Hey Jason,
The highest bandwidth 8:1 we have is the TMUX1308 with a bandwidth of 500MHz. What's the application here? Are we looking for 1GHz per line speed or is this something where 1Mb of information is sent across all the signal paths collectively?
if it's the former (1GHz per line) the following devices meet this criteria :
So if you take the TMUX1511, a 4 channel 1:1 switch w/ individual control pins for each channel and tie the outputs together, it would effectively be an 8:1 switch with 3GHz bandwidth, low leakage and low on-resistance. The caveat here of course is the additional select pins vs a traditional 8:1
Thanks,
Rami
Hey Jason,
We don't have test data on this and your worries are very valid and as I dug deeper into this, the off-capacitance would be an issue when tying these outputs together as well. The capacitance isn't an overbearing amount (2.5pF per) but you'd effectively be adding 7x Coff + Con (3.3pf) which would derate your bandwidth quite a bit. How much it would affect it i'm not entirely certain but looking at a comparable device in terms of the effective R(on) and Capacitance, I would estimate in the MHz.
Within our portfolio it looks like the best solution may be to cascade a few high bandwidth multiplexers. The TMUX136 would be a good device to start with since it's bandwidth is 6.1GHz. The low C-ON would ensure that the impact of the added load wouldn't be too high. I've drawn a very quick diagram. These are not drawn to scale to help show the alignment of the multiplexers/channels better.
So 4 devices could be used and each signal path would see 3 devices. The TMUX136 has 1 select line. So 4 would be needed to give enough support to multiplex to the 8 unique inputs. There will be loss with this method. However, how large this loss will actually be may only really be determined through actual implementation since they'll accumulate a lot of the loss from the board itself due to the added trace lengths. The mux itself shouldn't add too much loss since they should be located close to each other and TMUX136 will have a low R-On and C-On.
Let me know what you think of this solution.
Thanks,
Rami
