faster linear multiplexer ?

Question

Hello, all,

I just built a prototype of a board that has a multiplexer for a fast linear signal, and after some tinkering, found the slowest thing is the multiplexer chip. I am using a TI/Burr-Brown MPC506A. This was the only 16:1 analog multiplexer I could find that had any hint of a reference to speed/settling time value in the data sheet. I think I've got something that is useable, but it would be nice to do better. (This is for remote inspection of signals via oscilloscope, so there's no specific value I need, just "better".) Anyway, this is running off +/- 5 V supplies to be compatible with the rest of the board. One trick I came up with was putting a 1K load on the output of the mux, and raising the gain of the buffer amp. This got the performance up to the tolerable level. The signals in question have 5-10 ns risetimes, and an about 20 ns wide pulse leading a slower response from a detector. We pretty much lose all of that leading spike with the current mux chip, and would like to be able to see at least some of it on the scope.

If there is some other mux chip that has a much higher bandwidth, and especially if it is pin-compatible with the MPC506/DG506 pinout, that would be great!

Thanks in advance for any pointers,

Jon

whit3rd · Accepted Answer

The multiplexer uses a MOSFET switch, and there's tradeoff with conductance and stray capacitance. So, you could try circuit design that makes the switch connection at a non-slewing point (input node of a transimpedance amplifier). Or, you could go to a relay-based solution (if it's the signal that has to be fast, not the channel-switching). Packaged multiplexer/amplifiers might be of use (MAX4312 is an 8-to-1 video mux/amplifier).

Frank Miles · Answer

The 506 has a very high channel resistance, which gives you a nasty rolloff with internal and external C-loading. Your 1K load is simply reducing the Thevenin equiv. resistance of the source. The DG406 has 1/4th the channel R, which might be worthwhile. If your signals stayed within a 5V span you could probably find something with _much_ lower R (but probably a host of unpleasant circuit mods to make that work). This assumes that you're not trying to actually switch channels quickly - my memory is that this family all has some serious recovery problems on channel switching.

John Larkin · Answer

ADG1208 is an 8:1, rated for 550 MHz. --  John Larkin         Highland Technology, Inc jlarkin at highlandtechnology dot com  Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro   acquisition and simulation

Jon Elson · Answer

16 boards times 16 channels/board = 256 relays. No, we're not going to do it that way. Hmm, the non-slewing input has some appeal, but that ends up inverting the signal. But, maybe we could then invert again.

Jon

Jon Elson · Answer

Well, if we have to redesign the board, that might be an option. No indication it works at +/- 5 V, though. Jon

John Larkin · Answer

The data sheet says it does.

--

John Larkin Highland Technology, Inc

jlarkin at highlandtechnology dot com

formatting link

Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation

faster linear multiplexer ?

Join the Discussion

Didn't find your answer?