Maker Pro
Maker Pro

Grounding

E

Eeyore

Jan 1, 1970
0
Jim said:
Crap. Graham can't even give good advice on how to wipe ;-)

Let me explain for you then ! After wiping you 'wash'. A moist tissue is fine.

Graham
 
E

Eeyore

Jan 1, 1970
0
John said:
Just recently, I put a power supply (linear and switching regs), a
temperature controller (thermocouple-input, 150 watt PWM heater
output), a PPM-noise-level, 3-amp NMR gradient coil driver, and a uP
on a single small board, with a single ground plane. Works fine and
passed FCC and CE first time.

And another board has a VME interface, switching regs, uP, and twelve
12-bit ADCs clocked at 40 MHz, with sub-LSB RMS noise levels. Single
plane.

12 bit ! ? I should damn well hope there's no noise above an LSB !

20 bits here my lad ! Without split grounds the 'digital noise' was present in
the analogue signal at ~ -85dB wrt FSD ( which is +20dB wrt normal operating
level so S/N was degraded to 65dB ). Split grounds made it essentially inaudible
- easily better than 100dB down - not practical to measure it in fact, inside
the overall noise floor.

Graham
 
K

Keith

Jan 1, 1970
0
Sounds like you're working at somewhat higher freqiencies than me ! ;~)

CMOS with supplies approaching 75A with a 3GHz clock. ;-) Well,
that was last year...
 
J

Jim Thompson

Jan 1, 1970
0
Hello Graham,


Plus professors often teach that stuff. That, and those reference
designs, have kept my business humming. When I started out I really
didn't want to go into EMI but split grounds have kind of sucked me into
that funnel.



It can also mean that the chip designers didn't have much of a clue
about the system side of things ;-)

Come on, Joerg, It's to avoid the common inductance to ground if you
use a single wirebond.

They are meant to connect together to a single ground plane outside of
the package.
The worst case was a slew of field failures. Some pulses had hit into
the substrate diodes and much of the stuff across the AGND-DGND devide
fried. Nobody knew where the pulses came from but after a re-layout with
a common plane there were no more field failures.

In EMI we have an old rule: The EMI performance of a given design is
inversely related to the number of toroids it requires to make it behave.


...Jim Thompson
 
E

Eeyore

Jan 1, 1970
0
Jim said:
Yep. You don't want a common wirebond to cross-couple noise.

Now whyever might that be I wonder ? Something to do with resistance perhaps ?

Graham
 
J

Joerg

Jan 1, 1970
0
Hello Graham,
Sounds like you're working at somewhat higher freqiencies than me ! ;~)

That's not really the reason. There simply is no way to provide a low
impedance ground path on a chip, at least not at a reasonable cost. All
you have is a few teeny bond wires and those cannot connect the chip's
"plane" to the board plane very well. So you may need two returns there.
Very common on ADCs. But that certainly does not mean you have to keep
AGND and DGND split outside the chip.
 
Eeyore said:
Many reference designs use 'split' grounds.

There's a bit of a clue when the chip itself has dedicated AGND and DGND pins !

Graham

not really, the purpose of the pins AGND and DGND is not to suggest
that these pins be connected to the system analog and digital ground,
respectively, but to identify which pin is connected to the analog
gound of the IC and which pin is connected to the digital ground of the
IC.

It's up to the designer to decide which gets connected to which ground,
for low power digital mixed signal systems it's best to connect AGND
and DGND to the analog ground (i.e., the mixed signal device is
considered an analog part). For high powered mixed signal systems
(i.e., a fast DSP with built in A/D), then you split them.

See figure 9 for an example of the former and figure 10 for the latter,

http://www.analog.com/en/content/0,2886,760%5F%5F97529,00.html
 
J

John Woodgate

Jan 1, 1970
0
In message <[email protected]>, dated Tue, 8
Aug 2006 said:
Just recently, I put a power supply (linear and switching regs), a
temperature controller (thermocouple-input, 150 watt PWM heater
output), a PPM-noise-level, 3-amp NMR gradient coil driver, and a uP on
a single small board, with a single ground plane. Works fine and passed
FCC and CE first time.

