Author Topic: "5V tolerant" question  (Read 4682 times)

0 Members and 1 Guest are viewing this topic.

Offline c4757pTopic starter

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
"5V tolerant" question
« on: August 05, 2013, 03:15:04 am »
Warning: This may be a dumb question.

I am using an ADS1255 A/D in a mostly 5V system (not by choice; a bunch of older-design chips that require it). It requires no more than 3.3V on the logic side but specifies the following:

- 5V tolerant SPI

And in the pin table
- #CS: 5V tolerant input
- DIN: 5V tolerant input
- SCLK: 5V tolerant input

Should I make anything of the fact that they never mention whether DOUT is 5V tolerant when in high-impedance mode (#CS is high)? I'd assume it should be, since it's "5V tolerant SPI", but the fact that they mention it for three out of four pins makes me nervous to feed 5V up this expensive part. :-BROKE
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline David_AVD

  • Super Contributor
  • ***
  • Posts: 2863
  • Country: au
Re: "5V tolerant" question
« Reply #1 on: August 05, 2013, 03:23:36 am »
What about using a pull-up to the 3.3V rail for DOUT ?

EDIT:  The data sheet does specify that DOUT is only good for -0.3 to Vdd + 0.3 max so a pull-up to the 5V rail is out of the question.
« Last Edit: August 05, 2013, 03:26:29 am by David_AVD »
 

Offline c4757pTopic starter

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: "5V tolerant" question
« Reply #2 on: August 05, 2013, 03:27:29 am »
Hmm... How would that work? SPI's not open-drain...

EDIT:  The data sheet does specify that DOUT is only good for -0.3 to Vdd + 0.3 max so a pull-up to the 5V rail is out of the question.

Where? I don't see that. :-// You're not mistaking "D0/CLKOUT" for "DOUT", are you? I don't see it listed anywhere!
« Last Edit: August 05, 2013, 03:29:51 am by c4757p »
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10270
  • Country: nz
Re: "5V tolerant" question
« Reply #3 on: August 05, 2013, 03:37:14 am »
As i understand it, "5V tolerant" refers to inputs only.
ie, the input is tolerant of voltages above VCC.

Outputs, like Dout, will be 3.3V and will draw current if you try to pull them above 3.3V VCC
« Last Edit: August 05, 2013, 03:41:02 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline c4757pTopic starter

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: "5V tolerant" question
« Reply #4 on: August 05, 2013, 03:42:24 am »
Yeah, usually. But SDO/MISO on an SPI bus has to tristate when the chip isn't selected; shouldn't that mean that a "5V tolerant" SPI implementation would be able to tolerate 5V there when tristated?

If only they could put the good ol' input/output equivalent schematics like the microcontrollers always have...
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline Rufus

  • Super Contributor
  • ***
  • Posts: 2095
Re: "5V tolerant" question
« Reply #5 on: August 05, 2013, 03:44:42 am »
Should I make anything of the fact that they never mention whether DOUT is 5V tolerant when in high-impedance mode (#CS is high)? I'd assume it should be, since it's "5V tolerant SPI", but the fact that they mention it for three out of four pins makes me nervous to feed 5V up this expensive part.

I would say it isn't. So if you absolutely need to tri-state it with other 5v devices you will need a tri-state buffer running on 5v and a pull up/down for its input. You can probably get a single 125 type buffer in a 5 pin SOT-23 package.
 

Offline c4757pTopic starter

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: "5V tolerant" question
« Reply #6 on: August 05, 2013, 03:48:45 am »
74HC1G125;. Or two 2N7002; I had that going nicely on the breadboard a minute ago. Either way I definitely think I will buffer it - seems I'm the only one who thinks it ought to be 5V tolerant, wouldn't want to find out I'm wrong with $15 magic smoke when $0.20 will prevent discovering it the hard way...

Wow... takes a while to even find a tristate buffer that's specified for voltage above VDD on the tristated output. 74LVCE1G125 from Diodes Inc was the first I found, down a few pages on Mouser on a search for "1G125"... Though I might just do it the old-fashioned way with transistors. I've used that MOSFET bidirectional level shift circuit from the NXP (IIRC) app note many times, it's quite familiar and works well.
« Last Edit: August 05, 2013, 04:13:23 am by c4757p »
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline Guppzor

  • Contributor
  • Posts: 26
  • Country: au
Re: "5V tolerant" question
« Reply #7 on: August 05, 2013, 05:12:03 am »
I have a bit of a side question to this... When a datasheet states "tolerant", does this mean that it is perfectly fine to run the part at whatever spec it is declared "tolerant" to, or is it "tolerant" in that it can take that specced value, but it is not the intended mode of operation and there could be some consequences such as reduced life, change in other specced values, etc?
 

Offline c4757pTopic starter

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: "5V tolerant" question
« Reply #8 on: August 05, 2013, 05:14:12 am »
5V tolerant means it's perfectly happy interfacing with 5V logic. IIRC the two series ESD diodes are replaced by a Zener from ground to the input. The rest of the input circuitry is just a MOSFET gate, it doesn't care about VDD, just don't exceed VGS(max).
« Last Edit: August 05, 2013, 05:22:26 am by c4757p »
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline David_AVD

  • Super Contributor
  • ***
  • Posts: 2863
  • Country: au
Re: "5V tolerant" question
« Reply #9 on: August 05, 2013, 07:08:09 am »
Hmm... How would that work? SPI's not open-drain...

EDIT:  The data sheet does specify that DOUT is only good for -0.3 to Vdd + 0.3 max so a pull-up to the 5V rail is out of the question.

Where? I don't see that. :-// You're not mistaking "D0/CLKOUT" for "DOUT", are you? I don't see it listed anywhere!

Sorry, I did misread the D0 as DO.  The real question is whether or not you want to feed 5V into that pin while it's in Hi-Z mode.
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8429
Re: "5V tolerant" question
« Reply #10 on: August 05, 2013, 11:49:22 am »
You can use a current-limited supply to check what the input is like; just set the current limit to something like 100uA and slowly increase the voltage up from 3.3, and if it's truly 5V tolerant it won't take more than the specified leakage current (10uA) at 5V.
 

Offline David_AVD

  • Super Contributor
  • ***
  • Posts: 2863
  • Country: au
Re: "5V tolerant" question
« Reply #11 on: August 05, 2013, 10:06:49 pm »
The data sheet does mention the case of connecting DIN and DOUT for systems that implement bi-directional data.  It's a shame that they don't specifically state (I couldn't find it) whether this is valid with a 5V MCU.
 

Offline c4757pTopic starter

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: "5V tolerant" question
« Reply #12 on: August 05, 2013, 10:12:34 pm »
I'm not too worried about it, I can level-shift the line with a $0.05 BSS138 and two resistors, no trouble at all. I do wish they'd say, though... |O
No longer active here - try the IRC channel if you just can't be without me :)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf