Very informative, Mark. Thanks. I use Mouser #581-SA105E104MAR for
bypass. Multilayer Ceramic (MLC) Axial Capacitor SA10 0.1uF 50volts Z5U 20%
http://www.mouser.com/search/refine.aspx?Ntt=581-SA105E104MAR
$7.00 USD per 100
-Richard Brewster
http://www.pugix.com
Mark wrote:
bypass. Multilayer Ceramic (MLC) Axial Capacitor SA10 0.1uF 50volts Z5U 20%
http://www.mouser.com/search/refine.aspx?Ntt=581-SA105E104MAR
$7.00 USD per 100
-Richard Brewster
http://www.pugix.com
Mark wrote:
> On 4/25/07, Jorrit Tyberghein put forth:
>
>> > Ceramic discs are fine for decoupling.
>>
>> Hi,
>>
>> In general, which capacitor types are best for decoupling? Or doesn't
>> it matter that much?
>>
>
> Well, I'm going to assume that you asking about decoupling
> high-impedance IC's such as op-amps. Decoupling caps for tube
> cathodes requires much larger values.
>
> The short answer is that it doesn't matter that much. Decoupling
> caps should be physically small, inexpensive, and bipolar. Their
> precise value doesn't matter much, so temperature stability isn't an
> issue.
>
> Decoupling capacitors perform two functions.
>
> Firstly, the word "decouple" means to isolate or separate. As such,
> they act as filters to keep noise out of the circuit.
>
> Secondly, since PCB traces have some resistance, and are therefore
> subject to Ohm's Law, a sudden demand for current will result in a
> drop in voltage at that point. So decoupling caps also act as little
> power reservoirs.
>
> Ceramic monolithic capacitors are the best choice for decoupling.
> While they are available from a few picofarads up to as much a
> microfarad, generally, you want 100nF. In the United States, this is
> often written as ".1uF". While that might be bad math, such notation has
> the connotation of referring to a ceramic bypass cap. The caps
> themselves are often labeled "104". If you are using a particularly
> noisy chip, such as a timer or logic IC that can cause switching
> noise as it changes states, you might want a larger value. Although,
> with linear devices, such as op-amps, you are usually trying to keep
> the noise out of the chip, not the other way around. If you are
> using a particularly prissy op-amp, or a fast op-amp to drive a large
> load, or if the rails are particularly long or noisy, you might want
> to use two bypass caps, with one a hundred times larger than the
> other, in parallel. In this case, the larger cap would be an electrolytic.
>
> Ceramic caps are rated by their temperature coefficient. COG or NP0
> capacitors are the most temperature stable, but are also the largest
> monolithic ceramic caps. X7R capacitors have a greater temperature
> coefficient, but are a bit smaller in size. Z5U capacitors have the
> worst temperature stability, but are the smallest in size. That's
> what I use for power supply bypassing.
>
> The most important aspect of ceramic caps is that they have a low
> series inductance. This is also why physical size is so important --
> the smaller the cap, the smaller the leads and traces connecting it.
> The inductance of a capacitor raises its impedance, thereby limiting
> the frequencies it can filter. This is why bypass caps are placed as
> close as possible to the power terminals. This limits the series
> inductance, so it can filter high-frequency noise.
>
>
> To understand different capacitor values, here's a cute web page with pictures:
>
> http://www.kpsec.freeuk.com/components/capac.htm
>
>
> This page isn't as cute, but lists all the dielectric codes:
>
> http://xtronics.com/kits/ccode.htm
>
>