Hello Brad --
A related discussion occurred in mid-March on the Contest reflector, in the
context of receiver performance. I have
included a few abstracts. Hopefully other experience engineers can shed more
light on the topic, or even try it out on some
equipment.
Perhaps we can discuss it further at the topband dinner at Dayton.
73,
-- Eric K3NA
=========================================
from K3NA:
For some time now, ARRL and others have included in their collection of
receiver performance test the blocking dynamic range and
two-tone 3rd-order IMD, using two signals with some spacing such as 5 or 20
kHz. Of course, two strong signals doesn't emulate much
of the real world.
I recall a variation of this test that was used by AT&T to evaluate
performance of multi-channel radio receivers used to carry
large quantities of telephone channels. Naturally, one did not want a strong
signal in one telephone channel to contaminate the
signals being carried in other channels on the route. The test was performed
as follows:
-- instead of two signals being applied to the receiver under test, a
broadband noise was applied. The noise was modified by
notching out the bandwidth for one channel; i.e., essentially no noise in the
notched channel.
-- measurements were made in the channel corresponding to the notch.
-- noise power to the receiver was increased until the point at which the
measured channel started to exhibit degradation (e.g.,
increase in the noise floor).
This seems to be a more general test that corresponds more closely to what
a contest receiver experiences on a crowded band;
i.e., LOTS of signals attacking the receiver across the band.
Could those who are knowledgeable about receiver evaluation methodology
comment as to whether such a test would be more likely to
accurately characterize the ability of a receiver to hold up against strong
signals outside of the operating passband than the
two-signal test method?
Thanks.
-- Eric K3NA
=========================================
from W6LX:
Hi, Eric,
The test you described is called the Noise Power Ratio (NPR) test, and it is
used more and more to simulate the multi-carrier
conditions that an amplifier is likely to see. It is primarily for transmit and
receive amplifiers working at high frequency, but I
don't know if it has ever been applied to the entire receiver, RF-to-AF.
You need a very good signal generator to create that kind of a signal. The
noise has to be wideband and perfectly flat, with sharp
corners. The notch is very difficult to create. You want to get a notch about
one channel bandwidth wide, which in a typical ham
receiver would be about 2.5 kHz for SSB or 500 Hz for CW. An analog filter can
be used, but it is far from ideal. Nowadays, an RF
generator with arbitrary digital modulation is used and the NPR modulating
signal usually has to be downloaded from a computer into
the generator. The signal is usually in I-Q; that is, there are two channels
which are combined to give you the necessary
amplitude-and-phase-versus-time waveforms that describe the spectrum required.
You can see a glimpse of what I'm talking about at:
http://eesof.tm.agilent.com/docs/adsdoc15/examples/tutorial/NoisePowerRatio.html
Regards,
Al W6LX
=========================================
from VE3PN:
This is the NPR test frequently used to test Cellular PA etc look at the
following. I've never seen it used on Rx just Tx as it
needs a FFT or similar process on the final signal.
www.agilent.com
http://cp.literature.agilent.com/litweb/pdf/5965-8533E.pdf
or do a search on Noise Power Ratio for others or the Rohde&Schwarz page
http://www.rohde-schwarz.com/www/dev_center.nsf/show?OpenAgent&form=/www/dev_center.nsf/AppNotesFiles&ANNo=1MA29
Search for Noise Power Ratio or anything else on the page
http://www.rohde-schwarz.com/Homepage
(note links often change on these sites so do a search there's a lot of
interesting items
Also try
http://www.ifrsys.com/products/generators/2026q.htm
MY preference is to do a TWO or THREE tone test with the OUTER tone on the
slope of the filters being keyed ON OFF to make the AGC
> work I've found that vastly different IIP3's can be measured with even two
> tone signals if one is Keyed on off (NOTE it must be
shaped other wise clicks or cause too much junk.)
I'll see if my labs at work Is free one weekend and try it (NPR) on a HF rig .
A nice tool is Signal Studio (theres a demo download) just search for "Signal
Studio" if the links a lost one
http://we.home.agilent.com/cgi-bin/bvpub/agilent/Product/cp_Product.jsp?NAV_ID=-536887775.536883205.00&LANGUAGE_CODE=eng&COUNTRY_COD
E=ZZ
Peter Barron
Ve3pn@igs.net
=========================================
from W2VJN:
The testing you have described is called "Noise Power Testing" and the figure
of merit is the NP Ratio or NPR. This type of testing
is many decades old and has been used in the communications industry for
testing multiple channel equipment, including radio
receivers. Basically, the band of interest is filled with noise while one
channel is notched out, or free of noise. The radio is
tuned to the clear channel and the noise power is then increased until the
noise floor of the radio is degraded. Some simple
arithmetic then calculates the figure of merit.
In the early 70s I was involved with the design and construction of low power,
channelized SSB receivers. One of the specifications
which my customer required was a certain NPR. I obtained the equipment and
performed the test many times as each receiver design
progressed. The NPR test is an excellent way to optimize such parameters as
the mixer and product detector drive levels and
waveforms. Also, these receivers used 3rd order PLLs for the LO injection and
the loop parameters could be optimized by watching
the NPR as break points were moved.
A couple of years ago I was able to purchase a noise generator designed for NPR
testing on eBay. They are made by the Marconi
Company and can be picked up for $200 or so. HP also makes equipment, but it's
rather expensive.
The band stop filters used for notching out the noise are much too wide for our
purposes, but it's a simple matter to build a
quartz narrow band stop filter with crystals.
I have done this and have tested a few radios. The big problem here is that
it's not easy to relate NPR to 3rd order IMD tests.
Maybe some mathematician on here could work that out. I'd be happy to give him
the references needed to do the math. There are
books and articles. etc.
One day when there is not so much to do, I'll continue this study of radio NPR
measurements.
George, W2VJN
=========================================
|