Surge voltage rating on MLCC caps?

[PwrElectronics]

Been doing some searching and so far not finding much on this topic...

Situation:
A sensor supply tracking regulator goes out of the product and could be subjected to wiring harness faults.  The regulator IC can withstand I think it was 45-50V.  The system could have 60V.  So, the design had a 100V blocking diode to protect the regulator and a 25V MLCC was used as the filter cap since that diode would be there to protect it as well.

So, customer was complaining the sensor supply voltage is not precise enough (there is even a ADC on the micro to read it for ratiometric adj....)  The load regulation with the diode in series is pretty poor I agree.  So, a revision got rid of the diode and changed the agreed upon requirement spec so only a 40V short was allowed.

Now, I am circling back on documentation I see this 25V cap sitting there that got overlooked in the rush.  Ugh.     The test lab actually did the 40V shorts test and nothing failed...  The spec calls for a 1 min duration short.

What to do?

The proper solution is get a 50V cap but I cannot do a board layout rev now if I can't find one that fits the same space.  So far, seems like I can get a 50V part of the same C and size if I go with a wider tolerance on the dielectric.  The problem there is the minimum C value is below the regulator requirements.  Actually, even the current part fails that in the corner conditions.  Need to get with the application engineer for that regulator vendor for more details than the datasheet is telling me...

So, the bottom line question here at the bottom line...  Are there any surge ratings common for MLCC caps?  not mentioned in 2 of the 10 or so approved vendor datasheets for this part in my employers system.

[Vovk_Z]

Surge voltages are not stated for MLCC ceramics typically.
But from my experience almost all caps except electrolytes (I mean film, ceramics etc) have quite a good surge voltage capability. I guess ceramics can withstand up to 4x...5x rated voltage or so.

[jkostb]

I don't know what level of surge voltages you are thinking about. But what I do know is that ESD strikes can degrade/damage MLCC. The smaller package are more susceptible. A surge has much higher energy levels and could easily destroy MLCC's.
I don't think that MLCC will withstand surge pulses as specified in EMC standards e.g. IEC61000-6-1


https://incompliancemag.com/article/effectiveness-of-multilayer-ceramic-capacitors-for-electrostatic-discharge-protection/#:~:text=In%20reality%2C%20MLCC's%20exposed%20to,in%20excessive%20low%20frequency%20leakage.

[jonpaul]

Surge exposure calss 1..2..3...4?

Jon

[PwrElectronics]

On this design, there is a 100V 0.1uF MLCC for ESD right off the connector.  This filter for the regulator is a larger C (10uf) and physically larger and right at the regulator pins.  My thoughts on surge ratings are if similar to a electrolytic where there is usually a surge rating.  If this MLCC was such a cap, a 25V rated one would have a surge rating of 32V per one chart I looked at.

The short circuit requirements are the device must survive a DC short to X voltage lasting Y minutes.  In this case, my 25V cap getting 40V for 1 minute.  ESD test was a separate test.  We do have internal design guidelines on caps used for ESD at connectors since not all of those will survive.  The one lab usually have a test board made every so often and run ESD on it to see what lives and what dies.  Some ESD requirements we have to design for are pretty severe.

I did find one paper from TDK with some squishy wording about this suggesting caps could withstand significant surges but cautioned the designer about possibly reliability impact but then gave no suggestions on how to analyze it.

[jkostb]

Big problem with testing is that you don't know whether the applied surge damages the MLCC structure resulting in permanent damage. Your test maybe passes, but due to structural damage the capacitance changes.  Then later when second surge strike is applied it may fail in short circuit. There is a a lot of research papers on this topic e.g.

Analysis of ESD-Robustness of Multi Layer Ceramic Capacitors in System Applications

Question then is does this also applies for surge?

