Help: problem of v22bis software modem by Motorola SDK 2.0E

Are you using a loop connection, or direct? If direct, you may need to attenuate to obtain decent receive levels. I have seen that V.22bis modems I designed years ago would have problems during testing with high receive levels. Does either modem require loop current to properly terminate the line? Some designs will not operate well without loop current. Also, if you are using your own op-amp duplexer, try to optimize the impedance match.

regards, Darcy Roberts Ag > Hello dear all,

One other thing to check is the accuracy of the clocks. The software version may be correcting for clock slips, but this can be an issue. As you know, there is an async-to-sync conversion done somewhere in the code(and the reverse). I have found if the clocks are out enough, then you can get bursts of wrong characters until the modems synchronize again. Note: this is based on experience with a 73K224 based (hardware) modem design and may not apply to you.

Can you try connecting at an FSK rate (V.21 or Bell 103) ? This may help shed some light on the problem...

Can you configure the software to do an analogue loopback? If the problem goes away, it could be a clock issue.

Is there an AGC block in the software? Can it be disabled?

Is there an adaptive receive filter in the software? Can it be disabled or locked at a certain set of tap values?

regards, Darcy Roberts

Kelv> Thanks for your advice.

Hello Darcy and John,

This is my first modem project and thanks for your enolightening advice. I have tried all the digital and analog loopback tests on my target board. All passed.

There is only one crystal on my target board, which provides the dsp with about 60 MHz clock signal and help codec generate a 7200Hz Frame Sync of the serial communication between dsp and codec. I checked the clock for dsp and the frame sync signals. Both of them were very steady and I did not see anything abnormal.

Shall I change the dsp clock crystal to a more accurate one? I need also have to look at the application code which may cause the timing corrupt? Thanks.

Kelvin

You should strive for a carrier frequency of 1800 Hz

+/- 1Hz or better. The modulation rate accuracy should be +/- 0.01% or better.

Does this CPU use a PLL to generate the 60 MHz? Nudging the crystal a bit might help here to generate a more accurate carrier.

regards, Darcy Roberts

Kelv> Hello Darcy and John,

Correction:

The carrier for V.22bis is 1200 Hz +/- 0.5 Hz (low band) and 2400 Hz +/- 1 Hz (high band). I was thinking of V.32bis carriers...

The data rate is to be 2400 bit/s or 1200 bit/s +/- 0.01% with a modulation rate of 600 baud +/- 0.01%

Start off with a very accurate frequency counter, and get the software to generate the guard tone of 1800 Hz.

regards, Darcy Roberts

Kelv> Hello Darcy and John,

Sorry a mistake in my message:

The crystal is about 3.6864 MHz. The CPU PLL Module generates about

120 MHz for internal buses and 60 MHz for external. It seems that V22bis modem needs pretty accurate clocks for the communication. I will check the spec of the oscillator.

Thanks.