AoE x-Chapters - 1x.1 Wire-&-Connectors

Tim Williams <tiwill@seventransistorlabs.com> wrote:
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

But with visibility to where the currents are flowing, or with a
star-grounding scheme, those loop currents stay separate. Say the
connectors are clustered on a front panel, shields grounded to it. That
keeps outside noise out, good. Collect the interior cables (which are still
all grounded to the same point, preferably as coaxially as possible, no
weedy wire links), and bundle them into a single harness. Ferrite beads as
needed. They go over to the PCB, which no longer needs chassis ground at
all.

Or do the brute-force method and use welded chassis compartments to keep the
bignasty away from the quietstuff. Cost doesn't much matter in your test
equipment, but it's a very real tradeoff in production. Production even
likes to avoid metal if they can...
Click to expand...

My recent audio mix-amp project motivated me to read, or at least skim,
dozens of articles about /audio/ ground loops. Either Bill Whitlock or
one of the other gurus said something along the lines of ground is a
equal potential myth used by engineers to make their jobs easier.
The exact wording and source of that aphorism are temporarily lost
to me. It's Whitlock who definitely dispels other myths about trying to
use thicker grounds or multiple grounds to fight loops.

Thank you, 73,

--
Don Kuenz KB7RPU
There was a young lady named Bright Whose speed was far faster than light;
She set out one day In a relative way And returned on the previous night.
 
tirsdag den 6. august 2019 kl. 17.06.48 UTC+2 skrev Don Kuenz:
Tim Williams <tiwill@seventransistorlabs.com> wrote:
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

But with visibility to where the currents are flowing, or with a
star-grounding scheme, those loop currents stay separate. Say the
connectors are clustered on a front panel, shields grounded to it. That
keeps outside noise out, good. Collect the interior cables (which are still
all grounded to the same point, preferably as coaxially as possible, no
weedy wire links), and bundle them into a single harness. Ferrite beads as
needed. They go over to the PCB, which no longer needs chassis ground at
all.

Or do the brute-force method and use welded chassis compartments to keep the
bignasty away from the quietstuff. Cost doesn't much matter in your test
equipment, but it's a very real tradeoff in production. Production even
likes to avoid metal if they can...

My recent audio mix-amp project motivated me to read, or at least skim,
dozens of articles about /audio/ ground loops. Either Bill Whitlock or
one of the other gurus said something along the lines of ground is a
equal potential myth used by engineers to make their jobs easier.
The exact wording and source of that aphorism are temporarily lost
to me. It's Whitlock who definitely dispels other myths about trying to
use thicker grounds or multiple grounds to fight loops.
Click to expand...

when I took and emi class at uni one of the first thing the guy said was
something along the line of "ground does not exist"
 
tirsdag den 6. august 2019 kl. 16.18.38 UTC+2 skrev George Herold:
On Tuesday, August 6, 2019 at 3:55:13 AM UTC-4, piglet wrote:
On 06/08/2019 02:58, Phil Hobbs wrote:

I'll be 60 late next month, so a few strippers would be apropos--just
don't tell my wife. Maybe they could jump out of a cake. ;)

I generally use Xcelite flush cutters, which are far better than the
usual axe-blade dikes.  I also have the usual spade lug crimper / bolt
cutter/ wire stripper tools.  I don't like the big clunky strippers that
mechanically sense the insulation and adjust their blades accordingly--I
just want insulation to do as it's damn well told, and Teflon doesn't,
in so many, many ways.


I'll be 60 a few months after you. I already have my strippers! They're
Teledyne thermal and cope fine with Teflon and dialled back with any
other insulation devised.

piglet

Hah! I turn 61 in October. Respect your elders.*

George H.
*In grad school there was this smart Korean student. He'd been in the lab for years
and I was the newcomer post doc, who barely knew anything of the field.
(FIR spectroscopy of semiconductors) But I knew a bunch of electronics and low
temperature techniques. He would dismiss all my suggestions, then he learned
one day that I was a few weeks older than him. Like day and night I was now
his elder to be respected.. it struck me as weird, but totally part of Korean
culture according to the other Korean students at the school.
Click to expand...

