5-volt tolerant CMOS

.com

begrudge

. I'll be

so the one

nfiguration, or what is it?

bly approach the current levels quoted to make the MUX act up?

r-pack

spec is

the

e "5

iodes,

heets

Didn't you say you have a non-inverting configuration with the MUX selectin g X1/X10- that sounds like an R from OUT to IN(-) and then a resistor R/9 f rom IN(-) to MUX common with the B0/B1 open/GND or something similar, (or i s there too much leakage for that?)-seems like the currents into an over-vo ltaged MUX can be made real small.

It appears that particular one has diodes on the switch input clamping to Vcc/GND, but a zener-like structure on the control input for 5V tolerance, clamping to >+7 (presumably current flows to GND, not Vcc) and GND.

As an aside, that's quite a bit of dissipation if you take their 50mA figure seriously, max dissipation would probably limit the steady-state current.

The switch inputs clamping to Vcc is often inconvenient since it can cause Vcc to rise unless you add additional parts. A few of the higher end AD analog switches have zener structures on the analog inputs and disconnect for overvoltage exceeding Vdd + Vt (which could still cause problems in the following circuitry).

Look for switches that are guaranteed "off" with power off for this kind of functionality.

Say what you want about Maxim, but they do have this incredible beyond-the-rails analog switch technology:

Three internal 35MHz charge pumps. Their guaranteed max leakage is pretty high, but they'd be great for low-distortion switching of reasonably low impedance/low voltage circuits. Not so great for precision.

Hey, maybe John can use one of these as a bias generator for a PD-- they should output +35 and -27 from a 3-5V source. If they can actually be purchased.

I was planning to use circuit A, with R1 to limit the current into the mux. The opamp supplies are +-12.

Circuit B, which I think you are suggesting, also limits the current into the mux, but the mux resistance starts to matter and there are bandwidth penalties. Since R1 is free, I may as well use A.

Couple of things to watch out for:

a. Make sure the IN+ signal never exceeds the mux supplies, at least not driven from a hard source. Some opamps activate clamp circuitry or diodes if IN+ and IN- drift apart.

b. If the mux supply rail isn't otherwise loaded use a TL431 to shunt a little above. Else the current through R1 would raise that voltage until ... *PHUT*

:)

pically

re

me don't.

ive an

in a

he-rails analog switch technology:

The guaranteed leakage is 10x better than the FSA3157, and the isolation is about 30dB better, but the charge injection and terminal capacitance are w hoa way bad for anything but a low impedance application. Switching times a re immaterial for this app. It has a thermal shutdown capability, seems lik e a first for an analog switch. The charge pump noise is way down at -80dBm but not fantastic for precision signal conditioning. Maybe this not the be st part for the application.

Some sort of IC bias generator would be cute, if it didn't violate the Never Buy Maxim rule.

As an analog switch, that has way too much personality. I'd just as soon use an opto-SSR.

In my current project, I'll keep the pd bias supply out of the optical head. If I did put it on the same board with the photodiode, it would be a PV isolator or a soft, slow, low KHz charge pump.

Never Buy Miso!

Never buy Larkin? ...Jim Thompson