Comparing the KiwiSDR Against the RaspberrySDR Clone
The KiwiSDR is a 14-bit wideband RX only HF SDR which has up to 32 MHz of bandwidth, so it can receive the entire 10 kHz - 30 MHz spectrum all at once. Notably, the KiwiSDR does not connect to a PC directly, rather it is a cape (add on board) for the Beaglebone single board computing platform which similar to a Raspberry Pi. With most of the DSP processing done on the KiwiSDR's onboard FPGA, the Beaglebone serves a custom OpenWebRX browser interface which can be accessed over a network connection from anywhere in the world. If you're interested our initial KiwiSDR review from 2017 is here.
Over the years the KiwiSDR has brought some very interesting software developments out such as several new demodulators. However, our favourite is the TDoA feature, which allows users to leverage multiple public KiwiSDRs to locate the source of an HF transmission with remarkable accuracy.
KiwiSDR Clones
This year we've seen a number of cloned SDRs come out on the market, with almost all using LTC2208 ADC chips that have most likely been recycled from discarded equipment. One of those clones is the RaspberrySDR, which is a clone of the KiwiSDR.
The RaspberrySDR is not a direct clone however, as it brings some improvements. The biggest change is that the LTC2208 chip has a 16-bit ADC, and can provide up to 62 MHz of real time bandwidth. Also instead of a Beaglebone single board computer, a Raspberry Pi 3B+ is used instead. At the time of this post the RaspberrySDR retails for roughly $70 less than the KiwiSDR.
Comparison
Over on his blog KA7OEI has written up a comprehensive comparison between the KiwiSDR and RaspberrySDR. KA7OEI notes RaspberrySDR powers up and works with it's full 62 MHz bandwidth as expected. Measurements for sensitivity, dynamic range, image rejection are about the same.
However, there are some issues such as inconsistent RF level calibration, a broken s-meter at high SNR levels, "motorboating" on strong narrowband signals, and a broken firmware update button. Also interestingly, KA7OEI's tests show no improvement to the dynamic range. With two extra bits of ADC resolution on the RaspberrySDR we would have expected an improvement. Most of these issues are probably firmware bugs which could be fixed, but the dynamic range issue could be related to less care taken in the hardware design.
There has also been some discussion over on the WSPRDaemon forums here.
Ethics + Official Future Software Development
As the KiwiSDR source code is open source, it could be considered fair game to fork the code and make use of it in a derivative product. However, at the same time we should remember that the KiwiSDR developers have been working on this code and providing constant updates ever since the release. No funds from the clones will go to them and the success of a clone could spell the end of motivation for future software developments. In addition as KA7OEI notes, the code used on the RaspberrySDR seems to be somewhat obscured, and unlike the KiwiSDR, no open source schematic has been released. Any official long term support of the RaspberrySDR seems unlikely too.
John Seamons (ZL/KF6VO), the leader of the KiwiSDR project has announced that despite the clones KiwiSDR development will remain 100% open source with any future updates also being available to the cloners should they choose to implement them. He also mentioned to us that the clones will also be able to contribute to the TDoA service and can be listed on the KiwiSDR directory. However, the reverse proxy feature will be limited only for official products.
Here we are in 2023… the “ethics” issue of using the Kiwi version the SDR software for something like the RaspberrySDR is now moot. There are no “real” KiwiSDRs available to buy. There are no RaspberrySDRs to buy. The BeagleBone is a dinosaur. I do wish someone would make a commercial compatible cape for the Pi that would run the wonderful(!!!!) Kiwi-derivative of OpenWebRX. Sigh…..
Any comments on this:
For years, a backdoor in popular KiwiSDR product gave root to project developer:
See ArsTechnica.com published July 15 2021 for more information
Complaining about people forking kiwisdr’s software and not contributing to it is a bit rich, considering that kiwisdr devs forked openwebrx and did not contribute to it either.
So a tad “pot, meet kettle” to my ears.
