.  The classic differential amp configuration usually uses 4
resistors, as below:

                       RF1                     
               |-----/\/\/----|          GAIN = RF/R1
               |              |                 R1 = R2
         R1    |  | \         |                 RF1 = RF2
    ---/\/\/---|--| - \       |
                  |    |------|-------
    ---\/\/\---|--| + /
         R2    |  | /
               |
               \
               /
           RF2 \
               /
               | 
              ---

If you use this configuration, just hang your circuit on the circuit of
above, using R1 and R2 as the series resistors.  If you use a single supply
opamp running off 5 volts, and make the gain high enough, you will get a
rail to rail square wave out of the circuit (you could also just eliminate
RF1 and RF2 and run the circuit open loop, like a comparator, but you may
find that this is TOO much gain).  You can also add some positive feedback
by tying a high value (several Mohms) resistor from the output to the + in
terminal of the opamp.  This will give the circuit some hysteresis.


regards
dn
dnorquay at awinc.com