The Utah VHF Society

Analysis and recommendations of channel spacing for D-Star
operations on the VHF, UHF, and 23cm amateur bands

Purpose of this page:

Amateur radio has long been faced with the adoption of newer technologies:  It can be argued that innovation and experimentation are some of the main purposes for the existence of amateur radio.  As these new technologies come along, however, there is also the responsibility to accommodate these new systems into the existing framework.

D-Star is a fairly new digital voice system that is loosely based on other commercial standards.  It offers the potential advantage of ease of networking, the ability to send data, and the possible advantages inherent to digital modulation schemes in terms of signal quality.  Incorporating these signals amongst existing analog operations requires attention to technical details and some foresight in order to maximize the potential of this new technology as well as allow co-habitation of new and old systems.

Important Notes:

A bit of background:

There has been some mention of how spectrum-efficient D-Star is as compared with analog signals and, because of this, a lot more D-Star signals can be crammed into the same space as one analog signal:  One oft-cited instance is the simultaneous operation of several D-Star signals spaced only 6.25 kHz apart from each other.  While this sounds like an impressive feat, cursory examination of the bandwidths of the transmitters, receivers, and link margins will immediately reveal that this is NOT a good thing to do!  (It is worth noting that several Icom D-Star capable radios are not capable of tuning 6.25 kHz steps, anyway - see below.)

Important note:

Some of the recommendations on this page may apply only to the circumstances that apply in the Utah area.  It is the responsibility of the reader to study this and other available data in order to come to a reasoned and technically-sound conclusion appropriate to local conditions and patterns of usage!

Testing under "simulated real-world" conditions:

At the time that this page had originally been created, relatively little had been done to carefully analyze how D-Star signals will co-exist with each other - and with existing analog signals - in the real world, using real radios that people own.  In order to answer some of these questions, we decided to take a typical D-Star radio, an IC-91AD, and put it to the test.  To do this, we put together a test fixture.  The description and operation of this test system is as follows:


Figure 1:
Comparative spectra of D-Star (left) and typical analog (right) signals.  In each case, vertical divisions are 10 dB and horizontal divisions are 2 kHz.  An unmodulated carrier has been overlaid atop both images with the green line representing the level of the unmodulated carrier.
Click on image for a larger version
Spectrum analyzer comparisons of Dstar and Analog signals
Comments:

Figure 2:
Measured bandwidths of several ICOM D-Star capable radios in their wide and narrow modes.  Note that Icom specifies a -55dB bandwidth for "narrow" mode and a -60dB bandwidth for "wide" mode.
Radio
 -6dB Bandwidth
 -30dB Bandwidth
 -55dB Bandwidth (narrow) or
-60dB (wide)
IC-91AD
(Narrow FM/DV)
 8.6 kHz
 11.2 kHz
 13.7 kHz
IC-91AD
(Wide FM)
 10.7 kHz
 17.25 kHz
 20.6 kHz
IC-2200H
(Narrow FM/DV)
 7.7 kHz
 10.7 kHz
 13.9 kHz
IC-2200H
(Wide)
 12.9 kHz
 17.9 kHz
 21.2 kHz

Occupied transmit signal bandwidth and receiver bandwidth:

It is important to note that two major factors affect how two signals - whether they are D-Star or analog - interact with each other:
The relative "narrowness" of the D-Star signal is oft-touted as one of its strong points.  To be sure, more of the total transmitted energy is confined near the center frequency than is the case for the analog signal.  For the D-Star signal, the majority of the energy is constrained to within +-3.6 kHz of the center frequency.  In the case of the analog signal, the majority of the energy is constrained to within +-5 kHz of the center frequency.  This only tells part of the story:  If one looks at the -30dB points of the two signals, one notes that the bandwidth of the D-Star and analog signals are +-5 kHz and +-6 kHz, respectively - and it is the energy in these sidebands that, in part, dictates adjacent-channel concerns.  If one considers just the -30dB points of the transmit signals, a minimum D-Star to D-Star spacing of 10 kHz and a D-Star to Analog spacing of 11 kHz is suggested.
Figure 3:
Spectrum analysis of D-Star's baseband.  There is a null at 4800 Hz correlating with the bit rate and there are strong spectral components at intervals of 50 Hz  that correlate with the 20ms voice frame.
Click on image for a larger version
Spectram of a baseband D-Star signal