And another board has a VME interface, switching regs, uP, and twelve
12-bit ADCs clocked at 40 MHz, with sub-LSB RMS noise levels. Single
plane.

I also did the NIF timing system: about 2000 client points triggered to
within a few picoseconds accuracy, spread over the world's biggest
laser, the size of a football stadium. No split planes there, either.

This is NOT the way to gain friends among your colleagues.(;-) The
British English expression is 'jammy barsteward'.
 
J

John Larkin

Jan 1, 1970
0
Many reference designs use 'split' grounds.

Most of the eval board have a single chip, which the vendor considers
to be the center of the universe; they'll also tell you to bypass
every power pin with three capacitors, typically. A single plane-plane
connection under one chip gets messy when there are 10 ADCs on the
board. A bigger problem is that each of the ADCs (or whatever) has to
connect to the outside world, which can cause more ground loops and
EMI.
There's a bit of a clue when the chip itself has dedicated AGND and DGND pins !

That reduces ground bounce caused by wirebond inductance. Connecting
these pins to separate planes (at separate potentials!) can pump
serious noise currents through the AGND wirebond, which can be bad
news.

John
 
J

John Larkin

Jan 1, 1970
0
Not in the case of audio. You have to be careful where those ground currents
flow.

Well, of course you have to be careful. But audio isn't very
demanding.

John
 
D

Didi

Jan 1, 1970
0
Hello Joerg,
12 bits at over 100MSPS. Plus a time-gain control amp in front to cover
the required range.

that is of course challenging enough in itself, but is not the type of
situation I thought of. In the module I quoted, I do 10 MSPS, but
at 14 bits. If you have meant ground splitting will cause more
inductive
issues than it will solve, I agree with you. But at 14+ bits, with a
power supply switcher on the board, the sheer resistance of the
ground plane may be not low enough (on a small board)...
On the board you have seen on my website, you can probably well
guess where I have split the ground. One of the things that helped
was the fact there are no fast digital signals entering the analog
section
(just relay driving lines :), otherwise the split would have again
caused problems (signal "return" path related).

Dimiter
 
J

Jim Thompson

Jan 1, 1970
0
Well, of course you have to be careful. But audio isn't very
demanding.

John

As long as you use Monster Cable ?:)

...Jim Thompson
 
K

Kevin White

Jan 1, 1970
0
John said:
Don't do it. Use a single solid ground plane and handle any low-level
problems locally.

Splitting planes implies an impedance between them, hence high
relative AC potentials between "grounds". That will generally cause a
lot more trouble than it cures.

John

I agree, I have never found an advantage in splitting the grounds. I
try to keep the ground impedance as low as possible.

I do carefully manage the current flow from the digital and power
sections to minimize noisy current flow in traces used for analog
signals. Usually this means shaping the ground planes or cutting it in
various places - I keep the connection between these sections as broad
as possible though - definitely not a narrow trace or zero ohm
resistor.

Other techniques are to put resistors in series with the digital
signals driving A/Ds, so reduce the current flow in the input
capacitance of the A/D. Often it is useful to buffer the A/D outputs
with digital buffers so that the A/D does not drive significant
capacitative load and thus put high currents in its ground system.

I like to put lots of copper everywhere to keep as low an impedance
gorund as possible - for boards I do myself using Eagle CAD the polygon
capability works well for this. Admittedly this can increase
capacitance to ground which may be undesirable.

Be careful to avoid the digital signals passing over the ground region
for the analog signals. Analog Device has some useful app notes on
this subject.

Which approach is used depends upon the signal frequency. At low
frequencies the resistance of the ground system is the dominant effect
and using star wiring back to the power source may be the most
appropriate - at video frequencies and above (includes most digital
systems) the inductance is a major problem so putting lots of paths in
parallel and avoiding capacitative or inductive coupling between
sections will probably work better.

