[Mirrors]
Dealing with Computer generated RFI/EMI
Contents:
[Document Version: 1.40]
[Last Updated: Jan_30_1994]
Author: Daniel 9V1ZV
E-Mail: daniel%pandora@csah.com, 9V1ZV @ 9V1VS.SGP.AS,
or asirene@ntuvax.ntu.ac.sg
Date: 30 January 1994
Release: 1.4
One of the most frustrating problems about using computers with
radios, whether it be for controlling purposes or for decoding, is the
amount of RFI generated by these machines. Most of the time, the RFI
generated is enough to render certain bands useless and on other bands, it
may drown out any weak signals and distort or interfere with signals that
you want. This is totally unacceptable for working with digital modes and
even for CW. Thus one of the most frequently asked question is how this RFI
may be reduced or eliminated. The bad news is that, there is no way that I
know of to completely remove the computer generated RFI in most situations.
The good news is that there are definite steps that we can take to reduce the
RFI to a very acceptable level and in some cases, it will almost disappear
altogether. This document is a compilation of suggestions from various
persons and some of the things I have tried with my own system when dealing
with this problem. Many of the documents I have seen relate to situations
involving transmitters and how not to generate them (RFI). This document is
written from a receiving point of view.
I suppose most people would have already tried the basic steps to
improve signal conditions by having the receiver and antenna as far removed,
physically and electrically from the RFI source (computers and monitors in
our case) but I am also aware that sometimes there are limitations and
constraints as to how much distance can be had. One may have also tried
changing the orientation of the computer, monitor, receiver and antennas and
feeds to see if things get better. Having done all that, what else can be
done? This is the predicament I had and thus this document.
The standard disclaimer applies and I will not be responsible for
any accidents although I have tried my best to present the following
information in the best integrity.
Before we actually begin tackling the problem, it might be helpful
to know something about why computer generates RFI and how these get into
the receiver. The two main components of the computer is the main CPU and
the monitor (for simplicity sake). The computer runs at a certain clock rate
as determined by an internal oscillator. Most of the time the rates are
something like 4.77 mHz, 8 mHz, 12 mHz, 16 mHz, 20 mHz, 25 mHz, 33 mHz, 40
mHz, 50 mHz, 66 mHz and 80 mHz. This is not the only clock involved, there
is also another oscillator on the video generator card and sometimes a few
oscillators, plus those on other cards. As you can see, these clocks are all
oscillating in the HF and L-VHF regions which may interfere with signals we
would like to receive. To make things worse, these clocks are usually
sub-divided into a number of other frequencies within the computer. Since
the computer is a digital system, the characteristic waveform of these
signals are square-wave and square-waves tends to result in a lot of
harmonics. The video card also generates RFI because the data-pixel-rate is
often high enough to fall into the HF regions. All these reasons are why
computers and monitors are tops when it comes to generating RFI. This is
often made worse by computers with cheap plastic casings which do not shield
the system. Many people have the opinion that monitors are one of the main
RFI sources and this may well be the case. I have also noticed that the PC
keyboard also generates a considerable amount of RFI despite its innocent
look. This is because it contains a microprocessor on board which runs off a
clock in the 3 mHz range. This problem is compounded by the way the keyboard
PCB runs which makes like a pretty good loop antenna type radiator and so
the harmonics can be heard in the 2nd and 3rd harmonic range. The monitor
probably has an on-board crystal at about 14.316 mHz so you will find a
strong carrier there too.
The first thing we want to do is to determine how much RFI is being
generated by the CPU and the monitor. It may be that the CPU is not
radiating at all but the monitor is the culprit, or vice versa. This can be
done by switching off the monitor and leaving the CPU on just to see how
much RFI is getting into the radio. The following is a suggested procedure:
- Do a quick sweep across the bands to find out where the RFI is the
strongest. This is helpful because if we can reduce the RFI here,
there should normally be a corresponding decrease of RFI everywhere
else but not necessarily so.
- Disconnect the mouse, serial cables and printer cables, keyboard,
video cable, video power. If you can, run the receiver on batteries
for this part. Now turn on the computer and see if the RFI is
increased by any appreciable amount.
