Below the Noise Floor — Episode 10: "Receive Filters Explained"

Receive filters are the single most powerful tool for pulling a signal out of a noisy band - but only if you understand what they are actually doing. This episode covers the fundamentals of digital filters, passband width and shape, high and low cut controls, notch filters, and the audio peaking filter for CW.

Transcript

Copy

SPEAKER_00 0:00

We are now in phase three of the series. The receive and transmit basics are in place. This phase is about going deeper into Ether SDR itself, the controls that separate a workable station from a well-tuned one, and we are starting with receive filters, because filters are the most consistently useful tool on a noisy band and also the set of controls that most new operators either ignore entirely or adjust randomly without understanding what they are changing. Let me start from first principles, because the concept of a filter is simple once you have the right mental model for it. A radio receiver takes in RF energy across a range of frequencies. That energy includes the signal you want and everything else that happens to be present at the same time, other stations on adjacent frequencies, atmospheric noise, local interference, the random background noise that is always there. The job of the receive filter is to pass the frequencies you want and reject the frequencies you do not want. It is a frequency selective gate. In a software-defined radio like the Flex Radio with Ether SDR, this filtering is done in software using digital signal processing algorithms rather than physical components. This has a significant advantage over analog filtering. The filter shape, width, and position can be changed instantly and precisely with no hardware modification. You can narrow a filter by 50 Hz, shift it in frequency, add a notch at a specific point, and do all of this while listening to a signal without interrupting the receive chain. The filter in Ether SDR that you interact with most directly is the receive passband filter, the window of frequencies that the receiver lets through to your headphones. On the waterfall and pan adapter display, you can see this filter represented as a shaded region centered on your receive frequency. The edges of that shaded region are your filter boundaries. Anything inside the boundaries passes through. Anything outside is attenuated. Let me talk about width first, because it is the most important filter parameter. Filter width, also called bandwidth, is measured in Hz or kHz and represents the span of frequencies the filter passes. For SSB voice receive, typical filter widths range from about 1.8 kHz on the narrow end to about 3.0 kHz on the wide end. The default in most SDR software is around 2.4 kHz, which is a reasonable middle ground for voice. A wider filter, say 3.0 kHz, lets more of the audio spectrum through. The voice sounds fuller and more natural. High frequency consonants like S and T that live at the upper end of the voice audio range come through clearly, but a wider filter also lets in more noise from the parts of the band adjacent to your signal. If there is another station two or three kHz away, a wide filter may let some of their audio bleed into yours. A narrower filter, say 1.8 kHz, cuts out more of the surrounding noise and interference. On a crowded band where stations are close together, a narrow filter can meaningfully improve the readability of a weak signal by reducing the adjacent station noise that was competing with it. The trade-off is that a very narrow SSB filter starts to sound muffled. The high frequency consonants that were clear at 2.4 kHz begin to disappear at 1.8, and below about 1.5 kHz voice audio becomes noticeably unnatural. In Ether SDR, the filter width can be adjusted in several ways. You can click and drag the filter edges directly on the panadapter display. The shaded region has handles on its left and right edges that you can grab and move. You can use preset filter width buttons that jump to predefined values, or you can enter a specific width numerically. The visual drag method is the most intuitive because you can see exactly how the filter edges relate to the signal you're receiving. Now let me talk about the individual filter controls that make up the full filter set in Ether SDR. The high cut filter, sometimes called the high frequency cutoff, sets the upper boundary of the passband. For SSB voice, this is typically set somewhere around 2.8 to 3.0 kHz above the carrier. Moving the high cut lower narrows the filter from the top, which reduces high frequency noise and adjacent high side interference, but also cuts the upper end of the voice audio. Moving it higher widens the filter upward, which passes more audio and noise from that direction. The low cut filter sets the lower boundary of the passband. For SSB voice, a low cut around 100 to 200 Hz is typical. This removes very low frequency hum, power line interference, and the boomy low-end audio that does not contribute to voice intelligibility. Moving the low cut higher, say, to 300 or 400 Hz further cleans up low frequency noise but can make voices sound thin. Moving it lower toward zero passes more bass and more low frequency noise. Together, the high cut and low cut define the passband window. For voice, think of the low cut as cleaning up the bottom of the audio and the high cut as setting the top. A typical clean voice setup might be low cut at 150 Hz and high cut at 2800 Hz, giving a 2650 Hz total passband. On a noisy band, you might tighten to 200 Hz low cut and 2400 Hz high cut. On a quiet band with a