I believe too much respect for seniority between pilot and first officer
has been noted as a possible contributing factor to statistically high number
of plane crashes for Korean air in the 90's
 
On 2019/08/06 8:22 a.m., Don Kuenz wrote:
Tim Williams <tiwill@seventransistorlabs.com> wrote:
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

But with visibility to where the currents are flowing, or with a
star-grounding scheme, those loop currents stay separate. Say the
connectors are clustered on a front panel, shields grounded to it. That
keeps outside noise out, good. Collect the interior cables (which are still
all grounded to the same point, preferably as coaxially as possible, no
weedy wire links), and bundle them into a single harness. Ferrite beads as
needed. They go over to the PCB, which no longer needs chassis ground at
all.

Or do the brute-force method and use welded chassis compartments to keep the
bignasty away from the quietstuff. Cost doesn't much matter in your test
equipment, but it's a very real tradeoff in production. Production even
likes to avoid metal if they can...

My recent audio mix-amp project motivated me to read, or at least skim,
dozens of articles about /audio/ ground loops. Either Bill Whitlock or
one of the other gurus said something along the lines of ground is a
equal potential myth used by engineers to make their jobs easier.
The exact wording and source of that aphorism are temporarily lost
to me. It's Whitlock who definitely dispels other myths about trying to
use thicker grounds or multiple grounds to fight loops.

To avoid a potential misunderstanding, gurus generally advocate star
grounds, as you did above.

Thank you, 73,
Click to expand...

Star grounds work just fine until one of them doesn't. If it is a
current return line then all sorts of interesting things can happen. In
our industry all the logic in a series of pinball games was designed by
a company (who shall remain unmentioned, but starts with Rock and ends
.....) and they used a sort of star configuration for the ground returns.
This worked fine until one or more ground power connections developed a
bit of resistance (flat Molex 0.156 connectors mated to round pins)
which lead to power transistor bias drifting which then started melting
coils...intermittently.

John :-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."
 
On Tue, 6 Aug 2019 07:34:26 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

On Tuesday, August 6, 2019 at 10:16:04 AM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 00:32:26 -0500, "Tim Williams"
tiwill@seventransistorlabs.com> wrote:

"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.

Sure, but star grounding has a place at low frequency.
I cut my star grounding 'teeth' making a low freq. teaching lockin.
https://www.teachspin.com/signal-processor-lock-in

Where a uV of 'signal' getting into the wrong input would ruin my whole day.

George H.
Click to expand...

The way to avoid ground loop errors is to keep signals differential.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics
 
On a sunny day (Tue, 6 Aug 2019 11:31:10 -0400) it happened Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote in
<qic6g0$f57$1@dont-email.me>:

On 8/6/19 11:22 AM, John Robertson wrote:
On 2019/08/06 5:09 a.m., Phil Hobbs wrote:
On 8/6/19 4:07 AM, David Nadlinger wrote:
On 05.08.19 1:32 PM, Winfield Hill wrote:
  Here's a DRAFT copy of x-Chapter 1x.1, which starts
  things off, by dealing with wire and connectors.
  Comments and corrections please.

https://www.dropbox.com/s/kz335xwvl5v8gie/1x.1_Wire-%26-Connectors_DRAFT.pdf?dl=1


Regarding that circled capital L in footnote 21: Litzendraht as a
German noun should indeed be uppercase, so the current footnote is
correct (potential disagreements about capitalisation of the loanword
in the body copy notwithstanding).

  — David

Only in German.  In English we capitalize only proper nouns or those
at the beginning of sentences.

Cheers

Phil Hobbs


Well, maybe:

https://www.grammarbook.com/punctuation/capital.asp

Which opens with:

"Capitalization is the writing of a word with its first letter in
uppercase and the remaining letters in lowercase. Experienced writers
are stingy with capitals. It is best not to use them if there is any
doubt."

However it depends on the usage... New York City vs the city of New
York, and so on.

John ;-#)#


But we don't put die Capitals just so die dummkopfen Fremden can
recognize die Nouns. ;)

Cheers

Phil Hobbs
Click to expand...

