ESD protection on 1.8V

[sb_st]

Hi! I am trying to muddle my way through a design that revolves around an STM32U5 MCU, and am specifically trying to think about ESD protection. I have been referencing AN5612, which quite helpfully provides lots of specific examples for various applications (though it seems to perhaps be falling out of date, as a handful of the devices mentioned seem to now be marked NRND). 

One thing that I can't quite wrap my head around, though, is ESD protection as it relates to running some or all of the power domains of the MCU at 1.8V. For example, the datasheet for the stm32u59xx has an Absolute Maximum Ratings table that shows most FT (5V tolerant) pins have a max rating of either 5.5V or min(Vxx+4.0, 6) - so 5.8V in this case - with other pins having a max rating of 4.0. I know these are DC voltages, but I think I understand that it's these values I should use when trying to parametrically search for ESD protective devices, such that they have forward voltages above 1.8V and clamping voltages below 5.8, 5.5, or 4.0V, depending on the device. 

TVS diodes that fit this spec seem quite rare, though. Digikey, for example, lists over 100k different TVS diodes,  but when I filter for unidirectional diodes with reverse standoff voltages of 1.8-2.1V, I get only 7 results. This seems somewhat alarming to me, and makes me suspect I'm doing something wrong. 

Further, in trying to grapple with this with Gemini, it tells me that I actually need to be clamping at 2.1V, for reasons that are unclear to me. There are zero such devices on Digikey when I try to search for them. 

AN5612 doesn't make any mention about whether the devices it suggests in various application examples are valid choices regardless of whether the MCU is on a 3.3v or 1.8v rail, but I suspect this does matter somewhat? It seems like most devices presented in this document must be reasonably valid choices for 3.3V Vdd, but is this still true if we run at 1.8V?

All of this leaves me confused, so I wanted to ask for some advice. What do people use to, say, protect SWD lines on an MCU whose Vdd rail is 1.8V? What about slow interfaces like pushbuttons or potentiometers? 

Thank you!

[gregstm]

While you wait for an expert to reply, I will give you my opinion. I don't think it matters if the chip is run at 1.8V, protection specified for 3.3V would be fine.

From AN5612 - "An excessive electric field damages the dielectric. This occurs when the voltage across the dielectric thickness exceeds 80-100 V."

You just want to make sure you keep those damaging high voltages away from the IC.

More application notes can be found here - https://www.st.com/en/protections-and-emi-filters/esdalc6v1-1m2.html#documentation

[TDK]

> What do people use to, say, protect SWD lines on an MCU whose Vdd rail is 1.8V? 

There are plenty of ESD diodes intended for 1.8 V applications. You will find the clamping voltage to be significantly above 1.8 V. Here is one at random:

These are not on/off devices, but gradually turn on when voltage increases. The clamping voltage should not be read as "the voltage will never get above this". The diode lets more and more current through as the voltage increases.

They're devices for mitigation and they do a good job, but they are not perfect protection. Zap the device enough times and it will die. Imagine lightning hitting your house repeatedly. It's a high-energy event. Even with the best electrical prevention, eventually it'll damage something. (Not a perfect analogy, but a good one.)

 

> For example, the datasheet for the stm32u59xx has an Absolute Maximum Ratings table that shows most FT (5V tolerant) pins have a max rating of either 5.5V

This is for DC voltage values. For an ESD event, the "ESD absolute maximum ratings" table is the more relevant value. Pins are typically rated to 2 kV events. Adding an ESD diode increases that dramatically. The pins will generally see more than the "absolute max" dc voltage value during such events if they're connected directly.

 

> What about slow interfaces like pushbuttons or potentiometers?

Not sure I understand. Buttons and potentiometers are not typically ESD sensitive. If they're also connected to a pin with max 1.8 V, the same ESD diode would suffice. ESD diodes should be placed at the external interface, far away from the IC you're protecting (if possible).

[sb_st]

This is wonderful, thank you so much! 

I would really love to understand this more deeply, so I wonder if I can ask you to teach me a bit more. 

