1
\$\begingroup\$

I'm in the middle of a project to create a small development board using a 100Mhz Cortex M4 MCU and have to buy some decoupling SMT capacitors.

I've found the following options:

Cheap 10 cent from Mouser
Expensive 50 cent from Digikey

Both appear to have the same properties but the more expensive one is listed as being for "decoupling".

I did find this link about the expensive Johanson capacitor. It appears to be a special purpose built low ESR capacitor that is said to "replace" the need for multiple capacitors.

Will the cheap capacitor be fine or should I spend the extra money on the more expensive capacitors and why?

\$\endgroup\$
  • \$\begingroup\$ That part is an 0805, which means it is probably going to end up further from the pins than something smaller, quite possibly largely defeating the advantage. You might consider 0402's up close and 0603's just outside - it will probably work on the bench with just some of the 0603's manipulated into place by hand, but having the 0402's skillfully or automatically installed will quiet later units down. \$\endgroup\$ – Chris Stratton Jan 21 '15 at 1:42
5
\$\begingroup\$

The picture on page 12 of the Johanson datasheet shows the difference between it and a normal multilayer ceramic capacitor (MLCC):

enter image description here

The part on the left is a normal MLCC. The part on the right is the Johanson part.

The Johanson part is a 4-terminal part, actually a capacitor network with two separate capacitors with one shared terminal. By connecting the "A" and "B" terminals together, you obtain twice the specified capacitance. Because of the feedthrough shape of the ground terminal you also probably achieve lower equivalent inductance and a higher self-resonant frequency.

That said, 90% of designs can be done using normal MLCCs for decoupling. If you need higher self-resonant frequency, I'd recommend to first try using a smaller package than 0805, and only if that isn't adequate, consider using one of these specialized high-cost parts.

\$\endgroup\$
  • \$\begingroup\$ +1 on the last paragraph. I was doing some research before bringing it up, but the SRF of all the 0805 100nF caps I've looked at in the last few minutes is around 11 MHz. Useless for decoupling at 100MHz. \$\endgroup\$ – Matt Young Jan 21 '15 at 1:33
  • \$\begingroup\$ @MattYoung, Murata has a series, NFM15..., similar to the Johanson part, but the A and B terminals are internally connected instead of the G terminals. The NFM15PC104D0J3 is 0402 size, 0.1 uF, and SRF is between 100 and 200 MHz. Availablity, unfortunately, seems to be nonexistent. \$\endgroup\$ – The Photon Jan 21 '15 at 1:44
2
\$\begingroup\$

These appear to be two different parts.

The Mouser part is a bog-standard ceramic multi-layer chip capacitor (us old-timers call them "mono-caps"). The Digikey part is a feed-through capacitor. This type of capacitor provides better isolation of the input and output terminals. Note the the two end terminals are connected together inside the capacitor and the middle metalization is the ground terminal.

I would use the cheap capacitor from Mouser. Note that Digikey also sells similar capacitors, for about the same money. In other words, order the caps from whoever you are ordering the rest of your parts from.

One final note: be aware that even X7R capacitors can exhibit interesting but unpleasant effects as a result of temperature changes and voltage changes. In particular, certain ceramic capacitors can lose as much as 90% of their capacitance as the voltage across them increases. The data sheet for your particular device has the details. Also note that these effects differ between manufacturers and even between device families from the same manufacturer.

As they say, the devil is in the details.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.