SDR IF Experiments

The R820T tuner IC is used in the popular Airspy software defined radio (SDR) as well as many of the inexpensive RTL SDR dongles. [TLeconte] did some experiments on intermediate frequency (IF) configuration of the chip, and you’ll find his results interesting.

Using 5 million samples per second and the device’s real mode, the tests look at a what comes out when the IC reads a noise source. There are two registers that set the IF parameters, but the tests show the effects these registers have in precise terms.

According to the post, there are three things you can set:

  • Coarse IF filter bandwidth : narrow/mid/large
  • Manual fine tuning IF filter bandwidth from 0 (large) to 15 (narrow)
  • High pass filter frequency from 0 (high) to 15 (low)

Some of the settings don’t make sense — at least at the 5 MHz sample rate — because of aliasing. However, it is instructive to see what each setting does. [TLeconte] uses Octave to visualize the data.

If you want to know more about SDR in general, we have something to get you started. If you want to do your own testing, consider using GNU Radio.

23 thoughts on “SDR IF Experiments

  1. I just got into SDR a few months ago after reading about it here. It is great fun with very low cost as you can make your own antennas. I have been listening to commercial and military flights and I am shortly going to try receiving trunked transmissions from our local first responders, which takes 2 dongles so it gave me an excuse to get a second one.

  2. The first thought I had after reading this article was that someone realized that they hadn’t posted enough articles today, and said “Well, there’s 5 minutes left before we have to go, let me whip something up” and after writing the article spent the remaining 3 minutes pulling their vape pen out of their ear…

    It used to be, that when looking to hackaday you could count on meaningful content, with commentary about the outcome, and DETAILS.

    Soon after, this place became arduino central, and I was ok with that, because, well, it was still the same…. details, commentary, and pictures.

    Today: SDR IC photo… You might be interested…. Skip the details… Done.

    If things keep moving in this direction though, eventially, some article writer will just post the word “HACK” with a picture of a hatchet, duct tape, and some toothpicks, and that will be it for the day.

    This is hackaday.
    not Slackaday.

    1. look at hackaday posts from the very start. they put in lots of effort to bring lots of stuff to our attention.instead of 1 hack a day we now get minimum 5 hacks. Also hatchet, duct tape and toothpicks would make a good hacking competition

    2. ” “Well, there’s 5 minutes left before we have to go, let me whip something up” and after writing the article spent the remaining 3 minutes pulling their vape pen out of their ear…”

      Yeah, but this was by Al Williams, so I seriously doubt that applies in this case.

  3. And yet, people talk about “SDR receivers” but often it’s black box thing, a cheap wide range receiver using off the shelf software. ‘Software defined” only matters in terms of flexibility. They aren’t interested in building receivers, or understanding them, they just want to get some obscure frequency for some specific need.

    Yes, lots of people used to build regen or other simple receivers from articles in the magazines. And a relative few were very interested in receiver design, not just “good” receivers but great receivers. I don’t build them, but I’ve certainly been interested in good receiver design. SDRs have been around for a while, but mostly as expensive top end receivers, maybe fully digital but more likely a hybrid, analog to the point where digitizing could take over. Now SDRs are everywhere, but not that much talk about them. There are consumer radios, sometimes including shortwave, which use dedicated SDR ICs, but not a lot about how flexible they are (how inflexible they might be), What we are really seeing is consumer electronic devices, like those cheap DTV USVPB adapters, or cellphone ICs, stretched into something more. Obviously they work because they are SDR, but are they the right choices for good receiver design? You could have turned a tv set into a shortwave receiver in the old days, but that doesn’t mean it would have been a good SW receiver. What’s a good compromise between great receivers and limited cost hardware?

    The TV ones are “great” since they cover a lot of range, spectrum that often isn’t so accessible to many hobbyists. But for other hobbyists, they don’t cover useful range, ie the shortwave bands. Sure, you can modify them, but there’s fallout.