- Now, connect the keyboard (with the CPU on), then the mouse, then
the serial cable, the parallel cable, the video cable, monitor power
cable (don't turn on yet). As you reconnect these, note the increase
of RFI if any. You should now have some idea as to which is the main
contributor of RFI. If you should have the good fortune that none of
these result in any appreciable increase of RFI then you're in luck.
- Turn on the monitor and note the increase in RFI. Run both text and
graphics modes to see if the RFI is affected.
- Reconnect the radio to the power supply and again note the RFI
increase if any.
- Disconnect the radio from any antenna, both external and internal,
and note if the RFI goes away. Note that on some radios, when you
unplug the external antenna connection, the internal antenna is
automatically activated. To prevent this, plug in a dummy plug to
the antenna socket.
By now you should have a pretty good idea of which components are
contributing to RFI. Keep these notes, while we move on to another point,
RFI paths.
Earl Morse KZ8E suggest this for pin-pointing RFI source:
-
- "
In the event that this does not work you will need to investigate for cabinet
radiation from the monitor and computer. Make a loop from some RG58 coax by
stripping back the shield a few inches and making a couple of turns about an
1-2 inches in diameter. Solder the center lead to where you stripped back
the shield. Plug this loop into your receiver and use it to probe the
computer setup. A faulty computer cabinet can be easily closed up with fingerstock,
braid, or other conductive materials. Shielding a monitor cabinet is a lot
more difficult and could cause safety problems (high voltage) or overheating.
"
If you find the above method too troublesome, just get hold of a
battery operated portable AM/SW receiver with an internal ferrite rod
antenna and start moving around, rotating the radio as you go. Often you can
pinpoint the major RFI sources just by following the radio to the place
where the strongest hum is found. Of course, the radio should be tuned to a
section of the spectrum where the RFI can already be heard somewhat. Do this
with the radio near the suspected equipment and fine-adjust from there. This
will also give you an idea of what kind of RFI and on what frequency it is
focused on.
RFI gets into the receiver from the source through a number of
paths. This can be through the power supply, through the "earth" point of
the power supply, through direct radio emission, even through "shielding"
which are not properly designed or used. There are a number of ways to deal
with these situations. If in (6) above, your radio is still picking up RFI
with no antenna connected then, the shielding in your receiver is poor. If
in (5) above, the RFI increases as you reconnect the radio to the mains
power supply or adaptor, then RFI is coming in through the power supply. If
your radio is connected directly or indirectly to the computer through the
serial port or some other interface, that too could be a path for RFI. You
will know if this is a problem as you go through step (3) above. If the RFI
increases appreciably as you reconnect the external antenna, then the RFI
may be coming from direct emission or is being picked up from the antenna
feed.
Remember that in many cases, RFI is coming in from more than one
path so it is important to check out all possibilities. On the other hand if
one path is the overwhelming problem, you may want to deal only with that.
Other basic measures are to keep cables and connections as short as possible.
This will prevent these connections from becoming radiating elements for the
RFI from the equipment. If possible, have your radio equipment connect to a
different power circuit but watch the earth so that no dangerous voltage-
potentials are formed.
As you can probably see, RFI is a combination of problems and not
one problem. It can be classified into 3 categories:
- Shielding problems
- Filtering problems
- Design problems
These are not definitive categories but are used for simplicity. In many
cases, poor shielding is a prime suspect. So we will deal with this first.
Now to get on with the real action.
When we talk about shielding, there are 4 things to consider, the
radio itself, the monitor, the CPU and the antenna/feed combination. Poor
shielding can be an inherent problem to some designs but can also be due to
dirty connectors and old parts. Go through the equipment to make sure that
all the connection points are secure, this is especially in connection to
the antenna shielding, radio power supply, CPU casing, cabling. Make sure
that the wires are not old and all the connections are clean, no oxides on
surfaces. The presence of oxides makes for poor conduction and in some cases
it results in rectification of signals which can then lead to a host of
other problems.
