Hi Group,

One of our line driver chip went thru a lightning cable injection noise test which is modeled as a up to 2MHZ 10Vpp sine wave generator coupled to the chip thru a series of 37.5Ohm resistor and 0.1uF capacitor. At low noise frequency, the RMS impedance is limited by the capacitor but at frequency ~ 100KHZ, only the 37.5Ohm is limiting the current. The output is driven by a fully differential power amp. DC output resistance is low due to regulation of the output of the power amp. However, when the noise applied to the output, the output resistance of the power amp needs to inicrease to protect the chip from over heating since the noise source is forcing and injecting current from and into the chip. There are several solutions to this problem:

1. Insert a resistors ~100Ohm from the amp output to the load, but this reduce the headroom for supplies.

2. Use a thermal sensor to cut down bias current to the power amp but this will stop data from transmitting correctly.

3. Use inductor to the output but this is imposible to build on chip.

Anyone has any suggestions? If there is any confusion to the problem, please reply to me : davidday@deiaz.com

Thanks


David Day