server HPSDR hardware server
The HPSDR server handles communications with the HPSDR hardware and is designed to work with multiple receivers either on a single Mercury card or multiple Mercury cards. It demultiplexes the I/Q samples into separate streams for each receiver. The hardware server listens on port 11000 for TCP connections from clients. The client (dspserver) makes a connection request through port 11000, and sets up a TCP socket on on port 12000+rx, on which it sends commands to the server. IQ data is sent over UDP.
Commands to server
- attach rx
- attach client to receiver rx
- The receiver number rx is an integer from 0 to 3
- example: attach 0
- detach rx
- detach client from receiver rx
- example: detach 0
- frequency f
- Set attached receiver frequency to the integer frequency f (Hz)
- example: frequency 7056000
- start iq port
- start sending I & Q samples for the attached receiver to the UDP port specified
- example: start iq 13000
- start bandscope port
- start sending bandscope samples to the UDP port specified
- example start bandscope 12500
- stop iq
- stop sending I & Q data to the attached receiver
- stop bandscope
- stop sending bandscope data
The server sends a response to the client after parsing each command. The response is either OK or ERROR. Some responses include a short message: The response to an attach command includes the current sample rate.
A typical exchange would be:
|attach 0||OK 96000|
|set frequency 7056000||OK|
|start iq 13000||OK|
|set frequency 7056100||OK|
The I/Q samples are accumulated in the server until there are 1024 I/Q samples to send. The samples are sent in 512 byte UDP packets, to the port specified in the 'start iq' message, with 12 byte header and up to 500 bytes of I/Q samples.
The UDP packet header contains:
- 8 byte sequence number
- 2 byte offset within this set of samples
- 2 byte length of samples
Note that there are currently no commands to control setting the number of receivers, preamp, dither, random, samplerate or clock sources. As this was developed to support multiple receivers, I decided to not allow clients to change these as they would be changed for all receivers. They are all controlled by command line arguments when running the server.
server command line options (default value)
- --receivers <1..4>
- --samplerate [48000|96000|192000]
- --dither [off|on]
- --random [off|on]
- --preamp [off|on]
- --10mhzsource [atlas|penelope|mercury]
- --122.88mhzsource [penelope|mercury]
- --micsource [janus|penelope]
- --class [E|other]
./server --receivers 4 --samplerate 96000 --dither on --10mhzsource atlas
Note: As an experiment, there is also a softrockserver that supports the same protocol with the restriction that there is only one receiver and it uses the USB/si570 interface to control the frequency. The I/Q samples are taken from an Audio device specified on the command line.
dspserver is a client to the hardware server. In turn, it is the server for another client, typically a GUI, such as jmonitor. It provides a dsp processing service using DttSP. Typically one dspserver is started for each receiver. The dspserver listens for client TCP connections on port 8000+rx. i.e. rx0 listens on 8000, rx1 listens on 8001, etc. The client sends commands to the dspserver and dspserver sends data packets to its client on this same port. The first byte of each packet sent to the client identifies the type of data in the packet.
- 0 spectrum data
- 1 audio data
Commands to dspserver
- getSpectrum points
- Request points of spectrum data. The value sent for each point is an 8 bit unsigned integer. If the calculated spectrum contains more points than the client requests, adjacent points are combined, and the highest value in the group is sent.
- Example: getSpectrum 480
- setFrequency f
- Set the receiver frequency to f, in Hz
- example: setFrequency 7056000
- setMode mode
- mode is an integer specifying the type of demodulation
- 0 - LSB
- 1 – USB
- 2 – DSB
- 3 – CWL
- 4 – CWH
- 5 – FM
- 6 – AM
- 7 - DIGU
- 8 - SPEC
- 9 - DIGL
- 10 - SAM
- 11 – DRM
- example: setMode 0
- setFilter low high
- Set filter cutoff frequencies
- For LSB, both frequencies are negative
- For USB, both frequencies are positive
- For AM or DSP, typically low is negative and high is positive
- example: setFilter 150 3450
- setAGC n
- Select AGC time
- 1 - LONG
- 2 - SLOW
- 3 - MEDIUM
- 4 - FAST
- example: setAGC 3
- setNR n
- Set noise reduction
- 0 - off
- 1 - on
- setNB n
- Set noise blanker
- 0 - off
- 1 - on
- setANF n
- Set automatic noise filter
- 0 - off
- 1 - on
- setRXOutputGain gain
- Set receiver audio output level to the value gain
- The value is a floating point number in the range 0 to 1.
- example: setRXOutputGain 0.5
- startAudioStream buffer size
- start sending audio data.
- buffer size is optional, and is an integer, typically 480.
- If omitted, a value of 480 is used.
- stop sending the audio data.
data packet format Each packet has a 48 byte header.
|0||Packet type: 0=spectrum data; 1=audio data|
|48-end||spectrum or audio data|
Sample rate and Meter reading numbers are in ASCII characters.
Audio data is sent as one channel of 8-bit aLaw samples at 8000 samples per second.
dspserver command line options (default value)
- --soundcard SANTA_CRUZ, AUDIGY_2_ZS, MP3_PLUS, EXTIGY,
- DELTA_44, FIREBOX, EDIROL_FA_66, HPSDR
- --receiver 0, 1, 2, or 3
- --server IP address 127.0.0.1 (this is "localhost")
- --offset ???
summary of port usage
|dspserver||TCP||11000||connection request to server|
|dspserver||TCP||12000 + 2*rx||commands to server|
|server||TCP||12000 + 2*rx||responses to commands|
|server||UDP||13000 + 2*rx||I & Q data to dspserver|
|dspserver||UDP||15000 + 2*rx||demodulated audio to server for Mercury headphone/speaker output|
|dspserver||TCP||8000 + rx||spectrum & alaw-encoded audio to monitor|
|monitor||TCP||8000 + rx||commands to dspserver|