If the problem is a design one, such as poor shielding in a radio as
determined by step (6) in section 2.0 on RFI sources, or poor monitor or CPU
shielding, then we need to provide an adequate shield. This can normally be
done using tin/aluminum foils or conductive spray. The basic idea is that
we must line the casing of the equipment to be shielded with a barrier to
RFI. This must be done carefully because since the shielding is conductive,
it is possible to accidentally short circuit something and fry your radio or
whatever it is you are trying to shield. The spray is probably the easiest
to use but also quite expensive. My feeling is that metal foil probably
provides better shielding but is harder to apply. The conductive spray or
paint has a lower conductivity but spreads more evenly and gets into
difficult to reach locations more readily. What you need to do is to
carefully remove the plastic casing of the radio or monitor, paying
attention to cabling and eletrical contacts. Remember that opening the case
will almost certainly invalidate the warranty on the equipment. Clean the
insides of the case thoroughly and make sure that it is dry and free of dust
or grease before applying the spray. You may need to apply several coatings
to get better shielding. Make sure that at some point in the casing, the
applied shielding comes into contact with ground. Remember that in monitors,
chassis is not always connected to ground so check this out first. For the
shielding to be effective, you need to provide as complete a "wrap" as
possible but remember not to spray onto switches or anything that might
cause a short circuit. Also prevent blocking up ventilation holes. You may
at this point wonder about the screen itself. Well in most color monitors,
the mask inside the monitor acts as a shield of sorts. Wait for the coat of
paint or spray to dry before replacing the cover. Try to ensure good
grounding for the shield and avoid scratching off the paint. If arcing
should occur, use some insulating tape over the area. The kind of spray that
I use is known as EMI-LAC or EMV-LACK (by Cramolin) and is of German origin.
There are many other makes and most should work okay. In the event that you
cannot find either paint, spray or suitable foil, conductive tape will do
also. Remember, the conductive screening must NOT touch
the components. Pay
attention as there is always shock hazard when messing with monitors. Don't
do this yourself unless you know exactly what you are doing. :) Stay alive.
Gary Coffman KE4ZV has this advice about shielding:
-
- "
If the monitor is the culprit, there are conductive sprays you can
use on the inside of the plastic cabinet to reduce the hash. GC
Electronics makes one that works well. You have to strip the monitor
and spray the cabinet. Usually you want to spray inside for appearance
sake. When you put it back together, watch out that the HV section has
adequate clearance with the now conductive case. If it doesn't, glue
some fish paper in the proper spots to prevent arc overs. Unless the
monitor is a "hot chassis" design, bond the conductive coating to the
chassis, and bring a bond wire out from the chassis to station ground.
If the monitor is color, the shadow mask in the tube will form an
adequate shield, but if it's mono, then you may need to put fine
copper screening over the face of the tube. Spray paint it flat
black and it'll double as an anti-reflection screen.
If the cable is radiating, first make sure you're using a shielded
cable, then use some snap on ferrite chokes on the cable. These
chokes are good things to put on *every* external cable.
If the PC is the culprit, scrape paint so that the case halves can
bond properly, and add extra screws so that every seam has a screw
at least every two inches of it's length. That's what it takes to
get a good Faraday cage. Any openings in the case should be covered
with copper screen wire. The floppy drive opening is a problem because
you need access. The best way to handle this is to shield the entire
drive bay from the inside and accept the hash when the drive is in
use. Bring a bond wire from the case to station ground.
Of course it always helps to have the radio's antenna as far from
the PC as possible, and brought back to the radio via a well shielded
coax. The radio chassis should be bonded to the station ground. If
you're using a HT (so sorry) then you may have to locate the radio
and TNC far from the PC via a long RS232 cable. Note that the TNC can
also be a source of noise. In some cases it's better to leave the
TNC next to the computer and remote locate the radio with long audio
and control cabling. The TNC should be treated the same way as the PC.
Sand off the paint and add bonding screws. If it's plastic cased, either
use the conductive spray, or put it in a metal box.
"
If you are using the spray, make sure to apply an even layer and
you may want to repeat the process a few times to get a better screen. Do
not waste excessive spray on one spot but make sure that the coat does not
have "thin" spots. The same method can be applied to the CPU casing or to
the keyboard casing. The effectiveness of such a method on the keyboard is
debatable however, seeing that it is difficult to build a proper Faraday's
cage around the offending circuitry. Care must also be taken seeing the
tight enclosure of most keyboards.
Shielding of cables are also important. Try to make sure that the
video cable has a good shielding/screening. The same applies to serial and
parallel cables and, of course, the antenna feed should be well shielded.
For antenna feeds, avoid TV 75 ohm coaxes as these normally do not provide
sufficient screening. Try RG-58-C/U (which I think has better screening that
RG-58-A/U) or some other high-screening type cables like RG-8 and the like.
