Filtering supply lines

Stuart Woolford stuart at spechost.spectel.co.nz
Fri Jul 26 01:52:21 GMT 1996


> At 02:14 PM 7/24/96 +1100, Stuart Woolford wrote:
> 
> >>  The only way to guarantee you are at a true 0v ground is to run 
> >> a wire back to the battery terminal.
> >
> >and what defines the battery terminal as 0v ground?
> 
> The battery is the "power supply" for the entire electrical system.  I think
> it was mentioned earlier that both the supply + and ground be run back to
> the battery.  This helps in several key points:

no, the alternator is.. or at least, the job alternates ;)

> *The battery is extremely low internal impedence, and tends to "short out"
> high frequency noise.

not as low as a local ESR filter..

> *The battery acts as a large capacitor, absorbing both voltage spikes and
> supplying heavy load currents for short periods of time, and tending to
> average out the system voltage.

the system's + voltage, and reflecting these spikes onto the 
batteries local ground (scaled down, naturally..)

> *The alternator is theoretically connected essentially directly to the
> battery, therefore absorbing most alternator noise. 

the alternator is theoretically connected essenitally directly to 
nothing, as there are very few theoretical connections in a working 
car.. in real applications, the alternator connections are many and 
varied (and often quite stupid..)

> >really - the battery terminal can be a particularly noise prone 
> >ground point depending on the charging system design.. (from too much 
> >car sterio install experience..)
> 
> If this is the case, then the battery ground lead should be repaired or
> replaced.  Any voltage drop from the battery to the chassis will affect ALL
> devices attached to the chassis.  In my car I have separate 2Ga ground leads

this can also be seen as 'any voltage drop affects any 
battery-earthed items only..'

> going to both the engine and the frame directly from the battery.  This
> alleviated a lot of noise problems I was having with the stereo.  

wrong.. if 90% of the system defines chassis ground as 0v, then that 
is what it is, the battery is the item with the problem..

> Most grounding problems occur due to "ground loops", where there are several
> different paths back to the power supply (battery) from different devices
> within the car body.  Each of the paths has a different finite resistance,
> and current flowing through them will generate a voltage across each of
> these resistances.  This is what causes ground noise.  In an ideal world,

this is not a ground loop, a ground loop is where you have a 
circulating current in a finite resistance ground system, not a group 
of seperate resistive grounds..

> every device in the whole vehicle would have it's own separate ground wire
> running directly back to the battery, then any noise or voltage drop
> generated on any given ground lead would have no effect on any other lead.

They would still not have stable local ground potentials, but as long 
as you never connected them to each other, there would be less chance 
of ground loops..

> Unfortunately, this is not the case, and auto manufacturers rely on the
> assumption that the body itself is a single homogenous conductor, which
> unfortunately it is not.  

well, it is as close as you can get in a car...

it would be a REAL BAD THING for everything to run a lead back to the 
battery - as each of these leads, with MUCH higher resistance than 
the replaced chassis links, would both have a larger voltage drop 
than the old system, and also pick up a lot more noise from the many 
EMI generators carefully fitted to most cars..

remember, 90% of 'noise' problems in car electronics are EMI pickup, 
not ground loops, unless the systems are incorrectly 
designed/connected..

> The best you can hope to do is isolate your particular installation from the
> rest of the system by running it's own power and ground leads directly to
> the battery.  There is one caveat to this approach - you must ensure that

and then not connecting it to ANYTHING..

> your ground lead is the ONLY path back to the battery from your system, if
> there are any other "hidden" connections to chassis ground from the
> installed equipment (ie a radio chassis or speaker - lead), all bets are
> off, you've now created a ground loop instead of eliminating one...

This isolates nothing, it just gives you a new ground potential, with 
possibly more or less problems. unless the system has no other 
connections, but then, most must have..

> >looking for a 'true' ground is usually not the solution, finding a 
> >non-noisy one is quite a different job - I have usually found a good 
> >solid frame ground to be good - you pick up so much noise on you 
> >earth line to the battery that it gets real bad real fast. You just 
> >need to find a frame ground with good conduction paths to the other 
> >ground references you are linked to.
> 
> This is the standard method, unfortunately it is an approach that masks the
> symptom rather than cures the disease.  Sometimes it is the only thing you
> can do, to avoid, as you said, bad charging system design.  Grounding and
> ground noise is a whole black art unto itself.

not really.. just design things to avoid ground related issues, it is 
not that hard, just a tad more expensive, and impossible if you are 
using 3rd party systems..

The problem is not ground potentials, or loops.
Most single systems will operate at any ground potential, so long as 
all the internals share the same one, and then it does not matter if 
you have a local chassis ground, a batter ground, or anything, so 
long as they all see the same ground impedance, and it is within 
allowable tolerances (unless it is a high-power application, just 
about anything will be..)

The problem is with interconnected systems at different local ground 
optentials/impedances, couples with EMI pickup. a battery connected 
ground MAY help with the first (and may not, it depends what the 
other grounds are doing..) but will add to the second.
----------------------------------------------------------------------
Stuart Woolford, stuart at spectel.co.nz

                      >>>>In VI Where Available<<<<
----------------------------------------------------------------------



More information about the Diy_efi mailing list