Hi everybody. I'm relatively new to PCB design, so I hope somebody can help me. I designed a relatively large (12x12 inches) 4 layer circuit board with internal power and ground planes. The board is standard .062inch epoxy-glass FR-4 with 1 1/4 oz copper thickness. The circuit is a 52 channel current supply using a simple ADA 4610 quad opamp connected to PNP BCX-51 bipolars with a 10ohm sense resistor providing the current feedback voltage. I used the multi-layer board because I wanted to simplify the provision of power and ground access points across the board. I did NOT break up the internal planes - they stretch across the entire board. The circuit has a switch between "setup" and "operate" mode with an on-board resistor to provide a load for setting the current in advance. When the current is correct, the user switches to the external load circuit to start testing. Trouble is: everything works great in "setup" mode, but in "operate" mode, the active current driver oscillates like crazy at about 10 MHz - highly distorted waveform. Voltage across a 5ohm load is 30V p-p! I suspect the large ground plane of the ground plane is providing unwanted cross-coupling between input traces and output, and I'm planning to re-design and break the power and ground planes into smaller strips carrying power and ground only to the correct pins.
More information -- my client gave me a truly messy tangle of about 28 wires to connect to external loads for testing. You can easily demonstrate that the messy wires are involved in this problem by manipulating them and seeing the oscillation waveform changing amplitude and frequency. The actual test circuit will use better designed ribbon cables to connect to the test bench, but I want to make sure my re-design is as trouble-free as possible.
Can anybody offer any expert advice on this problem and my tentative solution? I'd be happy to provide more details.