On a multiple-output waveform generator box, I want to have a test connector that can be programmed to snoop one of eight analog signals. The frequency range is up to about 1 MHz, and I'd like to keep the crosstalk down.
So I want an 8:1 mux that either grounds the unselected inputs, or has the equivalent of t-switches inside. I don't think I know the proper search terms for something like this.
Any suggestions for parts?
for RF switches, absorptive is with 50R, reflective with 0R
if you can do you own selection logic, maybe adg902 ?
-Lasse
Yeah, something like the 902 would work, but an 8:1 mux would be nice, to save parts.
I was looking at ADG1208, which seems to have great crosstalk specs, until you see the test circuit, fig 36.
That's downright cheating. They get 6 dB for free with that extra 50 ohm resistor, and another 10 dB or so from the 100+ ohm switch "on" resistance working into the 50 ohm VNA.
maybe a tree of adg936s ?
having multiple layers of switching should also help with isolation
-Lasse
Shorting the intermediate steps (out of phase with the connection) helps a lot, too. Capacitance, and all that.
Why not bring out all eight channels to a rotary switch? Isn't that cheaper, after you add in the display of which-signal-is-being-snooped? MOS switches are nice, but capacitance and series resistance make them unappealing sometimes.
On a sunny day (Fri, 17 Oct 2014 16:01:28 -0700) it happened John Larkin wrote in :
Sounds like you need to search for crossbar.
Maxim used to makes some nice video crossbar switches. I haven't looked, lately.
Good old CD4051?
Yeah, it's hard to beat a real switch, or a relay, for electrical performance.
There are two main families of 8:1 mux's, the CD4051 types and the DG408 types. The 408s tend to be higher performance... wider bandwidth, higher voltage supplies, better protections, lower crosstalk.
It would be nice to mux +-10 volt signals, up to 1 MHz, with low crosstalk. Since this will be a scope monitor connector, the crosstalk can be whatever it is... I'd just prefer less.
signals.
has
Don't you need to analyze figure 38?
?-)
I can isolate the signals with opamps, so I don't care about cross-coupling between mux inputs. But when I select one input channel to drive the output, I don't want to see others leaking into the output signal.
So s-d off coupling matters more than s-s coupling.
The DG538A claims internal T switch structure. Not sure about input range, I think +/-5V.
piglet
If you are using opamps to isolate the inputs, can you use opamps with a shutdown feature and only enable the one opamp that's needed?
Allan
That's how I made a screamingly fast sample-and-hold back in my youth, patents...
3,638,041 Sample and Hold Trigger Circuit3,643,110 Sample and Hold Circuit
PDF's available on my website. ...Jim Thompson
I need four 8/1 test-output mux's, so that would be 32 opamps. It would work, but something simpler would be nice.
Perhaps roll your own T-switches using multiple 4053's? ...Jim Thompson
Again, lots of parts.
If I can't find a great 8:1 mux, I'll just live with the best I can find. These are scope test outputs, so some crosstalk can be admitted and tolerated.
Hard rotary switches are still a possibility, if we can make room on the small panel.
Nice part. Yes, it does have t-switches. Thanks.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required