Some time ago I had problems with
my ADSL connection. It was stable when my "speed"
was limited by the provider to 2 Mbps. After a service upgrade I had frequent
dropped connections particularly in the evenings. With the new service upgrade
the "speed" was limited only by the characteristics of my phone line
(around to 6 Mbps at that time).
This post describes the solution that I found to reduce the number of connections dropouts
due to the modem sync loss.
I checked through my modem that
the SNR (ration signal/noise) decreased considerably throughout the day, starting
around 6 dB in the morning, 3 dB at afternoon, 0 dB at end of the day and consequently a loss of synchronization.
I thought that if I could
increase the SNR via a connection "speed" reduction then I can improve the situation. As my modem does not allow the
parameterization of the maximum "speed" I tried to reduce my "speed" through an external circuit.
The trick is to "mislead"
the modem making it "think" that my phone line is worse than it
really is. Once the modem is synchronized, I can put the phone line back to the
"normal" state. In this way I could reduce my speed to about 5 Mbps
and have a SNR around 9 dB in the morning and keep a SNR around 3 dB at night
that gives some margin to avoid dropouts.
The working principle of the used
circuit is the following:
When the modem is powered ON with
the switch "S1" in the open position, the modem will be connected to
phone line with the inductor "L1" connected in series. This inductor
causes the modem to synchronize at a lower "speed" (compared with the phone line without the circuit). It reduces the bandwidth of the phone line and consequently the synchronization "speed". At this stage the synchronization
"speed" of my modem is approximately 5 Mbps and the SNR around 6 dB.
When the modem is already
synchronized then we close the switch S1. This switch "eliminates"
the inductor "L1" from the circuit and put back the phone line to its normal state. At
this stage the synchronization "speed" of my modem is approximately 5
Mbps (as before) but the SNR is around 9 dB.
When we close the switch "S1"
and if the "L1" inductance is relatively high, the transition will
be "abrupt" for the modem and we will have a synchronization loss. To
make this transition more smooth I used the following circuit:
L1 = 2 uH
L2 = 2 uH
Note: The inductance values are
approximate and obtained experimentally for my environment. If you want to try this circuit please check what is the best values of inductance for your environment (modem
and phone line). You can start with a high inductance value and after removes turn by turn of the coil.
I added the switch
"S3". It allows disconnect/connect the phone line instead of the
modem power.
The operation sequence is the
following:
1 - All switches OFF
The modem is disconnected from
the phone line. We have a connection loss.
2 - S3 ON
Wait for the modem
synchronization.
3 - S1 ON
The inductor "L1" is
"eliminated" from the circuit.
4 - S2 ON
The inductor "L2" is
"eliminated" from the circuit. The sequence is finished.
We need to repeat this process
whenever the modem is restarted (for example, after a power failure).
The prototype of the circuit was
built with material that already had at home.
Conclusion
With this circuit I reduced the
number of dropouts from around 3 to 5 by day to about 1 or 2 per week. Currently I have another provider
that uses better lines/equipment and my connection is stable at 7 Mbps without
the use of my external circuit.