I have a quite unusual design (high end test instrument in a big rack) with multiple PCBs, backplanes and connectors.
The customer now wants to add a USB C connection / functionality (because it is the future....) from a PC on one end of the device to the user interface pannel on the other side. And he wants to do that over the existing spare transmission lines while changing as few PCBs and connectors as possible.
I don't see it as an active cable, more as an USB C transparent extension with a USB C Plug on one side (connected to the integrated industrial PC) and a USB C receptacle on the user interface panel.
So all we have are 10 single ended 50Ohm (specified +-20%) high speed lines and an available 5V power rail with more than enough (>10W) power. Fortunately the alternative modes are not necessary. So no need for DisplayPort... USB PD is also not necessary, because only USB sticks or Webcams should be plugged in, 5V/900mA for the device without another choices is therefore ok.
The first idea would be to use the 10 single ended transmission lines as 5 "virtual" 90R (including tolerance ;-) ) differential lanes (2x RX, 2x TX and USB2.0 DP/DM) and to emulate somehow the CC1 / CC2 pins and the cable detect / powering procedure without an extra connection.
Would that be possible? I haven't found a solution which gets rid of the CC pins. As I said I would be willing to sacrifice compatibility as long as USB Sticks and Webcams are working.... Can I simply connect the CC1 and CC2 pins with 5k1 to GND on the upfacing port to tell the PC to expect a 5V/900mA device? Or can this create problems with the cable detection / enumeration?
The solution would look similar to 2 TUSB544RNQEVM, but with a lot of FR4 boards and connectors between them instead of the USB C cable and without a direct power / CC1/2 and AUX (SBU) connection
This would be a relatively simple solution with only the redriver and a few passive components.
The next Idea would be to convert / demultiplex the USB C port to USB A (only 3 high speed pairs + the rest of the lanes used for VBUS) with something similar to the TIDA-00891 (USB Type-C Plug to USB Type-A Receptacle), add a USB 3 redriver, go through the device, add another redriver and something similar to the TIDA-00890 USB Type-A Plug to USB Type-C Receptacle (I haven't found a schematic of the design, only a block diagram labeled as scematic is available) to have a USB C receptacle on the user panel.
This seems less "non standard" to me, but also more complex.
I have another question regarding this problem: I measured a maximum attenuation of the transmission line of 28dB at 3.8GHz. If this is to much are there any problems (timing, jitter, bias voltages) to expect if I split up the transmission line and insert additional USB redrivers? I haven't seen designs with multiple USB redrivers in series so far.