Well screened antenna feeds go a long way to remove RFI. I have been told
that the equipment should all be connected to a firm ground via heavy gauge
wire or braid. This is probably true and if you can, why not. Where cables
terminate, use good and appropriate connectors such as PL-259 or something
similar. In all cases, provide the screening with a good ground. A well
screened antenna feed may eliminate up to 80% of the RFI in some case, or
more. If you are listening on VHF and UHF however, the length of the feed
may need to be weighed against signal losses in the feed itself. Feeds such
as the 9913 have lower losses but are quite expensive. Use them if you can
however. Personally I still find the RG-58 to be the easiest to work with.
Sometimes, in some CPU units, there may be some parts which are
difficult to shield, such as the disk-drive. This is not too serious since
the drive is not operative most of the time, but if you elect to try to
screen that also, remember that the screen must all be well connected
electrically. The same goes for the rest of the computer casing. If you can,
use one with a proper metal casing on all sides. What we want is a Faraday
Cage so see that the whole case is well grounded. Scrape off some paint at
the screws so that there can be good electrical contact everywhere.
Other exposed connectors and splitters (which are not recommendable)
can also be letting in RFI so you may want to check out their screening as
well.
This represents the next major path for RFI. In some poor designs,
RFI is not properly filtered out of the computer or monitor power supply and
it thus leaks out into the mains, and from there into the radio power supply
and finally into the RF section (or AF sometimes) of the receiver. This type
of problems can normally be improved by using a line filter for the
equipment's power supply. Make sure that the line filter is rated for the
power the cable is meant to carry or it may burn up. There is a kind of line
filter which is essentially built as a socket which you need to replace the
one on the power supply with. I am not sure how much improvement this kind of
filter affords but if you elect to do this, do it with care! Other than using
line filters, you can also use ferrite beads and toroids on most any lines.
Experiment around with a bunch of these things. You could also put a bunch
of ferrite beads on the antenna feed line and this will act as a kind of
balun as well as preventing RFI from traveling on the screening. There are
many types of ferrite RF chokes, clip on types and ready made types, which
can be used. The ARRL Handbook has a description of some of these devices.
I have built myself a 4:1 balun (BAL-anced to UN-balanced) for my coax fed
dipole and now I get very much less computer RFI and other noise from my
antenna system due to the impedance matching and improved power transfer
characteristics. Better selectivity is also a benefit of a good antenna
system. Line filters and RFCs can do wonders so experiment with them at
various strategic locations such as power supplies, some audio lines etc.
Other than that, RFI often comes through other data lines and
control lines in parallel and serial ports as well as video ports. Besides
using ferrite beads on these, you may want to connect small value capacitors
between the lines to ground. For parallel and serial lines, use 0.01 uF
capacitors (multilayer, MKT if possible, others may work well too), and for
video ports, connect 100 pF capacitors from the RGB, H-Sync and V-Sync to
ground. These may have some other effects on the lines so experiment with
slightly higher or lower capacitor values.
If you are receiving on certain bands only, a bandpass, high-pass or
low-pass filter may help if used on the receiver front-end. This, however, will
not be effective for reducing in-band noise. A notch filter can be used on the
antenna feed to notch out certain strong emissions such as the 14.316 mHz
crystal oscillator on the video card or on the baud-rate generator. Note that
all kinds of filters (barring active ones) incur losses to some degree and if
your signal lands too near the unwanted emission then you cannot use the Notch
since the wanted signal may also get filtered out. A high-Q notch is often
preferred in this regards.
Concerning the use of clip-on filters (such as those sold by RS),
Steve Byan has this to say:
- "
The "clip-on choke" is a split ferrite bead in a plastic case, sold for RFI
suppression from personal computers. The two halves of the bead fit
around the cable; the plastic case has a hinge and a latch to hold the
two halves together. You could get the same effect by slipping a ferrite
bead onto the coax before attaching the connector.
This idea is similar to the ferrite bead baluns used in amateur radio.
The idea is to increase the impedance to RF current flow on the outside of
the coax shield - this ideally prevents the coax from becoming part of
the antenna.
I'm skeptical that one or a only a few of these Radio Shack beads
would have much effect when placed on the coax lead-in. I think
it would take quite a few beads to get enough inductance to help.
