Other Parts Discussed in Thread: LM2776
Hi,
We have been developing some sensitive (low noise) RF circuits for radio astronomy receivers and we need to use RF-over-Fibre which is supplied by a company in China. Unfortunately the company as helpful as they have been, have used a negative DC supply on their daughter boards. The original -5V chips (LM2776) are very noisy producing 2MHz charge pump spikes which couple into the RF band (50-250MHz). I designed a DC filter circuit using the LM27761 which should be better for noise. The circuit is shown below and basically uses a combination of common mode and normal mode chokes. The ground plane for the LM27761 chip is separated from the analog ground and only couples via chokes. Unfortunately performing sensitive measurements, we still pick up the spikes although they are dramatically reduced. We are now in a very difficult position because we need to go into high volume production soon and really need to fix this. Can anyone say what we are doing wrong. Is there an alternative TI part we could use to kill this noise? Help is much appreciated.
Kind regards, Nima
