Maker Pro
Maker Pro

Any criteria on bypass capacitor selection?

M

Myauk

Jan 1, 1970
0
Dear all,

I am now checking on a PCBA on which the power rail for digital
circuit (microcontroller and its peripherals) uses many 0.1uF, Y5V,
0805 capactiors in parallel between VCC and Gnd.

I wonder why 0.1uF capacitor was used there?

I could also see 4.7 uF used together with 0.1uF capacitors.

I would also like to know the criteria to choose voltage ratings on
bypass capacitors.

Is it safe to use 0.1uF, 50V bypass capacitor for 24V supply rails
which drive inductive loads?

Regards.
 
R

Raveninghorde

Jan 1, 1970
0
Dear all,

I am now checking on a PCBA on which the power rail for digital
circuit (microcontroller and its peripherals) uses many 0.1uF, Y5V,
0805 capactiors in parallel between VCC and Gnd.

I wonder why 0.1uF capacitor was used there?

I could also see 4.7 uF used together with 0.1uF capacitors.

I would also like to know the criteria to choose voltage ratings on
bypass capacitors.

Is it safe to use 0.1uF, 50V bypass capacitor for 24V supply rails
which drive inductive loads?

Regards.

You are after the lowest HF impedance between VCC and Gnd. Each
capacitor has an inductance. Parallel capacitors increases capacitance
and reduces inductance.

As a matter of principle I don't use Y5V, it may be OK in this
application but they have high voltage coefficients and temperature
coefficients of capacitiance. So the capacitance is always changing
dramatically. I would treat a 0.1uF Y5V as a 10nF.

I would fit at least a 100V caps on the 24V rails unless I had reason
to know a lower voltage would be OK.
 
M

Myauk

Jan 1, 1970
0
As a matter of principle I don't use Y5V, it may be OK in this
application but they have high voltage coefficients and temperature
coefficients of capacitiance. So the capacitance is always changing
dramatically. I would treat a 0.1uF Y5V as a 10nF.

Please let me know what type of dielectric material would be the best
for this application.

I would fit at least a 100V caps on the 24V rails unless I had reason
to know a lower voltage would be OK.

The reason that I am looking into the use of this capacitor is because
one of the 0.1uF, 50V, 0805, at 24V rails is found severely burnt.

The average current drawn in the circuit is around 3 to 4 Amperes.

Although 0.1uF and 4.7uF capacitors (50V rating) are placed in
parallel, only one particular 0.1uF is burnt, the rest of the
capacitors are still OK.

Please help me to give some suggestions on this issue.

As I am a newbie in this issue, I am totally lost. At least I would
like to request some help for the start. Any article discussing on
such kind of issue will be of great help for me.


Thanks and Best Regards
 
J

J.A. Legris

Jan 1, 1970
0
Ive seen high ripple current burn .1uf ceramic caps.  This was from a power
opamp oscillating and dumining 4 amps at 500Khz through the cap.  The ESR is
low on these caps but not zero.

Also take a look at the location of the cap relative to the load.  If the
one closest to the load got popped there is likely inductance between that
cap and the other ones so it was taking more than its fair share of the
ripple current.

I think X7Rs are less lossy than Y5Vs nd can handle more ripple current.

If they are SMT caps you could have had one cracked during soldering.  The
crack eventially turns into a short and releases the magic smoke.

Nice article with several worked out examples:

http://www-micrel.deis.unibo.it/~augusto/bypass1.pdf
 
E

Eeyore

Jan 1, 1970
0
Myauk said:
Please let me know what type of dielectric material would be the best
for this application.

One with less tempco. NEVER use use those cheap brown plate looking things.
They seem to be designed as fire starters.

The reason that I am looking into the use of this capacitor is because
one of the 0.1uF, 50V, 0805, at 24V rails is found severely burnt.

Not uncommon. It would appear to be an inrush current dominated effect.
I've seen this happen myself in a production environment where we had to do
a formal change.

Best result was to use 'box polyester' type capacitors instead. They seem
to be far more durable. I've even seen dipped wound mylar types burn up. I
bet it happened at power-on.

A smaller value cap may be more appropriate as well.

Graham
 
M

Myauk

Jan 1, 1970
0
Not uncommon. It would appear to be an inrush current dominated effect.
I've seen this happen myself in a production environment where we had to do
a formal change.

Best result was to use 'box polyester' type capacitors instead. They seem
to be far more durable. I've even seen dipped wound mylar types burn up. I
bet it happened at power-on.

A smaller value cap may be more appropriate as well.

Graham


But I wonder why only the 0.1uF was burnt but the other 4.7 uF
capacitor in parallel with it was not burnt under the same condition.

Reagards
 
M

Myauk

Jan 1, 1970
0
But I wonder why only the 0.1uF was burnt but the other 4.7 uF
capacitor in parallel with it was not burnt under the same condition.

Reagards

One more thing...

When I tried to capture the 24V power rail flucactions due to
different load turn on/off in PCBA, I don't see any spike large enough
to kill the capacitor.

I could only see a momentary voltage dip less than 1V on the 24V
supply rails.

Any suggestion for this?

Was my approach a wrong approach?

Regards
 
N

Nobody

Jan 1, 1970
0
Just a guess: smaller capacitor = less thermal mass.
One more thing...

When I tried to capture the 24V power rail flucactions due to
different load turn on/off in PCBA, I don't see any spike large enough
to kill the capacitor.

I could only see a momentary voltage dip less than 1V on the 24V
supply rails.

Any suggestion for this?

Was my approach a wrong approach?

What's the impedance of the power rails? A 1V drop could indicate a lot of
current, and it's the current spike which will kill the capacitor.
 
E

Eeyore

Jan 1, 1970
0
Myauk said:
One more thing...

When I tried to capture the 24V power rail flucactions due to
different load turn on/off in PCBA, I don't see any spike large enough
to kill the capacitor.

No surprise.

I could only see a momentary voltage dip less than 1V on the 24V
supply rails.

Any suggestion for this?

Was my approach a wrong approach?

You were looking at the wrong thing. Look at the charge current at switch-on. Sadly
very few manufacturers specify what the cap is capable of. If you connect a modest
value electrolytic cap across the supply as well, it may help limit the supply
dV/dt and hence the dI/dt into the decoupling caps. That would be one approach
that's nice and simple.

But the scenario you've described is one of those little dirty secrets the cap
manufacturers like not to talk about.

I would personally fit no larger than 47nF these days too.

Graham
 
E

Eeyore

Jan 1, 1970
0
Nobody said:
Just a guess: smaller capacitor = less thermal mass.


What's the impedance of the power rails? A 1V drop could indicate a lot of
current, and it's the current spike which will kill the capacitor.

Exactly. A relatively little known phenomenon.

Graham
 
M

Myauk

Jan 1, 1970
0
Now I am testing the same PCBA in the same working system with 0.33uF,
X7R, 16V rated voltage capacitor, while the supply is 24V. I turn the
3A load on and off for a few times, it has been more than an hour and
the capacitor has been alive until now!

Any suggestions to repeat the burnt capacitor failure?


One article from KEMET also says that it's FLEX CRACK, a mechanical
crack on the capacitor will ignite it occasionally.


Regards..
 
Top