I've recently constructed a simple but effective lock-in amplifier based on the ADA2200 and an AD8429 pre-amp that works remarkably well with the default RCLK frequency and IIR filter settings. However, I need to increase the modulation frequency to the 20-30 kHz range. To do this I've changed the configuration so the RCLK frequency is f_SO/4, and I've turned off the mixer so that I can sample all four data points and do the mixing programmatically. All of this works fine, but the signal is attenuated by nearly a factor of five in voltage, which is to be expected because the modulation frequency is now at f_Nyquist/2, whereas the default filter is still centered at f_Nyquist/4.
A perfect solution seems to be offered by the BP1 or BP2 filter register tables listed in Mini-Tutorial MT-234, both of which are centered at the correct frequency. However, when I load either of these sets of coefficients via SPI transfers, the ADA2200 analog output goes to zero and stays there. The same happens for the "All-Pass" coefficients. By contrast, if I load the coefficients from the "Default" column using exactly the same code on my microcontroller, the ADA2200 resumes normal operation with its default filter as expected. Is it possible that the table of coefficients is inaccurate? In particular, I've noted that if I change register 0x27 to 02 and leave everything else at the default values, this seems to shut off the output just like the BP1 and BP2 filter settings. Or is there something subtle about the data transfer process? The toggling process for register 0x10 is not very well documented, and might be causing me problems. I was hoping to do some tests using the separate list of All-Pass filter coefficients in Table 8 of the ADA2200 data sheet, but this table is clearly mislabeled, and is simply a repeat of the default filter parameters. Obviously, a little advice would be much appreciated.