Why isn't this filter filtering?

Discussion in 'Homebrew and Kit Projects' started by KE0VTF, Aug 15, 2020.

ad: L-HROutlet
ad: l-rl
ad: L-Geochron
ad: abrind-2
ad: L-MFJ
ad: Left-3
ad: Left-2
  1. KE0VTF

    KE0VTF Ham Member QRZ Page

    Forgive me for what must be a very sad attempt at a 14.6 MHz low-pass filter, it's my first one ever.

    It should have what it needs. There's a 330-ohm resistor in series, and a 33-pF capacitor to ground. But I'm not seeing any attenuation above 14.6 MHz. I've quadruple-checked my math & the circuit diagram, and I made sure it was facing the right direction, but it's still not filtering anything. How have I screwed up something so theoretically simple?

    Some background and details: I'm using an RTL-SDR to look at the HF bands. When it's using direct sampling, it mirrors the spectrum across 14.4 MHz (I learned this here: https://forums.qrz.com/index.php?th...g-noise-around-my-signal.719114/#post-5561603). So, I want to get rid of those signals above 14.4 MHz so they don't mess with my waterfall when I'm looking at anything at 20 meters and below. This filter is attached to a 20-meter dipole, with the other end in my SDR. I'm looking at WWV to see if the filter is attenuating anything above 14.6 MHz, but WWV is coming in as strong as ever, despite the filter.

    I'm also aware that I'm not using an ideal kind of resistor; these were all I had. I have some carbon composition resistors on the way, which I'll use to upgrade this if I can figure out what my mistake was.

    Attached Files:

  2. N5HXR

    N5HXR Ham Member QRZ Page

    I'm another newb, so I only know so much -- I hope someone will correct me if I say anything too stupid! I was curious how it would look if I simulated it. LTSpice suggests you're getting a really gentle slope, with less than 1dB attenuation between 13MHz and 16MHz (goes from about -2.5dB to -3.4dB). Looks like the -3dB point is about 14.6MHz. This result matches this online calculator, so I feel kinda' confident I did it right :).

    If you could reliably compare a signal at 13MHz to one at 15MHz through your filter, you'd see less than 1dB difference in attenuation from this filter, which means they both still show up really strong on the waterfall. It is filtering, just not much!

    You'll want a higher order filter that gives you a sharper drop off as you go up in frequency. You'd also probably want to make it an LC filter, and avoid having any resistive elements so you don't pay unnecessary attenuation cost across the board.

    Last week, I used this filter calculator to make a low pass filter for the 40m band, which has a much steeper drop-off. I set the "number of components" to 5, which means there are 5 reactive components working together. It ended up looking like this when I was done (those are T37-6 toroids for the inductors, which I think I found on Amazon at 10 for about $5):

    N0TZU, KE0VTF and KP4SX like this.
  3. KE0VTF

    KE0VTF Ham Member QRZ Page

    I see, that's super helpful! I'll give this a try with more components to make the slope sharper. Thank you so much for linking those calculators!
  4. KL7AJ

    KL7AJ Ham Member QRZ Page

    What is your load resistance?
  5. AI5DH

    AI5DH Ham Member

    With your construction technique all you need to do is lengthen the piece of twisted wire you call Ground. Additionally you can make the components leads longer. In no time it will not pass any HF
    Last edited: Aug 15, 2020
  6. N5HXR

    N5HXR Ham Member QRZ Page

    No problem. Before coming to ham radio, I'd done a lot of digital electronics hobby work, and feel pretty comfortable with it, but the analog side of things has been very mind-bending. Filter design is a deep well to fall into! Good luck with your experimentation.
    KE0VTF likes this.
  7. N5HXR

    N5HXR Ham Member QRZ Page

    Looks like the input impedance for an RTL-SDR is said to be 75 or 50 ohms, depending on front-end tuner and configuration. I think the direct sampling hack bypasses the tuner, though, and goes right to the input on the RTL2832, which seems to be something like 3,300 ohms. So if I'm understanding the post right, I guess it's that latter high impedance. (assuming I'm reading those articles right)
    Last edited: Aug 15, 2020
    KE0VTF likes this.
  8. VK4HAT

    VK4HAT Ham Member QRZ Page

    A single pole RC filter like that is only going to provide -10db of attenuation per decade or something like that, this means the slope of the filters shape is very low, you might not notice anything at all close in from a very strong signal. A decade is 10x the frequency, so maximum attenuation will be reached at 140mhz. for a 14mhz lpf.

    If you look at this LTspice simulation you can see that the maximum attenuation at 30Mhz is only 0.3 of a dB. In other words nothing that you will notice in an SDR.

    For what you are trying to do, a double tuned bandpass filter for 20m would be a better option. Actually, that is what I hang off the front of my SDR's to give them some reasonable front end.

    Last edited: Aug 16, 2020
    WB5THT, N1OOQ and KE0VTF like this.
  9. KE0VTF

    KE0VTF Ham Member QRZ Page

    This is all so helpful. In the future I'll make sure to get the slope of any filter I build. And I'm going to look into that double-tuned band-pass filter.
    N5HXR likes this.
  10. VK4HAT

    VK4HAT Ham Member QRZ Page

    I did not have time to elaborate yesterday, so i knocked these up for you while im sitting here having my morning coffee. The software for designing these kinds of filters is called DTC by W7ZOI it came with the book EMRDF, there is a copy of the software on my blog. I added some typical values to it and then simulated the results in LTspice. You can see that now 2Mhz away there is at least -30db of attenuation, which is reasonably good. Its a simple filter to design and build and gives very good results. The other benifit of using a bandpass filter here is that it knocks the low side frequencies down also, especially am broadcast that also can play havoc with the front end of many SDR's. Hope that helps some.

    dtcitcuit.png dtcitcuit_2.jpg
    WA7ARK, PU2OZT, KE0VTF and 1 other person like this.

Share This Page