strong signal, you can open it wider. CW operation, Morse code, uses very different filter settings. A CW signal is a single tone that turns on and off. The tone frequency relative to your receive frequency is called the side tone pitch, typically set somewhere between 400 and 800 Hz depending on operator preference. Because a CW signal is a single tone rather than a voice audio range, the appropriate filter width for CW is much narrower, often 200 to 500 Hz. A 500 Hz CW filter centered on the expected tone frequency passes the signal and rejects almost everything else on either side. On a very crowded CW band, you might go as narrow as 100 Hz, which is still wider than the CW signal itself, but tight enough to reject most adjacent signals. Ether SDR has CW-specific filter presets that set appropriate width and center position for CW reception. When you switch to CW mode, the filter defaults change automatically to something reasonable for Morse code. The audio peaking filter, sometimes called the APF or audio peak filter, is a CW-specific tool that deserves its own explanation. Rather than cutting frequencies outside a passband the way a band pass filter does, the APF boosts a very narrow frequency range, typically just a few Hz centered on the CW tone. The result is that the tone you are listening for becomes louder relative to the surrounding noise, improving its perceived readability even in difficult conditions. The APF is most effective when combined with a moderately narrow bandpass filter. Used together, the bandpass filter removes most off-frequency noise and the APF makes the remaining on-frequency tone more prominent. In Ether SDR, the APF control lets you adjust the center frequency of the peak and the Q factor, which controls how narrow and sharp the peak is. A high Q setting produces a very narrow and sharp boost. A lower Q produces a broader, more gentle peak. For most CW operating, a moderate Q setting that you tune to match the sidetone pitch of the station you are copying works well. Now let me talk about the notch filter, which is a different kind of tool. While a band pass filter passes a range of frequencies and rejects everything outside, a notch filter does the opposite for a single frequency. It creates a narrow rejection at a specific point in the passband while passing everything else. This is useful when you have a single interfering carrier, a steady tone, sitting inside your receive passband. A heterodyne whistle, a carrier from an adjacent station, a birdie from your own hardware. Any steady tone that lands inside your filter can be removed with a notch filter without significantly affecting the rest of the audio. In Ether SDR, the manual notch filter lets you place a rejection notch at a specific audio frequency within the passband. You listen for the interfering tone, identify its approximate frequency by ear, or by looking at the audio spectrum display if Ether SDR shows one, and place the notch there. The tone disappears or is significantly reduced without affecting the rest of the voice or digital audio passing through. Ether SDR also has an automatic notch filter that attempts to detect and remove steady tones automatically without manual adjustment. The automatic notch works reasonably well for single-strong carriers and is a useful default to have active during general operating. It can occasionally suppress something it should not, but it is generally a net positive for voice receive on a noisy band. One thing to understand about notch filters, they work on steady tones, not on voice or noise. A competing voice signal from an adjacent station cannot be notched out because voice is not a steady tone, it occupies a range of frequencies that changes constantly. The notch filter removes what is tonal and passes what is not. Let me close with some practical guidance on how to approach filter adjustment in Ether SDR when you sit down for a receive session. Start with the default preset for your mode. Ether SDR's defaults are reasonable starting points, typically a 2.4 kHz SSB voice filter, a moderate CW filter for CW mode. Listen for a minute and assess the conditions. If the band is quiet and signals are strong, leave the filter alone. If you are having trouble copying a specific signal because of adjacent interference, try narrowing the filter in small steps, a couple hundred hertz at a time, until the interfering signal reduces without making the target audio sound unnatural. If there is a steady tone interfering, enable the manual notch and place it on the tone. If low frequency hum or rumble is present, move the low cut up by 100 Hz and see if it helps. The most common mistake with filters is overadjusting. A filter that is too narrow for the mode makes voice audio muffled and fatiguing to listen to. A filter that is too wide on a crowded band makes weak signals harder to copy. The goal is the minimum width that lets you hear the target signal clearly given the actual interference present, not the minimum width possible. Filters reward a little patience. Set one parameter, listen, assess, then decide whether to adjust further. Chasing a marginal signal by progressively narrowing the filter often works. Randomly adjusting multiple controls simultaneously rarely helps and makes it hard to know what actually improved things. Next episode, we go deeper into two more receive tools noise reduction and AGC. These work differently from filters and are often misunderstood, especially the relationship between them and when each one is the right tool for the conditions you are facing. This is below the noise floor.