It might be better to use the beads on the offending appliances.
I'd recommend John Doty's method of grounding the coax shield at
the antenna, and running the coax buried or along the surface of the
ground. I think this will minimize pickup of local noise sources more
effectively than a few ferrite beads at the receiver.
"
Power line decoupling and filtering is essential for optimum
performance. The standard power supply of most PC's is a switching
power-supply. Such systems, while power efficient, tend to generate spikes
and unwanted harmonics due to the switching effect. Needless to say, if not
handled properly, these spikes will get into the mains and from there into
your radio. This is usually not severe because the design usually takes this
into consideration already. What is more worrying is the CPU clock leaking
into the mains through the power supply. Once again, the line filter is the
way out.
Often, when only the audio from the receiver is required to go into
the computer, you might want to consider some way of isolating the signal
from direction connection to the computer. One simple way is by the use of
1:1 audio transformers. Personally I have found a slight improvement here
but not to my satisfaction. I also tried connecting the audio from the
receiver to a adapted FM microphone and having another FM radio pick up the
audio which goes into the computer. This seems a lot of hassle but may be a
last resort or for people who really want distance between the radio and the
computer. Other forms of isolation can be implemented using IR or optics.
These will provide excellent isolation as far as the audio line is concerned
and no RFI worries from that path.
RFI problems are sometimes compounded by poor antenna or receiver
design. Antenna impedance mismatch, for example, can make things worse that
it needs to be. As such, use of baluns and transmatches may help. Although
the coax antenna feed is supposed to keep out RFI, a mismatched antenna and
feed may result in the coax screening itself picking up RFI. So if you are
using coax feed into a balanced dipole, try using a balun at the feed point.
Balun designs are really simple to build and can be found in the ARRL
handbook also. The impedance matching of the antenna to the feed can also be
handled by the balun. Baluns, being what they are, normally incur some
signal loss but in a good balun, this is insignificant compared to losses in
the feed itself. Besides, loss of signal strength is often made up for by
improved S/N ratio. I have not confirmed this but some out of band RFI can
be reduced by the use of antenna tuners which provide better selectivity.
Such tuners can easily be made from any number of designs found in amateur
circles because commercial ones tend to be rather expensive and come with
S.W.R. meters and other fancy things that the RX-only SWL does not need. In
my case, a coax feeding into an off-center-fed-dipole, the balun did
wonders to the S/N ratio and I actually say a signal strength improvement.
This does not mean that the balun improves gain but the selectivity it
provides may prevent the receiver AGC from kicking in and drowning out the
weaker station thus resulting in a higher apparent signal strength.
Kok Chen AA6TY has this to say about unmatched antenna systems:
- "
And, finally, finally, just as connecting a coax to a balanced antenna
will cause feedline to radiate, the reciprocal also happens. If you
feed a balanced receiving antenna with an unbalanced line like coax,
the outside of your coax will pick up noise and send it to the
antenna input of the receiver. Place a balun between the coax and
the antenna itself.
"
Excessive gain, such as may be the case with antenna pre-amplifiers
will not always improve the S/N because then the noise gets amplified along
with the signal. In this respect, normally a masthead pre-amp will perform
better.
Decoupling in some digital equipment is not properly done and can
lead to RFI being insufficiently suppressed. This can be remedied by placing
0.1 uF capacitors along all points in the power supply of the equipment
between supply and ground. This may sound silly to some because it may seem
easier to use one big capacitor, but while the electrical property at DC may
be similar, the distributed capacitance has a different effect on RF. You
want to ensure that your power supplies are properly decoupled and no
unwanted oscillations are taking place in the regulators.
The directivity properties of the antenna can also help improve
signal conditions under RFI. The loop antenna, for example, can be
positioned so that it nulls out at the RFI source. I haven't tried this
myself but in theory it should work but I cannot comment on the actual
performance you can get out of this. Similarly the dipole also exhibits
certain directive properties which can be taken advantage of to reduce RFI.
