[sdiy] some disturbing things i've read recently

[harry]

Hello Mystic....  (inline)

Gavin Russom wrote:

> well I'm moving along on my modular but I've recently run across a couple of
> technical "facts" that have alarmed me because i'd never heard anything
> about any of them and am not prepared
> 1. An IC will not work properly or at all if the power leads are too long, a
> possible solution for this is to put a .1 uF capacitor across the power
> leads (I imagine this is a filtering cap to suck off any transients that
> would be coming through the power lead?) Any body ever run into this
> phenomenon?

This "might" happen... but what is more likely is that if there are other ICs on

the same power lead the power drain of one might glitch the power supply of the
other, causing wrong operation.

Another thing is that some ICs may latch up or oscillate if the power leads are
"too long".

Too Long is a function of the inductance of the lead... or even the resistance
of the lead.  This is like using an AC appliance (electric drill) with a too
small power cord..
or with the cord coiled tightly.... it reduces the current flow and can kill the
motor.

In IC terms its more likely that the chip may oscillate... The 78xx and 79xx
series
regulators can and will do this... read the data sheets and check the
recommended
capacitor usage

Here is a reprint of a short reply I did before on the EFM group on this
subject.
email any further questions if you like...

Decoupling 101:

Circuits that are wired together are coupled. Sometimes there are connections

that you don't want coupled.  You don't want the lights to dim everytime you
open your garage door. So you decouple them.

Decoupling them (as Ed said) usually involves using capacitors as local power

storage devices.  Wires have resistance and inductance.  Resistance causes a
permanent voltage drop (usually small)... Inductance causes the voltage to
drop
when the current demand changes... when you 'suck' current, it drops... when
you
'stop sucking' it overshoots.  The fix is to put a large enough capacitor
RIGHT next
to the device that will 'suck' current... so that when it starts to suck it
gets the first
mouthful of electrons from the capacitor... and satisfies its thirst while
the inductance of the power supply wiring is overcome.  Eventually... the
power will get there.

When the current flow stops... the remaining current in the power supply
wiring will
keep on flowing to the last place it was headed... and will dump into the cap
instead
of the device. The cap should be big enough to supply the sudden demand, and
take up the additional charge when the load shuts off.

This DOES NOT happen at the speed of light... it is much slower... often 60%
or
less than light speed.  Thats why decoupling is necessary.

Decoupling assures that individual chips function properly... and that
unwanted interaction through the supplies does not occur.

Often cited is the 555 timer... this is one current HOG every time the output
changes state... it draws a huge current pulse from the supply.  In this case
its necessary to
have more than one decoupling cap... usually a large, slow electrolytic or
faster tentalum for the bulk of the charge... in parallel with a small
ceramic cap.

The two in parallel have a synergistic effect... the capacitor (with usually
very low series inductance) can supply a small current very fast.  The larger
cap (with higher
inductance) takes longer to get into action. The reverse occurs when the
current draw of the load stops.

These spikes... if allowed to rock the power supplies, can cause audible
noise...
unintended VCO sync, things like that.

Good decoupling is essential for the 78xx and 79xx series regulators. Follow
the
data sheets and use capacitors located right at (as close as POSSIBLE) the
regulator pins... even if the main supply cap is right nearby... big caps
have big
inductance... they are slow.

Another form of decoupling is sometimes used for unused pins... a good
example
is pin 5 (control voltage in) of the 555 timer. When it is not used... it can
pick up
unwanted signals (noise) and cause the frequency or period of the 555 to go
off.
Data sheet recommends a .01 or .1uF cap to ground if you are not using this
pin.

How much decoupling do I use ??? LOTS.  With op-amps I place them between
every package... for CMOS, at least every other package... for a 555, two
caps
in parallel... and I run the power supply wiring right back to the main
source (power
input to the card) separately.

There is one other form of decoupling... using a resistor in series with the
power supply leads... and a much larger cap. This forms a low-pass filter
that eliminates most low frequency noise from sensitive circuits (usually
things like preamps or
some filter stages).  You choose the resistor to be as large as you can...
allowing that the voltage to that chip is going to drop... maybe on a +/-12V
circuit you could run that preamp at +/-10V... then size the resistor to drop
2V.  Of course the more you drop... the less output power you will be able to
deliver...

Then you make the capacitors as large as you need to... to get rid of the
unwanted
ripple.  Thats one good reason to use full wave DC supplies (and not half
wave) but
that can be next weeks class, eh?  Again... if you use big caps... bypass
them with some little ceramic caps...

If your hobby takes you up to radio frequencies... sometimes three caps in
parallel are used, including surface mount types... to get rid of the last
little bit of lead inductance.

BTW:  inductance is minimized by keeping the path that the current flows in
SMALL, which is why you locate decoupling caps as near as possible to the
load.

When I do a layout... decoupling caps go in first... before all other
components

H^) harry



>
> 2. What is a system DAC? I've looked through all my books and can't find
> anything about it.
> thanks

In micro controlled synths... there is usually only one DAC that is multiplexed
to control all CV functions... that would be the "system" DAC... really its just
a DAC...