Automotive design - Ceramic capactor on battery supply line.

[halexa]

I have done a couple of HW designs for the automotive industry and been told that you should use double 50V ceramic capacitors on the supply line to omit single failure.
However when looking at other automotive designs I have seen the use of single 100v ceramic capacitor on the supply line.

What is your opinion and experience regarding this? I would prefer the single 100V cap from a design view.

[Dr. Frank]

Many OEMs require the usage of two CerC in series, each from different suppliers.
I think, 50V is sufficient.
We also use FlexiTerm capacitors in this place.

Frank

[Zero999]

It seems a bit unnecessary.

How about using a single X capacitor instead? It will be designed to withstand far higher voltages, than the normal 12V and will fail safely.

[SeanB]

That will work, but the designer should also place the capacitors such that they are at right angles to each other, so that flexing forces on the PCB will not cause both to fail shorted from stress, but instead will only cause the one to fail. Generally done because these capacitors will be on the edge of the board, and will be subject to flexing and stress of the connector they are close to as well, so having the 2 in series works better with placing, and the 2 capacitors are generally a lot smaller in board space that a class X film unit, plus you can place one on top and the other on the bottom of the board and connect using a via.

[Scrts]

Many OEMs require the usage of two CerC in series, each from different suppliers.
I think, 50V is sufficient.
We also use FlexiTerm capacitors in this place.

Frank

That is true. We always use two in series placed diagonally on a power supply, which is connected to battery. I think it's also related to getting short and cause fire. For really small applications like cameras, we use FlexiTerm and then put only one cap.

[max_torque]

What "failure" is this dual cap mean't to prevent?

[HighVoltage]

We usually use one 100V but only when required by the client, we use 2 x 50V

In your drawing it should be: +12V Kl.30
(Kl. 31 is usually ground in the Automotive industry)

[TimFox]

The failure that is prevented by a series connection is one of the two devices shorting out, which is always possible with MLCC stacked ceramic construction.

[Zero999]

The failure that is prevented by a series connection is one of the two devices shorting out, which is always possible with MLCC stacked ceramic construction.

I would have hoped there would be a fuse to protect against fire, if one of the capacitors failed short circuit.

[halexa]

We usually use one 100V but only when required by the client, we use 2 x 50V

In your drawing it should be: +12V Kl.30
(Kl. 31 is usually ground in the Automotive industry)

Yepp, noticed my mistake.

[Short Circuit]

This seems a rather useless feature. What is the point of preventing single point of failure if there is no way of detecting such failure?
Must MLCC caps crack on the production line (particularly during depaneling). With a single cap in place, this is detected. But with the series caps, no-one would notice... 

As a result of such single failure, also the EMI behaviour of the device would change.

[Dr. Frank]

This seems a rather useless feature. What is the point of preventing single point of failure if there is no way of detecting such failure?
Must MLCC caps crack on the production line (particularly during depaneling). With a single cap in place, this is detected. But with the series caps, no-one would notice... 

As a result of such single failure, also the EMI behaviour of the device would change.

You did not catch the problem..

These caps usually sit near the connector, therefore will be bended when the device is plugged in the car, and maybe will also fail due to vibrations.
This bending might cause cracks inside the capacitor, causing shorts.

These caps are directly connected to the battery, fusing is not possible, or is prohibited inside the application.
This creates a short circuit current, and heats the capacitor, the connector and the cable.
The short circuit current usually is not  so high, that the car fuse will break, therefore these components heat heat up and might catch fire, burning down the car in extreme situations.

So this series connection is a pure safety requirement, but mostly implemented only if the customer / OEM pays the bill, i.e. higher component price, double pick 'n place price, i.e. about 5 Cents more.
We try to mitigate the increase of cost by using FlexiTerm or similar "safe failing" MLCCs.

Frank

[Scrts]

Just out of general interest: what OEMs do not require this?

[Siwastaja]

This seems a rather useless feature. What is the point of preventing single point of failure if there is no way of detecting such failure?
Must MLCC caps crack on the production line (particularly during depaneling). With a single cap in place, this is detected. But with the series caps, no-one would notice... 

No, not at all. These MLCCs fail typically after hundreds or thousands of hours on field. Sometimes after five or ten years.

Typical failure mode is not a "short circuit" - it's a low value resistor. It's fairly typical that fusing doesn't help. The MLCC may dissipate a few watts for quite a long time, glowing red or yellow, until it burns out as open circuit, possibly taking something else with it during this time.

This is not super common, but it does happen.

Having two in series, possibly from different manufacturer, possibly in different orientation as suggested, would bring the probability of both failing during the lifetime of the product extremely low. Of course, the circuit must be designed so that the capacitance of a single cap, when put in series with the resistance of the "shorted" cap (a few ohms), still allows the device to perform. This issue indeed goes unnoticed, and if another of them does fail, then the product will fail in some way or another. But the chances of this happening are very small.

In any case, two caps in series prevents high current flowing when the first cap starts failing, preventing the quick chain reaction that makes the "short" worse.

[Short Circuit]

Ah, I see. I wasn't aware of such failure mode of MLCCs.

[T3sl4co1l]

As a result of such single failure, also the EMI behaviour of the device would change.

In a poor design, perhaps; in a good design, the circuit values will be chosen to allow for large tolerances, even just under normal circumstances (from C(V) effect and aging).  Cutting a cap in half (doubling the value) actually gives you double the headroom against all these value-reducing effects, which is... great!

That's most important for filtering caps, to deal with lower frequency EMI like SMPS ripple.  For high frequency EMI (and RFI), the values are usually "big enough not to matter", so that a large (~50%) change makes very little difference.

But it would be nice for this to actually be tested, to confirm the performance (because hey, if EMI pass/fail comes down to a couple dB, you need to know these things).  If the boards are designed for 100% test coverage and that includes series caps, that would provide an option -- also of confirming that the caps are okay without infantile / as-manufactured defects.

Tim

[max_torque]

ime, capacitor failures in automotive electronics are actually pretty benign. Because there is only around 14 or 28 volts driving the failure current, a small amount of resistance has a large effect, and the low voltage means there is not really any post failure arcing to worry about.  Typically, the component shorts, gets heated red hot and fails, sometimes taking the trace with it which acts as a fuse (depending on the thermal capacity of the failed cap)  As long as you haven't got a load of sensitive stuff packed tightly round the caps, they can fail all on their own and the device keeps working!

[gbyleveldt]

Heh, had a dead Bosch MEV17.2 in today with a failed ceramic MLCC. It went short circuit, took out the track feeding it from KL30 and popped the fuse in the car. Obviously, the car wouldn't start. I believe the car was jump started 30 minutes prior to the cap failing. This was a weird one...

[JiggyNinja]

I would have hoped there would be a fuse to protect against fire, if one of the capacitors failed short circuit.

Safety considerations might forbid the use of a fuse.

If something goes haywire in your radio and it blows a fuse, it's no big deal. Your ride is a little bit more boring, but safety is not compromised. You'll still get to your destination just fine.

If a fuse in your brake controller blows out...just use your imagination.

And it's not just in automobiles. I believe the National Electric Code in the US forbids using a fuse or circuit breaker on the supply to a fire pump.

Basically, there are some devices that must try to operate no matter what even if they destroy themselves or other devices, because disconnecting them is even worse.