Perhaps even more important is the detection bandwidth of the receiver.  Ideally, the D-Star's receiver's filter need only be wide enough to accommodate the primary "hump" that contains the majority of the energy - that is, out to +-3.6 kHz, or a total bandwidth of about 7.2 kHz, but practical considerations (manufacturing tolerances in the manufacture of the filter, achievable shape factor, group delay, expected transmit or receive frequency errors, etc.) require that the filter be wider than this.  As mentioned previously, the -6dB bandwidth of the IF in the IC-91AD is, in fact, 8.6kHz (+-4.3 kHz), dropping to -30dB at 11.2kHz (+-5.6kHz).  It is largely the combination of the receiver filtering plus the occupied bandwidth of the adjacent signal that dictates the minimum spacing of two D-Star signals.

Receivers designed for traditional analog FM use in amateur service are designed for a signal with a +-5 kHz modulation, so the receivers' filters are necessarily wider - typically 15 kHz wide at the -6 dB bandwidth and about 21 kHz wide at the -30 dB bandwidth.  For this reason - plus the fact that the analog signal is wider - it is necessary that the spacing between an analog signal and either another analog or even a D-Star signal must be wider than that between two D-Star signals.

"Equivalent SINAD" test:

Because the IC-91AD uses the FM-Narrow mode for demodulation, it is possible to relate the SINAD in FM-Narrow mode to the performance in D-Star digital voice mode.  This is is possible because the SINAD measurement tells us something about the amount of extraneous noise in the receiver's baseband - something that correlates well with data errors.  The SINAD readings noted in the tests below are unweighted, using audio taken from the speaker connector of the IC-91AD.

This test was conducted to provide a means to analyze the level of an interfering signal that would be required to degrade a D-Star signal.

For this test, three levels of D-Star signal disruption were investigated:
  1. "Clean" audio decoding:  No bit errors were observed over a period of 60 seconds or so.
  2. "Mostly clean" decoding:  One "bloop" (an unrecoverable bit error) occurred every 10 seconds or so.
  3.  Loss of D-Star sync:  At this error rate, not only has recovered speech become unintelligible, but the receiver can no longer maintain synchronization on the D-Star signal.
For this test, two types of situations were simulated using test equipment:
When each of the three levels of disruption were reached, the IC-91AD was switched to FM-Narrow mode while, at the same time, the test generator was switched from generating a D-Star signal to generating an FM signal modulated with a 1 kHz tone at +-1.5 kHz deviation:  At this point, an un-weighted SINAD measurement was taken using the audio from the IC-91AD's speaker connector.

As it turned out the SINAD readings for each of the "D-Star" signal disruption levels were the same whether the degradation was due to a weak signal or adjacent-channel interference.  The correlating SINAD levels were:
  1. "Clean" audio decoding:  At 17-18 dB SINAD was required in FM-Narrow mode to produce a signal that did not suffer audible decoding errors in D-Star mode.
  2. "Mostly clean" decoding:  At 15.5-16 dB SINAD or so, there was one audio "bloop" (an unrecoverable decoding error) in about 10 seconds.
  3. Loss of sync:  At 9-10 dB SINAD, synchronization of the digital signal was intermittent and no intelligible audio was recovered.
Comments:

D-Star to D-Star interference test:

The first test was to see how two D-Star signals interfered with each other, depending on relative signal levels and frequency separation.  In each case, the "weak" signal was monitored for errors while the adjacent signal was increased in strength.  A "solid" audio tone was transmitted on each D-Star data stream using a different tone for each transmitter (to tell them apart.)  In this way, bit errors were easily noted as "bloops" or disruptions in the received tone.  The levels below were those necessary to obtain "clean" tones with no obvious disruptions over a period of about 10 seconds.

In each case, one D-Star signal was being monitored while another D-Star signal (the one being varied in amplitude and/or frequency) was being used as the interference source.

