0603 vs 1206 cap case size

[shadewind]

Hi,

I'm making a design based around the TPS40211 boost LED driver and the data sheet says there should be a 1 µF bypass cap on the BP pin. I did add that to my schematic but I used an 0603 size cap instead of the 1206 size cap they used themselves in the example circuit. Is there any drawback to this? The advantage is, obviously, size.

Hmm... now when I look at Farnell, there seems to be very few 1 µF 0603 caps with X7R dielectric although X5R is very common.

Should I reorder the PCB with a 1206 footprint instead?

[ejeffrey]

No problem.

[shadewind]

Wouldn't an X7R cap be better then?

[Zero999]

It's a bypass capacitor so it shouldn't make any difference.

[shadewind]

Why is that? I thought bypass caps should always have as good dielectric as possible?

[mikeselectricstuff]

Depends on the temperature range. 0805 would be a default size for 1uf, 0603 for 100n and 1206 for 10uf - it's generally better to avoid the absolute smallest size for a value unless really necessary, if nothing else you tend to have a wider choice of suppliers/dielectrics and lower cost.

[ejeffrey]

If you have to do another revision anyway, go ahead and up the size if you have space.  Otherwise I wouldn't bother.

[Zero999]

Why is that? I thought bypass caps should always have as good dielectric as possible?

Characteristics such as stability and tolerance aren't important for bypass capacitors, all that's needed is a capacitor with a low ESR.

If it were used in an oscillator or filter then yes the tolerance and stability are important but a bypass capacitor isn't critical.

[alm]

Z5U caps lose a lot of capacitance if a DC voltage is applied, something that is quite common for bypass applications. Capacitance may be decreased by as much as 70% if the full rated voltage is applied.

[gxti]

All other parameters being equal, the smaller component will have less parasitic inductance and thus be more effective at decoupling. Xilinx app notes go on and on about how to squeeze out every last drop of capacitance, but for this application it likely does not matter at all. Use whatever you find easiest to work with.

[alm]

If parasitic inductance of the package is an issue, you're not likely to need 1uF or 100nF unless you're drawing huge amounts of currents. More like 100pF. Of course more capacitance doesn't hurt, but the largest value X7R in the same package as a larger value Z5U will likely perform better than Z5U as far as high frequency is concerned. And a smaller NP0 capacitor in a smaller package even better.

The 'grab a random 100nF cap, dielectric doesn't matter' mantra originates from the TTL (if not earlier) days when speeds were orders of magnitude slower than today.

[shadewind]

Yes, I see. The reason I'm using 1 µF is because the pin description specifically mentions that a 1 µF cap should be connected between that pin and ground.

[alm]

For a LED driver which will draw a fair amount of current and probably a fairly low frequency, any ceramic cap should be fine as long as you account for the voltage effect (decrease of capacitance as function of DC voltage). Since this is relative to the rated voltage, getting a cap with a higher rated voltage also helps.

[shadewind]

This is just the bypass cap for the driver chip itself. It uses an external FET so that is bypassed by other caps.