<h1>DSD (Digital Speech Decoder) demodulator and decoder plugin</h1>
<h2>Introduction</h2>
This plugin uses the [DSDcc](https://github.com/f4exb/dsdcc) library that has been rewritten from the original [DSD](https://github.com/szechyjs/dsd) program to decode several digital speech formats. At present it covers the following:
To enable this plugin at compile time you will need to have DSDcc installed in your system. Please follow instructions in [DSDcc readme](https://github.com/f4exb/dsdcc/blob/master/Readme.md) to build and install DSDcc. If you install it in a custom location say `/opt/install/dsdcc` you will need to add these defines to the cmake command: `-DDSDCC_DIR=/opt/install/dsdcc`
You can use a serial device connected to your system that implements and exposes the packet interface of the AMBE3000 chip. This can be for example a ThumbDV USB dongle. You may also connect to an AMBE server instance over the network.
DV serial devices are supported using the [SerialDV](https://github.com/f4exb/serialDV) library that is a mandatory requirement. Therefore you have to compile and install it in your system. Please refer to this project Readme.md to compile and install SerialDV. f you install it in a custom location say `/opt/install/serialdv` you will need to add this define to the cmake command: `-DSERIALDV_DIR=/opt/install/serialdv`
To effectively use serial DV devices for AMBE decoding you will have to add at least one device to the list of AMBE devices in use using the `AMBE devices control` dialog opened with the `AMBE` option in the `Preferences` menu. The list of devices is saved in the program preferences so that they are persistent across program stop/start. However if the device name or server address changes in between the corresponding reference will be lost.
Although such serial devices work with a serial interface at 400 kb in practice maybe for other reasons they are capable of handling only one conversation at a time. The software will allocate the device dynamically to a conversation with an inactivity timeout of 1 second so that conversations do not get interrupted constantly making the audio output too choppy. In practice you will have to have as many devices connected to your system as the number of conversations you would like to be handled in parallel.
Note also that hardware serial devices are not supported in Windows because of trouble with COM port support (contributors welcome!).
If no AMBE devices or servers are activated with the `AMBE devices control` AMBE decoding will take place with Mbelib. Possible copyright issues apart (see next) the audio quality with the DVSI AMBE chip is much better.
⚠ With kernel 4.4.52 and maybe other 4.4 versions the default for FTDI devices (that is in the ftdi_sio kernel module) is not to set it as low latency. This results in the ThumbDV dongle not working anymore because its response is too slow to sustain the normal AMBE packets flow. The solution is to force low latency by changing the variable for your device (ex: /dev/ttyUSB0) as follows:
DSDcc itself can use [mbelib](https://github.com/szechyjs/mbelib) to decode AMBE frames. While DSDcc is intended to be patent-free, `mbelib` that it uses describes functions that may be covered by one or more U.S. patents owned by DVSI Inc. The source code itself should not be infringing as it merely describes possible methods of implementation. Compiling or using `mbelib` may infringe on patents rights in your jurisdiction and/or require licensing. It is unknown if DVSI will sell licenses for software that uses `mbelib`.
If you are not comfortable with this just do not install DSDcc and/or mbelib and the plugin will not be compiled and added to SDRangel. For packaged distributions just remove:
- For Linux distributions: `plugins/channel/libdemoddsd.so`
- For Windows distributions: `dsdcc.dll`, `mbelib.dll`, `plugins\channel\demoddsd.dll`
For software built from source if you choose to have `mbelib` support you will need to have DSDcc compiled with `mbelib` support. You will also need to have defines for it on the cmake command. If you have mbelib installed in a custom location, say `/opt/install/mbelib` you will need to add these defines to the cmake command: `-DMBE_DIR=/opt/install/mbelib`
Use the wheels to adjust the frequency shift in Hz from the center frequency of reception. Left click on a digit sets the cursor position at this digit. Right click on a digit sets all digits on the right to zero. This effectively floors value at the digit position. Wheels are moved with the mousewheel while pointing at the wheel or by selecting the wheel with the left mouse click and using the keyboard arrows.Pressing shift simultaneously moves digit by 5 and pressing control moves it by 2.
With the DV serial device(s) amplification factor in dB is given by `(value - 3.0)*5.0`. In most practical cases the middle value of 5.0 (+10 dB) is a comfortable level.
Number of milliseconds following squelch gate opening after which the signal is declared open. There is a delay line for the samples so samples applied to the decoder actually start at the beginning of the gate period not loosing any samples. 0 means squelch is declared open with no delay.
If you right click on it it will open a dialog to select the audio output device. See [audio management documentation](../../../sdrgui/audio.md) for details.
When the display is active the background turns from the surrounding gray color to dark green. It shows informational or status messages that are particular to each format.
- at the right of the colon `:` is the bearing in degrees and distance in kilometers from the location entered in the main window `Preferences\My Position` dialog. The bearing and distance are separated by a slash `/`.
- Note 1: statuses are polled at ~1s rate and therefore do not reflect values instantaneously. As a consequence some block types that occur during the conversation may not appear.
- Note 2: status values remain unchanged until a new value is available for the channel or the transmissions stops then all values of both channels are cleared
- at the right of the `>` sign is the destination callsign. It is filled with stars `*` when call is made to all stations (similar to the CQCQCQ in D-Star)
This is the RAN number (0 to 63) associated to the transmission. RAN stands for "Radio Access Number" and for trunked systems this is the site identifier (Site Id) modulo 64.
This is the type code of the last message (6 bits) displayed in hexadecimal. The complete list is found in the NXDN documentation `NXDN TS 1-A Version 1.3` section 6.
