Highly Symmetric Oscillator - RF Cafe Forums
Post subject: Highly symmetric oscillator Posted: Wed
Oct 05, 2005 8:43 am
Joined: Wed Oct 05,
2005 8:23 am
I'm new here and have probably
a pretty easy question for you. I'm a physicist which means that
I hardly know anything about RF technology.
My problem is
that I observe a phenomenon in my experiment (a DC Voltage) that
depends on the polarity of my RF-Signal. (no kidding) I'm using
a commercial RF-Generator with a 50Ohm connector. I tried decoupling
with capacitors, used an RF-transfomer, but the problem stays. Funny
enough, if I take another RF generator (both Rohde & Schwarz)
the DC Voltage changes polarity.
It's not just the grounding!!!
Is has to do something with reflections on the outer conductor or
whatever. I mentioned the transformer I used: if I ground one of
the connections on the secundary (sample) side it doesn't influence
my DC-Voltage. But if I change the polarity on either the primary
or the secondary side my DC-Voltage flips.
I learned something
about balun-transformers but was wandering how complicate it is
to build an absolutely symmetric RF oscillator that would provide
me an RF Signal. I would like to transfer the RF over the inner-conducors
of two 50Ohm coax-cable to my sample - that would be a symmetric
100Ohm transmission line (?)
All the diagrams I've seen for oscillators
sofar did not seem to be completely symmetric. Do you have any suggestions?
The frequency should be somewhere between 1-100MHz and the signal
Post subject: Balanced signalPosted: Wed Oct 05, 2005 3:42 pm
The term you're looking for is "balanced".
You can get
a balanced signal either by using a balun (which is just a shortened
form of balanced-to-unbalanced) transformer, or by using a balanced
Most people use baluns, and they're both widely
available and easy to make (see the book by Jerry Sevick, "Transmission
line transformers" for details on how to make your own.
oscillators are also a possible choice. Googling on "balanced oscillator
circuit" should get you a good reference or two. (I'm at work right
now and don't have access to my books).
not guaranteed a 100 Ohm differential characteristic impedance by
using two 50 Ohm lines. Agilent has application notes on their Network
Analyzers which discuss this aspect of things.
Post subject: Posted: Wed Oct 12,
2005 4:21 am
Balanced or unbalanced, I am not
sure we yet understand what is going on here. A signal from a Rohde &
Schwarz (lucky you) generator coming down a coaxial cable does not
normally have a DC component to start with, and if it did, a series
capacitor would certainly remove it.
The signal can aquire
a DC component in lots of ways depending on what it is connected
to. Deal first with the "reflections on the outer conductor". These
depend on the the value of the termination impedance, ie, what your
kit looks like to the arriving cable, the frequency in use, and
the length of the cable. If the kit looks like 50 Ohms, and the
cable is 50 ohms type, the reflections don't happen. If the mismatch
is there, then they do happen, and the reflected signal going back
meets the signal arriving, resulting in a standing wave set of RF
voltages along the cable as the signals either add together, or
cancel each other depending on where you are along the cable. The
key thing to know here is that a coaxial cable in this condition
also has a standing wave on the OUTSIDE of the braid, and it will
be radiating all over the place like an antenna, using you, the
cables and kit around, and anything it is connected to as the other
half of its dipole.
This condition can mess up all sorts
of nearby instrumentation. I have seen a passive multimeter read
right across its scale from dc generated internally on rectification
diodes. Hums and buzzes everywhere. traces climb on the scope. If
there is enough power involved, the RF burn one gets just from reaching
to touch a knob can be painful. In this mode, your own body is the
counterpoise for the radiated field. Hams call it "shackitus" The
1. Match up. Provided your kit looks like a much
higher impedance than 50 Ohms, (say > 1000 ohms), a resistor
across the arriving cable will do.
If not, them make up a matching
network to suit. You will get lots of advice here on how, once we
know what the downstream kit is.
2. Suppress any current
on the cable outer, using small ferrite loading tubes threaded on.
A few turns, (say 10 or so) wound into the cable approximately saucer-size,
secured with tape can halt a braid current. Passing the cable several
times through a ferrite toroid will also do it.
You can even
get clip-on ferrites, but be aware that the types that suppress
EMI by lossy absorption do not work in the same way as ferrites
that genuinely to provide inductance at 100MHz. This trick is called
a "choke balun". Look on the ARRL site for cure techniques to TV
interference. It comes with picture HowTos.
3. Whatever you
do, if the arriving signal is not handled in a way that preserves
its waveform symmetry, you are "making dc". This includes overload
clipping on downstream amplifiers, samplers, whatever. Perish the
thought, but is the waveform symmetrical as it leaves the R &
S gen while loaded? Check using a T-piece connector and a scope
4. The phenomenon you describe .. that is.. you get
a different dc component when you "change polarity", is normally
that the shape of the wave is not the same on the positive side
as the negative going side. But here rises the question of exactly
how you "change polarity". If you started with a coax, you cannot
just connect it the other way about! Here is where the balun transformer
comes in. "Unbalanced" transmission line is not the same thing as
5. Finally, be very sure your attempts
to measure do not introduce the DC artifact. I have done this myself.
I just have to assume you know about differential scope measurements,
channel gain balancing, etc.
Good luck with this.
Post subject: Posted: Wed Oct 12, 2005
Joined: Fri Sep 02, 2005 7:25
Location: Hampshire UK
Dunno how it happened,
but that previous "Guest" was me