Circuit Filtering to Separate Legacy/TMCC and DCS signals

A lot of folks asked me about if it's possible to filter the 455KHz and 3.25 MHz DCS signal apart so I wanted to write a post with some good facts for everyone.

The short answer is not really.

Details:

 The DCS signal has a bit rate of 3.25 MHz but if you look at the actual spreading code MTH uses: "0100010111110110011100001101010" you can see there's a repeat of five 1s in a row meaning there is frequency content at 3.25MHz/5, or about 650 KHz. As this is very close to 455 KHz it means it's tough to build a filter that can reasonably reject one signal without affecting the other. To illustrate this I did some examples for everyone. In these I used the real DCS spreading code and then ran in through various filter configurations in ADS...

Example of Notch Filters:

Let's start with a notch filter. So the idea would be if we "notch out" 455 KHz the Legacy Carrier should go away leaving us with just the DCS.

Let's start with a sad filter that has QF=3 (QF = Quality Factor, describes how sharp the filter is). A Q of 3 is bad, it's typical of what you'd get if you hand-wound the coil.

So you can see the amplitude is sad, and the DCS is pretty distorted after the filter. 

Moving on to QF=20 (typical of a discretely implemented filter)...

The amplitude is better but the distortion is still pretty serious.

Finally a QF=100 is a pretty good quality filter with expensive parts. Once you get above QF=100 it becomes very very difficult to get the center frequency you want to align well.

Anyways the DCS signal still looks pretty awful.

Example of High-Pass Filters:

The second thing to try is to setup a high-pass filter that passes DCS while chopping out the legacy signal. For this I set the corner frequency up right on top of the legacy signal (450 KHz) so to minimize the effect on the DCS.

Here is a simple 1 pole filter (1st order RC)

The DCS looks sort of okay, but not great, some amplitude is lost, but it also only gets you about 3dB (factor of 2 rejection) on the legacy signal so I don't see it as being too useful.

Lets try a 2 pole system and move the second pole out a bit to get 10dB ( a factor of 10X) rejection at the legacy frequency since that would be more useful practically.

Now the DCS is really not looking so great.

Summary:

Although DCS has a bit rate of 3.25 MHz it has frequency content at much lower frequencies as bits within the code repeat. This content (down to ~600KHz) is very close to the legacy frequency (455 KHz) and so practical filters to separate the two signals are not easily or affordably realizable.