Hello everybody!
An interesting thread, once more!
I will contribute two examples of this problem, and how it has affected me.
My main HF transceiver for the last 22 years has been a Kenwood TS-450.
In the 1990's I integrated it into a pretty sophisticated station that
ran a packet radio BBS, combined with a fully automatic digital
satellite station, and was also usable on phone and other modes without
having to plug/unplug anything.
To get the HF transmission to work without unwanted feedback, I had to
apply three tricks:
- Bond the transceivers, power supply, computer, TNCs, etc, using short
pieces of copper pipe and car battery wire, to get them to a
close-enough potential. Thinner wire wouldn't work.
- Place a 470k resistor in series with the rear-panel audio input of the
TS-450, to form a voltage divider with the radio's 10k input impedance,
and modify the TNC to put out about one volt of audio instead of the
usual few millivolts, to override the audio signals induced between the
radio's box and and the TNC.
- Use a very well balanced feed line and antenna, to keep RF levels in
the shack low enough to avoid RF feedback.
This worked, although it was just at the limit when I used high TX power.
Five years ago I moved to a new QTH. I decided to use coax cable here
instead of balanced line. Of course I use a good balun at the antenna,
and an additional common mode choke on the feedline in the shack, but I
found that I was getting severe feedback when using the amp, bad enough
to totally garble my transmission whenever the cable into the TS-450's
rear panel connector was connected.
I reasoned that it was a "ground loop" (let's use that traditional
term!) despite all, and replaced my direct audio/PTT cable to the radio
by a totally insulated system: Two audio transformers for RX and TX
audio, and an optocoupler for PTT control. Surprise! The feedback was
still there, just like before!
Picture this: A well shielded, short (30cm) cable coming out of the
TS-450's rear panel connector, ending in a shielded box with those
transformers and optocoupler inside, and that box not connected to
anything else, the box connected to the cable shield, and the radio
picking up severe feedback while transmitting with the amp. Instead if I
unplugged that short cable and box from the radio, the feedback stopped.
Time to take the covers off the TS-450. Of course, inside I found the
accessory connector soldered to the IF/AF board, and the cable shield
goes into that board's ground! The board of course is mounted by several
screws to the metal structure, so there is excellent continuity to the
box - but at RF that's not good enough.
I modified my cable, lifting the shield from the DIN connector shell and
bringing it out of the connector. I then connected it to the radio's
box, at the closest screw. Problem solved. No more RF feedback during TX!
What's funny is that I could even return to transformerless, direct
coupling of the audio, without any harm. I prefer transformerless
coupling, because it has flatter frequency and phase response, and
avoids the transformers picking up magnetic fields.
I didn't know that this problem has been called "pin 1 problem", but the
problem is very real and widespread.
Now I will give you another example, a little less obvious than this one:
I have an audio setup for music, that consists of a commercial CD
player, equalizer, sat receiver and cassette deck (that one gets little
use nowadays), together with a homebuild AM/FM tuner, a simple volume
control (just a potentiometer), and from there two line-level audio
cables run to the homemade active speakers, which have active cross
overs and three amplifiers built into each speaker box.
For many years I had used this system just as described, but there was
always a tiny amount of line frequency hum. Not really much, not
bothersome, but it was there. So, the day came when I decided to attack it.
Firstly I checked the shield setup. It was all but optimal: While
everything except the speaker electronics was shielded, all these
equipment shields were interconnected only via the audio cable shields.
And nothing whatsoever was grounded. The whole audio system, every part
of it, was powered by two-contact AC plugs, so that the whole system was
floating off-ground.
I decided to attack the long cables first, the ones running to the
speakers. Since they come directly from 10k volume control pots, the
impedance there is high enough to easily pick up something. But worse
than that, any hum current coupling into the speaker electronics via
power transformer capacitance would propagate over the shields of those
long audio cables, which are thin, cheap ones, tending to produce
noticeable voltage drops even with small currents.
So I did two things: Turning the inputs of my speaker electronics from
single-ended into differential, to use the shielded wires in a
pseudo-differential system; and bond the speaker electronics "grounds"
(actually not grounded) to the shields of the rest of the equipment.
To do that, I installed three-wire power cables to my speakers, and
added a wire from an internal power plug earth contact to the box of the
equalizer, which is the last box in the chain before the volume control
and the cables to the speakers. The whole affair was still connected to
the wall plug by a two wire connector, without ground, so it was still
floating off-ground, but at least the long wires didn't have to serve as
"ground" bonding between the distant speakers and the rest of the system.
Once that was done, the hum was gone! But there was a new problem: A
quite intense, piercing whine. It came from the equalizer. Switching the
equalizer off, which puts it in bypass mode, the whine was gone. With
the equlizer on, by not active (EQ switch out), the whine was lower, but
still present.
Out came the equalizer, onto the surgery table. I found that it has its
20 potentiometers wired via long internal flat cables to the circuit
board. It also has a fluorescent screen that shows a spectral analysis
of the audio. And this fluorescent screen scans at 1.4kHz, and the whine
in the output was at 1.4kHz and its odd harmonics...
There was a little piece of flexible metal sheet inserted between the
audio wires and those wires connecting the display to the digital
section on the circuit board. And this little piece of metal shield had
a wire on it, connecting it to a screw on the metal box.
Disconnecting this wire, the whine got much stronger. Grounding it to
different places on the box produced totally random results: At some
places the whine was stronger, at some less strong. But it was always
quite weak, often only detectable by measurement - but as soon as I
grounded the metal box to earth ground, the whine got 50dB stronger!
I must confess that I wasn't able to find out in what paths the
different currents flow on the metal box of that equalizer. It's just
clear that the digital circuitry of its spectrum analyzer produces
significant currents in the shields, which couple to the high impedance
audio wiring (the equalizer pots are 100k each), and that the amount of
digital whine that couples into the audio circuit depends on what
connects exactly where to the box.
It's a great example showing that shields, even those made from
respectable metal sheet, don't have zero resistance, let alone zero
inductance!
My solution: I removed my ground bonding wire from the equalizer's box,
and instead put it to the box of my homemade AM/FM tuner. That bonds the
audio wiring shields just the same as when connecting to the equalizer
box, and my tuner isn't picky about being bonded that way, because it
has no digital logic inside, and also its box is made from thick
aluminium sheet instead of the thinner steel sheet of the equalizer.
Now I get neither hum nor whine, just music.
By the way, the equalizer doesn't have a "pin one problem". The RCA
audio input and output jacks are all directly grounded to its box! The
problem with the equalizer is that other parts of the circuit, like the
power supply ground return, and that little shield between the digital
and audio wires, are grounded to DIFFERENT places of the box. I guess
that the engineers at Marantz found out by trial and error where on the
box they should connect this wire! It works only as long as no other
grounding is added...
I remember my days as a kid, trying to build direct conversion
transceivers for 40 meters, that had ALL of the receiver gain at audio
frequency. That was when I learned that the "ground" on a printed
circuit board is definitely NOT a perfect conductor! Even using a
continuous ground plane is not good enough, when having 80-100dB of gain
at audio. Nor is it when wanting 100dB and more of dynamic range in HiFi
audio.
Manfred
========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|