---

There are plenty of ESD diodes intended for 1.8 V applications. You will find the clamping voltage to be significantly above 1.8 V. Here is one at random:

So - I feel silly asking this - how did you do this, exactly? For example, when I go to Digikey, and go to Circuit Protection->TVS->TVS Diodes, and I select 1.8V Reverse Standoff Voltage in the parametric settings, I see:

 Of the 19 results, only 6 are even stocked, and none show the specs you highlight in your screengrab. 

Further, if I go here to ST's site, and I visit either the ESD Protection page, or the TVS page, there are no products listed with a Vrwm of 1.8V (again, I'm just trying to replicate what you seem to be showing). 

I have this experience over and over and over again - TI's site, Nexperia, etc. etc., I try filtering by products that have a Vrwm of 1.8V (no other filtering), and come up with - at best - a single product, or - usually - no products. But I think "ST makes heaps of products for ESD, and they make heaps of MCUs that run at 1.8V! Why is there this apparent disconnect?!". I'm quite sure - given that you say there are "plenty" of these kinds of parts - that I must be doing something horribly wrong in my searching, because I'm totally unable to find "plenty" of options. The best I have ever found is this one, single product that indeed shows a Vrwm of 1.8v and a clamping of around 4V. But boxing myself into relying on a single product is terrifying, and I'm fairly sure the entire 1.8V world isn't relying on this one single diode. 

---

They're devices for mitigation and they do a good job, but they are not perfect protection. Zap the device enough times and it will die. Imagine lightning hitting your house repeatedly. It's a high-energy event. Even with the best electrical prevention, eventually it'll damage something. (Not a perfect analogy, but a good one.)

This I understand, thank you. I'm trying to bring a "real" product to life, and am trying to think through certification and general longevity (so, a person walking across carpet and picking up my product multiple times). I'm aware I'm not trying to create some sort of invincibility shield, but I'm imagining that there are a good many products out in the world that have suitable, professional consumer-grade ESD protection - that's sort of the 'target area of knowledge' I am trying to come up to speed on. 

---

This is for DC voltage values. For an ESD event, the "ESD absolute maximum ratings" table is the more relevant value. Pins are typically rated to 2 kV events. Adding an ESD diode increases that dramatically. The pins will generally see more than the "absolute max" dc voltage value during such events if they're connected directly.

---

Ah, ok this is a wonderfully helpful clue! So here is that table for the stm32u5:

I've read AN1709 and AN5612 several times over, to try to understand the various jargon/terminology often used to describe various ESD protections (JEDEC vs IEC 61000-4-2, HBM vs CDM vs air discharge vs contact discharge), and admit my grasp on all of it is tenuous at best. What, for example, would lead me to conclude from the above table that the figure I should be paying more attention to is the 2kV HBM value you mentioned, and not the 250V CDM value? And do I understand you to be saying that if I can find any ESD protection device that clamps below 250V, that would generally be sufficient for my application (regardless of whether my MCU is running at 1.8V or 3.3V, no less)?

My understanding is that that JEDEC standards seem to refer mostly to handling the individual component (the MCU in this case) in the factory during manufacturing/assembly, whereas IEC 61000-4-2 refers more to the system as a whole, as it is being operated after assembly (therefore, more relevant to "human walking across a carpet and picking up my device). But I may be mistaken about all of this (clearly I can't even do a parts search, so I'm not to be trusted much here), and am still finding it confusing. 

 

---

> What about slow interfaces like pushbuttons or potentiometers?

Not sure I understand. Buttons and potentiometers are not typically ESD sensitive. If they're also connected to a pin with max 1.8 V, the same ESD diode would suffice. ESD diodes should be placed at the external interface, far away from the IC you're protecting (if possible).

---

Sorry, by this I mean: I'm aware that selecting ESD protection devices varies depending on the use case of the exposed contact. So, I want to use something different to protect against a strike against a USB-C connector than I might against a SWD TagConnect footprint. I'm aware that - broadly speaking - there seems to be "low capacitance" and "not low capacitance" parts, the former being more useful for high-speed data lines, the latter being more useful for slower I/O or power lines. 

A pushbutton connected to a digital I/O on an MCU running at Vdd=1.8V is just my "base case" for trying to find an acceptable part - it's slow, doesn't need special current-shunting capabilities, doesn't need particularly low capacitance (at least, as I understand it), and needn't get into multi-line ESD array-type parts. So it just seems like a good example for conversation. 

