LVDS or RS 485 for lowest EMI

[Rene Tschaggelar]

As serial connection on desktop scale (few meter), what
would give the lowest electromagnetic interference, LVDS
or RS485 ? The speed is few 10kHhz to few 1 MHz.
There are no GND issues.

LVDS is 3mA into 100 Ohms while RS485 is 5V into 100
Ohms. Since LVDS is much faster, I'd assume it to have
less but shorter transients.

Rene

 

[Dave Garnett]

As serial connection on desktop scale (few meter), what
would give the lowest electromagnetic interference, LVDS
or RS485 ? The speed is few 10kHhz to few 1 MHz.
There are no GND issues.

LVDS is 3mA into 100 Ohms while RS485 is 5V into 100
Ohms. Since LVDS is much faster, I'd assume it to have
less but shorter transients.

Typical LVDS drivers have much faster edges, which means that you will see
EMI going up to hundreds of MHz - I speak from experience. In principle use
of a twisted pair should mean cancellation of most of the EMI, but in
practise driver skew and cable asymmetry mean that the cancellation is not
perfect. If EMI is an issue, I'd go for the slowest edges I could get away
with ...

Dave

 

[Tim Hubberstey]

Typical LVDS drivers have much faster edges, which means that you will see
EMI going up to hundreds of MHz - I speak from experience. In principle use
of a twisted pair should mean cancellation of most of the EMI, but in
practise driver skew and cable asymmetry mean that the cancellation is not
perfect. If EMI is an issue, I'd go for the slowest edges I could get away
with ...

I believe I've seen RS485 transceivers with controllable slew rates.
Don't remember where though (Maxim? Linear Tech?).

 

[Larry Brasfield]

As serial connection on desktop scale (few meter), what
would give the lowest electromagnetic interference, LVDS
or RS485 ? The speed is few 10kHhz to few 1 MHz.
There are no GND issues.

LVDS is 3mA into 100 Ohms while RS485 is 5V into 100
Ohms. Since LVDS is much faster, I'd assume it to have
less but shorter transients.

The common mode current out of an LVDS driver is
approximately constant. The transient component of
it is very short, which means that, if you filter it with an
LPF, the content below cutoff is low. I would expect
that you could build an RC filter to get residual EMI
below where it is normally a concern with either of
those standards, but it would be easier with LVDS.
However, if you do not want to build that filter and
plan to use unshielded twisted pair, a slew limited
RS485 driver would be a better bet.

 

[martin griffith]

As serial connection on desktop scale (few meter), what
would give the lowest electromagnetic interference, LVDS
or RS485 ? The speed is few 10kHhz to few 1 MHz.
There are no GND issues.

LVDS is 3mA into 100 Ohms while RS485 is 5V into 100
Ohms. Since LVDS is much faster, I'd assume it to have
less but shorter transients.

Rene

Hi Rene,
I just found this, while lookng for something completely different....

http://focus.ti.com/docs/apps/catalog/resources/appnoteabstract.jhtml?abstractName=slla169




martin

After the first death, there is no other.
(Dylan Thomas)

 

[Rene Tschaggelar]

Hi Rene,
I just found this, while lookng for something completely different....

http://focus.ti.com/docs/apps/catalog/resources/appnoteabstract.jhtml?abstractName=slla169

Thanks Martin,
the article was quite an eye opener.
How do I communicate the increased cost for the better cable...

Rene

 

[martin griffith]

Thanks Martin,
the article was quite an eye opener.
How do I communicate the increased cost for the better cable...

Rene

I only read it quickly, but there was no mention about cable emissions
with high data rates.

Time for a bit more digging around. But CAT5/6 etc. is installed
everywhere, it cant be that much of a problem, can it?




martin

After the first death, there is no other.
(Dylan Thomas)

 

[Rene Tschaggelar]

As serial connection on desktop scale (few meter), what
would give the lowest electromagnetic interference, LVDS
or RS485 ? The speed is few 10kHhz to few 1 MHz.
There are no GND issues.