Figure 4:  Interference between two D-Star signals
Situation
 Result
On-channel interference The interfering D-Star signal must be at least 12 dB weaker to avoid interference.
Equal signal strength
 A minimum separation of at least 6.25 kHz is required to avoid interference from another D-Star signal that is of equal strength.
10 dB differential
 A minimum separation of at least 8 kHz is required to avoid interference from a D-Star signal that is 10dB stronger.
20 dB differential A minimum separation of at least 9.5 kHz is required to avoid interference from a D-Star signal that is 20dB stronger.
30 dB differential A minimum separation of at least 10.5 kHz is required to avoid interference from a D-Star signal that is 30dB stronger.
40 dB differential A minimum separation of at least 12 kHz is required to avoid interference from a D-Star signal that is 40dB stronger.
50 dB differential (see note)
 A minimum separation of at least 15 kHz is required to avoid interference from a D-Star signal that is 50dB stronger.

Note:  Amplitude differences of 50 dB or greater are pushing the filtering and dynamic range limits of the receiver, as well as the ability of the test gear to simulate real-world signals.

Comments pertaining to this and other tests:
D-Star signal susceptibility to interference from analog signals:

Because D-Star signals inhabit the same amateur bands as analog signals, consideration must be given to how these should be spaced to avoid the analog signal's causing interference to the D-Star signal.  Unlike D-Star signals, the modulation and bandwidth of analog signals can vary widely - from being a CW carrier when there is no modulation, to a signal spread over a fairly wide bandwidth when fully modulated with voice energy.  Because of this, the interference to a D-Star signal from an analog FM signal can be somewhat transient and it the field it may not be immediately recognized as an interference source under uncontrolled conditions.

For this test audio was taken from an NOAA weather transmission to provide a source of voice modulation that was consistent and repeatable in terms of amplitude and spectral content.  The analog signal was modulated to +-5 kHz deviation, with limiting and pre-emphasis applied in the manner that is standard amateur practice.  Interference to D-Star was noted by the appearance of "bloops" (caused by unrecoverable errors) in the received signal, and more than one "bloop" in a period of 10 seconds or so was considered to represent a degraded signal:  It was noted that only slight (1-2 dB) increases in signal strength of the analog signal caused the D-Star signals to deteriorate very rapidly.

The results of this testing, using an IC-91AD for receiving, are as follows:

Figure 5:  Interference to a D-Star signal from an analog NBFM signal
Situation
 Result
On-channel interference Interfering analog signal must be at least 17 dB weaker to avoid interference.
Equal signal strength
 A minimum separation of at least 9 kHz is required to avoid interference from an analog signal that is of equal strength.
10 dB differential
 A minimum separation of at least 11 kHz is required to avoid interference from an analog signal that is 10dB stronger.
20 dB differential A minimum separation of at least 13 kHz is required to avoid interference from an analog signal that is 20dB stronger.
30 dB differential A minimum separation of at least 16 kHz is required to avoid interference from an analog signal that is 30dB stronger.
40 dB differential A minimum separation of at least 19 kHz is required to avoid interference from an analog signal that is 40dB stronger.
50 dB differential (see note)
 A minimum separation of at least 22 kHz is required to avoid interference from an analog signal that is 50dB stronger.

Note:  Amplitude differences of 50 dB or greater are pushing the filtering and dynamic range limits of the receiver, as well as the ability of the test gear to simulate real-world signals

Analog susceptibility to interference by D-Star signals:

The amount of interference caused by a D-Star signal to an analog signal is a rather difficult parameter to judge because, unlike with the D-Star signal, interference will gradually get worse as the interfering signal's strength increases and/or the separation is reduced.  The amount of interference experienced by the analog user also depends on the design of the receiver used and, in particular, the bandwidth of the filters in its I.F.  To provide some indication of the severity of the amount of degradation of the analog signal, two parameters were measured:
The analog signal used in this test was modulated at +-3 kHz with a 1 kHz sine wave.

For this test, several receivers were used, including:
It was noted that the filters in the IC-91AD used for "normal" +-5 kHz deviation were narrower than those typically seen in similar radios, around 10.7 and 17.25 kHz at the -6 and -30 dB points respectively.  The receiver filters in the other three radios were all about the same, approximately 15.0 and 21.0 kHz at the -6 and -30 dB points, respectively.

