Re: Proper size of capacitors for transient suppression?

Tom:

Thanks for the explanation. I am not having noise problems (haven't
turned on the switch yet - about a week away - so I've probably got
less noise right now than most). Paranoia, probably, but, after
reading the problems that others had, I figured that its better to do
it right the first time. The AC entering the main cabinet goes
through two large Corcom EMI filters. The recepticle that it feeds
(and the pendant is plugged into) has a built in surge supressor,
probably a small ferrite core and an MOV. The switching power supply
(according to the specification sheet) has built in EMI filtering.
Do you think I should do anything on the DC output side of the supply?

I figured, considering the rest of the effort (& expense ;-( that has
gone into this project, throwing a handful of small capacitors at it
is pretty cheap insurance.

I was looking at a couple of areas (and the .1uf capacitors may do
the trick). I have 14 low voltage lines leaving the pendant coprised
of the following types (ground is by a seperate #6 AWG Cu. bus):

DC voltmeter leads
Connections to remote Enable relay coil
+5, +12, +24 volts DC lines
E-Stop contacts
MPG encoder leads

Except for the MPG leads (not yet installed), the longest distance of
any line is less than 10 feet (3 meters). I used #18 wire for
everything to keep the resistance to a minimum.

The pendant also has the keyboard emulator embedded. I have heard of
people connecting each terminal through a capacitor to ground (56
terminals? Ee Gads!)

What are your thoughts and is .1uf about right? Should any of these
be different in your opinion?

Jeff

--- In CAD_CAM_EDM_DRO@yahoogroups.com, "caudlet" <thom@t...> wrote:
> --- In CAD_CAM_EDM_DRO@yahoogroups.com, "washcomp" <jeff@w...>
wrote:
> > I've decided to "beef up" the transient suppresion on my
cabling.
> > I'm trying to pick the proper size capacitor to use between
> the "hot"
> > leg and ground. I've seen people use .1 mf and Gecko uses 470 mf
> > across their terminals. Thats a hell of a range. Also Gecko
> > recommends using if the cable is longer than 18 inches (.5
meter).
> > They show the cactior next to the active equipment. My original
NC
> > controller had capacitors in series with resistors for this
funcion
> > on many of the longer lines.
> >
> > ANY RECOMMENDATIONS OF HOW TO SORT THIS OUT WOULD BE APPRECIATED
> >
> > Jeff
>
> Jeff: It all depends on the type signal you are trying to suppress
> and the function of the line. Power lines (DC) of any length over
> the listed 18 inches, should have larger electrolytics on the far
> end. The larger the cap value the lower the reactance (ac
> resistance) to a given frequency. A cap from the line to ground
acts
> as a "shunt" to decouple frequencies off the line. The larger the
cap
> value the lower the frequency it will shunt to ground. The large
> caps tend to shunt higher frequencies down to a "roll off" point.
> (Most larger electrolytics have charactersitics that cause them to
> have high frequency roll-off as well which is sometimes you will
see
> smaller caps in parallel with the big ones). On a DC line all you
> want is pure DC so the large caps react to everything down to even
> 60HZ. A large value cap on a signal line shunts the signal itself
to
> ground (bad thing!). The trick is to pick a value that gets rid of
> noise above the base signal but has little effect on the signal
> frequency itself. You will see values of .01 to .1 depending on
the
> signal. RF caps are a factor lower in value and can be in the
> picofarad range.
> Combining a series resistor with a cap to ground creates an RC
filter
> that has different characteristics than just a cap and the
> transmission line impedence. If you use to large a value for the
> resistor the signal level will be effected (depending on the source
> and destination impedence). Most logic level stuff uses 100 ohms
or
> less.
>
> Decoupling caps on your logic at the pendant end are important.
> *Each* chip needs a .1 MFD from the + pin VCC to a close ground.
> This decouples chip switching noise from getting into the local low
> voltage supply and effecting the other chips.
>
> Another good suppression technique is the use of ferrite cores.
> Typically the larger the core the lower the frequency it will
block.
> The magnetic material of the ferrite "absorbs" the frequencies and
> has a better roll off characteristic than simple caps. For power
> lines the bigger donut cores are good.
>
> There is a whole section of electronics involved with RC and LC
> filters. You can using various configurations build Low pass
filters
> (pass everything BELOW the roll off frequency), bandpass filters
that
> pass a given range of frequencies and reject lower and higher
> frequencies, and high pass filters that reject lower frequencies.
> There are entire books devoted to the design and use of filter
> networks. One of the issues with dealing with logic level signals
is
> that they run a pretty wide range of signals all the way from a
> single pulse per second up to several thousand in the IO circuits.
> You have to be able to pass the highest control signal but suppress
> the noise. Noise most often takes the form of higher frequecy
> impulse spikes that can be suppressed with smaller values of caps.
>
> Beware that caps conduct noise in either direction so if the noise
is
> actually on the ground it can inject it into the circuitry and make
> things worse!
>
> Are you having actual noise problems or is this just a level of
> paranoia? Put the DC electrolytics on the power lines and use the
> small bypass caps at the logic chips. Only start adding RC to
signal
> lines if you actually need it. Noise will most often manifest
itself
> as random problems...things that are hard to predict or duplicate.
> It can be difficult to troubleshoot under those circumstances.
>
> BTW transients are ususally fast rising edge spikes of voltage
> created from a component or switch being turned off and the stored
> energy in a circuit trying to disssipate as rapidly as possible.
> Large inductive loads generate nasty transients that can exceeded
the
> ratings of semiconductors by many times...never mind they only last
> for a few nanoseconds. Snubbers (typically a resistor in series
with
> a cap) gives it a path to ground. I only add this because you have
a
> switching PS in your remote controller and could be a source for
some
> noise. Make sure you have an AC input filter on the AC of the
> switching PS.