Zener diode on I2C lines

Hi,

paddy_p said:

5.1V zener diode on the I2C lines for safety purpose.

Click to expand...

What exact "safety" are you talking about?

***

The voltage will drop with 5.6V zeners, too. But not that much.
Now we need to know:
* how much did it drop with the 5.1V zeners?
* how much voltage drop do you / your application allow?

Technically the 5.1V zeners should still allow valid HIGH levels.

paddy_p said:

Any suggestions why these SCL and SDA voltage levels are dropped

Click to expand...

I don´t know what exact zeners you used.
To get the answer I´d consult the datasheet ... and check a chart: V_Z vs I_Z

paddy_p said:

The 1K resistor is added in the series with 5.1V zener diode.

Click to expand...

Why? It does not make sense to me.

paddy_p said:

MCU is from Cypress

Click to expand...

I doubt Cypress only produces ONE type of MCU.
If the exact part name is secret ... you need to look through the datasheet about:
* I2C lines operating voltages (HIGH / LOW threshods)
* I2C lines max allowed voltages (Absolute maximum)

If it´s not a secret: post links to all datasheets (directly at the manufacturer site)
***

paddy_p said:

Both are powered on 5V supply.

Click to expand...

All the time? 24/7?

I ask, because often the absolute maximum limits are with respect to VCC. Now, in case the device is not powered .. VCC is considered to be 0V .. then a zener does not give any protection on positive ESD pulse.

Because of this I rather use double schotty diodes (usually with a series resistor in direction to IC) .. or dedicated signal line protection devices.

And: I2C usually is meant for "on board" or at least "within an electronics device" ... so no external connection. In these cases usually no extra protection circuit is needed at all.

Klaus

You can review zener diode datasheet to see if it has detailed V(I) curves. Nexperia BZX84 datasheet e.g. has it, below curve explains well your observation.

Generally speaking, it's almost impossible to protect I2C bus completely against overvoltage. Which overvoltage source are you expecting?
Minimal Vih for SCL and SDA is 0.7 Vdd according to I2C-Bus specification, thus dropping bus voltage to e.g. 4.5 V, as expectable with 1k pull-up and C5V1 zener is no problem as such.

Leaving aside any actual protection for "whatever", zeners have a couple of features to consider.

One is that the reverse conduction "knee" can vary in its "softness" and leakage run up (to spec) as you approach the spec voltage. Key in zeners reference design, maybe not here.

Second, zeners have capacitance and this will scale with Rs and power rating (die size). Too stiff (like a simple spec-crawl looking for 1-2A rating) could end with you imposing more Cline than the pull-up can pull, in bit-time.

Third, zeners can be "programmed" by reverse over current. I have used "zeners zap" many times in on-chip trim networks. The damage threshold / programming level can vary with many process and layout features, so mfr & P/N. This means you have a safe and unsafe range of "whatever" protection (always so, but maybe closer-in than planned).

What a 1K series R would do, besides remove all effectiveness of clamping, I do not imagine.

check this link

you need to buy the 1% or 0.5% zeners - else you will have the issues described.

you can also buy 5.2 V zeners

Zener voltage tolerance is an additional problem, but too soft limiting characteristic is the basic issue.

Your issue likely stems from the clamping effect of the 5.1V zener diode. Even with a 5V signal, the 5.1V zener may conduct slightly, as its breakdown voltage is very close to the I2C operating level. This can cause signal degradation or voltage drops, especially with the series resistor in the circuit. Using a 5.6V zener works because it provides more headroom before conducting. A solution might be to use Schottky diodes instead, which can prevent over-voltage while not interfering with signal integrity.