LVDS is 3mA into 100 Ohms while RS485 is 5V into 100
Ohms. Since LVDS is much faster, I'd assume it to have
less but shorter transients.

Rene

More papers found :

Extending SPI and McBSP With Differential Interface Products
http://www-s.ti.com/sc/psheets/slla142/slla142.pdf

Comparing Bus Solutions
http://www-s.ti.com/sc/psheets/slla067a/slla067a.pdf


There is tons of documents after you have an idea.

Rene

 

[Rene Tschaggelar]

I only read it quickly, but there was no mention about cable emissions
with high data rates.

Time for a bit more digging around. But CAT5/6 etc. is installed
everywhere, it cant be that much of a problem, can it?

It was less about the emissions of bad cable but losses.
The common perception of RS485/422 is a mile at high baudrate,
so one is fine in all cases. Not so.
For longer cables and higher datarates, better cable are
required. Yes, there are CAT5 and CAT6 cable.
Installed everywhere ? No. I have them to become installed.
Your paper also compared different drivers, suitable for
different applications.

The paper
http://www-s.ti.com/sc/psheets/slla030c/slla030c.pdf
is concerned about EMI and LVDS and competing systems.

Rene

 

[Mac]

As serial connection on desktop scale (few meter), what
would give the lowest electromagnetic interference, LVDS
or RS485 ? The speed is few 10kHhz to few 1 MHz.
There are no GND issues.

LVDS is 3mA into 100 Ohms while RS485 is 5V into 100
Ohms. Since LVDS is much faster, I'd assume it to have
less but shorter transients.

Rene

After reading through the other posts, it occurs to me that you could
limit the slew rate of fast LVDS edges using series resistors and/or
shunt capacitors at the driver end.

Also, LVDS usually has very low skew and is often more differential (as
opposed to complimentary), so it might get better cancellation in a
twisted pair.

Given all that, I think LVDS would have lower emissions. But this is just
a gut feeling. I don't have tons of EMI experience or anything like that.

Whichever standard you use, you may want to use a common-mode choke to
put a damper on any high-frequency common-mode signal that may be trying
to get on the wire.

One of the reasons fast Ethernet and Gigabit Ethernet work with unshielded
twisted pair is that they employ multiple common-mode cancelling schemes,
including common-mode chokes. Since you need to be DC-coupled, you can't
do all that, but you could at least use the common mode choke.

--Mac

 

[Fred Bloggs]

Larry Brasfield wrote:
[...snip non-informational gibberish...]

However, if you do not want to build that filter and
plan to use unshielded twisted pair, a slew limited
RS485 driver would be a better bet.

Slew-rate limited RS485 does not come close to achieving a "few MHz" bit
rate, so once again you are a clueless p.o.s. shooting your mouth off
with a lot of terminology you don't know a thing about. You are a
worthless sack o'shyte as usual. Go away and stay away.

 

[Larry Brasfield]

Larry Brasfield wrote:

[Cut the usual Bloggs crap.]

Slew-rate limited RS485 does not come close to achieving a "few MHz" bit rate,

The OP stated "The speed is few 10kHhz to few 1 MHz."

Maxim has more than a few RS-485/422 transmitters
that do 2.5 MB/S. Some folks, not so intent on spew
as you, would call that close.

[Cut more of the usual Bloggs crap.]

Go away and stay away.

No, for the Nth time.

 

[Fred Bloggs]

The OP stated "The speed is few 10kHhz to few 1 MHz."

Maxim has more than a few RS-485/422 transmitters
that do 2.5 MB/S. Some folks, not so intent on spew
as you, would call that close.

Close is not good enough- it either makes or it doesn't- period. You are
a half-assed p.o.s. And do they have transceivers that do 2.5Mbps- you
don't say?- what a fountain of the mundane you are, idiot.

[Cut more of the usual Bloggs crap.]

Go away and stay away.

No, for the Nth time.

All you are doing is publicizing your ignorance and mediocrity- that
much should be clear even to you by now. You have nothing to offer
usenet in the way of substantial posts- you are too damned dumb to
understand the OP's question 99% of the time and then your answers are
worthless garbage- just like you.