Other Parts Discussed in Thread: USB-2-MDIO
Hi folks-
I'm working on an MDIO protocol decoder to be integrated into Keysight oscilloscopes. I have this EVM as a sample DUT and observe that on read transactions (using the USB-2-MDIO utility) that the MDC signal behavior is not compliant. For a basic read transaction there should be 64 MDC rising edges but there are only 63.
I've pasted a sample capture below. Note the pulse count measurement showing only 63 pulses. Also note the change in clock duty cycle about 15 bits into the address & command part of the packet. If you zoom in close enough then the setup & hold rules are no longer valid from that point on, the data edges coincide with the clock edges. I've put in a second screen shot showing the clocks occurring after the 32-bit preamble, with bookmarks showing the site of what appears to be where the clock goes wrong.
Has anyone ever reported this? Does anyone care enough to work with me to resolve it? I'd use this DUT as an example in our application notes if it gave a good example of the protocol.
Thanks a buch,
Scott Ferguson
Keysight Technologies
sferguson@keysight.com