Listen To 64 MHz At Once

We imagine that if [Tech Minds] told us he was listening to the HF bands, we might ask him which one? His reply might just be “All of them.” That’s thanks to the RX-888 MKII SDR he reviewed which delivers a 64 MHz window on the radio spectrum. You can catch the video review, below.

These are not especially inexpensive, but with that bandwidth and 16-bit resolution, it is worth it if you need that kind of horsepower. There is a separate input for VHF signals 64-1700 MHz where the bandwidth is only 10 MHz, but still.

Of course, making a very wideband front end for something like this is non-trivial, so we wonder how the performance is compared to similar-priced units with less bandwidth. On the other hand, it does seem to work well enough in the video. The software used limited the test to a 32 MHz bandwidth, which is still plenty.

Speaking of software, we noticed that the developers of SatDump and SDR++ are not happy with the state of the software for the RX-888. We aren’t sure if this remains a problem, but the device seemed to work well on the video, at least.

There are many options now when it comes to higher-end SDRs. We like the Pluto for both transmitting and receiving. Of course, the RTL-SDR kind of started everything with hobby SDR, but you can’t expect that much bandwidth with one of those.

27 thoughts on “Listen To 64 MHz At Once

    1. It would be more useful to add a quote from the video, though it’s really just citing from the AliExpress listing:

      Dual RF input, HF frequency range: 1kHz-64Mhz, maximum real-time bandwidth 64M.
      VHF frequency range: 64M-1700Mhz, maximum real-time bandwidth 10M.

      So it’s not always 64 MHz, just for the HF range.

      For people who want to make their own designs, or just need a good high frequency ADC, the used component is also interesting:
      LTC2208 16bit ADC at 130 MSPS

      1. At 13:00 he says that SDR console can just handle 32 MHz while the specs say 64 Mhz on HF. He does not explain why, if it’s a driver limitation, a SDR console only limitation, or anything else. I didn’t listen to the whole video when he was skimming all those radio stations however, so maybe he said more.

        But it shouldn’t be burried somewhere random in the video.

        1. If I had to guess…
          A really easy way to make I and Q data is to simply send alternating samples to each channel. This can be compared to mixing with a frequency exactly one quarter of the ADC clock. Those I and Q channels could, in theory, have 64MHz of bandwidth. In practice, the low pass filter means you can do a little downsampling before shoving data through the USB.
          If this is the case, you could (in theory) oversample 32MHz of bandwidth around a centre frequency of 161MHz.

  1. My first “Oh, that’d be neat for…. ” was to capture say 100-140MHz and listen to all the aviation chatter on there.
    When watching videos about aviation incidents you often get: “Audio on 132.7 not captured, we’ll hear them again on tower frequency”.
    That would be fixed in one swoop if these SDR devices were used and the raw data was kept for a few days or weeks to be able to tune in after the fact.

    Anyway. 10MHz bandwidth up to 1700MHz cancels all that. :-(

  2. I wonder how good it would be at generating random numbers.
    Johnson–Nyquist noise floor with 64 MHz of bandwidth is -95.7 dBm
    Johnson–Nyquist noise floor with 32 MHz of bandwidth is -98.7 dBm
    (ref: )

    From the linked document on google “Something else that bothers me is that SDR seems to be popular in the SWL community. A bunch of people recommend it when the performance can only be described as mediocre. Making a wideband HF frontend is an art, and you’re not gonna get any good result from something built down to a price like it. It’s a cool ham radio project, but not something that can be marketed as a commercial SDR. I’ve seen people claim that it has superior performance to Airspy and SDRplay SDRs, which is complete *censored*. ”

    The wider you open a window the more bugs will fly into your house! There is a reason that high quality RF receivers have a selection of band pass filters or a tracking filter in their front end.

    1. That’s a common fallacy. Provided the front end is linear(*), you can digitize wide-band and do all the filtering you want in the digital domain. You can implement much better filters there than are practical in the analog domain.

      (*) That’s the sticky part. A strong local signal, say a broadcast AM transmitter, can cause intermodulation products even in very good, linear front ends. A decent wideband front end SDR, like (e.g.) SDRPlay’s RSPdx, has selectable front end filters for this purpose.

      1. The funny thing is I would not consider a SDRplay to be a “decent wideband” receiver. Because at its maximum sample rate, the ADC is only 8-bit (In theory it would always be limited to the maximum throughput of a USB 2.0 High Speed port ~40MB/sec, but the silicon chip used, Mirics MSi2500, is unable to get anywhere near saturating the link fully. It is limited to 14-bit if sample rate is 2 to 6.048MSPS; 21.168MB/sec, 12-bit if sample rate is 6.048 to 8.064 MSPS; 24.192MB/sec, 10-bit if sample rate 8.064 to 9.216 MSPS; 23.04MB/sec, 8-bit if sample rate is 9.216 MSPS to 10.66MSPS; 21.32MB/sec). And then depending on the signal level you may need to attenuate your input enough to remain within the devices dynamic range (14-bit: ~86 dBFS; 12-bit: ~74 dBFS; 10-bit: ~62 dBFS; 8-bit: ~50 dBFS).

        I’m not saying the hardware is bad, just that I would not consider wideband where it shines.