It seem you don’t quite understand. Who’s complaining about them forking and not contributing? It’s open source. They don’t have to contribute anything. Only re-publish any modifications they make to covered code. I have a problem with them stealing my brand (name, logo etc). But they’ve mostly fixed that now.
As for KiwiSDR not contributing back to the OpenWebRX codebase: It never made much sense given our UI diverged so quickly and radically in support of our hardware. We did contribute a WSPR decoder and ability to playback IQ files three years ago. Go look. The pull requests are still pending, lol.
Also, both Andras (original OpenWebRX) and Andrew (Homemade GPS) have been KiwiSDR profit sharing partners since 2016 (and continue to be). A very substantial form of “contribution”. How many other open source projects do that?
Not going to be rewarding Chinese clones.
Surely the LTC2208 ADC is an export controlled item, what used to be called an ITAR (International Traffic in Arms Regulations) item ?
Commerce Control List September 11, 2020
CATEGORY 3 – ELECTRONICS
a.5. Analog-to-Digital Converter (ADC) and Digital-to-Analog Converter (DAC) integrated circuits, as follows:
a.5.a. ADCs having any of the following:
….
a.5.a.5. A resolution of 16 bit or more with an output rate greater than 65 Mega Samples Per Second (MSPS).
Or am I missing some detail ?
China is not an ITAR-banned country. How do you think, for example, Seeed gets away with manufacturing and selling the BeagleBone Green from Shenzhen? A devices that contains an ITAR-listed TI CPU chip? (due to its cryptography acceleration)
Besides, China has found a much better way to defeat the West than putting 16-bit ADCs in ballistic missiles. They simply sell Americans inexpensive, hastily-designed electronics that doesn’t last. Like the LED lightbulbs in my kitchen that keep burning out. Soon the consumer economy will be drained of cash. And when they can’t pay the rent or buy their favorite Travis Scott burger from Maccers they’ll be at each other’s throats. Problem solved. Remember China is not just a country, it’s a civilization. And they’ve got this all figured out.
Sorry, I probably did not make myself clear. What I’m saying that there is no problem selling a controlled item that fall under the EAR (Export Administration Regulations) to any individual (or company) in China , or any other approved country, with the correct documentation. What I am saying is that if an item is an export controlled item, that there is whole new level of documentation involved. That you can not sell any product containing a dual use item without to just anyone. From a legal perspective, a validated end-user authorization agreement (ultimately with BIS – Bureau of Industry and Security) needs to be in place before each individual sale can happen.
It is odd about the AM335x. But that restriction could also have been for, the PRU’s in the AM3356-AM3359.
From the datasheet:
Two Programmable Real-Time Units (PRUs)
– 32-Bit Load/Store RISC Processor Capable of Running at 200 MHz (~5 nanoseconds per instruction)
– 8KB of Instruction RAM With Single-Error Detection (Parity)
– 8KB of Data RAM With Single-Error Detection (Parity)
– Single-Cycle 32-Bit Multiplier With 64-Bit Accumulator
– Enhanced GPIO Module Provides ShiftIn/Out Support and Parallel Latch on External Signal
I suspect that it was this feature, even more so than the ability to do AES, DES/3DES, MD5, SHA with a builtin True RNG (which may have been “upgraded” since, to not be slightly more biased in it’s random).
From the Commerce Control List September 11, 2020 CATEGORY 3 – ELECTRONICS:
“e.1. Digital time delay generators with a resolution of 50 nanoseconds or less over time intervals of 1 microsecond or greater; or
e.2. Multi-channel (three or more) or modular time interval meter and chronometry equipment with resolution of 50 nanoseconds or less over time intervals of 1 microsecond or greater;”
I’m not even going to say the reason out loud why I think the part was covered. But because it is designed for industrial processes, with the possibility of being exposed to sources of ionizing energy, the part has party checks on the PRU’s the 8k for instructions, 8k of data and there is even ECC (error correcting code) in the L2 cache.
The hardware developer’s Twitter: https://twitter.com/fei666888
People seems to be more interested in his Pluto clone: “PlutoX” a 2T2R device with Gig Enet.
Or even his YouLoop clone. The guy is a machine.. lol