ESD Voltage for diode - Positive and negative transients

[FreshmanNewbie]

While selecting ESD protection diodes for, say, ethernet or usb datalines, we usually select the clamping voltage of the diode more than the working voltage of the signal lines and lesser than the absolute maximum voltage of those pins. We need to select a clamping voltage of the diode within that band, right?

For example, if the working voltage of the data lines is 3V and the absolute maximum rating of the pin is 5V, we need to select an esd clamping voltage diode between 3V to 5V, right?

But this would take care of positive transients, right? For example, if the absolute maximum rating of the pin is -0.5V to 6V, and I select a clamping voltage diode of 3.3V, this would take care of positive transients.

What about negative transients? For the negative transients, we need to restrict the clamping voltage to -0.5V, right? In that case, I searched a couple of ESD diode datasheets. Even though they are bidirectional, they don't have any graphs.

Is it right to expect that for a bidirectional ESD clamping diode, that says it has a positive clamping voltage of 3.3V, would it be correct to assume that its negative clamping voltage would be -3.3V?

If that's the case, what should we do if the pin's absolute rating is only-0.5V? What protection clamping diode's voltage we can use?

Edit:

This diode seems to have different clamping voltages for both ends of the diode. But both are positive clamping voltages only.

 

[FvM]

Above linked Semtech TVS arrays are unipolar, as clearly indicated by the schematic. Negative clamping voltage is silicon diode forward voltage, unfortunately not further specified in the datasheet.

 

[FreshmanNewbie]

Above linked Semtech TVS arrays are unipolar, as clearly indicated by the schematic. Negative clamping voltage is silicon diode forward voltage, unfortunately not further specified in the datasheet.

Yes, I know that the Semtech datasheet is unipolar. That's why I asked on what could be done for negative transient clamping?

 

[FvM]

The device does already best possible negative clamping.

 

[KlausST]

Hi,

As the datasheet shows there are two anti series zeners inside.
Thus in one direction the voltage is limited by 15V zener and in the other direction by the 24V zener.

Pin1 is the cathode of the 15V zener, pin2 is the cathode of the 24V zener.

Now it depends how you decide your test setup.
1)
In one test they connected pin1 to GND and get a positive limit at around +24V.
In the other test they connect pin2 to GND and get a limit of about +15V.
(Upper diagrams of fig 5)

2)
In one test they connected pin1 to GND and get a negative limit at around -15V.
In the other test they connect pin2 to GND and get a limit of about -24V.
(Lowrr diagrams of fig 5)

3)
You could also decide to always connect with pin1 = GND and get limits at +24V and -15V (positive and negative)

4)
You could also decide to always connect with pin2 = GND and get limits at +15V and -24V (positive and negative)

All 4 methods are redundant to each other and should give identical results

I'd say it's a rather uncommon protection part with asymmetrical limits. But the datasheet is clear.

****
So if you want to limit at -0.5V ... this (single) part is not suitable.

You may use it ... but use a parallel schottky diode.

You may use a single part with limits according your application.

*******
Btw: some ICs define their limit by "VCC + 0.6V" for example.
Mind that a protection device with 5.0V limit only protects if the IC is properly powered with at least 4.4V.
If the IC is not powered at all (switched off, Vcc = 0V) or VCC is less than 4.4V then the protection is not guaranteed.


Klaus

 

[betwixt]

I wonder if the confusion is over the basic operation of a Zener diode. The clamping function is in the reverse biased direction, they behave as normal diodes in the forward direction. So if two Zeners (like a TVS) are wired back to back the clamping is symetrical (assuming the knee voltages are the same) and is the breakdown voltage of one plus the forward voltage of the other.

Brian.

 

[KlausST]

Hi,

the clamping is symetrical (assuming the knee voltages are the same)

Yes. In the given datasheet the knee voltages are different.

Klaus

 

[FreshmanNewbie]

Hi,

As the datasheet shows there are two anti series zeners inside.
Thus in one direction the voltage is limited by 15V zener and in the other direction by the 24V zener.

Pin1 is the cathode of the 15V zener, pin2 is the cathode of the 24V zener.

Now it depends how you decide your test setup.
1)
In one test they connected pin1 to GND and get a positive limit at around +24V.
In the other test they connect pin2 to GND and get a limit of about +15V.
(Upper diagrams of fig 5)

2)
In one test they connected pin1 to GND and get a negative limit at around -15V.
In the other test they connect pin2 to GND and get a limit of about -24V.
(Lowrr diagrams of fig 5)

3)
You could also decide to always connect with pin1 = GND and get limits at +24V and -15V (positive and negative)

4)
You could also decide to always connect with pin2 = GND and get limits at +15V and -24V (positive and negative)

All 4 methods are redundant to each other and should give identical results

I'd say it's a rather uncommon protection part with asymmetrical limits. But the datasheet is clear.

****
So if you want to limit at -0.5V ... this (single) part is not suitable.

You may use it ... but use a parallel schottky diode.

