Suggestions/critique on 1MHz eddy/inductive prox. sensor.

Discussion in 'The Projects Forum' started by RRThygesen, Oct 31, 2012.

  1. RRThygesen

    Thread Starter New Member

    Oct 31, 2012
    Hi everyone.

    Im studying mechatronics, and have a habit that every semester needs to have a electronics project, and my current project has blown the gantt chart out of the water, but I still need to get this done right, so I was hoping for some comments before sending this of to the lab.

    What I have built so far, and what also seems to work with some noise added, is a inductive/eddy sensor, where a 1MHz sine drives a wheatstone impedance bridge, with a sensor coil and a ref. coil for temperature compensation. The imbalance of the bridge is measured with an inamp, with unity gain. This is gained by 20, and then 20 in a precision rectifier. The signal is the filtered in a 4th order butterworth LP, with a -100dB gain @ 1 MHz and roughly 40us settling time. Then the DC is picked up by another inamp, for offset and final gain. All ICs are decoupled with 100n from +15V and -15V to GND, and "reservoir" capacitors has been added in my this version, so I'm sure there's no power shortage. The system is driven with a function generator and a bench power supply.

    I have built a total of 3 prototypes, one on breadboard which convinced me that I could actually pull it off, and then 2 on PCB that had several faults, some related to design and other due to my inexperience with EAGLE. Therefore Ive had some difficulties finding the exact causes of the problems: Noise in my final signal, and the NOW IT WORKS/NOW IT DOESN'T WORK syndrome, where nothing is changed, except for the days left to deadline..

    If you would please look through the latest version, Ive also included the PCB layout since this obviously is a real performance killer in AC-applications (which I was totally unaware of initially).

    Any comment, proposal and help I can get to prevent noise, and make this robust will be really appreciated.

    Thanks in advance.
    Last edited: Oct 31, 2012
  2. RRThygesen

    Thread Starter New Member

    Oct 31, 2012
    I forgot to mention a important thing: The board will have ground plane on both sides, but that is not visible for practical reasons.
  3. Sensacell

    Well-Known Member

    Jun 19, 2012
    At 1 Mhz I doubt your PCB layout is the problem, this is a rather modest frequency and your board is small, currents are low, etc.

    I would look at the design fundamentals.

    What does your excitation signal look like? Amplitude and DC offsets?
    The sensing coil? Shielded with Faraday shield? How big is it? What are you sensing?
  4. Thygesen

    New Member

    Dec 2, 2010
    @Sensacell: Youre right, its only medium frequency, and from what I read designers start worrying at high frequency, with decoupling etc. Excitation is a sine with an amplitude of 1.2V, running aprrox. 10mA through the bridge. I have huge plan for my sensor coil, but so far I have only tested with ~10uH 3ohm coil on my pencil, which seems to work for testing purposes. I plan to make a more horseshoe shaped sensor, but round, so the target material closes the magnetic circuit if it touches. The target is steel.

    What about the resistors used for the opamps? It works flawlessly in simulation, but could the matching of the impedances cause some problems?
  5. nerdegutta


    Dec 15, 2009

    Perhaps a little off, but...

    Have you considered to use SMD resistors, since almost all of them are one the top layer?
  6. Thygesen

    New Member

    Dec 2, 2010
    Whoops, I apparently made a ****up with 2 accounts, dont know what I was doing last night.

    @nerdegutta: I'm primarily mechanical, so I have no idea how those little SMD work, I'm guessing theres only space for me to save with SMD components? I like the fact that I can use a drill to make PCB, its in my comfort zone :D
  7. John P

    AAC Fanatic!

    Oct 14, 2008
    Is an instrumentation amplifier really what you want there? I see that the AD620 gives an output that's 3dB down at 1MHz, though I don't know how accurate that figure would be. Generally an instrumentation amplifier is used to give a very accurate output representing the difference between two inputs, and this doesn't seem as if it ever can.
  8. RRThygesen

    Thread Starter New Member

    Oct 31, 2012
    @john. Thanks for the input, I chose the inamp for its high input impedance, and way back I think i started with a difference opamp, but changed for the inamp. I want the difference in the bridge, and youre right there is a slight attenuation @ 1mhz, but it produces a fine signal. Did you have other options in mind?

    General: I forgot to say that the opamps are actually ad844's, and ive checked that gain BW and slew rate are sufficient, and it also works - sometimes :D