For the susceptibility of an analog receiver to interference to D-Star, the performance of the IC-91AD (in FM mode) was worse in the on-channel and 5 kHz spacing cases than the other receivers tried.  For the list below, typical numbers are shown for the various receivers tested.  The typical signal level for the analog test signal was -93 dBm, a signal that resulted in a SINAD of about 30dB.  In certain cases, the levels of the two signals were varied by equal amounts to verify that the noted degradation was largely independent of absolute signal levels.

Figure 6:  Interference to an analog signal from a D-Star signal
Situation
 Degradation to 12 dB SINAD
 Degradation to 20 dB SINAD
On-channel interference D-Star signal must be > 3 dB weaker
 D-Star signal must be > 11 dB weaker
5 kHz spacing
 D-Star signal must be > 3 dB weaker
 D-Star signal must be > 7 dB weaker
8 kHz spacing
 D-Star signal must be > 3 dB weaker
 D-Star signal must be > 6 dB weaker
9 kHz spacing
 D-Star signal may be <= 1 dB stronger
 D-Star signal must be > 2 dB weaker
10 kHz spacing
 D-Star signal may be <= 8 dB stronger
 D-Star signal may be <= 4 dB stronger
11 kHz spacing
 D-Star signal may be <= 16 dB stronger
 D-Star signal may be <= 13 dB stronger
12 kHz spacing D-Star signal may be <= 26 dB stronger D-Star signal may be <= 22 dB stronger
13 kHz spacing D-Star signal may be <= 32 dB stronger D-Star signal may be <= 29 dB stronger
14-20 kHz spacing (see note) D-Star signal may be <= 40 dB stronger
 D-Star signal may be <= 40dB stronger
30 kHz spacing (see note)
 D-Star signal may be <= 60 dB stronger
 D-Star signal may be <= 60 dB stronger

Notes:
Analysis of D-Star <> Analog interference:

As can be seen from the above data, the D-Star signal was actually more susceptible to interference from the analog signal than the analog signal was to the D-Star signal.  This is likely a result of the "transient" nature of adjacent channel interference from an FM signal:  While, on average, the energy from an FM signal is contained fairly close to the center frequency, occasional peaks of modulation or in the spectra of the signal being modulated will cause energy to occasionally appear farther afield.  These occasional "peaks" will cause bit errors to occur in the received D-Star signal and if the number of errors gets to be too great, obvious decoding errors will result.

Note that in the analog domain, one has the obvious advantage in that the degradation increases more gradually as the interference worsens and this degradation is noted as the appearance of noise on the signal:  Even moderate amounts of noise does not necessarily result in the loss of intelligibility.



Channel spacing recommendations:

In real-world situations, it is recommended that at least 30dB of margin be designed into the systems when it comes to interference potential - and even more is preferred where practical.  It is perfectly reasonable to expect that two adjacent channels could have amplitude differences of 30 dB within their primary coverage areas, so suitable margins must be considered when frequency coordination is done.  In some cases, even more than 30 dB of margin will be required - as might be the case for repeaters with extremely large coverage areas, links, or in the consideration of frequency-reuse in some cases.

It should also be recognized that even if such a margin is designed into a system, a significant interference potential still exists, particularly when one considers that due to multipath and various propagation phenomena, signals from both the desired and undesired transmitters can be momentarily enhanced or degraded considerably - an effect that is most likely to be a problem in those areas with overlapping coverage.  In such situations, D-Star tends to fare worse, as the codec may take some time to re-synchronize after it has lost lock and several syllables may be lost.

Another consideration is that the normal tolerances of frequency stability for amateur gear may result in a transmitter (or receiver) being somewhat off-frequency:  It is not unreasonable for a UHF transmitter to be 1-2 kHz off frequency after normal component aging, when it is hot or cold, or if the radio has been exposed to severe mechanical shock - values that may still be within the manufacturer's specifications.  In these cases, degradation of the communications link can be expected and sufficient channel-spacing margin must be allowed for such occurrences - and those where radios, for whatever reason, are beyond the manufacturer's specifications.

Based on the above test data as well as frequency and spectral analysis, the following are recommendations of the Utah VHF Society:
 On 2-meters, the above recommendation is complicated by the fact that the channel spacing in Utah is 20 kHz - something that does not readily lend itself to the adoption of 12.5 kHz spacing.  This has two important implications:
The transmit bandwidth of a D-Star signal has been analyzed and measured to be over 60 dB down at +-10 kHz, so it may be possible to place a D-Star signal 10 kHz away from a band edge and maintain compliance with FCC rules pertaining to spurious and out-of-band emissions, but transmitter frequency tolerance considerations must still be observed!

