KiwiSDR Vs RaspberrySDR — A Tale Of Two SDRs

Once you move away from the usual software defined radio (SDR) dongles, you have only a few choices unless you want to drop some serious cash. One common hobby-grade SDR is the KiwiSDR. This popular unit runs Linux and can receive up to 30 MHz. The platform uses a dedicated A/D converter, an FPGA, and BeagleBone computer. Success of course breeds imitators, and especially when you have an open source design like the Kiwi, you are going to find similar devices with possibly different end goals. That’s how the RaspberrySDR came to be. This is a very similar unit to the KiwiSDR but it uses a Raspberry Pi, along with a handful of other differences. What’s different? [KA7OEI] tells us in a recent blog post.

Other than the obvious difference of the computer and all that it entails, the RaspberrySDR has a higher speed A/D (125 MHz vs 66 MHz) and 16-bits of resolution instead of the Kiwi’s 14 bits. This combines to give the Raspberry a wider receive range (up to 60 MHz) and — in theory — better performance in terms of dynamic range and distortion.

[KA7OEI] measures a few key parameters on both devices and arrived at some surprising conclusions. The Kiwi appears to boost signals near its cutoff frequency to compensate for losses in the system. The Raspberry — using adapted software — looks as though it does the same trick, but does it around the Kiwi’s cutoff frequency, which is lower. Probably a software fix could take care of that, of course.

There are also tests of image rejection and front-end overloading. The tests revealed a few problems with signal strength measurement and some other problems with the RaspberrySDR. The biggest issue, though, was that the 16-bit A/D didn’t seem to have better performance. Without proper design, throwing more bits at a problem isn’t always helpful and this appears to be a good example of that.

In the end, the Raspberry looks like a cheap clone of the Kiwi with some benefits, but also some drawbacks. The blog post also covers some open source issues where Kiwi is now saying some parts of their code will only be binary in the future and there has been some difficulty finding all of the Raspberry’s files. If you are looking to buy one, you might not find the name “raspberrysdr” but [KA7OEI] suggests searching for “New 16bit 62M real-time bandwidth network shared SDR receiver” which does turn up some results.

Of course, you can always use a Pi with a more conventional dongle, and that works well enough. If you want to make a Pi just transmit, you can do that with little more than a wire, although the quality might not be perfect.

37 thoughts on “KiwiSDR Vs RaspberrySDR — A Tale Of Two SDRs

  1. 125 MSPS 16-bit ?!?!?
    How can they sell a product using an EAR category 3 part sourced from the US, and avoid the pre-sale documentation required by law. Or is the 125 MSPS 16-bit ADC a pin compatible counterfeit in a package marked LTC2208 16-bit ?

    The odd thing for me is that mouser have the 16-bit “LTC2208CUP#PBF”, what I think is correctly, flagged as an EAR item:
    “This product may require additional documentation to export from the United States.”
    But mouser have not flagged the 16-bit “LTC2208CUP#PBF” as requiring any additional export requirements.

      1. I went to aliexpress and searched for “LTC2208UP” and most of the chips look different from each other, my guess is that they are all counterfeit:
        5x 1542 LTC2208 UP BT44726 (no LT logo)
        3x 1016 LTC2208 UP BT31623 (no LT logo)
        2x 0534 LTC2208 UP BT22611 (no LT logo)
        1x 1542 LTC2208 UP BT44726 (LT logo covered with black dot in images)
        1x 0534 LTC2208 UP BT22611 (missing L from LT logo)
        1x 1015 LTC2208 UP BT31565 (has the LT logo)
        1x 1425 LTC2208 UP BT41212 (no LT logo)
        1x 1731 LTC2208 UP BT61289 (LT logo hidden from view)
        1x 0637 LTC2208 UP BT24680 (LT logo half missing, different font for the text)
        1x blank chips with nothing on them, pin out looks right

          1. Goto ali, click on the first 15 matches on a search for LTC2208UP, look at, and compare, the image of each sellers chip and then tell me if you still think that none of then are counterfeit.

          2. Truth: don’t need to do that. If someone got hold of a bin of reject dice, they would have them packaged somewhere else, so yes, they would look different from the legitimate ones. This would be good for the counterfeiter, since most of these devices would appear to work, and they could sell more of them.

    1. The 125 Msps 16-bit ADC chips are probably made (or cloned) in China to begin with and they “fell off the back of the truck” only to be picked up by some rat-hole shop owner from SEG Plaza in Shenzhen.

      1. Counterfeits are often either a lower resolution version and/or rejects from QC and are already bonded (if needed). Someone runs a packaging line somewhere else after hours and labels them to look like better parts. Using a different packaging setup is why they often look different in the details.

        How they can market these high dollar items is beyond me, but consider the chips are probably free, the cost to make them is tiny and the $10 each they sell them for is somewhere around $9.90 profit. OTOH, if they are $10 and are good for 10 or 11 bits at 125MSPS?

  2. > If you want to make a Pi just transmit, you can do that with little more than a wire, although the quality might not be perfect.

    Obligatory warning: You really shouldn’t do that, though. Be kind to your spectrum, filter your outputs and make sure you’re only emitting energy on bands you can legally access.

