What would be the expected overall Insertion Loss for the worst case inputs (RF1/4) on 3 ADGM1304s chained in series?
The IL will depend on the frequency you are operating at. For lower frequencies the IL is very similar of all channels, see Figure 3 in the datasheet. So for example, a 2.5GHz signal has a max IL spec of 0.55dB so for 3 switches in series the max would be 1.65dB. But this would be very much a max figure with process margin included, typically you would only see 3 x 0.26dB typ which would be 0.78dB.
Thanks for the quick response.
We are working on a T&M product measuring 6.4Gbps signals so we would like
to have good bandwidth out to 10+GHz.
Are you sure that the IL function of chained devices is simply additive? I
thought it was more complicated and non-linear so that it had a greater
Torry Hack - GM
I was assuming an ideal 50ohm system with good return loss, also I am neglecting the loss of the pcb transmission lines themselves. For the additive insertion loss approach to be most valid you will have to have good return loss and that includes on your board at the frequency points of interest so things like pcb transmission line 50ohm design is really important. If the board/part doesn't have good return loss then indeed the IL will be greater than simply adding up the 3 switches.
To get a feeling for what 2 switches back to back look like with respect to insertion loss check out the CN-0377 reference design. That board goes to 2.5GHz and shows how the IL adds plus the addition of transmission line losses, see the dashed green trace in Figure 6 that shows the transmission line losses + 2 switches. Also look at the ADGM1304 evaluation board Figure 2, we try and target a board return loss of -20dB or better to make that contribution to insertion loss as small as possible, otherwise you start to see a ripple appearing on the switch IL plot.
Hope this helps.
Retrieving data ...