The Utah VHF Society Channel Spacing for Yaesu C4FM<>C4FM and
C4FM<>Analog FM signals
Purpose of this page:
As new technologies come into use on the amateur
bands, there is an increasing challenge to be able to evaluate and
support these technologies. In the past, conventional test
equipment has been used to maintain and diagnose such systems, but
with these new technologies there is a challenge to be able to
provide a means of being able to support such systems in a
meaningful way.
One of the recent additions to the list of technologies for
conveying low-rate voice and data is Yaesu's C4FM as found on
their DR-1 repeater, FT-1DR Handie-Talkie and the FTM-400DR mobile
radio.
Important Notes:
This page deals only with the narrowband
D-Star and Yaesu C4FM modes as found on the VHF and UHF U.S.
amateur bands.
The information on this page is based on
analysis that anyone with the proper test equipment and access
to a C4FM radio could perform.
What is C4FM and how does
it compare to D-Star?
As transmitted on-air, a C4FM signal - like
Icom's D-Star signal
- is simply FM: More specifically, it uses Frequency-Shift
Keying (FSK) to convey data. By properly shaping the modulating waveform and
appropriately choosing the amount of deviation, the transmitted
spectrum can be adjusted to minimize the occupied bandwidth while
still maintaining reasonable power efficiency in terms of being
able to transmit data.
The D-Star system implemented
on the amateur bands by ICOM uses a 2-level modulation
scheme: In terms of a raw signal, a low frequency might
represent a "0" while a high frequency may represent a
"1". What this means is that for each symbol transmitted,
we get ONLY a "0" or a "1":
In other words, we get 1 bit per "baud" and for 4800 baud
modulation used with D-Star, our raw data rate is 4800
bits per second.
C4FM's modulation is slightly more complex. Instead of
just a "0" or a "1" being transmitted by a low and high
frequency, respectively, there are actually four
possible states:
"00" - The
frequency is lowest.
"01" - The frequency is partway
between the lowest and the mid-point.
"10" - The frequency is partway
between the mid-point and the highest.
"11" - The frequency is highest.
During any single baud period,
the signal could be at any one of the above four
states. Like D-Star, the baud
rate is 4800 symbols per second, but since each symbol could
represent any one four bit combinations - 00, 01,
10 or 11 - we are actually conveying two bits per
baud. What this means is that at just 4800 baud, our raw
data rate is actually twice that of D-Star, or 9600 bits per
second.
Comment:
C4FM used by the Yaesu gear is
comparable to the C4FM used with Phase 1 P25 ("Project 25")
radios - See the Project 25
page on Wikipedia for more information.
We can't get something for nothing!
While we can send 2 bits per symbol using C4FM, that doesn't mean
that there isn't a compromise. When we send just one bit per
symbol - as in D-Star's MSK - we have only two
possibilities: A "0" or a "1". With C4FM we have four
possibilities which means that in the event of a noisy signal, our
receiver is going to have a more difficult time determining if the
symbol received was a "00", "01", "10", or a "11" and this
is due to the decreased "distance" between the four possible
states of C4FM as opposed to the two states of MSK used in
D-Star. According to Shannon's
Law, the more data we pack into a smaller period of
time, the more bandwidth and/or energy (power) we will need to
convey it if we wish to maintain the same error rate, so something
must be done with the C4FM signal to make it work well with weak,
noisy signals.
In order to enhance the ability to decode C4FM signals with fewer
errors, here are but two of possibilities:
Increase the modulation index. In
the simplest terms, if we "swing" our frequency more overall,
we are putting more distance between the four individual
symbol states, making it easier to distinguish between them.
Improve the receive modem.
Two-level FSK/GMSK is pretty easy to demodulate into 1's and
0's: If the signal is above "center" we can call that a
"1" and if below, we can call that a "0", but with C4FM, the
determination of the state isn't quite clear. Rather
than using a fairly simple decoder chip such as the CMX-589A
as used in some D-Star radios, we may have to throw a bit more
computing power at our signal to properly decode it. By
integrating the decision-making process with other aspects of
decoding - as well as running some intelligent noise reduction
and signal distortion compensation - it is possible to improve
the accuracy of our "guesses" and be more likely to come up
with the right answer.
Of those listed, above, I
am only certain of the aspects of the first - that is,
increasing the modulation
index as I have yet to look at a technical manual
for the Yaesu C4FM gear to see what is used for
demodulation.
Occupied bandwidth of the C4FM signal:
The question to be answered here is "How much bandwidth does a
C4FM signal need?" The answer, as implied by the discussion
above, is that more bandwidth is required than for D-Star, but how
much?
Figure 1 shows a composite of
several signals, from the same transmitter, to help one divine an
answer.
The Magenta (purple) trace is the C4FM signal.
The Cyan (blue-ish) trace is a +/-
5 kHz analog FM signal heavily modulated by a male voice.
The Yellow trace is the same
transmitter with no modulation present.
What is immediately apparent is the fact that the
occupied bandwidth of the analog FM and C4FM signals
are very similar.
If one compares these with plots of a D-Star signal (see Figure
1 on the page about D-Star Channel Spacing (link) on this web site)
it is apparent that the C4FM signal is significantly wider -
approximately 12.5 kHz at the -26 dB points as compared to
approximately 10 kHz at the -26dB points for D-Star, both
measurements being relative to the peak of the modulated carrier.
Comment:
Comparing the bandwidths of the
signals with respect to an unmodulated carrier would
require stating the resolution bandwidth of the measurements
owing to the power density of the bandwidth-limited noise
aspects of these digitally-modulated carriers. As long as
the resolution bandwidth is a fairly small fraction of the total
occupied bandwidth of the digital signal being measured we can
provide fairly consistent and repeatable measurements.
