Looking for switch that supports C-PHY protocol

Hey Archie,

Quick question here. Could you expand more on the 4.5GSPS? This is across the entire link (4 differential channels) right? What's the actual bit rate that will be seen per data line? Or Is this 4.5GSPS per line?

Thanks,
Rami

Hi Rami,

Thanks for your guidance. I received feedback from customer:

[1]Could you expand more on the 4.5GSPS---I didn't understand the question.

2]This is across the entire link (4 differential channels) right? --Yes,or 3 Trio.

[3]What's the actual bit rate that will be seen per data line? 4.5Gsps(Max)

Thank you.

Regards,
Archie A.

Hey Archie,

So for a short answer, the TMUX646 would have a higher bandwidth (6GHz) and would be a good option here and I believe it should work but would need some clarification on the bandwidth spec provided to be able to say this with more confidence. I have gone in further detail below. I do believe that the TS5MP645 may still work in this application, but once again, dependent on some clarifications.  
.

ArTzy said:

[1]Could you expand more on the 4.5GSPS---I didn't understand the question.

ArTzy said:

2]This is across the entire link (4 differential channels) right? --Yes,or 3 Trio.

[3]What's the actual bit rate that will be seen per data line? 4.5Gsps(Max)

There seems to be a little contradiction here. It's unlikely that it's 4.5Gbps transfer rate across the entire link (4 differential or 3 trio) and 4.5Gsps on each individual line. If each data line is 4.5Gsps then the rate would be significantly more unless they're encoding this is in a very inefficient way.

To be safe and clear on my definitions here, a data LINE is one individual signal path. 1 data LANE will be comprised of 3 data LINES. The LINK will be the all the lanes together. My feelings here are that the LINK should support 4.5Gsps but please confirm and correct me if i'm mistaken.
Please see [FAQ] How do I Select the Right Analog Switch/Multiplexer Bandwidth Based on the Data Rate of my Protocol?.  for more details on this. However, this may be slightly trickier with C-Phy given the transmission in bits to symbols. 
I've also drawn this figure to maybe help us with communication here as we talk about the spec

My interest is the bandwidth needed per LINE. As I mentioned above, it was stated in the response that 4.5Gbps was across the entire link and per line which wouldn't make sense. So if you could clear that up, it would be extremely helpful. 

I'm understanding this as they have 9 lines (3 trio) and a total of 4.5Gsps will be read across these 9 lines. I'm no expert in C-PHY but this is my understanding; C-phy will use encoding to pack 16 bits to 7 symbols so each cycle will encode 16/7 of a 'data bit per lane'. This will be our bit rate per lane. Therefore, 4.5Gsps / 3 lanes / (16/7) would be he bandwidth per lane. So you're already looking at <700MHz there. Across three lines per lane would give you more than enough bandwidth in the TS5MP645. 

If my understanding is wrong and it's 4.5 per line than the TS5MP645 wouldn't be able to support this. 

The being said, if it's 4.5GHz per link or per lane the TMUX646 bandwidth should be enough here. Per line may be pushing it a bit here. Typically I recommend selecting a device with a bandwidth 1.5x the signal frequency though so you would be slightly exceeding the recommendation (4.5Ghz x 1.5 = 6.7GHz bandwidth) by just a bit. 

Thanks,
Rami.