The use of antenna tuners and pre-selectors may help reduce out of
band noise and may provide a clearer signal especially if a high-Q tuner is
used in the front-end. If an active antenna is used, it is probably more
advisable to have that antenna mounted at the mast/feed-point. This is so that
only the signal gets amplified and not the RFI from the computer. This is
especially so when we are talking about broadband RF pre-amps which are more
susceptible to noise. I now have an MFJ-1020-A active antenna pre-amp which
also contains pre-selector section and it works well for me, removing a
considerable amount of out-of-band signals but at the same time, the noise
floor is also amplified. Judicious use of such a system can help improve
reception on many signals, especially of the continuous carrier types like
RTTY and FAX or SSTV.
If you are constructing your own interface equipment, try as best as
possible to build it into some kind of shielded casing, or a metal box. This
may help reduce some RFI input to the radio.
Besides the above, if one is using the computer for receiving
CW/RTTY or other digital signals, a good IF or AF filter will help in
removing unwanted noise. The advantages of the different filter types are
detailed in the ARRL Handbook. For those who are more well off, a DSP
adaptive filter can be of added convenience, aside from all the other SC
(Switched Capacitor) filters, Notch filters. Again, I do not have the
privilege to speak from experience. Bandpass filters are quite easy to
construct and you may want to experiment with certain designs before
investing in the real thing.
For those who are thinking of a new computer, I hear that notebooks
have very low RFI emission and may be very suitable for radio use.
In case you are like me and don't live near the ground floor, you
may want to use a balanced antenna system which does not require a ground or
you may want to get one of MFJ's Artificial Ground. Here is an excerpt from
Scott N3FI to me about this:
- "
I picked up this thing called the MFJ Artificial Ground, which consists of
(I think parallel) an LC circuit and a counterpoise wire. This allows you to
"tune" the reactance out of the counterpoise wire so that it "seems" like
an earth ground to the shack. No more zaps on your microphone!
Don't know if necessary on RCV only, though. Wouldn't imagine so. Just
remember that electrical outlet ground is ground at 60 Hz, but NOT in the
MHz. Your whole wiring system will act as an antenna!
"
As far as the wiring of the feed goes, remember to keep the feed
from running parallel with mains wiring and try to avoid fluorescent lamps
if possible. Some of these old lamps have faulty chokes which tend to
generate some noise. Most modern buildings should be okay but older places
tend to have this problem. While on the subject of wiring, remember that it
may not always be your computer/monitor which is generating the RFI.
Especially now that more and more people are staying in apartments or
terrace houses, your neighbor's computer system may also be a source of
noise. One way to go about this is to have a nice talk with your neighbor
or use some kind of grounded metal sheet or grid at the walls. In concrete
buildings, there are steel bars in the walls and so this may not be
absolutely necessary.
RFI has been a most perplexing problem and will continue to be so as
long as manufacturers of computers do not think if it as a problem. I
welcome contributions and other suggestions so that we can advance in our
fight against RFI. Please direct all suggestions to:
asirene@ntuvax.ntu.ac.sg
I would also like to thank persons who have offered their advice and
experience in RFI problems. Tks.
Kok Chen AA6TY kchen@apple.com
Earl Morse KZ8E e.morse@zds.com
R. R. Scott NF3I ham@wam.umd.edu
Gary Coffman KE4ZV gary@ke4zv.atl.ga.us
Tom Bruhns K7ITM tomb@lsid.hp.com
Andrew Moore dmoore@viper.cis.upenn.edu
Steve Byan steve@hicomb.hi.com
others whom I may have lost their names ...
The ARRL Handbook offers many helpful information concerning baluns,
transmatches, filters and chokes. These will come in helpful for those who
want to find out more about antenna and transmission line theory. Information
can also be obtained from the ARRL e-mail server. Simply send a mail
containing nothing but "HELP" to info@arrl.org to get started.
The server contains some more information on RFI and related problems.
Have fun, down with QRM and RFI. A final word of caution, be sure to
know what you are doing or you may fry your equipment in the process, or fry
yourself (which will arguable solve all your RFI problems! :). In case of
doubt, get an ELMER or someone experienced with electronics to help you out,
or to walk you through the steps.
73 de Daniel 9V1ZV
+-------------+-------------------------------------+
| Daniel Wee | daniel%pandora@csah.com |
| 9V1ZV | daniel.wee@f516.n600.z6.fidonet.org |
| UUCP1.12j | Packet: 9V1ZV @ 9V1VS.SGP.AS -- |
+-------------+-------------------------------------+
A public-opinion poll is no substitute for thought
-- Warren Buffet --