Don,
Thanks for the help on this. The project is called Mako, and is a new marine chartplotter. They are using the PCM1804 in the sonar receive chain, downmixing the signal from the transducer into the frequency of our high res PCM1804. Volumes for this single project are up to 240ku/yr, starting in December. If the new sonar topology works, it will be the basis for their new designs moving forward. (This is highly confidential to Garmin.) Hence we need to help them solve this to not only win this socket, but many other projects coming behind it. Here's a description of Garmin's problem:
We are seeing some ‘noise’ relating to the differences between the + and – outputs of the differential path (THS4521 op amp to PCM1804 ADC) and find that we need to ‘ac couple’ the path. This AC coupling is adding in another ‘filter’ that is of concern to us, so we were wanting to model and simulate these effects. (we are guessing that these differences are coming from the different input/offset currents and biasing of the 2 parts – since we are trying to use >100dB dynamic range, these slight differences are showing up as ‘noise’ - - the AC coupling is working, but we need to become comfortable with/fix the frequency response effects)
Maybe if you have a good Spice model of the PCM1804, then that may answer our questions – but it needs to include the input impedances and effects so that we can understand the effects of the ADC sampling times on the rest of the system.
We have built most of this system into a simulator, and are trying to find the proper R and C values to make all of our signals work. Steve’s questions are to finish getting the components modeled such that we can determine the frequency responses that we are needing.