Re: [Discuss-gnuradio] GMSK at 768kb/sec and other good stuff

[Dawei Shen]

Hi, Eric

Thanks for your detailed and prompt reply. I will try that. At the same time, could you please provide more pointers to understand the GMSK2 code?

If I understand correctly, the code uses the Mueller and Muller timing error detector, right? The output of the Mueller and Muller should be a quantity which has the same sign with timing error and proportional to the absolute value of the error. Then this output is given to the feedback interpolator to get the intermediate value between two samples.

However, I could only understand the general framework so far. Could you please provide more details? Especially which book or paper did you refer to and if possible, could you point out which chapters I should read more carefully and how the codes match them?

It seems to be a boring question. Thanks for your time.

Dawei

On 11/30/05, Eric Blossom <address@hidden > wrote:

On Wed, Nov 30, 2005 at 10:47:22PM -0500, Dawei Shen wrote:
> Hi, Eric and other dudes
>
> Hope you  have enjoyed a nice Thanksgiving.

Thank you, I did.

> We have bought two Flex boards and we are doing the experiments with the
> GMSK2. We do TX and RX using two machines, but at the receiver, we can't see
> anything. So could you please help me clarify a few things?
> 1. There are two SMA connectors on the Flex boards, one is called TX/RX, one
> is called RX2. What are their functions? Should I use the TX/RX one at both
> receiver and transmitter?

The code is currently set up to use the TX/RX port for both transmit
and reception.  It is possible to use the TX/RX port for transmit
only, and the RX2 port for receive only, but there's currently no high
level interface to control this.  Using two separate antenna ports is
useful if you've got an external LNA, power amp, etc.

> 2. What kind of Antennas do you use for the experiments?

You need some kind of antenna appropriate for the 70 cm band.  Matt's
got some available on his web site, or you could try a local ham radio
supply (ask for a 70 cm handi-talkie antenna with an SMA connector), or
you could try sticking a piece of wire into the SMA connector.  The
wire should be approximately 17-18 cm long.

> Do we need to use an external amplifier?

Nope.  The transmitter puts out approximately 100 mW into 50 ohms.

On one machine, try:

  $ ./benchmark_gmsk_tx.py -M 10

On the other one, try:

  $ ./benchmark_gmsk_rx.py -g 30

The two boards shouldn't be sitting right next to each other.
Put them at least 3 or 4 meters apart.

The output on the rx side should look something like this:

    address@hidden gmsk2]$ ./benchmark_gmsk_rx.py -g 30
    RX d'board A: Flex 400 Rx
    RX d'board B: TV Rx
    Using RX d'board A: Flex 400 Rx
    len(rx_chan_coeffs) = 63
    >>> gr_fir_fff: using 3DNow!
    bitrate: 1Mb/sec
    Rx gain_range:  (0.0, 65.0, 1.0)
    ok = True  pktno = 4627  n_rcvd =    1  n_right =    1
    ok = True  pktno = 4628  n_rcvd =    2  n_right =    2
    ok = True  pktno = 4629  n_rcvd =    3  n_right =    3
    ok = True  pktno = 4630  n_rcvd =    4  n_right =    4
    ok = True  pktno = 4631  n_rcvd =    5  n_right =    5
    ok = True  pktno = 4632  n_rcvd =    6  n_right =    6
    ok = True  pktno = 4633  n_rcvd =    7  n_right =    7
    ok = True  pktno = 4634  n_rcvd =    8  n_right =    8
    ok = True  pktno = 4635  n_rcvd =    9  n_right =    9
    ok = True  pktno = 4636  n_rcvd =   10  n_right =   10
    ok = True  pktno = 4637  n_rcvd =   11  n_right =   11
    ok = True  pktno = 4638  n_rcvd =   12  n_right =   12
    ok = True  pktno = 4639  n_rcvd =   13  n_right =   13
    ok = True  pktno = 4640  n_rcvd =   14  n_right =   14
    ok = True  pktno = 4641  n_rcvd =   15  n_right =   15
    ok = True  pktno = 4642  n_rcvd =   16  n_right =   16
    ok = True  pktno = 4643  n_rcvd =   17  n_right =   17
    ok = True  pktno = 4644  n_rcvd =   18  n_right =   18
    ok = True  pktno = 4645  n_rcvd =   19  n_right =   19
    ok = True  pktno = 4646  n_rcvd =   20  n_right =   20

If you get over or underruns on either machine, you'll need to reduce
the computational load.  You can fiddle with the -i <interp> and -d
<decim> command line parameters to reduce the data rate.

[Yes, we ought to have sometime more intuitive like --bitrate 500k]

Eric