AnsweredAssumed Answered

Vastly differing results with AD7747?

Question asked by jchang16 on Nov 18, 2015
Latest reply on Dec 4, 2015 by EricC



I am using the AD7747 to measure the height of a MEMS thermal bimorph.  I have a pad on the substrate and a bimorph that moves out of plane.  I connect the pad on the substrate to the Cin+ and the thermal bimorph to ground of the AD7747.  When actuate my device by passive heating with a hotplate, the bimorph bends closer to the substrate and I have been able to get measurements with the AD7747 on the change in capacitance between the pad on the substrate and the thermal bimorph, letting me measure the change in height of the bimorph.  The capacitance is well correlated to the height of the bimorph.  I'm getting changes on the order of hundreds of femtofarads from the highest point of the bimorph to when it is flattened, almost directly on top of the pad on the substrate.  This range is expected from calculations and simulations


The problem occurs when I try to actively actuate the thermal bimorph.  The bimorph is designed like a U so that one end is my voltage bias and the other end is my ground.  When I bias the device (<1V), a current passes through the device, heating it up and it moves closer to the substrate.  I would have expected that the AD7747 measurements should match between the passive and active actuation for the same height of the bimorph.  However, with active actuation, the capacitance changes are less than 5 femtofarads from the highest point to when it is flattened.  Even worse, when I actively actuate the bimorph to nearly flat, I see the 5 femtofarard increase.  But then I passively ramp the hotplate on it while the active actuation is still on, the AD7747 still measures the hundreds of femtofarad increase.  The data looks just like the passively heated scenario, just offset by 5 femtofarads.  This doesn't seem to make any sense since the bimorph already starts out flattened so I would expect barely any change in capacitance when passive heating starts.


I'm wondering if my voltage potential is offsetting the +/- Vin of the AD7747.  If so, is there a way to access +/- Vin so that I can account for my biasing?  Also, any other insight would be extremely appreciated!


Attached is a very crude schematic of my setup.


Thank you,