Ideas for Inductive Communication Receiver

Discussion in 'General Electronics Chat' started by Phil_84, May 21, 2015.

  1. Phil_84

    Thread Starter New Member

    May 21, 2015
    6
    0
    Hi All, I am building a system to send data using a magnetic field over a range of about 1 - 2 meters max. I have a transmitter coil resonating at 140KHz approx 30cm diameter, built using H-bridge driver and micro-controller. The receiver coil is a 15mH wire wound SMD inductor, with parallel cap to tune it to 140KHz. I have a resistor in parallel too to act as a small test load of 500K ohm.

    Using the oscilloscope I can detect the clean 140KHz sine wave easily around 1m away which is around 20mV pk-pk. Getting closer to the transmitter the amplitude shoots up to around 2V pk-pk.

    So far so good. My next step is to amplify this signal, which is where I am struggling. The receiver needs to operate at 3.3V, and I need to amplify the signal to ideally a on/off square wave. Then I can easily filter and modulate the 140KHz to transmit some data.

    I have tried a op-amp microphone circuit, but it didn't work. It needs a high input impedance of around 500K to 1M ohm so as not to dissipate the received energy. C2 couples the signal to the +ve input via a voltage bias since it is AC.

    With a 1NF cap between R4 and ground I can almost get the input signal out, but it has a small negative gain. I need to try and make this positive. The IC is wired up correctly I have checked many times.

    circuit.png

    Any ideas would be welcome. Even if I can get the amplifier working, I will prob then need a comparator stage to get the desired on/off signal out.

    Cheers
    Phil
     
  2. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,516
    1,246
    The DC potential between the opamp + and - inputs is 1.65 V, and the opamp is trying to multiply this by 6.

    1. Add a 0.1 uF capacitor between R4 and battery-. This reduces the opamp gain to 0 at DC, so the opamp output should be sitting a 1.65 V + 6 x Vio (opamp input offset voltage).

    2. Increase R1 and R2 to at least 1 M. This presents a Thevenin equivalent resistance of 500 K to the input circuit. What you have now is loading the tank with 50 K.

    3. Check you opamp's input bias current spec. With such a high resistance bias network, bias current through the resistors will create offset errors.

    3. Make sure your opamp has enough gain-bandwidth to handle the signal. Your requirement is 16 dB at 150 kHz, plus you want at least 10 dB and should have 20 dB over that for good closed loop stability. So the desire is for an opamp that has an open loop gain of at least 36 dB at 150 kHz while running on 3.3V. Combined with low input bias current, this will be difficult to find.

    ak
     
    Phil_84 likes this.
  3. alfacliff

    Well-Known Member

    Dec 13, 2013
    2,449
    428
    how about getting rid of r1, r2, and c2. then connecting the tuned circuit directly to + and - inputs? the input impedance of the opamp should be sufficient to not spoil the Q of the tuned circuit.
     
  4. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    What the Kid said...
     
  5. MikeML

    AAC Fanatic!

    Oct 2, 2009
    5,450
    1,066
    There still needs to be a path from the opamp inputs for the opamp's input bias current, and the inputs still need to be biased to about Vdd/2, so there have to be some resistors, somewhere.
     
  6. alfacliff

    Well-Known Member

    Dec 13, 2013
    2,449
    428
    wouoldnt there be a path from the output and ground for both the + and - using r3 and r4?
     
  7. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    This might be worth a try. Input impedance=1meg.
    InverterAmp.gif
     
    Phil_84 likes this.
  8. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,516
    1,246
    Alec - don't think so. When acting as an inverting linear amplifier, the input terminal is a virtual ground just like with an opamp. That is not necessarily bad; the input could be run in "current mode" as a transimpedance amp. It is a tradeoff governed by the Q of the inductor.

    Alfaclif - maybe. But on top of all of the other demands on the opamp, it now would need a common mode range that extended below its negative rail. A CMOS opamp could work - except for the absolutely terrible bandwidth.

    ak
     
    Phil_84 likes this.
  9. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    You may have difficulty with the SMD inductor, one common application is in switch mode power supplies, so they're generally designed to minimise radiation - that particular quality works both ways.
     
