Which 74xx logic families are overvoltage tolerant on their outputs?

Here's my dilemma: I have a 3.3v active-low logic line, and I want to use it to drive a 5v active-low logic line. The 5v line already has a pullup, so my plan was to use an open-collector buffer IC like the 74LCX07, with its VCC at 3.3v, as a buffer.

This seems fine, but I've read that many 74xx logic families have ESD protection diodes on the output, meaning that applying >VCC to their output pins - even open drain ones - will feed current into the chip's VCC via the ESD protection diode.

Is this the case? If so, which if any of the 74xx families are safe to use like this?

• I think the datasheet of the specific chip that you will use will reveal most of the information that is needed to answer your questions. Otherwise, the answers will not be so specific and may not be satisfying, in my opinion. May 10 '12 at 8:05
• @abdullahkahraman Either they don't, or I'm unable to interpret them sufficiently well to understand. For instance, many parts list a "max output voltage", but it's not clear if that applies when the output is in HI-Z mode or not. May 10 '12 at 8:12

The 74LCX07 will not clamp the output to $V_{DD}$ (the datasheet talks about $V_{CC}$), since its particularly targeted at interfacing between different supply voltages. Note that the inputs are 5V tolerant, also with $V_{DD}$ = 3.3V.

The datasheet says maximum output voltage is 5.5V (never use the 7V mentioned under Absolute Maximum Ratings), but should have mentioned for what supply voltage, for instance also for 3.3V.

The only characteristic I could find which actually indicates that a higher output voltage is allowed is off state current on page 4, which gives a value of 10$\mu$A at $V_{CC}$ = 2 to 5.5V and $V_O$ = 5.5V.

edit
That could have been clearer. Especially since it contradicts what it says under Absolute Maximum Ratings: "DC Output Diode Current, for $V_O > V_{CC}$: 50mA."

The NXP 74LVC1G07 seems to be a solution.

To level shift up from a 3.3V system to drive 5V CMOS input levels, simply connect the LVC output to a 5V termination voltage through a pullup resistor as shown below. The outputs of these devices are 5V tolerant and provide a simple solution to drive 5V CMOS input levels.

From here.

It also only mentions output clamping current for $V_O < 0$, so it won't clamp 5V.

• Good find on the off state current stat - that pretty clearly implies that it's 5v tolerant on the output. Thanks! May 10 '12 at 10:35
• Steven, I don't find the Specs ambiguious at all. In fact it is very precise and clear. YOur criticism of their specifications is unjustified. Let me quote; Vo = DC Output Voltage (HIgh or Low)(NOTE2) -0.5 to 7.0V ... Io = DC Output Current = +/- 50mA (NOTE2 says Io Absolute Maximum Rating must be observed. ~ FWIW both conditions indicate exactly what I said below. Also note "some margin" means exactly that! Absolutely, you do not design to run at the Absolute Maximum Rating... I dont need to shout this....it is clear and accurate May 10 '12 at 11:44
• @Tony - You shouldn't look at the 7V, but the 5.5V. It doesn't say it only applies for Vcc = 5V. And how about the contradiction with off state current? May 10 '12 at 11:47
• Thanks again for the updates. Is it safe to say that this part from other manufacturers will be the same? Because, for instance, Diodes version of the same part (diodes.com/datasheets/74LVC1G07.pdf) in their datasheet lists, under absolute max, "Voltage applied to output in high or low state" as -0.3 to Vcc+0.5, which seems to imply a 5v pullup would not be okay. Also, is it safe to assume other LVC parts are also safe? May 11 '12 at 8:35
• @Nick - Oh, that's shit! Parts with the same type number should be compatible, save some minor detail, but this is a major difference! No, it looks like the Diodes won't do. Bad! I'd stick with the NXP, that clearly shows the application. It looks like you'll always have to check the particular datasheet for the manufacturer. But we knew that already, didn't we? May 11 '12 at 8:41

I fully realize you specifically wanted to use a 74xx family device, but if you're not tightly constrained (why?), a member of the optoisolator family might be the better solution. Your grammar was all "a" so if you only need one I/O an indestructible 4N25 could be the better solution. A 74LCX07 provides six io ports... but for about the same price you can get a 4N25 optoisolator and a resistor or two and get 7500 volts of isolation on one port...

There is also something to be said for the simplicity of design of an opto solution... nothing you do involving less than 7500 volts or so on the output has any effect on the input devices... makes things a little simpler, time is money, etc.

Finally a peculiar logic family could disappear in the future, leaving you with a future design problem. On the other hand the venerable 4N25 opto will be around as long as 555s 741s 8051s who knows. LCX limits you to, what, Fairchild and STM? Not that there's anything wrong with that, but there must be at least a dozen generic opto makers. Opto is pretty well future proof, they'll never be a day when you can't buy a 4N25 or similar off the shelf.

• I don't think so. The 4N25 costs twice the 74LVC1G07, is way bigger, requires an input resistor and consumes a lot of power. And to replace the 74LCX07 you would need 6 of them! May 10 '12 at 11:41
• Vince, I agree the opto's will be around for a long time and their use should be limited to where you need up to 7.5kV isolation. Unfortunately that does make a better design the one used in this question. There are many other factors to consider. This design has better margins for noise margin, when you consder the off state of the opto is around .2nA base current, it forces me to consider stray voltages and circuit leakage from moist dust. In fact the specs say in order to guarantee 0.5V max V0 with 2mA out, you need to drive the LED with 50mA (worst case). Maybe that didnt occurred to you. May 10 '12 at 12:12
• ( so I agree with Steven) May 10 '12 at 12:13
• Steven, again there is no ambiguity or contradiction in the Absolute Maximum Rating on this device 74LVC1G07. May 10 '12 at 12:32
• Seems a bad design solution to me. May 11 '12 at 6:16

As long as you give some margin below the MAX SAFE LEVELS , you are safe.

that means .. +/- 50mA for output diode current and -.5 to 7.0 Vdc for Vout as long as above is observed,

Running at any Vdd (that) is allowed is ok, even 3.3V (this is a tuff little part)

no worries for this part,same for other parts.. observe all MAX SAFE LEVELS but they may be lower.

. Good question tho...

• You're not supposed to operate a device at Absolute Maximum Ratings. Keep away from them. Don't even mention them. May 10 '12 at 10:59
• "Don't even mention them". Use normal operation conditions. May 10 '12 at 11:35
• To Not mention critical specs is to be ignorant of the safe operating margin. He wants to know if it is safe, not what is clearly recommended. Thats why I mentioned it. to explain the spec as it is well defined. I think specs need to be fully stated and restated not ignored. I think you are being hostile, when he wanted to know if there was a hidden risk. May 10 '12 at 11:50
• I don't agree. If Normal Operation Conditions say +5.5V maximum you should stick to that, and not say "What the heck, it says here I can go to 7V!" BTW, those 5.5V maximum already give you a margin, because the idea is a 5V +/-5% supply. May 10 '12 at 12:08
• @TonyStewart: The device will not be destroyed "instantly" below the Absolute Maximum Ratings, but on some devices lifetime may be degraded by an unspecified amount by conditions which are within an unspecified distance of those ratings; data sheets give no indication whether that means that operation at 5.6 volts (AMR==7.0 volts) will wear out in 30 seconds a part that would otherwise have lasted 30 years, or whether 6.9V would reduce lifetime from 30 years to 29. Mar 17 '16 at 20:53