I just wished people would no longer use filenames like
ThisSchematicIsTheLatestAndGreatest.PDF
3x.20_Precision-1.5kV-1us-Ramp_WH.pdf

Or even worse:
This\ Schematic\ Is\ The\ Latest\ And\ Greatest.PDF
Hurley\ in\ de\ Waterkampioen\ feb\ 2008.pdf

as the \
sucks in Linux


But use like this:
this_schematic_is_the_latest_and_greatest.pdf
 
On a sunny day (Tue, 06 Aug 2019 09:31:58 -0700) it happened John Larkin
<jjlarkin@highlandtechnology.com> wrote in
<4qajke1phctjjthdjmaaer81dducc35edg@4ax.com>:

On Tue, 6 Aug 2019 07:34:26 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Tuesday, August 6, 2019 at 10:16:04 AM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 00:32:26 -0500, "Tim Williams"
tiwill@seventransistorlabs.com> wrote:

"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.

Sure, but star grounding has a place at low frequency.
I cut my star grounding 'teeth' making a low freq. teaching lockin.
https://www.teachspin.com/signal-processor-lock-in

Where a uV of 'signal' getting into the wrong input would ruin my whole day.

George H.

The way to avoid ground loop errors is to keep signals differential.
Click to expand...

Go optical.
 
On Tuesday, August 6, 2019 at 12:32:07 PM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 07:34:26 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Tuesday, August 6, 2019 at 10:16:04 AM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 00:32:26 -0500, "Tim Williams"
tiwill@seventransistorlabs.com> wrote:

"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.

Sure, but star grounding has a place at low frequency.
I cut my star grounding 'teeth' making a low freq. teaching lockin.
https://www.teachspin.com/signal-processor-lock-in

Where a uV of 'signal' getting into the wrong input would ruin my whole day.

George H.

The way to avoid ground loop errors is to keep signals differential.
Click to expand...
Hmm, OK I guess that would help... In the end thinking is required.
I think most of the time I have ground issues, I don't think about it
enough ahead of time. Or maybe you just have to bump your head into a
ground problem before you know enough to think about it. And then it's a circuit
redesign or adding a crutch* to the instrument.

George H.
*I'm thinking of this piece of plastic I had to add to our diodle laser controller
to isolate the DB9 connector shell from the back panel ground.
--

John Larkin Highland Technology, Inc

lunatic fringe electronics
Click to expand...
 
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:5l2jke9d7464mbgfu3kthjl92sjef9vgjm@4ax.com...
Not really. Suppose there's a huge noisy inverter in the box. Ground
loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.
Click to expand...

I'm not spotting anything, can you add an arrow to the "huge noisy inverter
in the box"?

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/
 
"Jan Panteltje" <pNaOnStPeAlMtje@yahoo.com> wrote in message
news:qicbcm$ckr$1@dont-email.me...
I just wished people would no longer use filenames like
ThisSchematicIsTheLatestAndGreatest.PDF
3x.20_Precision-1.5kV-1us-Ramp_WH.pdf
Click to expand...

My favorite is the,

Customer Project 12345_dingleBerryPOWERSUPPLY_REV_E.SCH

A true blend of DOS-style ALLCAPS (but still using LFNs?), caps-avoidance,
underscores (when a space apparently isn't available?) AND spaces (oh..).

Working with other people has only reinforced my understanding that the
90-10 rule applies no matter how "educated" a group you're sampling...

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/
 
On Tue, 6 Aug 2019 10:15:59 -0700 (PDT), George Herold
<gherold@teachspin.com> wrote:

On Tuesday, August 6, 2019 at 12:32:07 PM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 07:34:26 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Tuesday, August 6, 2019 at 10:16:04 AM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 00:32:26 -0500, "Tim Williams"
tiwill@seventransistorlabs.com> wrote:

"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.

Sure, but star grounding has a place at low frequency.
I cut my star grounding 'teeth' making a low freq. teaching lockin.
https://www.teachspin.com/signal-processor-lock-in

Where a uV of 'signal' getting into the wrong input would ruin my whole day.

George H.