On 70cm, with 25 kHz analog channel spacing being used in Utah, it is perfectly reasonable to place two D-Star channels within one analog channel:  One D-Star signal would have a frequency 6.25 kHz below and the other would be placed on a frequency 6.25 kHz above the center frequency of the analog channel.  Such spacing would also afford protection between adjacent D-Star and analog users.  The caveat to this recommendation is that not all D-Star capable radios are able to tune in 6.25 kHz steps - see the warning below.

Why 12.5 kHz minimum spacing instead of 10 kHz?

Why 12.5 kHz D-Star to D-Star spacing when others have said that even 10kHz might be wasteful?  An oft-overlooked consideration is transmitter and receiver frequency stability.

For example, the specifications for the IC-91AD are +-2.5ppm - and this implies that the transmitter or receiver could be a bit over 1 kHz off-frequency on 70cm.  With adjacent channels, this means that two channels could be 2 kHz closer to each other (if, say, the lower one was 1 kHz high and the upper one was 1 kHz low) and reduce the spacing to less than 10.5 kHz - a difference that reduces margins somewhat.

Conversely, if a 10 kHz spacing is used, frequency variances could reduce the spacing to less than 8 kHz under worst-case conditions - a separation that pushes against the skirts of receivers' IF filters, not to mention the transmit signal spectra!

Remember:  The above are minimum spacing recommendations.  Depending on the specific situation, there may need to be other considerations based on the necessity to protect existing systems.

Note:  These recommendations assume that the primary mode of operation is to be voice.  D-Star data transmissions tend to be more susceptible to errors than voice transmissions, owing mostly to the inbuilt FEC in the voice coding as well as the redundant nature of human speech and the ability of the listener to mentally "fill in" missing pieces:  Data transmissions may not be so forgiving to errors in reception and require greater margins.  If time permits, similar tests may later be run using "data-only" transmissions.

Other D-Star Co-channel and adjacent channel tests:

Mark, N5RFX, has also done some adjacent-channel testing for D-Star signals:  The results of this testing are in his paper, found at the link below:

http://home.roadrunner.com/~mdmiller7/images/dv/ch_sp/Dstar_Co.pdf

Please note that the presentation, methods, and criteria of these tests were slightly different from those that we have done, so one must read both writings carefully before making comparisons.  If anything, Mark's results show a greater tolerance of D-Star signals to the various interfering sources than what we observed.  At least some of these differences are due to the fact that the the ID-800, the receiver used by Mark, seems to be better at tolerating adjacent-channel signals/interference than the IC-91AD, the radio that we used for our testing, plus the fact that the threshold of acceptable degradation to the D-Star signal may have been different.

Channel spacing for 128kbps D-Star ("DD" mode) on 23cm:

Another D-Star standard may be found on the 23cm (1200 MHz) amateur band. On this less-crowded band it is permitted to run much higher symbol rates than is permitted on 2 meters and 70cm and a 128 kbps mode is available:  One radio that can operate using this protocol is the Icom ID-1.  Also capable of the "standard" 4800bps DV mode found on the 2 meter and 70cm band, the addition of 128kbps makes higher-speed links practical.  The ID-1 has its own Ethernet interface, allowing standard internet IP protocols to be passed around over the air using half-duplex with a reported throughput of up to 90kbps.

Figure 7:
Spectrum analyzer plots of a 128kbps D-Star signal on 23cm in a span of 1 MHz (left) and 250 kHz (right).
Click on image for a larger version
Spectra of 128kb DD mode D-Star signals on 23cm
With this higher speed comes a much wider bandwidth, but how wide, exactly?  Figure 7 shows the transmitted spectra of an Icom ID-1 in DD mode (128kbps.)  As can be seen from this plot, the signal is about 150kHz wide (at the -26dB points) as is mentioned by the specifications, but one can also see that sidebands extend beyond this, albeit at much lower levels.