A design I am working on currently uses 17 LTC1403A 14-Bit 2.5MHz A/Ds
in an optical measurement system with bandwidth of around 1MHz.. I
have surrounded the pre-amp sections with the A/Ds such that the
control signals (isolated with resistors at the driving ends) pass
around the periphery of the analog section. I have flooded the analog
section with copper to reduce the ground impedance.

These A/Ds are pretty nice to work with - they have differential inputs
and serial outputs - reduces the number of digital lines causing noise,
and a pretty small MSOP-10 package.

In initial measurements the noise level seems to be pretty much as
expected.

cheers

kevin
 
E

Eeyore

Jan 1, 1970
0
Joerg said:
Hello Graham,


Very low frequencies is where star-grounds and stuff like that can
indeed work. Ultrasound guys consider audio to be DC ;-)

Quite so. My own view about most of the stuff I do is that it's 'moving DC' too !

Problems can still creep up when you install such split ground gear in
the vicinity of a strong RF emitter. Radar bases, coast guard stations,
AM talk radio towers, 13.56MHz "blasters", you name it, I have seen them
pretty much all. Mostly I have to come out there to find out what's
causing EMI and then sit down with the client to gently break the news
that they need a few re-layouts and chassis mods. Usually the engineers
take that in strides but often the enclosure designers don't. Frequently
they are artists rather than engineers and they can become quite livid
when I suggest to move away from an injection molding scheme.

Ah ! I ensure I'm involved in enclosure design right from the start.

Graham
 
E

Eeyore

Jan 1, 1970
0
John said:
Well, of course you have to be careful. But audio isn't very
demanding.

110+ dB dynamic range isn't demanding ?

Graham
 
E

Eeyore

Jan 1, 1970
0
Jim said:
As long as you use Monster Cable ?:)

Monster cable is for the 'aftermarket' clowns. It's not even especially well made
I'm told.

Graham
 
P

Phil Hobbs

Jan 1, 1970
0
Joerg said:
Hello Graham,


Very low frequencies is where star-grounds and stuff like that can
indeed work. Ultrasound guys consider audio to be DC ;-)

Some of us consider millimeter-waves to be dc too!

Low frequency ground currents aren't nearly as well localized as at high
frequency, due to the lack of inductive reactance, so they tend to smear
out all over the board unless you do some strategic cutting.

Cheers,

Phil Hobbs
 
D

Didi

Jan 1, 1970
0
John,

I know you have enough experience. My "advice" related comment
was (hopefully obviously) directed at those participant(s) of the
thread
who were looking for and advice.

As for "not a single split plane", my latest reply to Joerg says what
I think of it. It is all a matter of small and precise enough until you
run into plane resistance problems. Have a look at
http://tgi-sci.com/tgi/hs_top.gif , the split plane sure does not hurt
and I am quite sure unsplitting it will yield a measurable (although
not dramatic) worsening effect.
Again, it is a matter of size/precision/speed.

And many years ago (15....), I was mad enough to do a similar
thing with a _flyback_ convertor on board (these here are just
step-down). On that design, without the split planes with all
the peak currents flyback convertors have, I would just have
been lost, it would have not been just about being "measurable".

Dimiter
 
E

Eeyore

Jan 1, 1970
0
not really, the purpose of the pins AGND and DGND is not to suggest
that these pins be connected to the system analog and digital ground,
respectively, but to identify which pin is connected to the analog
gound of the IC and which pin is connected to the digital ground of the
IC.

It's up to the designer to decide which gets connected to which ground,
for low power digital mixed signal systems it's best to connect AGND
and DGND to the analog ground (i.e., the mixed signal device is
considered an analog part). For high powered mixed signal systems
(i.e., a fast DSP with built in A/D), then you split them.

See figure 9 for an example of the former and figure 10 for the latter,

http://www.analog.com/en/content/0,2886,760%5F%5F97529,00.html

And for audio you split them too. See page 9 also. Nice PCB film

http://www.wavefrontsemi.com/UserFiles/File/AL_Info/11/WavefrontEV1101A.pdf

Graham
 
Top