The way to avoid ground loop errors is to keep signals differential.
Hmm, OK I guess that would help... In the end thinking is required.
I think most of the time I have ground issues, I don't think about it
enough ahead of time. Or maybe you just have to bump your head into a
ground problem before you know enough to think about it. And then it's a circuit
redesign or adding a crutch* to the instrument.

George H.
*I'm thinking of this piece of plastic I had to add to our diodle laser controller
to isolate the DB9 connector shell from the back panel ground.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics
Click to expand...

I most always bolt the connector shell to the box, and also solder it
to the main PCB ground plane. You never know what nasty stuff might
come in over the shield. Best to stop it soon.

VME is an exception. Dsub shells are bolted to the module front panel
(rack ground) but not connected to the PCB/VME ground plane. Caps are
OK.
 
On Tue, 6 Aug 2019 13:35:04 -0500, "Tim Williams"
<tiwill@seventransistorlabs.com> wrote:

"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:5l2jke9d7464mbgfu3kthjl92sjef9vgjm@4ax.com...
Not really. Suppose there's a huge noisy inverter in the box. Ground
loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.

I'm not spotting anything, can you add an arrow to the "huge noisy inverter
in the box"?

Tim
Click to expand...

Pretty much the entire upper-right quadrant. The prime power is 24
volts DC which we were required to isolate. It's huge and noisy by the
standards of a 12-bit 250 MHz ADC. Not "reading trash."

The CW multiplier string is to make up to 75 volts for an EUV
photodiode.
 
Lasse Langwadt Christensen <langwadt@fonz.dk> wrote:
tirsdag den 6. august 2019 kl. 17.06.48 UTC+2 skrev Don Kuenz:
Tim Williams <tiwill@seventransistorlabs.com> wrote:
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.

Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

But with visibility to where the currents are flowing, or with a
star-grounding scheme, those loop currents stay separate. Say the
connectors are clustered on a front panel, shields grounded to it. That
keeps outside noise out, good. Collect the interior cables (which are stil
l
all grounded to the same point, preferably as coaxially as possible, no
weedy wire links), and bundle them into a single harness. Ferrite beads as
needed. They go over to the PCB, which no longer needs chassis ground at
all.

Or do the brute-force method and use welded chassis compartments to keep th
e
bignasty away from the quietstuff. Cost doesn't much matter in your test
equipment, but it's a very real tradeoff in production. Production even
likes to avoid metal if they can...

My recent audio mix-amp project motivated me to read, or at least skim,
dozens of articles about /audio/ ground loops. Either Bill Whitlock or
one of the other gurus said something along the lines of ground is a
equal potential myth used by engineers to make their jobs easier.
The exact wording and source of that aphorism are temporarily lost
to me. It's Whitlock who definitely dispels other myths about trying to
use thicker grounds or multiple grounds to fight loops.


when I took and emi class at uni one of the first thing the guy said was
something along the line of "ground does not exist"
Click to expand...

Found the formerly lost aphorism! It is indeed from Whitlock, part of
his AES presentation. [1] The pertinent excerpt is shown below.
Allow me to note that my mix-amp utilizes two "grounds" at different
potentials. The mixer uses a virtual "ground" that's halfway between +6V
and -6V. The single supply amp uses +12V and 0V on its power rails (AKA
Vcc and Vdd) so the amp "ground" sits at -6V (Vdd) relative to the mixer
"ground."

--------------------------------------------------------------------
What Does "Ground" Mean
* Also known as "Earth" in the rest of the world
* Utility Power: an actual electrical connection to SOIL
* Electronics: a common return path for various circuits, whether or
not actually connected to soil
* A FANTASY invented by engineers to simplify their work
* The "uni-potential" fantasy assumes all ground symbols in
schematics are at exactly the same voltage
* Truth: Real-world conductors have resistance, causing small
voltage drops
* Truth: Ground circuits most often serve, either intentionally or
accidentally, more than one purpose
* Meaning has become vague, ambiguous, and often quite fanciful

The use of ground symbols in schematic diagrams (actually just
a convenience for avoiding more lines in the drawing) lulls us
into thinking that they’re all at the same potential. That's the
essence of the fantasy ... but far from the truth. Until room-
temperature super-conductors become a common reality, "grounds"
are connected by wires, PCB traces, or sheets of metal - all of
which have both resistance and inductance. So much for the fantasy!