You may use a single part with limits according your application.

*******
Btw: some ICs define their limit by "VCC + 0.6V" for example.
Mind that a protection device with 5.0V limit only protects if the IC is properly powered with at least 4.4V.
If the IC is not powered at all (switched off, Vcc = 0V) or VCC is less than 4.4V then the protection is not guaranteed.


Klaus

Thank you for your answer.

I have 2 follow-up questions:

1. So, if I use the Semtech part in my application, it is a unipolar device and used for only protection against positive transients and not negative transients, right?

In that case, I need to select a bidirectional diodes. In those diodes, the clamping voltages would be the same in both the directions, right? Can you provide me with an example in this case?


2. When you say, "
Mind that a protection device with 5.0V limit only protects if the IC is properly powered with at least 4.4V.
If the IC is not powered at all (switched off, Vcc = 0V) or VCC is less than 4.4V then the protection is not guaranteed."

You mean that the ESD device doesn't help if the device to which it is connected is not powered? Why does it require that the device must be powered for protection to kick-in?

 

[KlausST]

Hi,

1)
I don't know how I could write it more clearly: it limits 15V in one direction and 24V in the other.
So for sure it limits both ways. Negative as well as positive.

2)
If ... if the device allowes "VCC + 0.6V" maximum voltage ... and now if not powered, then VCC = 0V.
Then obviously "VCC + 0.6V" = "0V + 0.6V" = 0.6V.
--> you must not apply more than 0.6V to the IC's input. This can not be fulfilled with the above protection device. No protection.

Klaus

 

[dick_freebird]

The point of ESD protection is to return the threat energy to its source
as losslessly as possible. The signal diodes are only part of this current
loop. Draw out that loop in full. If you can't then you don't know the
whole picture. The "missing piece" might be important - and is often
the "some assembly required" bit.

Your "from ground" diode could be a plain diode with breakdown
exceeding signal+margin, or could be a zener / avalanche /
reach-through (e.g. GGNMOS) diode of closer-to-signal-max
that can supplement or replace rail clamps to complete the
return loop. I prefer this in a "star ground" ESD return, myself.
Keep supply transients out of the chip core instead of plumbing
them right into it and betting on the rail clamp(s) to keep it all
safe.

 

[FvM]

1. So, if I use the Semtech part in my application, it is a unipolar device and used for only protection against positive transients and not negative transients, right?

Completely wrong. Do you read the answers in your thread? As stated, the Semtech device protects perfectly against negative transients, but doesn't allow negative signal greater than diode forward voltage. If (and only if) your application requires negative signal swing, you'll choose a bipolar TVS device.

 

[KlausST]

Hi,

Sorry for the confusion. After reading the whole thread I recognized that there are different types of diodes / datasheets in post#1.

In my posts I replied on the "unsymmetric zeners" from nexperia only.

Klaus

 

[FreshmanNewbie]

Hi,

1)
I don't know how I could write it more clearly: it limits 15V in one direction and 24V in the other.
So for sure it limits both ways. Negative as well as positive.

2)
If ... if the device allowes "VCC + 0.6V" maximum voltage ... and now if not powered, then VCC = 0V.
Then obviously "VCC + 0.6V" = "0V + 0.6V" = 0.6V.
--> you must not apply more than 0.6V to the IC's input. This can not be fulfilled with the above protection device. No protection.

Klaus

Thank you.

Could you please answer the 1) question for the Semtech part?

Also, for the 2? question, so, there can never be protection when the device Vcc is OFF, against ESD even if diodes are placed?

 

[dick_freebird]

You won't understand it until you think in terms of the current
that is imposed by that delivered charge (squoze through a tube
of your choice) and the voltages and branch currents it will create.
These voltages you are talking about are products of dynamic
charge redistribution. Here comes the tsunami, where's the storm
drain? Erm....

You have two requirements on the electrical side. With sub-demands
of course. One is, don't mess up the operation when anyone's watching.
No leakage, no capacitance, no delay, have headroom.

Two, don't let anyone else mess up the plant when nobody's watching.
Voltage held below pin-ESD-safe (per waveform as it will be received).

There are now (I gather) powered-ESD threat models and spec regimes
for some classes of equipment, These might flow down to a device
or maybe you push back and tell the customer to put a proper clamp
out where the threat is. Because catching chainsaws is not your business.
Unless you took the job with that in the spec. Uhh....

 

[D.A.(Tony)Stewart]

The Semtech 3374N uses a Zener with 4 diodes to make the interface bipolar protected

It is similar but has better speed than back-back zeners, which are asymmetric for some reason..

Secondary ESD protection in the IC is mainly to protect itself from lower energy discharges during manufacturing.

other info
uni vs bipolar https://www.mouser.com/pdfdocs/AND8424-D.PDF

https://www.analog.com/en/technical-articles/hfta160-esd-protection-for-bipolar-integrated-circuits.html

https://www.onsemi.com/pdf/datasheet/nup1105l-d.pdf

I realize this does not answer your question.