    Phil_84 likes this.
  10. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    Oops. My bad.
     
  11. Phil_84

    Thread Starter New Member

    May 21, 2015
    6
    0
    Thanks for all your feedback and ideas. I will have a go at making a few changes to the op-amp circuit you have suggested. I also tried feeding the ac signal biased to 2.5V into a LM339, with 2.5V on the other input. This sort of worked but was quite a noisy digital output. Might try a few different comparators, think the 339 is quite old now.

    I considered the fact the SMD inductor probably isn't the best a small air core coil worked better, although the one I am using is less shielded than many others. It does work quite well for the sort of range I am after + benefit of off the shelf / small form.
     
  12. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    Here's another option. This time I'm convinced the input impedance is ~1meg :). Discrete components avoid the probable difficulty of sourcing a wide band-width opamp workable with a single supply at only 3.3V.
    FET-bufferedAmp.gif
    J1 is a high-impedance unity-gain buffer. Q1 provides sustantial gain and is base-biased so that its output is centred around mid-supply. U1a provides further gain, sharp switching and a logic-level output.
     
  13. RichardO

    Well-Known Member

    May 4, 2013
    1,230
    382
    Doesn't the 2N3819 have too high a "threshold voltage" (Gate-Source Cutoff Voltage) for a 3.3 volt circuit?
     
  14. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    Simulation shows it's ~-2V but current dependent. Given the sample-to-sample variation, you might need to strike lucky. I daresay there are more suitable JFETS with lower thresholds (I haven't checked).
     
    Last edited: May 22, 2015
  15. Phil_84

    Thread Starter New Member

    May 21, 2015
    6
    0
    I have made some progress. Tried a few JFET buffer circuits with no luck, so went back to the op-amp etc.

    First interesting solutions was to feed the LC circuit directly into a micro-controllers AD converter. Sample rate is not high enough, but you can use it to detect when it is around 60cm from the transmitter as it is quite sensitive like Oscope. Its input can also cope with the -ve part of the waveform.

    Secondly I built a two stage op-amp circuit below which seems to work. Total gain is only about 10.

    Amp 1 is basically a buffer. I am not sure about the following;

    1 - I only added C1 after I found the circuit didn't work when i removed the Oscope probe from this point (C2 / R8 junction). Seems 68pF ok value here. Not sure why this is needed to make the circuit work.
    2 - C3 should be 68pF for resonance at 150KHz, but only works with a 15pF cap of different type.
    3 - Making R3 any bigger has no effect on the gain of the second stage.
    4 - Making R1 smaller has no effect on the gain either.
    5 - Would like a gain greater than 10, but that seems to be the limit for some reason.

    circuit2.png
     
  16. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    Its possible to do better with a low VGSthr TO92 style MOSFET, but you need a source resistor to develop a voltage proportional to current, and a bipolar transistor to sense that voltage - the collector terminates the bottom of a bootstrapped voltage divider that feeds stabilised bias to the gate.

    As the source current increases, the bipolar transistor starts to conduct more and pull down the voltage divider that feeds the gate.

    It also needs a strategically placed electrolytic, otherwise you get AC nfb and not much gain.
     
  17. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
  18. Phil_84

    Thread Starter New Member

    May 21, 2015
    6
    0
    Hi, I have been going over some op-amp basics and what AnalogKid said about gain-bandwidth product which went over my head at the time.

    The op-amp I am using only has a gain bandwidth product (GBWP) of 1MHz or 20dB at my freq of operation, probably due to its low power / low 3.3v operating voltage. This is prob why I am having difficulty amplifying this signal past x10 at best.

    I can get away with lower carrier freq, so will re-calibrate the transmitter and receiver LC circuits to 10kHz to test the theory. Should have have a gain upto 40dB at that freq.
     
  19. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
  20. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,516
    1,246
    If the circuit is built exactly as drawn, it will not work. The + and - inputs of both opamps are reversed. The Vcc/2 bias voltage goes to the + inputs.

    Also, add a coupling capacitor between the first and second stages to prevent your gain stage from amplifying DC errors.

    With these corrections you should be seeing a gain of about 15, not 10.

    ak
     
Loading...