Misguided Strategies
* Reduce unwanted ground voltage differences by "shorting them out"
with massive wires or bus-bars
* Reduce noise experimentally by finding a "better" or "quieter"
ground
* Skillfully route noise to an earth ground, where it disappears
forever!
* Is an earth ground for electronic systems really necessary? Think
about aircraft electronics ...

Random experimentation with grounding connections is the worst
possible way to solve the problems. Not only is little learned in
the process, but the experiment can be lengthy ... usually ending
only when someone says "I can live with that." The idea that
"mother earth" will simply absorb noise is often called the "sump
theory" of grounding. And when was the last time you saw an airplane
dragging around a ground wire?
--------------------------------------------------------------------

Note.

1. https://centralindianaaes.files.wordpress.com/2012/09/indy-aes-2012-seminar-w-notes-v1-0.pdf
Thank you, 73,

--
Don Kuenz KB7RPU
There was a young lady named Bright Whose speed was far faster than light;
She set out one day In a relative way And returned on the previous night.
 
On Tue, 6 Aug 2019 08:41:09 -0700 (PDT), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

tirsdag den 6. august 2019 kl. 17.06.48 UTC+2 skrev Don Kuenz:
Tim Williams <tiwill@seventransistorlabs.com> wrote:
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

But with visibility to where the currents are flowing, or with a
star-grounding scheme, those loop currents stay separate. Say the
connectors are clustered on a front panel, shields grounded to it. That
keeps outside noise out, good. Collect the interior cables (which are still
all grounded to the same point, preferably as coaxially as possible, no
weedy wire links), and bundle them into a single harness. Ferrite beads as
needed. They go over to the PCB, which no longer needs chassis ground at
all.

Or do the brute-force method and use welded chassis compartments to keep the
bignasty away from the quietstuff. Cost doesn't much matter in your test
equipment, but it's a very real tradeoff in production. Production even
likes to avoid metal if they can...

My recent audio mix-amp project motivated me to read, or at least skim,
dozens of articles about /audio/ ground loops. Either Bill Whitlock or
one of the other gurus said something along the lines of ground is a
equal potential myth used by engineers to make their jobs easier.
The exact wording and source of that aphorism are temporarily lost
to me. It's Whitlock who definitely dispels other myths about trying to
use thicker grounds or multiple grounds to fight loops.


when I took and emi class at uni one of the first thing the guy said was
something along the line of "ground does not exist"
Click to expand...

It is usually in the form "universal ground _potential_ doesn't
exist", all ground potentials are different. Absolute ground potential
doesn't exist even in Sevre outside Paris, where they used to keep the
standard kilogram prototype and the standard meter prototype :).

While there has been a lot of discussion in this thread about
grounding topologies inside the box, things get nasty when boxes are
connected together with perhaps a large distances between boxes.

A good practice is to have three separate grounding networks on site,
one for mains neutral N, one for the PE network and a separate
Technical/functional earth TE/FE networks. Each of these networks have
dedicated bars at the mains entry point of the building or complex. At
exactly this single point only the three bars are joined together with
jumpers as thick as your thumb. When the system is down, the jumpers
can be removed and measurements can be made to check that the networks
are truly separate.

The N network is polluted by mains return current of a single phase
system or the current imbalance in a three phase system. Even in a
balanced three phase system the third harmonics due to rectifier load
will pollute the neutral.

Mains power supply filters are usually connected to the PE, so it is
polluted by power supply noise.

Signal grounds are usually connected to the hopefully clean TE/FE
network. This requires that no equipments which has a signal ground
connected to the chassis and then to the PE are connected to TE/FE,
since a single mains filters will pollute the whole TE/FE network.

For this reason N, PE and signal grounds (TE/FE) should ne kept
separated and all three brought to the connection pins of a box, so
that they can be connected to three separate ground networks.