    I shudder at what seems to be bad receiver design, wondering if I’m missing details because they turn to digital. Wide bandwidth is seen as a good thing, “look at that display of the band”, but good receiver design puts some filtering close to the antenna.

    I could go on, but instead, I remember when SDR was “the future”, a wonderful device that would eventually trickle down. But it was presented in terms of flexibility, design then becoming the software you write for it. But now, it’s generally off the shelf software, people not interested in inventing new modes (done easily because the receivers can be adapted via software). Or even just get incredible filters because of the software, so cheap once you have the hardware compared to a bunch of hardware filters (often optional in commercial equipment to save cost) or more complicated design so the filters could be simple LC filters.

    A long way to say we need more tutorials about both the hardware and the software. This isn’t a great one, but it hints at that direction. The datasheet for the IC would reveal similar findings, It’s for tv, the different bandwidths is relative, wide as tv has always been. Once you have information, you can improve, Take or discard mass consumer ICs, and improve on them, and lets see some of it being about building SDR receivers, rather than adapting off the shelf consumer equipment. Or software to get that high performance.

    But then for most people “radio” is the local AM or FM broadcast stations


  4. “Some of the settings don’t make sense — at least at the 5 MHz sample rate — because of aliasing. However, it is instructive to see what each setting does. [TLeconte] uses Octave to visualize the data.”

    Now, this makes me wonder: what settings? What does not make sense? Why? “Because of aliasing”??? Could just as well gone full Terry pratchett and proclaimed “its probably beacuse of quantum”.

    “The Discworld, of course, operates on magic. Magic flies the Turtle, spins the Disc and reclaims the Rimfall; it’s natural and normal and unremarkable. What sometimes perplexes people is the odd way in which things sometimes happen without apparent magical influence. These inexplicable phenomena are given the collective name quantum.”

      1. Did you read the posted article? At some point, you wind up over 2xf for 5 MHz so you get aliasing. As far as the writer understanding, first are you kidding? Have you read some of his other stuff? I’m going to guess he understood it fine he just didn’t explain to you what aliasing meant. Second I have noticed this with a lot of the short articles here. At the end of day they are pointing out some other article or video for us to go look at. Do you really wnat/need them to completely summarize the OP for you? A fine line but the whole point is hey this is interesting go read it. And if you did, all your questions were answered if you knew how to understand them. I wish HAD had some comment voting or even commenter voting. I was looking around this morning and a lot (not this one) people add links that are already in the post, so it is clear a lot of commenters don’t read the whole writeup or like this one they might read it but they don’t read the original one.

  5. The R820T2 (and R820T which was never used in any of the testing, if an Airspy was used) are EOL parts. What would be interesting would be to carry out similar testing on the current production tuner (which has different registers) the R836 (ref: ).

    “Some of the settings don’t make sense — at least at the 5 MHz sample rate — because of aliasing”
    I would see this statement as wrong, if the coarse IF filter bandwidth is 2MHz or less, then 5MSPS sample rate is perfectly good.
    The R820T2 has IF bandwidth that can be tuned from 1 to 15 MHz –

    My suggestion would be to carry out additional tests using more sample rates (none of which are officially supported, from what I can see):
    12MSPS IQ
    6MSPS IQ
    4.096 MSPS IQ

  6. I don’t see any explanation of how these tests were run from an RF system point of view, just: “…we will use a noise source directly connected to the airspy and airspy_rx to save the received data to a file in raw real mode…” What the heck does that mean? No block diagram, no source stepping to check linearity, no single and dual frequency (intermod) tests, nothing! The plots are pretty, but don’t trust anything you see without more (missing) details about how these tests were done. How do we know the noise generator is even matched to the DUT input? Are there at-least passive in-line resistive attenuators to swamp mismatches (better yet an isolator)? In-general using a noise source instead of a swept single frequency source without exercising great care can lead you down a rabbit hole filled with bad data. Maybe the Author of the study did take care with the measurements, but where are the details? Or is this just another case of “Makers” going out of control due to ignorance.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.