Bypass caps (was:Re: [sdiy] Silly question about dual vs. quad op amps

[harrybissell]

let me think... uhhh... NO   :^P

It is a matter of risk management. If I was doing a production run of
10,000 or more units... I'd study VERY carefully how many caps I might
be able to save.  Or... if the units had to be very small or lightweight
I'd do the same (to save size, or weight).

If it was only a few units, or if the cost of recalling those units was
very high...
I'd use one cap per IC per power pin, minimum... and add some larger
tantalum and or 'lytics as Bob Pease recommends.

Sequential Circuits went through five software releases in the Rev 3.2
because they
didn't use any decoupling caps on the POT board, which does not have a
single
active component on it.  They had to make the pots less and less
sensitive to try and
cover the noise issues.  Boy a few caps there would have helped. When I
saw there
were none, I added them myself (back at Rev 9.2) and fixed the problem.

What price failure ?    More time is wasted by NOT using sufficient
bypass caps
than is saved by not designing them in (in small production runs). It is
really hard
to OVERDO the caps, unless the power supply is unable to start because of
the peak
current :^P

I work (by day) designing industrial controls.  A failure here will mean
men in the
field for days - weeks replacing every control at our expense. One man
for one day
pays for every decoupling cap in the whole plant.

(now if you do FCPs... then desing them out, they will be out of service
by the end of the month anyway :^)

H^) harry

limor wrote:

> adding -too many- caps is not necessarily a good idea and its
> certainly a waste of money. lets be reasonable, people: a single .1uF
> ceramic cap per IC is just fine unless there are multiple power lines
> and the datasheet requests more. and, for high-current spike ICs, a
> 2uF tant or 20uF electrolytic for every 3-5 ICs to reduce resonant
> ringing on spikes (Pease, pg 45)
>
>    limor
>
> On 12/5/05, Samppa Tolvanen <samppa.tolvanen at gmail.com> wrote:
> > On 12/5/05, Harry Bissell Jr <harrybissell at prodigy.net> wrote:
> > > I use .1uF caps for decoupling, one per IC pin.
> > >  I would use more if the current draw of slew rate was
> > >  very high, usually not an issue with TL07x series.
> > >
> > >  I do not share decoupling caps between duals.
> > >  I run power supply traces to the capaciotr pins, then
> > >  to the IC.  Sharing with two duals will probably mean that
> > >  the ICs are too dar from the cap to be effective.
> > >
> >
> > Lets try: 100nf  per each pin? 8 times per dual and 14 for quad?
> > Strange shit, man :D
> >
> > S
> >
> >