Channel spacing:

Given that the bandwidth appears to be on the order of 150 kHz, one might believe that 150-200kHz channel spacing would be adequate - but this would be incorrect.  According to ICOM's own specifications, the receive bandwidth characteristics for the ID-1 are:
It is this latter figure that dictates the minimum channel spacing.  Clearly, 150kHz spacing is far too narrow, so allowing for a reasonable degree of adjacent-channel isolation, the Utah VHF Society recommends a channel spacing for these carriers of 500 kHz.


Comment on the IF filtering used in the ID-1:

Figure 8:
Passband and group-delay plots of the 10.7 MHz 2nd IF filters used in the ID-1 for 128kbps DD mode.  These plots are for single filters:  The ID-1 uses two such filters in cascade to set the receiver bandwidth.  (Source:  Murata)
Click on image for a larger version
Murata-supplied plots of the bandpass filters used in the 128k DD mode of the ID-1 (plot is for one filter - the ID-1 uses two filters in series.)
It was noted that the ID-1's final IF frequency is 10.7 MHz.  Inspection of the ID-1's service manual reveals that it uses a pair of cascaded 10.7 MHz ceramic filters of the sort used in commercial FM broadcast receivers.  These particular filters (Murata SFELA10M7HA0G-B0) are designed specifically for data use and have a slightly poorer shape factor than standard ceramic filters used in FM broadcast receivers to optimize group delay response to minimize distortion of the data.  According to Murata (and as can be seen in Figure 8) the stated specifications for these filters are:
Note that two of these filters are used in series in the IF chain to improve the response, with an MC3356 used as a demodulator.  In the first IF (at 243.95 MHz) there is a SAW filter that provides "roofing" filtering for all digital and analog modes:  The nominal bandwidth of this filter appears to be on the order of 750 kHz, but further specifications are not yet known.)

It is hoped that we will be able to aggregate several Icom ID-1's and perform more detailed tests to determine adjacent-channel tolerance at various spacings and signal levels.  Being that 23cm isn't a heavily-utilized band in Utah and that presently-available D-Star systems are synthesized, the 500 kHz spacing seems to be a "safe" value and, if further testing warrants that a narrower (or wider) spacing is more appropriate, changes can be made at that time with little inconvenience.



The GMSK modem used in the ID-1 and IC-91AD:

The ID-1 uses a CMX589A GMSK modem chip for recovering data from the GMSK baseband signal from the MC3356 demodulator in both the low-speed (DV) and high-speed (DD) modes:  The ID-1 uses a separate modulator to generate I/Q signals for transmit, leaving half of this chip unused.  The ID-91AD, on the other hand, uses this chip for both reception of and generating the GMSK baseband waveforms.

The CMX589A is an integrated receiver/transmitter that is designed to receive and generate GMSK baseband waveforms.  (A data sheet for this chip may be found here.)  Interestingly, this chip has an "RX S/N" pin that outputs a signal that can be used to approximately estimate the signal-noise ratio of the received signal, but alas this connection (pin 23) is left disconnected in the ID-1 and IC-91AD:  This is a pity, as the use of this pin might have proven helpful in determining optimal signal quality when setting up D-Star links, not to mention in everyday use by the casual user!

Comment:  The ID-1 uses this chip only for receive while the IC-91AD uses it for both receive and transmit.  It is worth noting that BT (the ratio of the transmit filter's -3dB bandwidth and the bit rate) is set for 0.5 in the IC-91AD's modulator, a reasonable compromise between occupied bandwidth and ISI.


A Warning about the selection of channel spacing and center frequencies:

It should be remembered that not all Icom D-Star capable radios have the ability to tune in the same size of frequency steps.  In order to better-fit into the bandplan of existing systems, it might be tempting to pick a frequency that is based on a multiple of 6.25 kHz.

Not all Icom D-Star radios are capable of tuning in 6.25 kHz steps!  An example of a radio that cannot tune 6.25 kHz steps is the Icom IC-2200H:  This radio can tune in 5 and 12.5 kHz steps and various multiples of of those step sizes.

Before deciding on a frequency plan for your D-Star channels, make sure that the center frequencies that you pick are, in fact, based on multiples of 5 or 12.5 kHz or you will leave people out!



Disclaimers:

Other Utah VHF Society links related to D-Star:

Misc. links related to D-Star:

The above list is, by no means, exhaustive:  Other information may be found via web searches.


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.

Return to the  Utah VHF Society home page.


Updated 20080617