This
page relates our experiences with installing a
D-Star repeater at a very busy site, the
problems encountered, and their resolutions.
Getting ready to install the D-Star
stack:
The purpose of this was twofold:
This latter point is very important because,
as you might realize, trying to use D-Star gear - whether it be
as a user or repeater - is a disaster when trying to diagnose
link problems. Without any built-in tools (ones that could have
easily been built into the software/firmware!) diagnosing a
problem of any sort is a significant challenge, typically
requiring that the operator(s) switch to analog in an effort to
"hear" what might be going wrong with the path - problems that
could include interfering signal, multipath, or simply a weak
signal. The repeater gear is arguably more difficult to use when
diagnosing problems as, unlike the portable gear, one can't
simply switch to an analog mode to "hear" what's going on!
What we learned was that the site noise raised the effective
noise floor of the receiver by a mere 10-15dB. This type
of measurement may have been possible to ascertain with the
D-Star gear, but measurement of this parameter is comparatively
trivial when done using FM analog gear! This amount of
site noise might sound terrible, but in actual practice one can
still achieve very good coverage considering that this site is
at the top of a 5000-foot tower made of rock, so HT-power
coverage from 80 miles away is still quite practical - if you
are using a good mobile antenna.
Installing the D-Star stack - and a
problem:
After several months of operation, the time
and equipment became available and the D-Star stack was
gradually installed in September-November 2009.
The VHF repeater didn't work.
Fortunately, John, K7JL, who was doing the install had the
foresight to install RSSI and Discriminator test jacks on the
VHF, UHF and 23cm gear. ANYONE who plans to put
up a D-Star repeater on ANY band is well-advised
to make these modifications and complain to Icom about their
having been left off the gear in the first place!
These modifications are documented by N5UD at this link. The original link as stopped
working so this is a locally-archived copy.
In testing with an Iso-Tee - a simple device
that lightly couples ones signal generator to the receiver's
feedline - the service monitor being used (an IFR 1200
without the booster amp) could not output enough
signal to register ANY quieting on the discriminator output,
indicating an effective desense of at least 40dB. Thinking that
something was wrong, the cables were moved back to the Kenwood
TK-740 and everything measured out fine. When the D-Star
repeater was connected directly to the service monitor to check
sensitivity, it tested within specs.
Hmmm...
Well, we knew that the duplexer
itself wasn't at fault owing to the fact that the Kenwood
repeater seemed to work just fine, and even if the Icom was
being overloaded by its own transmitter, that wouldn't explain
the severity of the desense - especially since it was still
deaf with the transmitter was disabled!
Fortunately, due to work on another
amateur repeater project going on at the same time elsewhere
on the site we had a 2-meter bandpass cavity (a 4" DB
Products DB-4001) to try out, and that made all of the
difference, reducing the desense from some
immeasurably high value (at least with the service monitor's
available output and the Iso-Tee's coupling) to something on the
order of 20dB or so. Because this cavity was in poor shape
and had been pulled from service for rebuilding after having
been used for 25 years or so - suffering heat, cold, moisture
and direct hits from lightning - it was quickly replaced on the
next trip with a pair of 6" Sinclair bandpass cavities in
series, each set for 1dB coupling.
It took BOTH of the 6"
Sinclair cavities to reduce the desense of the Icom VHF
repeater to the same level that the Kenwood analog
experienced WITHOUT these two bandpass cavities - that is,
down to the site's noise floor.
It should be noted that the Discriminator output - while a
useful diagnostic tool - was only partly helpful in
diagnosing and quantifying the problem: An important clue
came about by monitoring the RSSI output - which was noted to be
nearly "pegged" when the repeater was on the duplexer alone, but
the discriminator output just showed normal "noise". Once
the extra bandpass cavities were added the RSSI dropped to the
1.3 volt area with no signal, not too terribly far above the "no
antenna" reading. (Note that RSSI values vary from
radio to radio and depend on the site noise. Having said
that, different radios should still be "sort of" similar in
their responses.)
Something that you may not know about your
duplexer:
One of the most important things to know is
that most duplexers - even those marked or marketed as
"Bandpass/Band-Reject" - are REALLY only Band-Reject
once one moves very far away from the design (notch and pass)
frequencies. The duplexer used on the 2-meter D-Star
repeater - a brand new, 4-cavity TX/RX brand "BP/BR" type
costing >$2000 - was no exception!
If were to assume that "Bandpass/Band-Reject" automatically
means that everything way off-frequency will be filtered out,
you would be WRONG, WRONG, WRONG!
While it is true that SOME
duplexers have honest-to-God bandpass responses - that is, a
coupling loop on one side of the resonator and another loop on
the other side - most do not!
