Decoupling capacitors between supply pins of ics

[Polymer]

It may be a silly question, but I remember being taught to put a decoupling capacitor between the supply pins
of an ic. You could even get DIL-Sockets with them already soldered in underneath. The nominal value was usually 100nF.
That was over 25 years ago.

I do not remember seeing any in the Baracuda schematic. Is this an obsolete thing nowadays, is my memory playing cruel jokes or
have the aliens invaded and I haven't noticed?

[kt315]

nowaday tantalums are better. I did foresee to set them and in King Cobra board and in Delta Pulse board anywhere I could. also good ceramics
you may take from old PC boards, they are 4-5 uF and more (not in course how much voltage they are).

[Carl-NC]

Yes, still a Good Idea. I continue to use 100nF ceramic, heavier bypassing on critical analog or ADC chips.

[Silver Dollar]

Many schematics don't show them even though they are on the board.
On some layouts they are missing. It doesn't hurt to add them.

[Polymer]

Hi again, I am perusing an online distributor to get a batch of de-coupling capacitors in 100nF and 1uF to be connected in parallel on each chips power pins.

Here are the types I have found and then comes my question to that:

0.1uF, 63v, electrolyte, 0.04 Euros/piece
0.1uF, 100v, ceramic X7R +-10%, 0.08 Euros/piece

1.0uF, 63v, electrolyte, 0.02 Euros/piece
1.0uF, 63v, ceramic, 0.54 Euros/piece

All parts are with leads radial - smd versions are cheaper I have found.

My question:
Do the electrolytic capacitors when used for de-coupling have
any advantages/disadvantages when compared to the ceramic caps?
Rated voltage and price being equal of course.

Thank you!

[kt315]

ESR meter is working on frequency 80-90kHz, square(-wave) signal . it gives absolutely FULL info WHAT IS BEST. tantalum much much better than ceramic 100nF. USE ESR METER, not
LC meter and you ll see full clear picture.

[greylourie]

I think electrolytic are good for filtering around power supplies. Maybe not a good idea on things in the analogue signal chain. There you are looking at ceramic caps.
Some even suggest a range of values and types depending on the application. Or using a cap and an inductor bead/ cap and resistor. Thats for the pro's.. Some good info can be found in app notes by fpga manufacturers - giving guidelines for low noise layout in digital applications.

For info on analogue side some good threads on bypassing cap types and values can be found in diy audio forums.

Maybe forum members with more knowledge can give us tips on type/values, as I am also very interested in this subject.

[greylourie]

Found a link. Might be of help to members. http://www.analog.com/library/analog...09/layout.html

[Polymer]

Hello kt315, Thank you for your reply.

At first I was confused by what you wrote and read many articles about it and how to measure.
A whole new world has opened up to me and I will be getting an ESR meter, most run @ 100 kHz anyway, so no problem.
Which one do you have? The red or blue or homebrew?

What amazed me when researching capacitors is that ceramic caps except COG's have a helluva loss of capacity dependent upon applied voltage.

I intend to de-couple the Baracuda PI and Surf PI and see if it makes a noteworthy difference. At the moment I intend to using 0.1uF Ceramics and 1uf Electrolyte
or Tantalum in parallel.

[Polymer]

Hello greylourie, a great link & read, thank you!

I have been reading up like crazy on this topic and yes, diy audio forums have great info on this as well.
I'm happy to know that others like yourself are interested in this topic.

Just learning the good/bad of each capacitor type is an adventure for me, let alone how to apply that knowledge adequately.

[greylourie]

Hi,

I think Carls post above is a good guideline.

I've seen other members posts recommending x7r type caps. Also a few of photos of higher performing circuit boards here on Geotech show examples of bypass caps.
Quite a few IC manufacturers have guides covering aspects of this subject. It can all be very complicated, especially when using more than one cap/component for bypassing duty - very easy to actually create and introduce more noise. Beyond my scope, or knowledge/skill level.

[Polymer]

Hi again,

It looks like really good decoupling is an art & technique and really important. Here's two links I also found that help me understand decoupling and different types of capacitors:

http://www.vagrearg.org/content/decoupling

http://conradhoffman.com/cap_losses.htm ("Dielectric absorption is a particular problem in capacitors used in integrators...")

- We have those in our metal detectors, don't we?

In a german electronics forum there is mention of purposfully using the "bad" non-linear characteristics of cheap ceramic caps to not
accidentally get an oscillating circuit when using them for de-coupling.

So using high quality capacitors (foil ones like fkp) for de-coupling would be nonsense as far as I can understand.

Still learning and enjoying ... Have a nice day

[greylourie]

Hi,

Really good links in your post above. That first one shows clear graphic examples.

I have a couple of pages to link :

http://williamson-labs.com/480_byp.htm

http://williamson-labs.com/480_dec.htm

Hope they are use.

[Qiaozhi]

You need to keep the subject of decoupling capacitors in perspective, and don't get too carried away. Most of the examples quoted are for high-speed digital design, which can often seem like almost a black art. For the frequency range encountered in the metal detector designs featured in the Geotech forums, decoupling to the same extent is not necessary. During the development of the MiniPulse Plus I tried adding decoupling caps (as the original MiniPulse design didn't have any) but there was no improvement in operation, or any discernible reduction in noise on the power lines. However, I feel that separating the analog and digital power supplies would be beneficial for the next revision. Next time you read one these decoupling articles, look carefully at the scale of the frequency axis, as usually they start from 1Mhz upwards.

Although I will add that adding decoupling (where you think it may be required) won't cause any issues, even if it doesn't make any significant improvement. It will just increase the cost of your design, and take up more real-estate on the PCB.

[Silver Dollar]

I tried many different things to decouple the stages in my TGSL and none had
much effect there so I agree with Qiaozhi. It is a good practice but don't expect
miracles!