        1. Honest question: which SNR are you trying to achieve over the wide band? Or about which bandwidths would you care if you look at SNR?

          8 bits would usually be called “plenty”; as you say: without any processing gain, that’s about 50 dB QSNR; if your *whole-observed-band* SNR is worse than 50 dB, then you don’t have any trouble with the bit depths.

          It gets better: there’s no signal that spans the whole observable band. Instead, you’d extract individual transmission bandwidths, which is a filtering operation, and in itself leads to processing gain that gives you actually more bits per sample, and an SNR boost (uncorrelated noise vs correlated signal being added up coherently in a filtering operation). This literally means that you can lift signals out of what would be quantization noise floor. Note how this contradicts your claim you get 50 dBFS dynamic range. Typical application of this are by the way ubiquitous GPS receivers, which have 4 bit (maybe by now even less) ADCs, and still receive signals with a power density lower than that of thermal noise. Processing gain is real, and we mustn’t forget it when we do an end-to-end description of what our systems can do :)

          1. There is a special application of listening wide bandwidth at once: solar radio bursts monitoring. 8 bits is enough but at least 1000 channels/second on a 80-100MHz or more bandwidth is needed. The solar burst cover many octaves, have a frequency drift and fine structure you can see in the spectrogram (real time FFT is needed). Is this receiver capable? TNX. 73 from Romania.

      2. (Disclaimer: I know zilch about topic, so forgive me) can’t intermodulation products be predicted and canceled in processing, if the front end transfer function has been characterized and modeled in the system?

        1. The problem is that you don’t know before where your signals are, so you cannot actually predict and subtract them.

          Generally, any even harmonic mathematically erases information, which you can not get back. :(

    2. You can do an excellent direct sampling receiver, or an excellent heterodyne receiver, or an excellent homodyne receiver – depending on your application’s needs. Frontend design is *always* the challenge when dealing with weak analog signals. So, that part’s a rant, not an argument :) It’s OK to rant sometimes!

      > The wider you open a window the more bugs will fly into your house! There is a reason that high quality RF receivers have a selection of band pass filters or a tracking filter in their front end.

      That analogy is not very helpful. Whether you filter in analog or in digital makes absolutely no difference, assuming the amount of things on the outside doesn’t affect the fidelity of the analog to digital conversion.

      We can actually quantify that, quite easily. Unless you can show me your receiver actually runs into nonlinear behaviour due to too much noise, or that your ADC clips, or doesn’t have enough number of bits to represent the SNR your signal has *including* the processing gain of filtering in digital domain, then you’re right, you need to reduce the analog bandwidth. Otherwise: be careful to not convert an easy-to-misunderstand rule of thumb into a technological rumor.

      1. ah that last sentence was missing an negation. Should have read:

        Unless you can show me your receiver actually runs into nonlinear behaviour due to too much noise, or that your ADC clips, or doesn’t have enough number of bits to represent the SNR your signal has *including* the processing gain of filtering in digital domain, then you’re not right, you don’t need to reduce the analog bandwidth.

        What I’m trying to say is that while it’s good to know that having a narrower analog frontend reduces the noise and dynamic range your digitization stage has to deal with, it’s also important to remember *why* we’re doing it, and that there’s a concrete problem that solves. If the problem is not there, there’s nothing you need to solve.

      1. > Couldn’t you attack that by placing a wide-band transmitter.
        Yes and no.
        The quality out randomness out of any radio is low quality (I’ve fed samples into the dieharder test suite – it is far below abysmal for cryptographic use – well any use really), so you would need to feed it through a hashing algorithm anyhow.

        Oh, and I never explicitly said anything about connecting an external antenna, I could use a 1 megaohm or even a 1 gigaohm resistor inside an oven (500 kelvin ; ~227 °C ; ~440 °F) inside a Faraday cage as my input. (64MHz bandwidth would be a RMS voltage of to 0.0013 volts or ~0.042 volts respectively for megaohm and gigaohm).

    3. Your point is valid, but in todays world, outside of america, few live next door to a powerhouse transmitter. In my neck of the wood the closest sender is 200+ km away and its transmitting power is 30 kW. One of the strongest broadcaster in europe today aside from BBC, is probably Radio Romania with 300kW. The only other broadcasters that by ear are on the same levels are the chinese and some arabic stations.

  3. Guys that “support” RX888 are so lazy, that thye do not made universal driver (module) for Linux. And said “if you want driver, do it yourself.” I said if you want sales, buy yourself. DONT BUY THIS PIECE IF BULL$#1T

  4. I’ve got a few old prototype 5G mMIMO boards from my work that have 4 of the Xilinx RFSoCs on them. wish there was some software i could load on them to do some SDR stuff. I’ve contemplated designing some custom boards to liberate the RFSoCs onto a nice generic package, but that is a project i simply just do not have the time for

  5. In this write-up it has the link to the Pluto article, which was then said to be easily available at $100-$200.
    So I checked on aliexpress and the cheapest there now is 330. And the one used in that article is 400.
    So inflation is hitting here too I guess.
    Also the Pluto starts at 30MHz at best the article said, so isn’t comparable, but that’s another point.

    @Sulio Pulev well said!

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.