-
Ceramic capacitor will get damaged if repeatedly overvoltaged by spike?
Posted by
ocset
on 14 Nov, 2019 20:55
-
Hi,
I exposed a 630V rated ceramic capacitor (1812, X7R) to a spike voltage of 680Vpk. (the spike voltage as in the attached)
It did not blow up. In spite of the fact that I exposed it to this spike repeatedly, every 4 seconds, over a 5 minute period. Do you think I have damaged this capacitor in any way?
How long would it be before this capacitor did get damaged, when treated like this?
-
#1 Reply
Posted by
Gyro
on 14 Nov, 2019 21:08
-
Hi,
I exposed a 630V rated ceramic capacitor (1812, X7R) to a spike voltage of 680Vpk. (the spike voltage as in the attached)
It did not blow up. In spite of the fact that I exposed it to this spike repeatedly, every 4 seconds, over a 5 minute period. Do you think I have damaged this capacitor in any way?
Yes treez, that's why it has a
Maximum 630V voltage rating.
Ever heard of de-rating?
P.S.
How long would it be before this capacitor did get damaged, when treated like this?
You seriously expect an answer to that one?
-
#2 Reply
Posted by
ejeffrey
on 14 Nov, 2019 23:39
-
Maybe? It's only about 10% overvoltage, but it is above the rating.
-
#3 Reply
Posted by
vk6zgo
on 14 Nov, 2019 23:58
-
Hi,
I exposed a 630V rated ceramic capacitor (1812, X7R) to a spike voltage of 680Vpk. (the spike voltage as in the attached)
It did not blow up. In spite of the fact that I exposed it to this spike repeatedly, every 4 seconds, over a 5 minute period. Do you think I have damaged this capacitor in any way?
How long would it be before this capacitor did get damaged, when treated like this?
Probably not.
The "630v" rating is a standard value, so there is a fair chance that the cap is really capable of operating at higher voltages.
In addition, the duty cycle is very low, so again, unlikely to cause damage.
All that said, it is very poor practice, &, no, you can't extrapolate from damage being unlikely in this scenario, to one where the cap is subject to over-voltage either continuously, or intermittently, over much longer time periods (days, weeks, years).
-
#4 Reply
Posted by
ocset
on 15 Nov, 2019 06:37
-
Thanks,
I must admit, I once worked in a huge multinational aerospace company. I was doing a board where I needed more capacitance but didn’t have room. We weren’t allowed electrolytics, had to be ceramics. I was publicly trashed in front of the whole office by the Chief Engineer who told me that I should have used ceramic capacitors which were only rated for about half the rail voltage (because they were smaller and I would fit more of them on) …because as he said, ceramic MLCC capacitors can easily withstand this, and he said it was a mark of my ignorance that I was not aware of this.
-
#5 Reply
Posted by
AndyC_772
on 15 Nov, 2019 07:06
-
The voltage rating of an X7R cap has just as much to do with its capacitance-voltage characteristic as it does with the stress the part is physically able to withstand.
Applying a high voltage to a ceramic cap effectively saturates the outer layers of the conductive plates with electrons, which means any further electrons have to occupy space which is further inside the conductive material. This boundary layer effectively increases the separation between plates, which reduces the capacitance (recall C = εA/d).
It may not be good practice, but the chances of damaging the part with such a relatively small spike are tiny.
Depending on how you're measuring the spike, it may not even be as big as you've seen on the scope. A small overshoot can easily be the result of some stray inductance or EMI picked up on the cable.
Your Chief Engineer is wrong, but not not because the caps will necessarily be damaged by a 2x over-voltage. They just won't actually work.
-
-
Thanks,
I must admit, I once worked in a huge multinational aerospace company. I was doing a board where I needed more capacitance but didn’t have room. We weren’t allowed electrolytics, had to be ceramics. I was publicly trashed in front of the whole office by the Chief Engineer who told me that I should have used ceramic capacitors which were only rated for about half the rail voltage (because they were smaller and I would fit more of them on) …because as he said, ceramic MLCC capacitors can easily withstand this, and he said it was a mark of my ignorance that I was not aware of this.
sounds like a chief bullshitter
-
#7 Reply
Posted by
dmills
on 15 Nov, 2019 12:38
-
The guy was a tool, you might get away with it, but it only takes having to do a recall ONCE to eat a hundred times any savings you might have made.
Also, even if it works today, it might not work on a future batch of notionally the same part, and the manufacturer will laugh at you.
The datasheet tells you what the manufacturer is prepared to put there name to, going outside datasheet means you have to do all the qualification on the parts yourselves (Including potentially per batch qualification). Now sometimes (Down hole, high or low pressure, cryo, some space) you have really no choice but it is always better not to go there if you can avoid it.
-
-
I think if its what he said it is, they should have got the manufacturer to re-rate with actual testing and go by a real MTBF (would someone still buy it though?) or to make a new part (wow, pushing the industrial bar), its cowboy behavior
and to make fun of engineer following the absolute maximum.. damn
-
#9 Reply
Posted by
T3sl4co1l
on 15 Nov, 2019 14:38
-
Well, C(V) is a useless argument here, as all type 2 caps will be long gone at this voltage.
C0G stores far more energy at higher voltages, say over 200V or so I think. Costs more, but you aren't getting anywhere near the nameplate value in X7R which ends up more expensive, and bulkier besides.
treez: read the datasheet, capacitors are rated, and tested, for momentary overvoltage. The datasheet will say how much and for how long.
Tim
-
#10 Reply
Posted by
T3sl4co1l
on 15 Nov, 2019 14:45
-
Applying a high voltage to a ceramic cap effectively saturates the outer layers of the conductive plates with electrons, which means any further electrons have to occupy space which is further inside the conductive material. This boundary layer effectively increases the separation between plates, which reduces the capacitance (recall C = εA/d).
