Digital IC bypassing

[hollowman]

For commonly-used digital-audio ICs (see below) what are some of the most-effective bypass components and bypass techniques? (I'm trying to create some sort of all-in-one reference/guide on the topic).

Example: I've included some values for various Philips chips -- used in myriad upgrade mods -- of their mid-late-1980's CDPs. I'm not sure if these values are "ideal", but they are better than stock (unmodded) component.

DAC chips:

- Philips: TDA1541(A): Vdd to gnd: 100uF electro (high-quality e.g. Panasonic FM) + 100nF film; use ferrite bead for input power; as close as possible to IC

- Philips (others):

- TI/Burr-Brown (various):

- Cirrus (various):

Digital Filters (oversamplers):

- Philips: SAA7220: Vdd to gnd: 100uF electro (high-quality e.g. Panasonic FM) + 100nF film; use ferrite bead for input power; as close as possible to IC

- Pacific Microsonics: PMD100, PMD200

- NPC (various):

- Wolfson (various):

Input receivers:

Cirrus: CS8420, CS8416, CS8412

Other ICs (e.g. servo/controllers; 74HC inverters, flip-flips; etc.):

- Philips: SAA7210: Vdd to gnd: 100uF electro (high-quality e.g. Panasonic FM) + 100nF film; use ferrite bead for input power; as close as possible to IC

Philips: 74HC74: 10uF (high-quality electro, e.g. Pana FM) + 47nF film; use ferrite bead for input power; as close as possible to IC

- Others:

[rossl]

Surface mount packages for the 100nF are superior for bypassing. The lead inductance of a radial lead part reduces it's effectiveness at higher frequencies.

Of course, retrofitting is a problem. Ideally the surface mount cap should be designed in to the PCB as close to the power pins as possible.

To retrofit an older PCB, place the SMT cap directly on top of the IC and use the shortest possible bits of wire to attach to the power pins. You will need a good pair of tweezers.

Ceramic X7R in 0805 package is OK for digital but some purists will want to go with PPS. 1206 may be easier to handle.

[jackinnj]

Texas Instruments shows a 100nF ceramic and a 1uF electrolytic on the power pins for their Class-D amplifiers.

Ferrite beads for the input are available in 0603, 0805 etc. They can also be used as the inductor in the output of lower wattage class D amplifiers.

[Eric Juaneda]

You can read this,

Supply Decoupling by Guido Tent
How (Not) to Decouple High-Speed Operational Amplifiers by Bruce Carter from Texas Instruments

Eric

[mrjam]

Hi guys

Here is an archive containing "few" plots concerning caps.

I hope you'll enjoy it! ..and have a BUSY day!

PS. vertical axis is 10dB\division, so each square from top means 10dB more attenuation.

http://www.audiofaidate.it/forum/upl...caps_tests.zip

[Bernhard]

Originally posted by Eric Juaneda
You can read this,

Supply Decoupling by Guido Tent
How (Not) to Decouple High-Speed Operational Amplifiers by Bruce Carter from Texas Instruments

Eric

Guido Tent: Use lytics, no need for ceramics.

Bruce Carter from Texas Instruments: Use ceramics, no need for lytics.

[mrjam]

Check the step response of your bypass solutions, parasitics included.

http://www.mit.edu/~6.331/an47fa.pdf
pages 11-12
...but you should read the entire AN!

[hollowman]

Are transport servo ICs -- particularly in Philips units -- especially sensitive to component-value changes? Recently, I upgraded (or though I had) a few components in my CDB650's aging servo board. And some values of replaced components did not match (I can only stock so much!). E.g.:

(2) Orig PSU pin 47uF 16v --> Panasonic FM 100uF 25V
(2) Orig PSU pin 220uF 16v --> Panasonic FM 220uF 50V
(2) Orig 5-ohm PSU pin resistors --> ferrite beads

Usually, the above "procedure" works well on digital-section ICs (e.g. TDA1541, SAA7x10 and SAA7220). But in the servo section, post-“upgrade”, the sound of CDB650 in question lost some "focus" (but increased in detail). Also, and randomly, some tracks near end of certain discs take a long time to cue up.

Could be those Pana FMs caps breaking in (??) – I’ve heard they take a while. What effect do larger voltage ratings have on cap performance? IAC ... for ICs such as TDA1541 and SAA7220, very-noticeable improvements are noted both immediately and as components break-in.

[spzzzzkt]

Originally posted by Bernhard


Guido Tent: Use lytics, no need for ceramics.

Bruce Carter from Texas Instruments: Use ceramics, no need for lytics.

Hi Bernhard,

Did you just skim the Bruce Carter article ?!?!?!?

The primary use, then, for electrolytic capacitors, is at frequencies below 1 MHz. If a designer is sure that these frequencies are present, then they are necessary. If, however, there is no reason to believe that these frequencies are present, electrolytic capacitors are not necessary.

That is a little different to saying "no need for lytics". I suspect following his criteria most audio gear require the use of lytics for decoupling.

cheers
Paul