|
Adi Post subject: balanced current supply: noise considerations
Posted: Mon Aug 21, 2006 11:54 am Lieutenant Joined:
Mon Aug 21, 2006 11:52 am Posts: 4 Hi! I am working on
a project in which I need to characterize the hysteresis properties
of superconducting wires. The measurement is done by injecting a current,
with risetime of the order of ns, into the wire and by monitoring the
voltage across the wire. The latter shows a jump when a transition occurs.
Given the fact that the timescales involved are ns and that the
distance to the device will be relatively large (~ 1-3 m), I would need
to work with transmission lines. However, using a coaxial cable would
be a possible problem, due to ground loops or ground potential fluctuations
(the shield of the coax needs to be grounded at the source and also
at positions closer to the sample). One solution would be to use a balanced
configuration, with an arbitrary waveform generator that delivers from
50 ohm single ended outputs two voltages opposite in polarity along
two 50 ohm coaxes. Close to the device, each line is terminated in a
50 ohm resistor. The voltage between the signal points of each resistor
is used as a source for the device (current bias through a large resistor).
This configuration should eliminate ground related problems both at
low and high frequency. I would like to mention that this method
is described in an old book on noise isolation techniques: Henry Ott,
"Noise reduction techniques in electronic systems". However, I am not
sure whether it should work unmodified at high frequencies. I
would greatly appreciate any suggestions and comments on this problem.
Adi Top nubbage Post subject: balanced current
supply: noise considerationsPosted: Tue Aug 22, 2006 3:26 am
General Joined: Fri Feb 17, 2006 12:07 pm Posts: 218
Location: London UK Hi Adi I do not know for sure whether such
an approach would reduce noise, but it sounds reasonable. However,
to derive a wideband balanced line from an unbalanced 50 ohm input,
this is best achieved with a balun (balanced to unbalanced transformer).
There are many varieties of baluns. One simple type takes a copper
50 ohm coax air-line, and at the end remote from the 50 ohm source,
cut 2 parallel slots a quarter wavelength long axially in the outer
conductor, 180 degrees apart. For a 1 nS pulse, this would be about
15 cm long. Short the inner to the outer on one side of the two ears
formed by the slots. Take the balanced output from the ends of the two
ears. If you use a search engine for "balun" you will find a
large number of designs. Top Adi Post subject:
Posted: Tue Aug 22, 2006 7:31 am Lieutenant Joined: Mon
Aug 21, 2006 11:52 am Posts: 4 Dear Nubbage, Thank you
for your reply. I looked briefly into "baluns" and I intend to do it
more later. There was one detail which I did not understand: are
these transitions frequency independent? Because my signal contains
pulses with risetimes of a few ns, so it should have flat transmission
from DC to ~1 GHz. Adrian Top nubbage
Post subject: balanced current supply: noise considerationsPosted: Tue
Aug 22, 2006 10:58 am General Joined: Fri Feb 17,
2006 12:07 pm Posts: 218 Location: London UK That point concerned
me as soon as I had posted my suggestion. The Fourier spectrum of
the pulse with a rise time of 1 nS for sure extends to say 2GHz.
The majority of baluns are intended for RF applications over say a 20%
bandwidth of the centre frequency of the carrier, so many will be unsuitable
for your application. The widest bandwidth from a few tens of
kHz to 2GHz would be achievable with a toroid transformer with a single
primary winding one side of which is to ground and the other to the
source. The secondary comprises a winding that provides the balanced
output. Several ferrites are suitable for UHF and low microwave frequencies.
Mini-circuits produce a range of these balun transformers at low
cost, but they have power limitations. Alternatively you could wind
your own. The design basis is to ensure there is enough inductance in
the windings at the lowest significant frequency in the spectrum.
Another approach might be a Class A low gain, high gain/bandwidth
product transistor amplifier, dc coupled, where the pulse source drives
the base, and the balanced output is taken from the collector and emitter.
If the balanced feeder is open circuit that might work, but if short
circuited you would again face having no transmission at dc, otherwise
you would be short circuiting the collector and emitter. There
are also video amplifiers now available that are configured for driving
balanced cables, I think from national Semiconductor Inc. They accept
an unbalanced coax input drive. Top Adi Post
subject: Posted: Tue Aug 22, 2006 11:05 am Lieutenant
Joined: Mon Aug 21, 2006 11:52 am Posts: 4 Thanks a lot,
Nubbage! I will look into the solutions you propose and let you
know. Adrian Top fred47 Post subject:
BalunsPosted: Mon Aug 28, 2006 11:09 pm General Joined:
Wed Feb 22, 2006 3:51 pm Posts: 104 Hi! In my experience,
if you need very broad band performance, only transmission-line-based
baluns will do the job. Transformer-based solutions can't reach the
extreme bandwidths that transmission line ones do. Of course, the low
frequency end of your requirement is set by the pulse repetition rate
- which might be very low - you didn't mention what it is. The
classic book on the subject is Transmission-Line Transformers, by Jerry
Sevick, now reprinted by Noble Press. Good Luck! Fred
Top Adi Post subject: Posted: Tue Sep 12, 2006 1:22
pm Lieutenant Joined: Mon Aug 21, 2006 11:52 am Posts:
4 Hi, I checked, following the suggestion of Nubbage, for
solutions using active components. I did not manage to find them at
National Semiconductors, but Linear Technology has some models that
do the job. Thanks for the tip Nubbage. Fred47, thank you for
your reply. Using your method is indeed an option. The repetition rate
will be larger than 1 MHz. one problem which remains though is that
I do need to control also the offset level. This could probably be done
by adding a DC signal, also from a differential source, ina way which
does nor disturb the high frequency properties of the balun (with inductors
and/or resistors), so you get some sort of balanced biasT. The inconvenience
is then the more complicated RF design and the addition of a DC signal
source. Adi Posted
11/12/2012
|