Re: Star Grounds

[Shiva]

"tony.r" > wrote in message
...
> Star Grounds
>
> What a lot of fun and games I have had with locating a squeal problem in
my
> project amp. I have done everything.
>
> 1/ the microphonic test tapping the preamp tubes with an eraser to see if
> you hear a reaction from your speaker.
> 2/ Replaced a bunch of cable with shielded cable.
> 3/ Grounded the centre spigots on my preamp
> But when I separated the input jack ground out of the star I got a big
> result.
> I grounded to input jack separately and the problem has been located.
> Star Ground or Bus ? The star is to difficult to locate problems from.
> The problem is still their but at least now I know where it is.
>
> http://www.ncable.com.au/~tony.r/
>
> tony.r
>

Hi Tony.
The "pure"star system never worked for me - the slightest miscalculation
will set up ground loops that you'll never be able to track down - to many
variables. The easiest grond system (for me, at least), is a modified star,
or star->bus->star>bus>star. Most comercial gear uses that, but a *really*
clear example of it is a Fender amp. Look at one, and see the way the
filter caps are grounded to the brass plate behind the pots, and the general
grounding layout. It's simplicity itself, but people kind'a underapreciate
it, or not notice it at all. I've tried pure star, with the sam e results as
you - spent hours on end chasing groundloops, and then... just junked the
layout. Oh, house wiring is a great source of easily-worked bus wire -
just tin the whole wire...
'luck,
-dim

[Tom Schlangen]

Hi tony.r,

> Star Ground or Bus ? The star is to difficult
> to locate problems from.

I don't think so. If star ground is _consequently_
used, it is a good as any other _consequently_
used grounding strategy, like plane ground.

Problems only arise when one mixes several strategies.

BTW, a bus ground is just a stretched star ground :-)
with the exception, that sometimes it isn't as
easy to predict in which direction the electrons
flow between the taps to the bus ...

Personally, I prefer star grounds with sub-stars:
Each channel has its own star (or 2, for the driver
and the power stages, at convenient points, where the
low level signals are fed in), and each of this
stars is connected to a central ground star at the big
PSU filter caps with _one_ wire (so no star-to-star
conections, execept to the central star).

It is good practice to give low-level high-gain
stages their own sub-stars, with ground connections
as short as possible to the star placed at strategic
position near by.

All ground stars are insulated from the chassis,
and there is only _one_ connection from the central
star to the chassis plane via a 10 ohms / 10 watts
resistor (tnx to Patrick Turner for that tip), and
the chassis plane being connected to mains earth wire.

Regarding inputs and volume controls, I just follow
Fred Nachbaurs "patch cable approach" with great
results.

Except for the missing chassis shielding effect
I get even very hum-free breadboard level circuits
this way :-)

I think with a consequent star ground strategy you
simply can't go wrong and I don't mind that this
approach will need some feet of black insulated
wire even for medium comlexity projects, since this
wire is cheap when you use 1 mm^2 solid copper wire
as used in home mains installations.

Tom

--
A consultant is a man who knows 40 ways to make love,
but doesn't know any women.

[Tom Schlangen]

Hi Shiva,

> The "pure"star system never worked for me - the
> slightest miscalculation will set up ground loops

No, it won't, if you really keep it "pure" :-)

> to many variables.

No, quite contrarily, consequent star ground
has not many "variables" except for maybe
pacing the substars where they make sense, but it
likely has some more wires than a clever mixed
scheme or plane ground. Essentially, pure star
ground (with substars) is nearly a no-brainer.

> The easiest grond system (for me, at least), is a
> modified star, or star->bus->star>bus>star.

I would only recommend cleverly "mixed" grouding
schemes if one really knows what one is doing (i.e,
one is clever and/or has patience to try out
by trial and error), or for highly commercial
projects, where some inches less or more copper
wire per amp sum up to kilometers by thousands
of built amps.

Tom

--
Anyone who considers protocol unimportant has
never dealt with a cat. - R. Heinlein

[Tom Schlangen]

Hi shiva,

> Whoa. Pure star with substars. That would be...

A tree (without the roots). This would be a better
description or name for what I use, but I suppose
you surely knew what I meant already.

Sorry, but often I am lacking the right/correct terms;
english isn't my native language, and RAT is the only
occasion for me to write in english. I don't speak it
actively; the last time was half a decade ago when
traveling Scotland.

