So why wouldn't other chip vendors move to POF instead of offering PHY solutions for copper? We asked.
"There is a lot of know-how in getting the signal right on POF, because there are more losses than on glass fibres. We use pulse-amplitude modulation with 16 discrete levels (PAM-16), we must filter out noise and equalize the channels. At the beginning, there was a couple of other companies working on this topic, but they left the market", Ciordia said.
At the Ethernet & IP @ Automotive Technology Day, KDPOF announced two new transceivers, KD1051 and KD1052, implementing the physical layer of Ethernet to transmit data at 100Mbps over standard SI-POF (Step Index), MC-POF (Multi-Core) or Plastic Clad Silica (PCS). Both chips support different standards for the host digital interface, including RGMII, RMII, MII and, for the KD1052, 1000BASE-X and 100BASE-X. A Serial Management Interface is integrated for configuration and monitoring.
What's more, these chips can also operate in 1Gbit/s mode for evaluation purposes and the company has already unveiled the specs of its next chip, an automotive GE-POF transceiver designed for normal operation at 1Gbps (over 40 meters of POF). Designed in a 65nm CMOS process, the automotive grade KD1053 will be sampling in 2017 in a 56-pin QFN 7x7mm package.
On its roadmap, KDPOF sees 10Gbps on POF the next logical step. "It is a bit more complex and we'll have to use graded index plastic optical fibres instead of stepped index fibres", admitted Ciordia, but we see the need for 10Gbit/s automotive Ethernet by 2025 and we hope to be ready by then."
Without naming them, Ciordia told us that a German car OEM would use KDPOF's transceivers in production vehicles starting 2018, while another unnamed OEM had planned its move to POF for 2020. "Even at 100Mbps, they are all coming to us and asking if we can solve their EMC problems. One of the projects we are involved in is an electric/hybrid vehicle."