I am working on a design where a DP83822H PHY is capacitively coupled to a KSZ8567 Ethernet Switch. The ethernet link works correctly at ambient temperatures but fails at -40C so i am trying to get to the bottom of why. Our software team is using Linux and forcing the interface to 100 Mbps fixed full duplex. However, when I measure the signals I see a peak to peak of 2.5 V pk to pk.
I was reading AN-1159 for a different TI PHY for transformerless ethernet operation and the application note discusses the expected voltage levels, roughly 2.5V pk to pk for 10 Mbps and roughly 1V pk to pk for 100 Mbps. Since the voltage level I am seeing is contradicting the Ethernet Spec and this application note, I am wondering if the PHY is really still using 10 MBps. Are there extra steps to perform by setting the bias voltage and Link on the PHY?
I have attached a scope shot of the Rx+ vs Rx- I captured, the application note and the portion of the schematic of interested. the PHY is U31 on the left corner of the first page.