I have noticed a problem with some installers in foreign countries
that when they notice these three ground terminals they simply connect
jumpers between these terminals outside the box and use a single wire
for a connecting to mains neutral, now actually PEN :-(
 
Mikko OH2HVJ wrote:

Some metric comments; many metric people don't have the gut feeling of
AWGs or mils.
Click to expand...

I concur, the circular mil per square foot times ounce parlance is
hardly compehensible to a person grown up in a SI-only environment.

Best regards, Piotr
 
On 2019-08-06 18:31, John Larkin wrote:
On Tue, 6 Aug 2019 07:34:26 -0700 (PDT), George Herold
gherold@teachspin.com> wrote:

On Tuesday, August 6, 2019 at 10:16:04 AM UTC-4, John Larkin wrote:
On Tue, 6 Aug 2019 00:32:26 -0500, "Tim Williams"
tiwill@seventransistorlabs.com> wrote:

"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

Here's a 12-bit, 250 MHz ADC a couple inches from a multi-output
switching power supply. The 6-diode CW multiplier string is about an
inch away.

https://www.dropbox.com/s/opa02hou39tduiq/ESM_rev_B.jpg?raw=1

Noise is below 1 lsb RMS.

One solid layer 2 plane. Everything is grounded.

Sure, but star grounding has a place at low frequency.
I cut my star grounding 'teeth' making a low freq. teaching lockin.
https://www.teachspin.com/signal-processor-lock-in

Where a uV of 'signal' getting into the wrong input would ruin my whole day.

George H.

The way to avoid ground loop errors is to keep signals differential.
Click to expand...

Or using a current not a voltage as signal.

Arie
 
On 06.08.19 1:09 PM, Phil Hobbs wrote:
On 8/6/19 4:07 AM, David Nadlinger wrote:
Regarding that circled capital L in footnote 21: Litzendraht as a
German noun should indeed be uppercase, so the current footnote is
correct (potential disagreements about capitalisation of the loanword
in the body copy notwithstanding).

  — David

Only in German.  In English we capitalize only proper nouns or those at
the beginning of sentences.
Click to expand...

Really? I somehow hadn't realised this in all these years. ;)

Jokes aside, the footnote I mentioned specifically mentions/quotes a
German noun, which needs to be capitalised. Lowercase-"litzendraht"
wouldn't make sense in either language.

— David
 
On 8/6/19 2:43 PM, Tim Williams wrote:
"Jan Panteltje" <pNaOnStPeAlMtje@yahoo.com> wrote in message
news:qicbcm$ckr$1@dont-email.me...
I just wished people would no longer use filenames like
ThisSchematicIsTheLatestAndGreatest.PDF
3x.20_Precision-1.5kV-1us-Ramp_WH.pdf


My favorite is the,

Customer Project 12345_dingleBerryPOWERSUPPLY_REV_E.SCH

A true blend of DOS-style ALLCAPS (but still using LFNs?),
caps-avoidance, underscores (when a space apparently isn't available?)
AND spaces (oh..).

Working with other people has only reinforced my understanding that the
90-10 rule applies no matter how "educated" a group you're sampling...

Tim
Click to expand...

I use camel case e.g. because it helps a lot finding the right file from
my archives, e.g.

2019-08-05-InvoiceABC05.pdf
TLV431AshuntRefOnSemiUnstableHorribleAvoidAvoid.pdf ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com
 
On 8/6/19 3:57 PM, upsidedown@downunder.com wrote:
On Tue, 6 Aug 2019 08:41:09 -0700 (PDT), Lasse Langwadt Christensen
langwadt@fonz.dk> wrote:

tirsdag den 6. august 2019 kl. 17.06.48 UTC+2 skrev Don Kuenz:
Tim Williams <tiwill@seventransistorlabs.com> wrote:
"John Larkin" <jjlarkin@highlandtechnology.com> wrote in message
news:cavhkeppudjiempknhs0jb84jpkau3sgo8@4ax.com...
Signal integrity is easy. PCB layer 2 is solid ground plane. Bolt it,
and grounded connector shells, to the metal box in as many places as
possible.

Cable shields to the box ditto.


