Hello,
Background - I'm using a AIC3254 configured as a simple audio preamp/processor. The design consists of a 11.2896mhz oscillator connected to MCLK (the other clock inputs are not connected), a small PIC microcontroller that transfers code from an eeprom to the '3254 via I2C, serial/programming ports for downloading the code and not much else. The idea was to make something that could run code produced by PurePath for a '3254-U EVM as-is with no changes besides scripted conversions to convert the .cfg file into the format I used for eeprom storage (strip comments whitespace etc but otherwise still the same script). On boot the PIC reads code from the eeprom one w30xxyy.. sequence at a time and sends it to the '3254 (including interpreting w30fexx as a millisecond delay). The MCLK oscillator runs all the time. The circuit is built on perfboard using an adapter board for the '3254 - bad I know but for testing I don't care about noise, just want to get the design right before committing to foil.
All that stuff works fine, with serial debug functions I can dump the '3254 registers and it matches the script values. Except it doesn't work. Here's the odd thing and the question - discovered that if I touched the BCLK pin it suddenly came to life and worked fine, and was surprisingly quiet considering the way it was built. Further discovered that to get it working without human digit intervention I had to connect BCLK to anything that oscillated for a bit (doesn't have to continue oscillating), I have an LED that flashes while booting so used that. Here's the really strange part - needed a 100K resistor in series with BCLK. Direct connection or a 470 ohm resistor did not work. I'm glad it works, but I'm puzzled as to why I need to do anything with BCLK at all, and especially puzzled about needing a series resistor.
First question is why? Has anyone heard of anything like this? Possibly due to the poor layout using the DIP adapter?
Next question is what is the proper way to treat MCLK and the clock inputs I don't need?
Sorry for the mystery :-) but I'd kind of like to figure out the observed behavior before moving forward with the design, thanks for any insights.