[RFI] ethernet EMI revisited - questions for the pros
k8ri at rogerhalstead.com
Sun Nov 16 21:27:50 EST 2008
aa8ia at aa8ia.org wrote:
>> From: "Martin Ewing" <martin.s.ewing at gmail.com>
>> Subject: Re: [RFI] ethernet EMI revisited - questions for the pros
>> This is an interesting discussion. I would add a few points/questions:
>> -Does anyone know of published "typical" spectra of 1000baseT vs 100 vs 10?
>> I agree that 10bT is much quieter on HF than 100bT, but it is possible that
>> 1000bT pushes the spectrum up enough to be quieter on HF. I did some work on
I doubt that is what is happening. You have the basic clock frequencies
in the chips generating or regenerating the basic data frame. Typical
systems that run Gigabit also run 10 and 100 base-T. Whether the clock
actually changes I don't remember. The data stream as you have surmised
generates a noise that is pretty much random. That noise not only
depends on the basic speed but character string. Typically the highest
frequency would be generated by a stream of alternating 1's and 0's.
Unfortunately life is not simple in the network world and much depends
on the chip set and the implementation.
Wired networks normally use twisted pair which is very good at
discriminating against common mode ingress, but as they are wound around
each other it can make a good antenna for radiating common mode. However
the inductance and capacitance from this method also limits the distance
and speed. Theoretically the 1's and 0's should be "square, but in real
life they are rounded off and the waveform slopes across the
transition. The longer the run of CAT5e the more they are rounded off
and attenuated until you reach a point where the amplitude and
transition have degraded to the point where the receiver can no longer
recognize the transitions reliably. This is *about* 300' in a straight
line with no bumps in the impedance. Then the chip sets do not generate
a perfect squared off transition. Some of the older chip sets and
implementations were terrible, not only with the wave form but they
generated lots of noise.
I once made the mistake of stapling a run of CAT5e to the garage
ceiling. To say its range was limited would be an understatement and no
I didn't stick any through it. <:-))
Two things I remember about Networking. We had to figure the data rate
for a given string of characters using different base speeds such as 10
and 100 base-T. That's a lot of Calculus that I sorta vaguely remember.
The second was that class was without doubt the toughest one I had in
college and that includes writing a full blown C compiler in "Compiler
Design", that worked using an input scanner..
>> this some years back: http://aa6e.net/rfi/ether_details.html .
> After hearing what Roger had to say regarding his all-gigabit network,
It's not exactly the typical home installation.<:-))
> I was wondering the same thing. I'll check out your link.
It's quite interesting.
>> -Do you know if your RFI is coming in through your antenna? It may be
>> coupling in to your feedlines locally in the shack. If this is so, you need
>> to look into balancing your antenna system. In-shack RFI sources should not
>> matter very much if you're only listening to what's coming from your
>> (distant?) antenna.
> I probably should have mentioned that my antenna is only up about 20
> feet right now - It is outside and the closest point is about 10 feet
> from the house. The RFI can be heard, but to a lesser extent, from
> inside the house using a portable antenna. When I hook up my
> receivers to the antenna outside, via the feedline running next to the
> routers, the signals are definitely stronger. My feedline is LMR400
> and runs into the basement to within a foot of the DSL router.
I had 5, 200 foot runs of LMR-400 plus 28' pigtails at the top. It's
about 74' from the basement to the base of the tower. From the rigs to
where the lines ran through the conduit to the tower they all ran in one
bundle for about 6 to 8 feet. This is the equipment box and run to the
Unfortunately I have no photos of the "rats nest" prior to leaving the
basement. You'll note two conduit runs to the base of the tower. The
equipment at the ends has two different power feeds, but a common ground
system. So essentially the grounds for the shop and the house are
eventually tied into one large network of 32 or 33 8' ground rods
cadwelded to many, many feet of bare cable.
> wouldn't have expected the intereference to be picked up by the
> LMR400, but I'm guessing that is a possibilty? If I hook the receiver
> up to the antenna outside, bypassing the feedline that runs into the
> basement here, the signal levels of the interference are much less.
> So yes it does sound like it is getting in through the feedline.
The shield might possibly carry the signal out to where the antenna
hears it and at that point it is no longer common mode due to frequency
and phase changes.
>> -I have found that much of the Ether RFI is generated by switches/routers
>> and passed on CAT 5 as common mode signals.
> I think that is my experience as well - specifically from the ethernet
> transceivers. If I have nothing plugged into the ethernet of either
> router, there is no interference from those devices.
I'm wondering what generation of chips and equipment you might be
running? The newer equipment is much cheaper (or less expensive) than
the older when it was on the market and the signals are *IN GENERAL*
much cleaner. My router suffered a catastrophic accident and is
currently pretty much PC board hanging on a bunch of leads.
>> -The "right" solution for wiring around the house is WiFi in my case.
>> Security is a minor issue in my situation, and all the high-speed traffic is
>> between my systems in the shack. Long runs of any copper cabling are an
>> invitation to HF troubles, esp. if you're QRO.
I have KW stations on each end, plus the CAT5e runs to the shop pass
within 10 feet of the base of the 100' 45G where the antennas top out at
130 feet. For 6 to 8' it runs in a bundle with all of the cables. That
tower takes on average 3 *verified* direct strikes a year although there
were none seen this past summer. I have seriously considered optical
cable, but the pieces that go at the ends are just a tad pricey for me
> I agree with you, as long as you don't have a need such as Rogers' for
> high speed backup.
Guess I haven't changed much then: <:-))
Reminds me of when I was working. I had a developmental Laboratory
Information Management System, better known as LIMS. The data from the
running system was backed up to my system daily and tested programs were
uploaded/transfered to the running system when I approved them. Thing
is, this was at a very large chemical corporation with many labs. There
were many millions of records that ended up being transferred at each
daily update. The system architecture was changed and all of my traffic
ended up going across the bridge that carried all of the office and
records traffic. It almost brought things to a stand still.
> The wireless adapters I currently have in use are
> running at 54 mbit and are "quiet."
I believe the newer wireless will run almost twice that fast, but not
quite the equivalent of 100 Base-T.
> I'm going to get another USB
> wireless adapter and a PCI wireless card (for an older machine) and do
> away with ethernet for normal operation. But there will always be
> times when I have to plug some ethernet device in for a day or two in
> order to work on it, and so I want to clean things up.
The one router here has 5 RJ45 ports and can still handle quite a few
Wi-Fi connections when turned on. Of course with multiple computers it
really slows down. That's one of the things I like about the switch. I
can have two computers exchanging files without having much effect on
the rest of the network including connections to the Internet.
> RFI mailing list
> RFI at contesting.com
More information about the RFI