# Do capacitor packages make a difference in MLCCs?

I have a circuit which has an RC dependent timer. I got two types of ceramic capacitors, both of them have literally the same value, but different voltage ratings:

• 1206  4.7 μF, 25 V capacitor

• 0805  4.7 μF, 50 V capacitor

They are connected with an 18 MΩ resistor in parallel to form the RC circuit which is tied to an op-amp.

When I use the 1206 package, the timer stays for 4 minutes approximately, while when using the 0805 package, the timer stays for 3 minutes.

Why is that happening, even though both of them have the same capacity?

How can I calculate the time precisely? Is there any equation for this especially used with an op-amp, because $$\\displaystyle \tau =RC\$$ doesn't give a precise timing for this circuit?

Notes:

• The circuit is powered through a capacitive power supply.
• I used the same circuit but just swapped the two types of MLCCs so it is just one capacitor connected at a time, not both of them.
• Unfortunately I don't have any datasheet for these types of capacitors, I bought them from a local store.

Edit:

I checked the two types of the capacitors, it seems that:

• 1206  4.7 μF, 25 V capacitor is X7R.
• 0805  4.7 μF, 50 V capacitor is X5R.
• Do the manufacturer(s) provide characteristic data? What type, exactly are they, X7R? Chip thickness? Part number? If not, they can be basically anything. Commented Jun 29 at 1:39
• Aluminum electrolytic capacitor have a more constant capacitance with voltage, so if you goal is a somewhat predictable RC timer, some 10% tolerance electrolytics can be bought for a few cents on Digikey. Commented Jun 29 at 2:22
• You are better off using a plastic capacitor like polyester or polycarbonate.
– qrk
Commented Jun 29 at 2:37
• @user1850479 Most any capacitors other than type-II (X7R, X5R, X6S, etc) or type-III (Y5V, Z5U) ceramics have a pretty constant capacitance with voltage. You could also use type-I ceramic, film, tantalum electrolytic, or something weirder and they'd probably perform just as well as aluminum caps. Commented Jun 29 at 13:22
• (continued) Meaning the opamp will dominate, and thus the theoretical RC curve and τ will not hold. Note that the bias current will depend heavily on the device and temperature, so you'll likely won't get a precise timer even if you fix the capacitor. If this doesn't suit your needs, you could try reducing the R by (say) 100x and increasing your C by 100x - provided you have the space for the larger cap. Commented Jul 1 at 14:28

It is likely that the capacitors are Class2 ceramic capacitors. The capacitance is voltage dependent. The voltage rating has little effect.

Here is a graph showing this effect from Analog Devices. For signal processing and timing purposes, C0G/NP0 ceramics and film capacitors are preferred. Notice how the 1206 size performs better than the 0805 size for the same voltage.

So:

Do Capacitor Packages makes difference in MLCCs?

Yes, very much so!

• MAN THIS IS JUST MIND BLOWING! I had never thought that Capacitance can vary dramatically based on its package but after reading the link you had included it seems that there a lot happening in the industry of the Capacitors. Commented Jun 29 at 3:05
• @HazardousVoltage It's not the package size that matters, but the size of internal structures. E.g. 1206 10µF might have the same structure size as a 0805 2.2µF and they would behave in a similar way (relative to their nominal capacity) Commented Jun 29 at 16:04
• @asdfex understood as stated in the link : "In fact, any material that allows a device to meet or exceed the X7R temperature characteristics, ±15% over a temperature range of -55°C to +125°C, can be called X7R." Correct so it is based on the structure. Commented Jun 29 at 17:44
• @asdfex it is based on the number of available dipoles in the dielectric which in turn is based on density and volume. Commented Jun 30 at 0:37

Ceramic capacitors in that value and size are (at best) something like X7 or X5 dielectric, and hence have a large voltage coefficient (as well as temperature coefficient and aging). The capacitance is typically measured at a tiny bias voltage such as 1V. The capacitance at rated voltage can be a fraction (like a 1/5 or even a 1/10 fraction) of rated. Physically smaller capacitors tend to have worse voltage coefficient. You can think of this as analogous to inductors, where a small and large inductor may have the same inductance at low current, but the smaller one will tend to saturate first, all other things being equal.

Some manufacturers have detailed information on the typical behavior of their parts- Murata in particular has very good information. If you are not concerned about the resulting nonlinearity of voltage vs. time at constant current, you may not care too much, but note that these characteristics are variable in many ways.

Here is an 0402 (!) 25V 4.7uF X5S capacitor:

A 1206 X7R

A different 1206 X7R from the same maker:

You could also consider a film capacitor or a low-leakage electrolytic capacitor (aluminum or tantalum), both of which will tend to be larger but have less variation of capacitance with voltage.

If you continue to use the ceramic types, physically bigger ones rated for higher voltage will tend to be more repeatable between manufacturers, models and from sample to sample. For less detail-oriented makers, for example, both the second two curves meet the datasheet limits for each other so they could switch from one to the other without telling anyone. They are only 'typical' characteristics and if you depend on them and get burned, it's on you.

P.S. As a reminder that temperature and aging have effects, here is the 0402 "4.7uF" capacitor at 5V bias over temperature:

And the typical aging characteristics of different dielectrics from Digikey/TDK:

The high capacitance, low ESR and small size of ceramic capacitors may seem like magic to those of us who were around before they were developed, but as many stories remind us, magic often comes at a price.

• I feel like I was fooled for my whole life by knowing this, thinking that all packages were the same. That's literally called magic Commented Jun 29 at 3:09
• @HazardousVoltage It's not the package size directly, but the number of barium titanate unit cells available (which is a function of volume, so it's indirectly package size). Commented Jun 30 at 13:20

Ceramic capacitors with dielectrics other than C0G/NP0 are varicaps. Their internal structure as well as dielectric will affect how sensitive their capacitance is to DC voltage bias.

Why is that happening, even though both of them have the same capacity?

They don't have the same capacity because they don't use the C0G/NP0 dielectric. They are variable capacitors with capacitance mostly decreasing with DC voltage (not always!). They are different parts (due to different package), so they simply are varicaps with a different capacitance vs. voltage response.

In short: their capacitance is a function of the voltage on the capacitor, and is not a same across devices. Different cap designs/sizes are expected to react differently to DC bias.

When I use the 1206 package, the timer stays for 4 minutes approximately, while when using the 0805 package, the timer stays for 3 minutes.

The package is a bit of a red herring. Different packages dictate different internal structure - number of layers, layer thickness, etc. So changing the package gives you a different capacitor even if the nominal specifications are the same. These differences affect the capacitance response to DC bias.