ADA4940 Model misbehaving.

Hi,

I have a problem with modelling a circuit using the ADA4940 as an ADC driver, with a AD8235 instrumentation amplifier preceding it.

Whilst each device models correctly by themselves, done with ADA4505 pre-amp stages being driven by sine & cosine waveforms (simulating accelerometer inputs, in a constant velocity cyclic pattern), these being combined at the inputs of the Inst Amp or ADC driver.

When the Inst Amp is used to drive the ADC driver (for S/E to bidirectional adaptation for ADC), the result is that for the ADA4940, the +ve node output, +ve feedback pin and +ve node input are pulled low, and the the current draw on some nodes in the ADC driver are showing extremely high (kilo-amps....) in the node list from the simulation tool.

Attached are .jpg files of the cricuits (ADC-driver alone & combined with Inst Amp), screenshots of the output waveforms for both scenarios and the nodes "voltage & currents" results lists of both scenario's from the DC Conditions analysis.

I have not been able to figure out why this is happening, not sure if there is something strange with the Analysis tool (TINA Ver 9), or someting with one or both of the models (ADA4940 & AD8235). Both .cir files used were downloaded from the Analog devices website in the past 24 hours.

Any advice or assistance in solving this would be greatly appreciated.

Cheers.

AD4940_ADA8235_problem_1.zip
  • Hi Garth,

    I'm not sure if I completely understand your issue, but from looking at your schematic,

    Here is what I see: The dc voltage at the junction (R22 and R32 should be the same as

    the dc voltage at the junction (R21 and R25). If this is not the case the two outputs of the ADA4940

    will move away from each other from one direction of the other depending on the voltage orientation.

    This can cause possible railing which can lead to high current draw.

    Can you check if this is part of the issue?

    A good place to check what I'm talking about is using our diffampcal tool, which is very easy to use and you can download it for free.

    Chalry

  • Harry, the circuits shown are only test circuits, to simplify the problem - not the full circuit. RL for the 4505's is approx 200k in the main design.

    The LTC1078 is used as a simple simulation of the front-end of the ADC, LTC1408, the give a load for the ADA4940.

  • Hi Charly, I assume you mean R23, not R32.

    Using Thevenin calcs and following the recommendations in the datasheet (page 24), for the termination of a single-ended input for the device, I got a value of Rts (R25) of 220 Ohm, but then by experimenting to bring the two node voltages as being the same, it is in fact around 2,2k, ten times larger.

    Not too surewhat it is, but withthe larger R, I have got the combination to work in the simulation, now to try and get a similar result in the full circuit.

  • Yes you are correct I did mean R23.

    The point I'm making is because I don't know what the DC voltage you have at IAIN- and IAIN+ and because you have R22 tied to Vcc, you might end up with difference DC voltage then on the ground side of R25.

    So what you can play with is adding a DC source between R25 and ground and adjust that voltage until the differential

    output is correct.

    Charly