This is a 16 bit collection of flags to indicate which services are available displayed in hexadecimal. The breakdown is listed in the NXDN documentation `NXDN TS 1-A Version 1.3` section 6.5.33. From MSB to LSB:
This is the RAN number (0 to 63) associated to the transmission. RAN stands for "Radio Access Number" and has a different usage in conventional or trunked systems:
- Conventional (RDCH): this is used as a selective squelch. Code `0` means always unmute.
- Trunked (RTCH): this is the site identifier (Site Id) modulo 64.
This is the type code of the last message (6 bits) displayed in hexadecimal. The complete list is found in the NXDN documentation `NXDN TS 1-A Version 1.3` section 6.
The display shows 16 points as yellow crosses that can be used to tune the center frequency (A.1) and FM deviation (B.17) so that symbol recovery can be done with the best conditions. In the rest of the documentation they will be referenced with numbers from 0 to 15 starting at the top left corner and going from left to right and top to bottom.
There are 4 possible points corresponding to the 4 possible transitions. x represents the current symbol and y the previous symbol. The 4 points given by their (y,x) coordinates correspond to the following:
As you can see the pointer can make all moves except between (-1, -1) and (1,1) hence all vertices between the 4 points can appear except the one between the lower left corner and the upper right corner.
There are 16 possible points corresponding to the 16 possible transitions between the 4 dibits. The 4 dibits are equally spaced at relative positions of -3, -1, 1, 3 hence the 16 points are also equally spaced between each other on the IQ or (x,y) plane.
The X input is the discriminator signal and the Y input is the symbol synchronization signal that goes to the estimated maximum discriminator signal level when a zero crossing in the symbol synchronization control signal is detected and goes to mid position ((max - min) / 2) of the discriminator signal when a symbol period starts.
The symbol synchronization control signal is obtained by squaring the discriminator signal and passing it through a narrow second order bandpass filter centered on the symbol rate. Its zero crossing should occur close to the first fourth of a symbol period therefore when synchronization is ideal the Y input should go down to mid position in the first fourth of the symbol period.
Ideally the figure should show a cloud of persistent points at the locations marked by points 0 to 3. Each one of these points represent an ideally decoded symbol.
<h4>B.2: Symbol (Baud) rate</h4>
Here you can specify which symbol rate or Baud rate is expected. Choices are:
-`2.4k`: 2400 S/s used for dPMR and 4800 b/s NXDN
-`4.8k`: 4800 S/s used for 9600 b/s NXDN, DMR, D-Star and YSF.
Normally you would always want to have a matched filter however on some strong D-Star signals more synchronization points could be obtained without. When engaged the background of the button is lit in orange.
Since dsdcc version 1.7.1 the symbol synchronization can be done with a PLL fed by a ringing filter (narrow passband) tuned at the symbol rate and itself fed with the squared magnitude of the discriminator signal. For signals strong enough to lock the PLL this works significantly better than with the ringing filter alone that was the only option in versions <= 1.6.0. Version 1.7.0 had the PLL enabled permanently.
However with marginal signals the ringing filter alone and a few heuristics work better. This is why since DSDcc version 1.7.1 the PLL became optional.
You can use this button to toggle between the two options:
- with the locker icon in locked position: PLL is engaged
- with the locker icon in unlocked position: PLL is bypassed
When in lock position the button lights itself in green when the PLL lock is acquired. Occasional drops may occur without noticeable impact on decoding.
<h4>B.6: Symbol synchronization zero crossing hits in %</h4>
This is the percentage per symbols for which a valid zero crossing has been detected. The more the better the symbol synchronization is tracked however the zero crossing shifts much not deviate too much from 0 (see next).
With the PLL engaged the figure should be 100% all the time in presence of a locked signal. Occasional small drops may occur without noticeable impact on decoding.
This is the current (at display polling time) zero crossing shift. It should be the closest to 0 as possible. However some jitter is acceptable for good symbol synchronization:
This is the estimated median level (center) of the discriminator input signal in percentage of half the total range. When the signal is correctly aligned in the input range it should be 0
<h4>B.9: Discriminator input signal level range in %</h4>
This is the estimated discriminator input signal level range (max - min) in percentage of half the total range. For optimal decoding it should be maintained close to 100.
Toggle button to select slot #1 voice output. When on waves appear on the icon. The icon turns green when voice frames are processed for this slot. For FDMA modes you may want to leave only this toggle on.
Toggle button to select slot #2 voice output. When on waves appear on the icon. The icon turns green when voice frames are processed for this slot. For FDMA modes you may want to leave this toggle off.
<h5>B.12 (3): TDMA stereo mode toggle</h5>
- When off the icon shows a single loudspeaker. It mixes slot #1 and slot #2 voice as a mono audio signal
This button tunes the stroke of the points displayer on B.1. The trace has limited persistence based on alpha blending. This is the 8 bit unsigned integer value of the trace alpha blending. Default value is 100.
This button tunes the persistence decay of the points displayer on B.1. The trace has limited persistence based on alpha blending. This controls the alpha value of the black screen printed at the end of each trace and thus the trace points decay time. The value is 255 minus he displayed value using 8 bit unsigned integers.
This is the one side deviation in kHz (±) leading to maximum (100%) deviation at the discriminator output. The correct value depends on the maximum deviation imposed by the modulation scheme with some guard margin. In practice you should adjust this value to make the figure on the signal scope fill the entire screen as shown in the screenshots above. The typical deviations by mode for a unit gain (1.0 at B.18) are:
This is the gain applied to the output of the discriminator before the decoder. Normally this would be set at unit gain 1.0 while the FM deviation is adjusted. However this can be used to extend the range of FM adjustment.
