LMH0340: Advice for connecting to LMH0318

Hi Zach,

LMH0340 is 3G SDI serializer with integrated 75-ohm cable driver. You are planning to connect this output to LMH0318 100-ohm differential input over 12" twisted pair?  Can you please provide a block diagram of this application.

Secondly, please let us know dB loss at 3Gbps for 12" twisted pair.

Regards ,, Nasser

Unfortunately I cannot provide a hard number on the dB loss, the twisted pair passes through an interconnect with an unknown and unfixed impedance, so I'm counting on the 0318 to to clean up the signal before outputting it from my system.  I also recognize that there is a mismatch between the input and output, but the 0318 appears to be my best bet for useable signal coming out the other end of my system.  I am principally wondering on which side of the twisted pair should I place the AC coupling caps, both chips call for them, and I can place them on either board, but I'm not sure if it'd be better upstream or down.  My design is giving both chips the best chance of working that I can, ultra-low-ripple power supplies, fine tolerance passives, adherence to recommended layout, but the twisted pair is just a messy way to move high frequency signals that I can't avoid.

Hi Zach,

1). Between 75 and 100 ohm differential pair there is no adaptor? Let's first take care of this and then we can discuss AC coupling Cap placement. 

2). Also, please check online and there should be passive lumped-sum RCs that you can use to go from single ended to differential 100-ohm. Is this an option?

3). Another idea is that FPGA have integrated serdes - LMH0340 functionality is built into the part - would this be an option? Using this option we can make this a bit easier.

Regards ,, Nasser

Alright, sorry for the confusion, I am executing the design choices of a more senior engineer and he forgot to mention a point before.  We are planning on exchanging the 75Ohm resistors on the output for 100Ohm resistors to try to make it match the 0318.  We are fixed on the 0340 because it outputs real SDI, not a proprietary serial format that has to be deserialized at the other end.  

Is there any way to avoid putting an adapter between the two? I don't know what you mean by lumped-sum RCs.

While the FPGA does have an internal serializer I would prefer not to use it because the implementation is very complex and this is a somewhat time limited project.  We really want to use the external IC.

Thanks,

- Zach

Hi Zach,

Got it. This now makes more sense. I believe you can change 75-ohm pull up on LMH0340 and replace with 49.9-ohm R to drive 100-ohm differential pair for the LMH0318 input. However, i must say i haven't come across a similar case in the past. 

If you have s-parameter model of your twisted pair cable, you can do signal integrity analyses using LMH0318 IBIS-AMI model. 

Also, if you have a LMH0340 design already you can make this change - using 49.9-ohm R instead of 75-ohm R - and check output eye diagram on scope. 

Regards ,, nasser

Excellent, I will do that, only one question left then: Both the 0340 and the 0318 specify coupling caps, should I actually put both? Or jus one set? And on which side of the noisy twisted pair if just one set?