Looking for 3Gbps 1:5 LVDS fanout buffer

[Yansi]

Hello,

for an application I am looking for a 1:5 fanout buffer, up to 3Gbps data rate, that could be interfaced to an AC coupled and terminated device inputs.

The signal source I need to fanout has a 750 mV differential swing (into a 100 ohm load).

The devices i need to route the signal to have inputs AC coupled and terminated, accept 200 - 350 mV differential swing. So it seems an LVDS fanout buffer such as the DS25BR204 (datasheet) would fit nicely, but I need not only 4, but 5 outputs really.

Do you know of any other fanout buffer, that would fit the above, with probably LVDS output swing and accepting LVDS/LVPECL input swing and would work as a simple 5 output fanout buffer?

So far I have only found some with LVPECL outputs, but I have no idea if or how I could use those with the AC coupled terminated device inputs.

Thanks for any hints for devices.
Y.

//EDIT: This is the LVPECL 1:6 fanout IC: Renesas 8S58035I (datasheet). Can these even be used for data? They seem to be marketed just as clock buffers.

[Yansi]

Okay, so after a bit more research, I have found this application note, that shows how to AC couple LVPECL logic. Seems some weak termination (140-220 ohm) is used on the transmit side in that case. See the appnote here on page 13, Figure 11: SLLA120 - Interfacing Between LVPECL, VML, CML, and LVDS Levels. Co AC coupling the LVPECL is in fact possible, and the higher voltage swing probably manageable with an attenuator. But I'd better find something more appropriate for the job.

After some more digging on the Renesas website, I have found this pair of 1:8 fanout devices, probably the only LVDS ones having more than 4 outputs:
8SLVD1208NBGI (2.5V variant)
8SLVD1208-33NBGI (3.3V variant)

These have LVDS outputs (yaaay!), and in case of the first one (2.5 V variant) also very good price.  The 3V3 variant seems to be unobtanium currently and expensive. But 2V5 LDO costs peanuts... 

But the question remains, are these devices really capable of passing data, or why are they marketed just as clock buffers? I am confused as to why is that and where the difference is.

[Alex Eisenhut]

It's the 1:5 that's the killer.

https://www.microchip.com/ParamChartSearch/chart.aspx?branchID=9988715

[Yansi]

Awww, awesome! How come I did not remember to check Microchip?

Found this one: SY89847U, 1.5 GHz, 1:5 fanout, 2.5V (datasheet)

It is even cheaper than the IDT part above. No power eating unused outputs flapping in the breeze, it is exactly the 1:5 I was looking for.
Although this one has slightly worse jitter spec, for my evil prototyping, it is more than enough. And if it is not, I have problems elsewhere
Also much better input design, that is easier to use, than the other IDT part, where one needs external termination, external bias and what not.

[Alex Eisenhut]

I still can't believe Micrel is Microchip now, but here we are.