[langwadt]

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

[brumbarchris]

I think this is the best piece of advice:

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

[PwrElectronics]

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

A good idea but did not pan out yet.  At least in the same mfg series as the current part, a 50V does drop less but not enough less to matter much.  Different brands sometimes have different curves so need more investigation.  This unit can be setup on board population to put out a nominal 5v or nominal 12v for the sensor supply.  So, the 12v case at min C is the corner.  Regulator data sheet implies 7uF min but the way they do it is not stated as some actual hard value so hoping to get with the app engineer for that part.  The next incremental higher C would 15uF from my looking but those don't seem to be available at 50v w/o a change in package dimension.

But, I almost forgot that for the 12v application, I left in the blocking diode so I don't need to survive shorts to anything.  I think I need to be more detailed with the design docs to account for more of these permutations!

One mitigation factor and why the customer agreed to the 40V short instead of 60V in their system is the sensor supply wires run on a different connector and harness bundle from the 60V power supply so odds of a short to that voltage not significant.  However, with the spec saying 40V, this could get picked up by another product application group to put in a 24v system where the short spec is 36V and maybe their harness bundle is different.

At least I have a little time to figure something out.  Otherwise, the initial application is literally headed to production in a couple weeks and BOM changes and especially PCB changes not welcome!  We made prototypes on our own SMT line but the production is being sent out to contract mfg and they don't do changes quickly.

[Someone]

I think this is the best piece of advice:

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

That is nice advice and all, but nothing to do with voltage withstand/surge. Despite being offered in the same case, nominal capacitance, C vs V curve, and from the same manufacturer, failure at voltage stress is not equal. Above or below the rated voltage.

Some capacitors are 100% tested to some withstand voltage such as 2x or 3x the nominal, but it's only a screening test during manufacturing and not a warranted performance parameter in use.

I did find one paper from TDK with some squishy wording about this suggesting caps could withstand significant surges but cautioned the designer about possibly reliability impact but then gave no suggestions on how to analyze it.

Politely reminding the customer to do their own reliability/quality trade off, if its not in the datasheet then you're on your own to do that work. Don't want to take on the risk of failure or the costs of extensive testing? then buy a part rated by the manufacturer for the job.

[langwadt]

I think this is the best piece of advice:

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

That is nice advice and all, but nothing to do with voltage withstand/surge. Despite being offered in the same case, nominal capacitance, C vs V curve, and from the same manufacturer, failure at voltage stress is not equal. Above or below the rated voltage.

the point was that a 5uF 50V @12V might might have just as much capacitance as a 10uF 25V @12V, so getting a capacitor with the voltage rating that meets spec might not lose any capacitance even with a smaller uF rating

[Someone]

I think this is the best piece of advice:

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

That is nice advice and all, but nothing to do with voltage withstand/surge. Despite being offered in the same case, nominal capacitance, C vs V curve, and from the same manufacturer, failure at voltage stress is not equal. Above or below the rated voltage.

the point was that a 5uF 50V @12V might might have just as much capacitance as a 10uF 25V @12V, so getting a capacitor with the voltage rating that meets spec might not lose any capacitance even with a smaller uF rating

Examples? Most capacitors are specified for their (almost) peak capacitance, and most capacitors of the same composition/material follow the same C vs V curve.

[thm_w]

Could squeeze in a larger package type into the existing layout, if there is space? Maybe have to manually solder it in if its a really big difference.
I tested a 25V 0805 X7R and it hit 358V before it popped. No guarantees of course.

I think this is the best piece of advice:

check the C vs. voltage rating for the 25V and 50V cap, a nominally smaller value 50V cap might have just as much C as the 25V cap at the running voltage

That is nice advice and all, but nothing to do with voltage withstand/surge. Despite being offered in the same case, nominal capacitance, C vs V curve, and from the same manufacturer, failure at voltage stress is not equal. Above or below the rated voltage.

the point was that a 5uF 50V @12V might might have just as much capacitance as a 10uF 25V @12V, so getting a capacitor with the voltage rating that meets spec might not lose any capacitance even with a smaller uF rating

Package size is what matters, generally.
https://www.eevblog.com/forum/projects/is-ceramic-capacitor-package-relevant/msg4480996/#msg4480996