The biggest clue to this was that putting a wattmeter on the RX coax (along with a dummy load where the receiver would have connected) showed 10's of milliwatts of RF coming down the pipe from the single-band Telewave VHF stick mounted at the 60 foot level, and this energy was the combined, intercepted power of the FM and TV broadcast transmitters on site. Admittedly, it is a bit much to ask for any receiver to deal with, maybe, 1/10th of a watt of garbage coming down the coax - even if none of it is anywhere near 2 meters - but it was interesting to note that the lowly Kenwood TK-740 seemed to have no problem (and was not really helped by the addition of the bandpass cavities!) while the Icom was totally demolished by the same RF configuration.
One of the reasons for this very poor
strong-signal performance is that portions of the Icom repeater
itself consist of modules from standard ICOM mobile D-Star
radios: As you are likely aware, many mobile radios are
designed more for absolute sensitivity rather than strong-signal
performance and, unfortunately, the Icom repeater's front end is
no exception to this rule.
Actual responses of a typical
duplexer:
For a graphic example of a typical response of
a "Band-Pass/Band-Reject" duplexer, refer to the spectrum
analyzer plots in Figure 3. The duplexer being
tested was a "6-can" Phelps-Dodge unit capable of over 95
dB of TX/RX isolation. While this is not
the same one as used at the repeater site, its response is very
typical of such duplexers from different
manufacturers - including the "4-can" TX/RX unit on-site.
In each of these pictures, the plots are as
follows:
"Close-in" response:
In the Top image of Figure 3
we see the response of various combinations in the range from
140 to 155 MHz.
If we look at the Yellow trace - the
duplexer - we can clearly see the "high-side" notch. Note
that because of the setting of the resolution bandwidth on the
analyzer, the true depth of the notch isn't visible (it was
really more than 95dB) but the relationship between the two
frequencies is very apparent, with the peak response (minimum
loss) being at the "low-side" transmit frequency: It is
from this combination of deep notch and minimum loss at these
two frequencies that the "Band-Pass/Band-Reject" designation
arises. It is our opinion, however, that this
"Band-Pass/Band-Reject" designation is misleading - as we
shall soon see!
Now, take a look at the Magenta
(purple) trace. This is the response of the single 4"
bandpass cavity tuned to the "transmit" frequency. As you
can see, the insertion loss is minimal at center frequency and
the loss increases as one moves away from that peak. Its
rejection at the "receive" frequency (where the notch is) is
only on the order of 10dB - not nearly enough to provide
adequate TX/RX isolation of a typical system.
Finally, look at the Cyan (blue)
trace. This is the response of the "3-can" duplexer leg
and the 4" bandpass cavity in series. Around the transmit
frequency there is only slightly more loss than with the 3-can
filter by itself - which is understandable. You can also
see that just within the span of this plot that the combination
of the two types of filters greatly increases the
off-frequency response: Note in particular how the Yellow
trace - the response of the 3-can duplexer alone - is increasing
with frequency while, with the combination of the two, it's
actually decreasing! Finally, although it is not apparent
from the plot, the notch depth is greater! You would
expect this, as the band-pass cavity alone has 10dB of rejection
at the notch frequency, but the insertion loss of the two sets
of filters cascaded actually measured out to being more than
15dB greater than that of the "3-can" duplexer alone owing to
the added "magic" of the interconnecting coax cables and their
impedance transformation properties: Proper selection of
these cables and their electrical length can allow the two
filters to reinforce each other!
One important thing to notice is that if you
were to locate your repeater on a site that has other
VHF users - say, those in the 150-174 MHz region - or even
another 2-meter repeater on-site, the 3-can duplexer alone will
probably not help you when it comes to keeping those
other transmitters out of your receiver! The addition of a
single bandpass cavity can reasonably provide at
least another 35-40dB of isolation from those "other"
transmitters. Since we already know that the Icom's front
end is quite fragile, this extra filtering is arguably more
important!
"Wide-band" response:
Now, look at the Bottom image of Figure
3. This is a span that covers from 30 MHz at the low end
to 1 GHz at the high end.
First, look at the Yellow trace - the
response of the "3-can" duplexer leg. Although not noted
on the plot as such, the tops of the peaks on the
yellow trace represent an insertion loss of only about 1-2dB
while the analyzer's vertical scale is set at 10dB per division.
One alarming fact is that within much of the
FM broadcast band (included in the first peak on the left edge)
falls within a portion of the response curve where the rejection
is less than 10dB! What this means is that
FM broadcast energy intercepted by your antenna will not be
much-hindered by the duplexer! If your repeater antenna is
on the same tower as an FM broadcast station - or even if it is
on a tower that is near an FM station - you could
be intercepting quite a bit of power which, as we know, will
absolutely demolish the Icom receiver!