> To the alternative, how would you diferentiate
> "star->bus->star" from the modified star you're
> suggesting?

A tree with roots, to keep the picture.

> you must be a heck'o a bright feller',

Why? Just the contrary, I KISS by using star ground
(or tree ground, if you prefer). I just made some
few experiences concearning grounding of selfmade
amps and did some reading on that subject. There are
quite some interesting articles out there on that
subject. None of them suggests suggests mixed schemes
for fool prove results, but you surely knew that
already.

> What about a plane?

Yes, what? Consequent plane ground surely is another
good grounding scheme, featuring probably shortest
gound connections, if one uses the chassis for
the plane. But that approach has drawbacks, too, and
you surely know them. It was often used for gear produced in
high numbers, main reason being to save some copper/money
because wires to gnd-plane (mostly the chassis being
used for that) are short. But you surely knew that already.

Tom

--
Knowledge is power - knowledge shared is power lost.
-- A. Crowley

[Kirk Patton]

"Tom Schlangen" > wrote in message
...

> Yes, what? Consequent plane ground surely is another
> good grounding scheme, featuring probably shortest
> gound connections, if one uses the chassis for
> the plane. But that approach has drawbacks, too, and
> you surely know them. It was often used for gear produced in
> high numbers, main reason being to save some copper/money
> because wires to gnd-plane (mostly the chassis being
> used for that) are short. But you surely knew that already.

Hi Tom . . .

In grounding schemes, it is very important to remember that just because a
piece of wire has a low impedance at DC, it doesn't necessarily have a low
impedance above 1 MHz, and oscillations responsibile for symptoms like those
described in the original post are frequently found at pretty high
frequencies. The reason for grounding leads, wires, and traces to be short
in critical nodes is absolutely to improve performance, not to save money by
saving wire. This is precisely the reason for local power-supply bypass
capacitors near the stages they serve . . .

A comprehensive analysis of both the AC and DC ground CURRENT paths is
necessary; looking at grounding in terms of (hypothetical) voltage will get
you nowhere. The most important part in terms of preventing oscillation is
keeping a short, clear path between the active stages and their AC ground
returns, which is usually different from the DC ground return, which can be
different from the signal reference ground.

An example would be the typical plate-loaded inverting input stage of a
typical preamp or power amplifier. The AC ground return for the input stage
is the negative side of the power-supply filter capacitor that smooths the
power used in the input stage. Thus, AC performance will be optimized when
the components that flow current to ground (grid-leak resistor, cathode
resistor, cathode resistor bypass cap) have short leads to the AC return.
The DC return is the power-transformer's secondary center tap (2-diode
full-wave config.)

While I'm not sure that this is what you were suggesting, let's say that the
power transformer's center tap and power cord ground are both easy to attach
to the chassis near the tranny through short wires. Now if you took a
misimplemented "star ground" approach and connected the input stage's
grid-leak resistor to the same point through an 8" piece of wire, the
cathode resistor through another 8" piece of wire, and the cathode resistor
bypass capacitor through another 8" piece of wire, and that stage's filter
cap through yet another 8" piece of wire, you would have, at the very least,
a situation far from optimum, as the HF impedance through an effective 16"
of wire between any two nodes can be quite significant. (Don't laugh, I've
seen exactly this in a couple of commercially-produced pieces of equipment.
Both of them (power amps) had very interesting tendancies to become HF
transmitters under load . . .)

If, however, all of the above-mentioned components were located the same 8"
away from the PSU transformer, and their leads simply attached individually
to the chassis after snipping the leads fairly short, the AC ground return
for each of them would be much shorter, and consequently have lower AC
impedance.

In short, while not necessarily glamorous, the old Fender Blackface strip of
copper at the front of the amp for a grounding bus was a pretty good
approach. If one were simply to move the preamp stage's filter caps up to
the top of the amp (easy with today's smaller caps), they would have a
circuit/grounding layout that would be pretty tough to improve upon.

Regards,

Kirk Patton

[Tom Schlangen]

Hello Kirk,

thank you very much for elaborating, I've saved your
interesting article for future re-reading.

Tom

P.S.: I know next to nothing about tubed guitar amps,
but owned (and used) an Orange stack (4x EL34, I remember)
about 20 years ago, when playing geetah and bass in some
local bands. That was long before I was interested in tube
circuits per se ...

--
Do not meddle in the affairs of wizards,
for you are crunchy and good with ketchup.