This analysis and
comparison allows us to come to several conclusions and
speculations as to the rationale behind these differences:
By using GMSK
(Gaussian Minimum-Shift Keying) in D-Star, the occupied
bandwidth is reduced, but at the expense of signal-noise
ratio. With a modulation index of approximately 0.5
(e.g. the deviation is half the symbol rate) our recovered
baseband is "noisier" than it would be for a wider deviation (e.g.
somewhat akin to speaking quietly into the
microphone when the signal is noisy) but with
the reduced bandwidth, we can use slightly IF filtering on our
receiver and gain back a little bit under weak-signal
conditions.
By using C4FM, Yaesu has somewhat increased
the "fragility" of the received signal, but by running a bit
more deviation (a modulation index of approximately 1)
more baseband data is recovered, elevating the digital
waveform out of the noise somewhat and recovering some of this
lost margin. In doing this, the transmit bandwidth is a
bit wider than for D-Star (e.g. approximately 12.5 kHz versus
10.0 kHz at the -26dB points)
"Digital signals make better
neighbors"
When comparing a digital
signal to an analog one, there is more than just occupied
bandwidth to be considered, but also the way that the bandwidth
is actually used. When either a C4FM or D-Star signal is
active, it is, for practical purposes, bandwidth-limited noise
and owing to the nature of the digital signal, the way the
energy is spread over the occupied bandwidth is fairly constant
over time.
An analog signal, on the other hand, is quite variable. As
can be seen from Figure 1, an unmodulated carrier (e.g.
silence) consists of a "spike" with most of the power sitting on
the nominal transmit frequency. With the wildly variable
nature of the human voice and the sounds that it can make, the
occupied bandwidth of an FM signal can vary from just that
single spike to broad bursts of noise over quite a wide
bandwidth from the sounds of consonants and this can be
difficult to represent on a static spectrum analyzer plot!
What this means is that with normal (analog FM) voice there will
be periods of very narrow bandwidth corresponding to silence and
quiet parts of speech as well as very wide bandwidth bursts that
last only milliseconds - but this has an implication for
neighboring frequencies: While an adjacent analog signal
will occasionally have bursts of "splatter" that will very
briefly impact a close, adjacent-frequency signal, a digital
signal - with its fairly consistent modulation - will deviate
very little from its nominal bandwidth and will thus be a better
"neighbor" than an analog one!
Receive bandwidth:
While the bandwidth occupied by the transmitted signal is
important, on a radio system we must
take into account the bandwidths of the receivers typically being used. If we had a
hypothetical digital signal that was just 5 kHz wide - but our
receiver was 20 kHz wide - while we could place two of those 5
kHz signals 10 kHz apart with room to spare, our 20 kHz wide
receiver would not be able to distinguish the
two!
Through empirical measurements of various D-Star gear, we
determined that the typical receiver in D-Star mode was
approximately 11 kHz wide at the -30dB points. Considering
that the vast majority of the transmitted energy of a D-Star
signal is contained within less than 10 kHz of bandwidth, this
was a reasonable compromise between practical and affordable IF
filtering and minimizing degradation of the received signal by
the filters themselves. In our analysis, we determined
that a "safe" spacing for D-Star signals - considering the
occupied bandwidth of the transmitted signals, the bandwidth of
the receivers that would be used - as well as expected frequency
tolerances - was 12.5 kHz.
Clearly, the C4FM signals are wider than the D-Star signals -
quite comparable to analog FM signals - so this means that we
require a wider channel spacing. In measurements with
actual C4FM gear, it was determined that the IF filtering
bandwidths appeared to be the same as those for normal
"wide-band" (+/- 5 kHz) FM signals. (Specific
measurements of actual C4FM gear receive bandwidth will
follow.)
(Preliminary) Conclusions:
At the present time we have yet to run actual co-channel
interference tolerance tests to determine how adjacent C4FM
signals interact with each other or, more importantly, the
degradation of an analog signal by an adjacent C4FM
signal (and vice-versa) - This testing will follow.
At the present time we can conclude the following with good
confidence:
20 kHz channel spacing should be
observed with C4FM signals for geographically adjacent
systems. That is, if you
have a strong, nearby C4FM or analog FM signal on an adjacent
channel while you are receiving a weak, distant C4FM or analog
FM signal, 20 kHz spacing is adequate to prevent interference
issues.
15 kHz channel spacing should be done
with careful consideration of mileage and geographical
coverage. At the present time our preliminary
measurements indicate that the same precautions taken with
analog systems at 15 kHz spacing should be observed with C4FM
and analog FM signals. It may be possible that
the interference of between two C4FM signals 15 kHz apart - or
even a C4FM signal to an analog signal - may be
acceptable in certain geographical adjacent situations, but
this is only conjecture and a conclusion as to the degree of
susceptibility is pending analysis.
This page is a work in progress will be updated.
Disclaimers:
The above procedures have been tested
using available test gear and Yaesu radios and are believed
to be valid. It is likely that this information will,
in the future, be updated and techniques refined.
It is up to you, the reader, to
verify that this information is, in fact, correct and
suitable for your needs. We cannot be held responsible
for your use of the above information!
If you find that the above information
is incorrect or incomplete, please contact the frequency
coordinator using the link below.
Your mileage may vary!
This matter is open for
discussion: If you have concerns or opinions one way
or another, please make them known to the frequency
coordinator at the email address below.
Questions, updates, or comments pertaining
to this web page may be directed to the frequency coordinator.