You might also note that there are plenty of
responses in the area of 450 MHz where there are often
high-power paging systems as well as throughout much of the UHF
TV band. In the case of Farnsworth Peak, we were also
getting contribution from those transmitters as well!
Also note that that throughout the range shown
on the Yellow trace you'll see that for most
frequencies, the 3-can duplexer leg does not
offer much isolation! As you might expect, there are many
spurious notches scattered throughout the frequency range (again,
owing
to
the resolution bandwidth of the analyzer, the true depth of
the notches is not apparent) the 3-can assembly is, for
the most part "wide open!"
Now look at the Magenta trace - the
single 4" bandpass cavity. As you can see there is a
strong, narrow peak at the 2 meter frequency for which it tuned
(the first peak of the Magenta trace at the far left)
but there are a number of spurious peaks as well. As it
turns out, a 1/4 wave bandpass cavity has a natural tendency to
respond to odd-order harmonics as well as the
fundamental frequency - that is, if the cavity is tuned to, say,
146 MHz, it will also have peaks at approximately 438,
730 and 1022 MHz. In spite of these peaks you can see
that, for the most part, the signals are attenuated across the
band.
Finally, look at the Cyan trace.
As with the upper image in Figure 3 this is a
combination of the 3-can duplexer leg and the single 4" bandpass
cavity. You'll notice that compared to the Magenta
trace (the single bandpass cavity) the insertion loss is
higher-still for most frequencies. If you compare the Cyan
to the Yellow trace you'll see that there is a
tremendous difference in what passes through and gets to the
receiver!
Reducing higher frequencies with a low-pass filter:
As with the bandpass cavity alone, the 3-can
duplexer leg also has responses at its odd-order harmonics - in
addition to lots of other places - which is why we still see
those frequencies coming through. Practically speaking, we
can eliminate those "harmonic" responses very easily by
adding a low-pass filter - such as that from a
junked 2-meter transmitter - to our duplexer/cavity
arrangement. If we were to do that, there would be
essentially nothing getting through above 200 MHz - or
anywhere else!
How about adding more cavities?
Many amateur 2-meter repeater installation use
"4-can" duplexers (2 cavities per leg) and they exhibit much the
same response - although the "off-frequency" responses may have
even less attenuation than that of the "6-can"
duplexer - but only by a couple of dB overall. This
reinforces the fact that if you have a "4-can" duplexer and have
interference, a "6-can" duplexer will probably not help
- unless the problem is really too-little TX/RX
isolation!
As shown, if you do have a "4-can" duplexer,
you can improve its performance with the addition of a bandpass
cavity: You would not only get vastly superior
"off-frequency" rejection, but your TX/RX isolation would
improve as well. The caveat in adding a bandpass cavity
(or cavities) is that sometimes there can be an interaction
between the notches and the bandpass cavity response, so you'll
want to readjust both of them to achieve both minimum insertion
loss and greatest notch depth - and be prepared to experiment
with different-length interconnect cables between the duplexer
and the bandpass cavity, trying cables that are multiples of
both 1/4 and 1/2 electrical wavelength long and seeing which one
works best. If you do this, also be prepared to check and
re-tune the cavities as necessary to maxinimize notch
depth/minmize bandpass loss as the tuning may change slightly
with these different configurations.
One important measure of
performance of any duplexer is the amount of isolation
between the Transmit and Receive ports: Too little
isolation, and your own transmitter will deafen your
receiver! As it turns out, for 2-meter ham use with a 600
kHz spacing, a good-quality 4-cavity duplexer will have
plenty of isolation for a typical repeater - assuming it
provides 85dB or more TX/RX isolation. There are a few
reasons why one might need more isolation than this,
however, including:
We aren't done yet...
Because we ran out of time (and bandpass
cavities) we had yet to do one more important thing and the time
of originally writing this article: Put a proper bandpass
cavity and lowpass filter on the TX port. While we have a
2-stage isolator there now, it is worth remembering that a
high-band VHF isolator isn't going to work very well at keeping
the FM-band or UHF DTV signal that might be 10's to 100's of MHz
away, so one has to "pre-filter" before applying the TX antenna
to the isolator.
In our testing, we have found that the UHF Icom D-Star repeater
isn't being demolished by the extra signals coming through the
duplexer, but we consider that to be mostly a matter of
luck: This receiver will also sport at least one bandpass
cavity when we get the opportunity to do so.
Adding "test equipment" to the
D-Star stack:
In light of all of this, we are considering
adding another piece of test equipment to the D-Star
repeater: An analog repeater.