Not really. Suppose there's a huge noisy inverter in the box. Ground loop
currents once again. Cables simply tied "shields to the box", and PCB
"bolt[ed] ... to the metal box", draws those currents into both. Now your
precision ADCs are all reading trash.

But with visibility to where the currents are flowing, or with a
star-grounding scheme, those loop currents stay separate. Say the
connectors are clustered on a front panel, shields grounded to it. That
keeps outside noise out, good. Collect the interior cables (which are still
all grounded to the same point, preferably as coaxially as possible, no
weedy wire links), and bundle them into a single harness. Ferrite beads as
needed. They go over to the PCB, which no longer needs chassis ground at
all.

Or do the brute-force method and use welded chassis compartments to keep the
bignasty away from the quietstuff. Cost doesn't much matter in your test
equipment, but it's a very real tradeoff in production. Production even
likes to avoid metal if they can...

My recent audio mix-amp project motivated me to read, or at least skim,
dozens of articles about /audio/ ground loops. Either Bill Whitlock or
one of the other gurus said something along the lines of ground is a
equal potential myth used by engineers to make their jobs easier.
The exact wording and source of that aphorism are temporarily lost
to me. It's Whitlock who definitely dispels other myths about trying to
use thicker grounds or multiple grounds to fight loops.


when I took and emi class at uni one of the first thing the guy said was
something along the line of "ground does not exist"

It is usually in the form "universal ground _potential_ doesn't
exist", all ground potentials are different. Absolute ground potential
doesn't exist even in Sevre outside Paris, where they used to keep the
standard kilogram prototype and the standard meter prototype :).

While there has been a lot of discussion in this thread about
grounding topologies inside the box, things get nasty when boxes are
connected together with perhaps a large distances between boxes.

A good practice is to have three separate grounding networks on site,
one for mains neutral N, one for the PE network and a separate
Technical/functional earth TE/FE networks. Each of these networks have
dedicated bars at the mains entry point of the building or complex. At
exactly this single point only the three bars are joined together with
jumpers as thick as your thumb. When the system is down, the jumpers
can be removed and measurements can be made to check that the networks
are truly separate.

The N network is polluted by mains return current of a single phase
system or the current imbalance in a three phase system. Even in a
balanced three phase system the third harmonics due to rectifier load
will pollute the neutral.

Mains power supply filters are usually connected to the PE, so it is
polluted by power supply noise.

Signal grounds are usually connected to the hopefully clean TE/FE
network. This requires that no equipments which has a signal ground
connected to the chassis and then to the PE are connected to TE/FE,
since a single mains filters will pollute the whole TE/FE network.

For this reason N, PE and signal grounds (TE/FE) should ne kept
separated and all three brought to the connection pins of a box, so
that they can be connected to three separate ground networks.

I have noticed a problem with some installers in foreign countries
that when they notice these three ground terminals they simply connect
jumpers between these terminals outside the box and use a single wire
for a connecting to mains neutral, now actually PEN :-(
Click to expand...

That fancy approach works great until you find yourself working next to
a big VF motor drive. :(

At that point, squeezing everything onto one board and keeping the loop
areas tiny is a big help.

Cheers

Phil Hobbs

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com
 
On 8/6/19 5:05 PM, David Nadlinger wrote:
On 06.08.19 1:09 PM, Phil Hobbs wrote:
On 8/6/19 4:07 AM, David Nadlinger wrote:
Regarding that circled capital L in footnote 21: Litzendraht as a
German noun should indeed be uppercase, so the current footnote is
correct (potential disagreements about capitalisation of the loanword
in the body copy notwithstanding).

  — David

Only in German.  In English we capitalize only proper nouns or those
at the beginning of sentences.

Really? I somehow hadn't realised this in all these years. ;)

Jokes aside, the footnote I mentioned specifically mentions/quotes a
German noun, which needs to be capitalised. Lowercase-"litzendraht"
wouldn't make sense in either language.

 — David
Click to expand...

It's a matter of usage. For instance, in English there's such a thing
as a cousin-german. (Not cousin-German.) I'd never write "litz wire"
with a capital.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top