Hey all,

I am currently doing a project that uses a 1.25Gbps SGMII Ethernet signal that I get over to other boards through 50 ohm SMA and coax. I'd like to switch over to carrying that signal with USB 3.1+ Type-C cables since it would also allow for power distribution to other boards and a few other benefits. The trouble is in the impedance though.

The signal that I am using is 100 Ohm, SERDES LVDS while USB uses 90 Ohm. Does anyone know of a simple solution (think buffer or a couple passives and an IC) that can address this 10 Ohm Differential impedance mismatch? If not, a complex one? To be clear, I don't want an Ethernet-to-USB converter IC. I will also need to convert it back on the receiving side to 100 Ohm Differential.

Any help or ideas you can provide is greatly appreciated!

-Mezz

 

Are you sure the mismatch is actually a problem? This mismatch may cause some mild reflections which may or may not be a problem. You really need to understand what is actually happening. I suggest either simulation, or experiment if you have access to a VNA.

 

I agree with Papabravo The mismatch should be a problem. I would try it out and nothing shoud be damaged.

 

Corrected about that.
I agree with Papabravo The mismatch should not be a problem. I would try it out and nothing shoud be damaged.

 

Are you sure the mismatch is actually a problem? This mismatch may cause some mild reflections which may or may not be a problem. You really need to understand what is actually happening. I suggest either simulation, or experiment if you have access to a VNA.

Wow, thanks for such fast responses guys! My gut told me no, especially since I saw USB 3.1 has a +-7% spec but I figured I'd go do my due diligence. Glad to see you guys don't think it will be an issue. I might be able to use a VNA from my work, after work hours of course Another question or two if you don't mind, approximately how much current do I need to drive a 1m USB cable and about 6 inches of impedance controlled trace? I'm going to use a TI DP83867IS PHY getting switched by a passive TI HD3SS3212 switch IC, but I'm struggling to find the output current necessary for the job. Originally, I was going to use an 8 inch SMA so I guessed it would since I read that the PHY can support 18 inches. In my path to do the calculation myself, I thought I would find an IC for USB and then see how much they use, subtract the current from the PHY output, and then get a buffer to fill the rest. However, I failed to see how much the output current of the PHY is since I'm not entirely sure what the datasheet is saying under the section 7 "power consumption" part. Any thoughts are greatly appreciated whether from experience or aid in reading the datasheet spec; let me know if I left a detail out or something.

Might be worth noting that I am feeding the Tx into the Rx of another DP83867IS and will be running VDDIO = 1.8V
Links to datasheets are hyperlinked.

-Mezz

 

Wow, thanks for such fast responses guys! My gut told me no, especially since I saw USB 3.1 has a +-7% spec but I figured I'd go do my due diligence. Glad to see you guys don't think it will be an issue. I might be able to use a VNA from my work, after work hours of course Another question or two if you don't mind, approximately how much current do I need to drive a 1m USB cable and about 6 inches of impedance controlled trace? I'm going to use a TI DP83867IS PHY getting switched by a passive TI HD3SS3212 switch IC, but I'm struggling to find the output current necessary for the job. Originally, I was going to use an 8 inch SMA so I guessed it would since I read that the PHY can support 18 inches. In my path to do the calculation myself, I thought I would find an IC for USB and then see how much they use, subtract the current from the PHY output, and then get a buffer to fill the rest. However, I failed to see how much the output current of the PHY is since I'm not entirely sure what the datasheet is saying under the section 7 "power consumption" part. Any thoughts are greatly appreciated whether from experience or aid in reading the datasheet spec; let me know if I left a detail out or something.

Might be worth noting that I am feeding the Tx into the Rx of another DP83867IS and will be running VDDIO = 1.8V
Links to datasheets are hyperlinked.

-Mezz

At the speeds we are talking about I would hope that you have an estimate for the distributed capacitance and inductance of the 1 m of cable. It won't take much to slow the edges down. Again simulation and measurement are your friends. Leave no stone unturned.

 

I have found working with USB that sometimes the USB standards on paper do not work. I have had many problems with USB transmissions even though all the data said it should work. I again would go with what Papabrove said simulation and measurement is your best bet.

 

Out of curiosity, what do you guys recommend/use for SI analysis? I have access to altium but have heard a mixed bag of reviews regarding this type of analysis using their engine.