woensdag 29 april 2015

Low pass filter added.

To do's of the previous post I have worked on:
- Make graphs of the spurious level with normal and shifted oscillator.
- Make a proper low pass filter between oscillator and DVB-T dongle to see if I can reduce noise (the signal into the mixer is not a clean sine wave, as the Si5351a outputs a square wave).
- Make a script to use or alter rtl_power to make frequency plots with spurious-bypass option.

As I mentioned the clock from the Si5351a board has a square wave output.
Here is the signal at the input of the R820t mixer at an 8mA drive level, and without filtering (just the decoupling capacitor and 82 ohm load resistor).

With the help of rtl_power and Matlab (or Excell) you can see the amount of noise you will have with no antenna connected. I've used the exact same command as on the http://kmkeen.com/rtl-power/ page. The antenna input was loaded with a 50 Ohm load.

rtl_power -f 24M:1.7G:1M -g 50 -i 15m -1 noise.csv 

Noise at 8mA drive level, without lowpass filter.
Wow lots of spurious. But bare in mind that I did not alter anything else to the DVB-t dongle yet. No metal case. USB not filtered, etc...

Time to add in a low pass filter.
On the website http://www.wa4dsy.net/filter/filterdesign.html you can model a low pass filter in a second. I've asked for a 3-pole filter with a cutoff frequency of 30MHz and a load of 50 Ohms. If you click the link below you will see what filter was calculated.


Part Values
PartButterworthChebyshev 0.1 DBBessel
L10.2653 uH0.3801 uH0.0895 uH
L20.2653 uH0.3801 uH0.5845 uH
C1212.21 pF169.10 pF102.97 pF

The butterworth type filter had values closest to what I had in my SMD 0805-size assortment I recently bought on Ebay. With two 270 nH inductors and a 220 pF capacitor this is wat I build. Two SMA edge connectors back to back with the components in between. This filter was added between the Si5351a board and my local oscillator SMA-connector on the DVB-t dongle.

And connect it.

Hey the signal now starts to look way more like a sine wave. Still some distortion in there, but that was to be expected. Perhaps I'll make another with more poles in the future, but I'm just experimenting now, I don't have to go all the way yet.

Now let's see if it made any difference in the spurious levels.

Noise at 8mA drive level, without lowpass filter.
Noise at 8mA drive level, filter added.

Yes it did. The spurious levels have dropped significantly. Time to lower the drive level all the way back to 2mA.

Spurious at 2mA drive level with low pass filter.

YES !!! This is the way to go. If you add an external local oscillator to your DVB-t dongle and you want no spurious, you have so make sure that you:

A: Supply a clean sine wave to the R820t
B: Supply the correct signal level to the R820t (to be determined)

Now for all you folks that have soldered a 3.3Volt TCXO directly to your DVB-t dongle. Did any of you do any filtering or level correction? Please answer in the comments.


8 opmerkingen:

  1. hi Milan,

    How I can to reduce level to 2mA, from software or resistive attenuator ?

  2. Software: Register 16: CLK0 Control Bit D1&D0. (CLK_IDRV). See page 30 of the datasheet: https://www.adafruit.com/datasheets/Si5351.pdf

    si5351.drive_strength(SI5351_CLK0, SI5351_DRIVE_2MA);

  3. Great info, you must have good eyes and even better soldering skills.
    I assume "With two 270 Ohms inductors" is a typo ?

  4. Yes, that is a typo. I will change it. Thanks.

  5. Hi Milan,
    Just made a couple of filters using your method and component values, filled the void between the connectors with hot glue to insulate the components then wrapped the joint with copper tape to screen the connection. I get an impressive looking sign wave.
    Thanks for the ideas and info.

  6. Hi Milan,

    We (Tom-W5KUB and myself) are working on vfo for the BITX.
    Tom has the BITX and SI5351 board working but is hearing a buzz.
    IF is 11Mhz.
    VFO(Bfo/Vco) is at 4.8-5Mhz.

    Any suggestions with LPF to use? E.g. which cutoff frequency to use for best results?

    Thanks in advance

    Ron - PA3FAT (Valkenswaard)

  7. Sorry for the late reply. I would aim for a 3dB point at a few megahertz above 12MHz in this case. Suggestion: around 14-15 dB.