Hi All,

In one of our design we have manufactured LED strip PCB.
In the design, there is LDO to regulate input voltage to 7.5VDC(V_MBI) from 24VDC(VIN) to feed an IC.

As seen,
F4 : 48V 120mA OZCJ0012FF2E as poly fuse,
D9 : SMAJ26A as TVS.
C7 : CC0603KRX7R9BB104 0.1uF 50VDC 0603 MLCC capacitor
PTC-TVS block used to protect voltage input line for overcurrent & overvoltage conditions.

** There are 24 meters LED strips fed by one PSU.
** PSU unit (MW LRS-350-24) is 4meter far away and it is connected with 2x2.5mm² cable to LED strips.
** One power feedback cable is also connected to end of strip to reduce stress on PCBs.
** One meter LED strip is 12W.

Under given conditions, rarely C7 capacitor fails or blows.
What could be the cause, because there are overcurrent & overvoltage protection devices.



Pleased your valuable responses.

 

C7 is of poor quality.
No other ideas

 

How much power is being dissipated in the linear regulator? This has to be the least efficient design I've ever seen?

 

I have never encountered a blown 0.1μF capacitor.

 

How much power is being dissipated in the linear regulator? This has to be the least efficient design I've ever seen?

LDO feeds just one IC and it needs max 10mA.
According to this, it is like 0.165W.

 

As Pyrex says, the capacitor C7 may be of poor quality; or there is excessive current/voltage ripple present shortening the life of the capacitor. Put a scope probe on the C7 voltage and check that any resulting capacitor voltage variations/currents are within the component’s specification.

Many years ago the company I worked for noticed that they were suffering a high failure rate of a capacitor rated at 35Vdc, used on a 24Vdc rail. On contacting the capacitor manufacturer they advised that with a constant 24Vdc applied to the capacitor, they would expect to see such a failure rate (despite there being more that 10V headroom from the applied voltage to the component’s rating).

 

I have never encountered a blown 0.1μF capacitor.

Occasionally I find 1uF 50V 1206 MLC capacitors that fail short - and we only buy them from reputable suppliers.
Very occasionally the end caps fall off, but that's only when I solder them by hand.

 

Many years ago the company I worked for noticed that they were suffering a high failure rate of a capacitor rated at 35Vdc, used on a 24Vdc rail. On contacting the capacitor manufacturer they advised that with a constant 24Vdc applied to the capacitor, they would expect to see such a failure rate (despite there being more that 10V headroom from the applied voltage to the component’s rating).

Amazing
Can you tell us the manufacturer of those capacitors?

 

Amazing
Can you tell us the manufacturer of those capacitors?

I got my capacitors from known distributor and manufacturer is Yageo.

 

LDO feeds just one IC and it needs max 10mA.
According to this, it is like 0.165W.

I thought you were powering the LEDS from the regulator.

 

With regards the 35V rated capacitors failing on a 24V rail, I was not aware of the manufacturer/supplier – but when I suggested my employer sue the company it was pointed out that there were all sorts of potential issues such as proving that the capacitors were used within temperature/peak voltage/current etc. The only likely winners being the lawyers.

 

With regards the 35V rated capacitors failing on a 24V rail, I was not aware of the manufacturer/supplier – but when I suggested my employer sue the company it was pointed out that there were all sorts of potential issues such as proving that the capacitors were used within temperature/peak voltage/current etc. The only likely winners being the lawyers.

Yes it's like that...

In my practice there were many capacitor faults in the switch mode PSU with power factor corrector. A capacitor battery is charged up to 380V from 230V mains , capacitors were rated to 400V , 220uF . PSU operate 7/24 . Faults started in two years . Anyway, it's predictable...

But if a 35V capacitor fails at 24V

 

Under given conditions, rarely C7 capacitor fails or blows.

I read this as you were surprised that the capacitor was not failing more often, and your question was why not. The tone of others here suggests they read it the opposite way, that you're surprised it ever fails. This seems more likely what you meant.

Choosing a cap with a rating at least double the maximum expected voltage seems a decent rule of thumb. A 35V rated cap in 24V service sounds acceptable, but you really need to look at the nominal voltage as a statistical thing. How often does it take an excursion to much higher voltages? Even a one-in-a-thousand chance will soon kill your cap.

More generally, we all need to be aware that ratings don't mean ANYTHING unless the test method is specified and the quality control system defined. Some things (like fuse performance) are defined by industry standard test protocols. We come to rely on such ratings and only get bit if the manufacturer is outright fraudulent. But many "ratings" are defined under conditions chosen by the manufacturer and not something you or I would consider remotely realistic or meaningful. Maybe that "35V" capacitor can survive a nanosecond at 35V and three nanoseconds at 30V.

 

I don't know how manufacturers test and spec their components.
I would expect a capacitor rated for 35V should be able to withstand 45V or even 50V indefinitely. I would expect manufacturers would have built in a margin of safety.

On top of that, as an engineer, I would add my own margin of safety. Therefore, I would not use a 35V component in a 35V circuit. I would select one rated for 50V or 63V or even higher. As @wayneh says, go for double the working voltage.

 

Wh does everone keep talking about a 35V caiacitor when the TS stated:

C7 : CC0603KRX7R9BB104 0.1uF 50VDC 0603 MLCC capacitor

 

Wh does everone keep talking about a 35V caiacitor when the TS stated:

Because they are talking about the headroom, not the absolute rating. The 35V capacitor was in a circuit (not the TS') where there was an expected peak of 25V. Yet, with 10V of headroom it was still failing, and the manufacturer said this was expected.

The case of the 35V cap is an analogy, and suggests that the 50V cap might be suffering from something similar. The voltage of the TS' capacitor doesn't change the potential value (no pun intended) of discussing the 35V cap raised as an example.