More broadly, my design includes an SD card, USB, SWD headers, and a headphone/microphone jack, all of which I think will require different ESD protection solutions (maybe - I don't know, that's what I'm trying to understand). I think some of the examples in AN5612 would work fine for my situation (USB, for example, is USB, and I don't think 'cares' whether the MCU is at 3.3V or 1.8V?). But the SWD lines are confusing to me - can I use the same ESD device on all the lines? AN5612 implies not, that two different parts are required. But AN5612 is part marketing, part education, so I take it with a grain of salt. Youtube videos from Phil's Lab show him using a single part on all lines...but the MCU is running at 3.3V in those examples.

You can probably sense that I am tumbling down a rabbit hole here, and am trying to grasp for roots to hold onto. 

I also note that - in other forum conversations about ESD protection - there tends to be some generally-accepted group knowledge (like "just use parts rated for 3.3V, you'll be fine"), but making a choice without understanding it at least a little bit makes me uncomfortable, and I don't own a spark gun to be able to test my choices, so coming at this academically is kind of the best I know to do currently...

Anyway, thank you for your patience with all of this, I really am interested in understanding it more deeply. 

[TDK]

Hi @sb_st, I'm afraid we're reaching the limits of my expertise here. That is the primary reason for my lack of response.

Yes, there are considerably fewer 1.8 V ESD parts out there than 3.3 V. But there are some, and you only need 1. There are fewer than I thought. This also gives some credibility to the idea that lots of people out there are just using 3.3 V parts because at the ESD level, it's not a huge distinction between the two. I don't think there are any issues with your parts search. No doubt an expert in ESD could weigh in more here.

> What, for example, would lead me to conclude from the above table that the figure I should be paying more attention to is the 2kV HBM value you mentioned, and not the 250V CDM value?

If your intention is to shield the device from humans applying the shock after shuffling their feet on carpet, or whatever, then HBM is most appropriate as it most accurately mimics the type of shock that is applied. There's not a fixed equivalence between them. You shouldn't think "2 kV HBM" is more or less protective than "250 V CDM". But there's certainly a correlation.

 

Sorry I can't weigh in more here. I don't want to overstate my expertise in this area.

 

At a higher level, sometimes engineers talk about ESD and spend a lot of time on it. Sometimes it's useful. But here, the MCU is already fairly ESD resistant and (my opinion) chips being killed by ESD is far less common than you would be lead to believe by reading stories on the internet. Yes, ESD is real, but it also serves as a convenient "catch all" for when engineers don't know why something failed. "Oh, it must be ESD", when in reality it's some voltage spike or transient during turn-on that is slowly killing the chip.

I have killed many chips over the years. I don't believe I've ever killed one through ESD despite pretty lax precautions during development.

I've had customers kill many chips over the years. Same story. They're far more likely to plug in the wrong cable to the wrong connector than to damage something with ESD.

You have a fixed amount of time and energy to spend. Focus on functionality. ESD testing is relatively straightforward once you have a design.

[sb_st]

Hey @TDK, this is terrific - I appreciate the practical knowledge as much (if not more) than the academic/technical knowledge! Thank you so very much, I'm quite grateful for your insights!

[LCE]

Some things have been said before - but sometimes repetition helps... ;)

1) Each device usually has some basic ESD protection, so 3.3V protection even for 1.8V devices should be okay.

2) If want / must protect your product from st000pid users (plugging 230 VAC into the 1V input) or surge things get much harder: bigger TVS / surge capable diodes, fuses, ...

3) Think about what might really get exposed to ESD, surely not each pin / device on the PCB - leading to the next and most important point...

4) Many (inexperienced) designers just put some ESD / TVS stuff somewhere without thinking about the most important point: where do the pulses / superfluous energy go? 
It's so important to take care of input and "dissipation" *** paths - because a badly placed TVS diode might not help at all if some high voltage pulse is already in the system, might create some ground bounce, etcpp.

 

E.g. you have an input connector to the external world, so a TVS diode would best be placed not only as close to the connector, but also with its GND pin maybe not to the regular PCB's GND but better directly to the device's metal case (which should / must have some low impedance GND / earth connection) if possible.

 

*** no native english speaker here, although I do so much in English, sometimes I don't find the right words...