  3. Both [KA7OEI] and [Al Williams], and even to some extent, [John Seamons], developer of KiwiSDR, show some signs of having a bone to pick with the developer of RaspberrySDR. First, KA7OEI says a number of snarky things in his blog. For example: saying that RaspberrySDR “borrowed” much of the code from KiwiSDR. Putting the word borrowed in quotes tends to imply that he believes that some other, less generous word might be more appropriate. To this I say, OPEN SOURCE. The thing about open-source software is, no, you don’t necessarily get a return from your project that’s proportionate to the amount of work you put into it. Unless, that is, you ALSO utilize FOSS software in developing it. Open source just leads to resentment if you expect to be paid back. That is not its nature – open source is a PAY-FORWARD thing. You take advantage of other people’s work that they gave you for free, and so you offer your own work to others for free. Until you somehow feel slighted and start talking about hiding essential parts of your work inside binary blobs. So let’s knock off the snarky comments. These only lead to bad feelings everywhere, and are NOT in the spirit of FOSS.

    Maybe I’m going too far in including [Al Williams] though – all I see is a misrepresentation of the facts that is most likely just an honest mistake. Al, you say that “The biggest issue, though, was that the 16-bit A/D didn’t seem to have better performance”. Please read the blog again: the RaspberrySDR DOES have better performance, like about 5 dB improvement in noise floor and dynamic range, whereas the addition of two bits of resolution can, ideally, result in a 12 dB improvement. This is either misleading or misinformed. That 12 dB improvement is only possible if quantization noise is the ONLY source of noise in the output. Any other noise source (such as front-end noise, as KA7OEI points out) would not be reduced by adding bits to the A/D conversion. Also, ADCs themselves have a certain noise level, and many have “effective bits” listed in their spec sheets, which is always less than the actual number of bits. What’s the bottom line? Probably that sixteenth bit didn’t help at all, but the fifteenth certainly did. Anyway, the complaint here is that it isn’t ENOUGH of an improvement. Give me a break. In all of the RF work I’ve ever done, 5 dB was a HUGE amount.

    I don’t know how open the developer of RaspberrySDR is being, but clearly he is working to ADVANCE what [John Seamons] started, both in improving the performance of the design, and in adapting it to the more popular Raspberry Pi platform. I would suggest to Mr. Seamons that he is equally free to incorporate those improvements to his design if he wishes, free of charge.

    1. When I was working on a different ADC used in a basestation I found that noise floor increased (sensitivity decreased) with temperature. A squirt of freezer spray on the ADC (and not the LNA’s) brought the sensitivity back, and another squirt improved it past room temperature measurements. This is a very quick experiment to perform – even with the shielding cans in place. Well worth a try. If you can identify the culprit, stick a peltier cooler on there.

    2. Indeed, If you put your stuff out as open source you only have a valid complaint if whichever licences terms are violated by the other project(s). And if you can’t respect the variety of open license when you use it – and perhaps open source your tweaks and upgrades to the code as the license demands use something else!

      Nobody is going to care if you tweak some open source project to do exactly what you want for a personal project and don’t open-source your modifications – in all likely hood they are not of any use to anybody else anyway.. But when you are selling a product, or publishing ‘your’ work for profit please do as the damn license asks you to and give credit etc. Its good for everyone!

      Does seem like the RaspberrySDR is better than this article suggests, certainly looks like a good value very functional proposition.. I think I might just have to pick one up.. Not sure what I’d do with it yet though..

    3. Even funnier, KiwiSDR it self forked the open source openwebrx and did not contribute back patches and enhancements back then either :D
      So complaining about kiwisdr’s fork of openwebrx being forked is a bit amusing.

      1. FALSE. Pull requests pending since 2017 for a WSPR decoder, OS X support and IQ file playback. To be fair these were complicated PRs and Andras was way too busy with PhD work to be considering them. Also, I’m a Python noob and the implementation quality was likely low. Fortunately all this has been addressed by the current OpenWebRX developer.

        https://github.com/ha7ilm/openwebrx/pull/87
        https://github.com/ha7ilm/csdr/pull/29

        (I miss the old days. You know, before the web and social media gave idiots a greater voice than they deserve. I include myself in that category.)

        1. “(I miss the old days. You know, before the web and social media gave idiots a greater voice than they deserve. I include myself in that category.)”

          I laughed, (it is all too true about me as well!)

    4. “To this I say, OPEN SOURCE.”

      The usual problem with the Chinese thieves (see below) is that they use Open Source software and then make significant modifications to it which they never publish AND often use it in a commercial product. That is a no-no for open source anything unless it is specifically stated that it is OK in the open source license.

      IP, copyright and trademark theft is a long-standing CULTURAL thing in China which corporations were warned about long ago in scholarly papers about doing business in China… which they ignored for the sake of capturing a share of the potentially huge Chinese market which, for them, will evaporate, just like the Chinese market for iPhones is now, as soon as the stolen tech and manufacturing methods are perfected by China. IP theft is NOT frowned upon culturally or legally punished in China. Our corporations along with our bought government allowing this have proved the great truth of the very old commie adage, found in various forms, “A capitalist will sell you the rope you hang him with.”

      Just one of the books on this: To Steal a Book Is an Elegant Offense: Intellectual Property Law in Chinese Civilization (Studies in East Asian Law, Harvard University, January 1, 1997)

  4. I’ve read several times about this strange device, the RaspberrySDR … but i’ve been unable to find out where to buy it, where to read its “product page”. Google is saturated with blog entries about using Rpi with RTL-SDR dongles and didn’t helped… any clue ?

    1. So then, how do you define IP theft, if it includes actions that are legal? Seriously, it looks to me like many people are all for open-source when THEY use other people’s work without asking (which is “legal in terms of open source”), but then when somebody does the same with THEIR declared open-source work, they get all bent out of shape. Look: I know that many people would prefer that you work with them in making changes you want to their software, rather than forking the project, but that’s just a courtesy. If that’s what you want, you don’t want most of the open-source boilerplate licenses out there. Maybe what people really want is a “free for anybody to use, unless they’re in China” license.

Leave a Reply to Foldi-One Cancel reply

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