5-volt tolerant CMOS

Overdriving an analogue mux can have a lot of nasty effects without blowing up the part, e.g. connecting all the inputs together. How about a nice voltage divider to prevent the overdrive in the first place?

Cheers

Phil Hobbs

I don't mind the mux getting weird, I just don't want to fry it. If I absolutely have to add a part, it could be a tiny schottky diode to +5. I just begrudge every part on this schematic; the PCB will be under 2 square inches. I'll be using a dual opamp and the other half needs to be powered from +12, so the one that I'm gain-switching x1/x10 with the mux is +12 too.

But I've been generally interested in finding out what's inside those "5 volt tolerant" cmos parts. The positive direction must have some sort of zener-y thing, for ESD.

Den søndag den 20. oktober 2013 21.24.53 UTC+2 skrev John Larkin:

pically

re

me don't.

ive an

in a

olutely

dge

be

e one

volt

-y

the datasheet for the old xilinx fpgas that had 5V tolerant pins for 5V pci it said a zener like structure to ground

afaict from the datasheet for that mux it is only the control input that can handle 7V independent of supply, the analog is Vcc+0.5

-Lasse

Usually there will be parasitic substrate diodes. One to the negative supply (mostly that would be ground) and one to VCC. There are usually no zeners involved on CMOS logic gates. If I deliberately bang into these I try to stay under 10% of abs max. But don't do that in multi-pack device for lots of sections at the same time because there will always be a total chip maximum to avoid latch-up or vaporizing a bond wire.

If in doubt, ask. I have use the SD5400 mux in unorthodox ways after they gave me some additional information.

Since they give an upper limit for Vs of VCC + 0.5V that kind of hints that a parasitic diode would come on. Strangely, then they don't give a parasirtic diode current rating for the signal path. For a 7ohms Rdson device that should be beefy but I'd ask.

"5V Tolerant" parts do not have an ESD diode to VDD. They instead have what is called a "snap" diode to ground... actually a tricky NMOS structure that avalanches and eats the ESD pulse.

So... As long as you don't approach +7V with an ultra-fast pulse (*) you should be good to go.

(*) A _really_ fast edge may be interpreted as an ESD event and trigger the "snap" diode... it's triggering is rate dependent. ...Jim Thompson

Are you sure this also goes for the signal path here? The datasheet looks like that may apply for the logic input but the signal range is stated 500mV beyond either rail abs max. Top of page 3:

Good point, but the footnote is still ambiguous. Maybe the analog pins have esd diodes to Vcc; I can test for that.

Could be. Looks like the "5V Tolerant" only applies to the logic pins

They could have made the signal path "tolerant" as well. Maybe the Fairchild engineers didn't study those old chip designs I did for them

Body floating is downright trivial to implement on the VDD side.

Signal mangling will occur above VDD. I doubt that physical harm will occur... it says 50mA max.

Who knows. The more I read of the data sheet, the more forked tongue I see. ...Jim Thompson

What are you switching, the feedback resistor in a basic inverting configuration, or what is it?

Non-inverting, opamp and two resistors and the mux.

All CMOS chips use Schottky diodes to both +/- rails to prevent SCR latchup and provide ESD protection. A series current limiting resistor is adequate.

From his imagination. I know of NO situations where Schottky's are used for ESD. ...Jim Thompson

No. The "5 volt tolerant" parts usually allow +7 on pins, regardless of VCC.

--

John Larkin Highland Technology Inc

Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators

Is it a current feedback amp or something, otherwise how could you possibly approach the current levels quoted to make the MUX act up?

Careful with the signal path on mux chips. That "regardless of VCC" often only applies to the control inputs.

Even the CD4049UB?

Cheers

Phil Hobbs

-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics

hobbs at electrooptical dot net

AD8034 is rated for 60 mA short-circuit current, typ. I have an unused 1K r-pack section nearby, so I can limit the current into the mux. But the FSA3157 spec is fuzzy about what happens in the pullup direction.

As Lasse points out, the analog pins probably have ESD diodes to Vcc, but the datasheet footnote is ambiguous.

I was more generally interested in what the input circuit is like in those "5 volt tolerant" parts and I was hoping somebody would know.

As Phil notes, most analog mux's behave badly if you do turn on the esd diodes, and the manufacturers must hire the NSA to make sure none of their data sheets or appnotes reveal that fact.

I know... I was bitten by that in my youth. In most CMOS mux's, if the analog switch pin voltage goes beyond either rail, a supposedly "off" switch turns on in common-source mode.

And sometimes ESD diode current gets sprayed all over the chip.