This is an interesting observation, but I'm not sure you are aware what the electron density is in metals--?
It is an important figure for semiconductors, though.
I wonder if silicon capacitors are heavily enough doped that this isn't a factor? Nonlinear capacitance would be apparent (a MOS or SOS capacitor has less capacitance near zero bias, IIRC, because the depleted surfaces effectively thicken the capacitor; under bias, the boundary layer (holes or electrons) comes right up to the insulator).
Tim
-
#11 Reply
Posted by
AndyC_772
on 15 Nov, 2019 15:51
-
I've no doubt the reduction in capacitance is a complex function involving the combined electrostatic field of all those electrons, and the definition of the Volt.
Once upon a time, I could probably have calculated it. Nowadays, I think I prefer a simplified explanation, possibly backed up with some real world measurements to get the point across.
I do have an impedance analyser which could do it, but sadly real work always seems to get in the way.
-
#12 Reply
Posted by
T3sl4co1l
on 15 Nov, 2019 17:16
-
Saturation of ferroics seems to be a statistical process, averaging to a simple power-law asymptote. I don't think I've seen what function a typical ferrite B-H curve approximates, but the typical curves for capacitors gives an approximately linear E(V) curve. Which implies C ~ 1/V or thereabouts. (Or probably more like sqrt(1 / (V^2 + Eo)) since it's symmetrical and has a linear range near zero. Also an offset in V for electrets.)
Tim
-
#13 Reply
Posted by
SeanB
on 15 Nov, 2019 19:29
-
Thanks,
I must admit, I once worked in a huge multinational aerospace company. I was doing a board where I needed more capacitance but didn’t have room. We weren’t allowed electrolytics, had to be ceramics. I was publicly trashed in front of the whole office by the Chief Engineer who told me that I should have used ceramic capacitors which were only rated for about half the rail voltage (because they were smaller and I would fit more of them on) …because as he said, ceramic MLCC capacitors can easily withstand this, and he said it was a mark of my ignorance that I was not aware of this.
Now we know who the Boeing designer of MCAS was........
There is a reason why you can get MLCC units with a built in fuse, and just because they cost more is not one of the reasons.
-
#14 Reply
Posted by
Gyro
on 15 Nov, 2019 20:22
-
Yes, MLCCs do blow. At low voltage, high current they will burn up and damage the board. At 630V+ with reasonable reservoir capacitance and/or a relatively direct link to the mains involved (it looks like the steady state voltage is about 340V), it is likely to be rather more catastrophic.
-
#15 Reply
Posted by
TimFox
on 15 Nov, 2019 23:16
-
In my youth, I noticed that mica or paper capacitors in (relatively high voltage) vacuum-tube equipment often had two ratings: working voltage and test voltage, where the latter was usually twice the former. Obviously, the manufacturer did not recommend operation at the higher voltage, but thought it safe to operate at the lower voltage after they tested it at the higher voltage for a short time without failure.
-
#16 Reply
Posted by
AndyC_772
on 16 Nov, 2019 08:15
-
Yes, MLCCs do blow. At low voltage, high current they will burn up and damage the board.
Agreed. I've seen a number of boards fail over the years with a short across the power supply rails, which turned out to be a failed decoupling capacitor. A thermal camera is often a good way to spot which one has died.
In my experience MLCCs are much more susceptible to mechanical damage during handling and soldering than they are to electrical stress. It's very easy to fracture one and cause the plates to short if you're careless.
-
#17 Reply
Posted by
BravoV
on 16 Nov, 2019 08:22
-
I guess this is more a language problem thread, rather than tecnnilcal, as the OP doesn't fully understand the words mentioned in component's datasheet like -> "Rated voltage" or "Absolute maximum rating".
-
#18 Reply
Posted by
Gyro
on 16 Nov, 2019 10:43
-
For Electrolytics you will often find a time specified 'surge rating', probably dating back to the time taken for the valve heaters to warm up, but probably still relevant where the supply 'peaks' before something starts up.
Manufacturers 'Test' or 'Withstand' voltage is a one-time test under specific test conditions, you can't rely on it for normal operation.
As far as I'm aware, "Absolute maximum rating", only applies to semiconductors - and then only ICs. Other factors, such as SOA are much more important for discretes.
Operating above Rated voltage is a no-no as far as I was taught. De-rating is the order of the day for reliability. With MLCCs (in fact all caps) dissipation is also a critical factor. Thermal dissipation and eventual breakdown due to current waveforms is every bit as important. The OP didn't specify the capacitance value that he was pumping these voltage spikes into, and hence the current and thermal dissipation involved.
-
-
KOA/Speer has a good document which goes into great detail on the very subject of this thread:
http://www.koaspeer.com/pdfs/MLCC.pdfFrom said document it is seen that Class 1 dielectrics (C0G, U2J, CH, etc.) will tolerate a way higher peak voltage than Class 2 or 3 dielectrics (X7R, Y5V). Figures 2-5 show the spread of DC breakdown voltage vs. initial voltage rating, and a 1nF/1206/50V C0G type will withstand nearly 6kV while an X7R version will crack (perhaps literally) at around 3.5kV*. The difference in breakdown voltage for AC is even more dramatic as the higher dissipation factor (ie - AC losses) of the Class 2/3 dielectrics add more self-heating to the mix from ESR. Finally, the pulse withstand capability is summarized in Figs. 17 and 18 where Class 1 dielectrics prove superior yet again.
* - These are "instantaneous failure" voltages, so not something the capacitors are expected to withstand. Still, you can reasonably conclude that brief excursions of a few % above rated voltage are unlikely to cause any damage whatsoever.
EDIT - added * comment.