16

I was thinking decoupling, but you would have to overdimension them because of the high tolerance and temperature stability. And doesn't a 1uF capacitor (instead of 100nF) have the same inductance problems as a 1uF X7R for decoupling?
Are there other applications where the tolerances and variations are so little important that a Y5V or Z5U could be preferred over X5R or X7R? I realize they're somewhat cheaper but that doesn't count if the quality is too bad to be useful, IMO.

Federico Russo
  • 9,658
  • 17
  • 68
  • 117
  • They're great for saving money, when you don't care too much about reliability. – Connor Wolf Dec 21 '13 at 08:23
  • Wait until you look up the obnoxious voltage coefficient of capacitance for Z5U and Y5U capacitors and you will re-ask the question, "What are Z5U and Y5U capacitors good for"! :D –  Nov 12 '14 at 17:50

5 Answers5

12

I would say the uses can vary depending on your initial project goals/specifications (e.g. what temperature range you want the circuit to perform under, voltage range etc)
You decide the specs/tolerance limits for a particular project, so if you run the numbers and the circuit will function under worst case scenario with slacker tolerance on certain components then all should be well.
This may mean in one project you avoid them completely and in another you use nothing but.
Generally I would agree that they are commonly used as cheap decoupling/bulk capacitance, but there is no reason why you couldn't use them for e.g. a rough timer/oscillator if it still works as intended within your specifications.
Check the datasheet for graphs over temperature, frequency, voltage, etc and decide whether the part will be suitable for a particular use.
Monte Carlo SPICE analysis is a useful tool for determining how a circuit will perform with component variations.

Oli Glaser
  • 54,990
  • 3
  • 76
  • 147
  • 1
    +1 for the hint that the capacitance decreases not only with temperature, but also (and drastically) with voltage. – zebonaut Oct 07 '11 at 14:53
7

I suspect that in many applications, if a "10uF" cap with an inferior dielectric, paralleled with a good 0.1uF cap, will work as effectively for bypassing as would an ideal 1uF cap, but will cost less than a 1uF cap with a good dielectric.

On the other hand, I've sometimes thought that for bypassing devices which will be switched on and off fairly often, having a cap whose capacitance dropped off sharply with voltage could actually be an advantage. Suppose one has a 3.3-volt device which draws 1mA, needs 1uF of bypassing, and is needed for 1ms once per second; the device will completely drain the cap between uses. Charging the cap to 3.3 volts will require 3.3 microcoulombs of electricity, every time the cap is switched off, that energy will be wasted. Every second, the device will require one coulomb of energy during the 1ms that it's "on", and burn 3.3uC uselessly after it's "turned off". In effect, the cap would be wasting three times as much energy as the device was actually using.

Now suppose that one could get a cap with a capacitance of 3.3uF at below 0.1 volts, and zero capacitance above that, and one wired that cap in parallel with the power switching device; assume further that the input to the power switching device has 100uF of usable capacity. To allow for inductance in that cap or the 100uF board cap, the device also has 0.1uF of "normal" capacitance in parallel with it. In that scenario, each on/off cycle will require charging the 0.1uF cap to 3.3 volts, requiring 0.33uC, and charging the 3.3uF cap to 0.1 volts (no energy will be spent charging it from 0.1 to 3.3 volts) using another 0.33uC. So energy wastage would be cut from 3.3uC (or 330% of the current usefully employed by the device) to 0.66uC (or 66% of the current usefully employed). Wastage would be cut by 80%; energy consumption would be cut by over 60%.

In practice, I doubt one can get caps of suitable values with such a sharp fall-off of capacity versus voltage, but if one could, it would be possible to greatly enhance the efficiency of some battery-powered devices.

supercat
  • 45,939
  • 2
  • 84
  • 143
6

In most home consumer electronics, rated to operate from only (say) 10C - 35C, the temperature coefficient doesn't matter that much.

The poor tolerance could be compensated for by using multiple low cost Y5V/Z5U capacitors. Also sometimes the standard 100nF decoupling capacitor can be made smaller without significant performance loss.

Thomas O
  • 31,546
  • 57
  • 182
  • 320
4

Yet another answer, but no one mentioned it..

While y5v's seem whimpy, from an emi standpoint they can have a slight advantage over x7r in some applications, which is in regards to their self resonance. x7r's are quite peaky, and y5v's are somewhat flatter. Play with this tool for example - http://www.avx.com/SpiApps/#spicap

Erik Friesen
  • 1,851
  • 1
  • 11
  • 21
3

bulk bulk bulk bulk bulk bulk bulk bulk....

Bulk capacitance, where you need to store as much energy as possible in a given package. You supplement it with smaller capacitors that have better high-frequency characteristics if you want good overall bypassing.

Jason S
  • 13,950
  • 3
  • 41
  • 68