This may sound odd or even sacreligious, but it makes sense in a
way: On a shared site such as this there are a lot of
things that can go wrong once you rule out a problem with the
gear itself. For example, a malfunctioning FM broadcast
transmitter can throw lots of garbage across the spectrum that
is often transient in nature. Unfortunately, the very nature of
D-Star complicates using the tried-and-true diagnostics
previously used for track on-site QRM (like "hearing" what the
QRM sounds like) and especially on an all-digital system,
determining the cause of system degradation can be particularly
difficult - especially if it is intermittent and occurring only
when it is inconvenient to drop everything and race to the top
of the mountain, hoping that it is still happening by the time
that you get there!
The plan would be to have the analog repeater normally commanded
off, using a T/R relay to switch between the two transmitters.
Because of the site noise we have the option of simply using a
splitter on the receiver and ignoring the 3-4dB hit on insertion
loss - which would make no difference in receive sensitivity,
anyway. If problems show up that appear to be RF-related,
we'll simply command the analog repeater online and do "normal"
tests to determine the possible cause. Fortunately, we
have enough rack space and gear around to pull this off.
Another thought was to simply install a GE Exciter and an NHRC-4
repeater control (and a simple squelch circuit) inside the
repeater's box to accomplish the same thing, using the Icom's
discriminator output as the audio source - which would be more
representative of the system's receive state, anyway: The
200mW or so of TX output from the exciter board would be more
than enough for testing.
In our brainstorming we have also thought about having an analog
repeater in which one takes the discriminator outputs of the
various D-Star repeater's receivers and selecting among them,
providing a way to remotely "hear" what the D-Star receivers are
hearing to see what sort of QRM may be present - a valuable
diagnostic tool especially if the QRM is intermittent in nature.
Finally, one suggestion made on the Yahoo
Group by Greg, N6LDJ, was to use a computer on-site with its
sound card connected to the discriminator port of the D-Star
repeater. In this way the on-site audio can be piped
(something possible using a number of available programs - or
even via an on-site IRLP or Echolink node) back, via the
internet to a monitoring point. In this way one can
monitor the "analog" world that the D-Star repeater sees and if
some sort of interference appears, there's some hope in
discerning its origin!
A bit more about the tools and techniques used:
Using the RSSI indication:
As was mentioned, the RSSI
(Received Signal Strength Indicator) and discriminator
output (audio) signals from the D-Star's receivers were
brought out via added rear-panel BNC connectors. The
RSSI provides a voltage that is more-or-less proportional to
the input signal level in dB.
As a baseline, one would measure
the RSSI voltage without an input signal (but with the
receiver connected to a dummy load) and then, using an
unmodulated carrier, measure the resulting voltage as one
increased signals and note the readings. This would
provide a ready-made "calibration" chart for that
radio (remember - they will all be a bit different!) that
one should keep on-hand - preferrably taped to the radio!
Having this information is very
useful for a number of reasons:
Because digital transmitters are becoming increasingly common in FM, TV and even land-mobile use, it is increasingly likely that these types of interfering signals - which are, in essense, just bandwidth-limited white noise - will show up only as noise in a receiver. This is in contrast to the interference that one might expect from analog TV and FM signals in which there may be tell-tale "sync buzz" or some identifiable audio component!An extremely useful piece of diagnostic equipment that was wielded was the bandpass filter! To be fair, we already knew that the receiver was being overloaded simply because we had observed that the Kenwood repeater that had been used for testing didn't have any problems, and this simple fact ruled out any problems with the fundamental operation of the duplexer and antenna system as well as eliminating extremely high on-frequency site noise levels as the culprit. Had we not known this, we would have used the bandpass filter which was, in our case, in the form of a bandpass cavity. The fact that installing this cavity in the receive leg of the duplexer - which was tuned to the input frequency of the repeater - considerably improved performance, the diagnosis of receiver overload was confirmed and the resolution of this problem was clearly spelled out.
Using conventional test gear to evaluate and test D-Star systems
Disclaimers:
The above recommendations are based on experience, analysis, and the testing described. They also take into account current Utah frequency coordination policies, which are based on previous and ongoing experience and geographical considerations.
The above recommendations should not be applied in other areas of the world without due consideration of local operating practices, needs, and conditions to determine if they are appropriate.
Other Utah VHF Society links related to D-Star:
Using conventional test gear to evaluate and test D-Star systems - This page covers some aspects of D-Star and analog signals and related test equipment that may make it easier to evaluate the performance of D-Star systems and links.
Observations of the codec used for D-Star - How does the codec used for D-Star respond if subjected to sounds other than those of the human voice? We decided to find out.
The following are FAQ's provided by the Utah VHF society. Note that these may topically overlap the links above:
Questions, updates, or comments pertaining to this web page
may be directed to the frequency
coordinator.
Return to the Utah VHF Society home page.
Updated 2011220