This article was written by Taylor Killian on his blog in August 13. He contrasted the three parity SDR platforms, thinking that the three products would be the most popular SDR platform of the coming year. I think this article is very valuable for reference, a simple translation of a turn around.
The original is here:
Http://www.taylorkillian.com/2013/08/sdr-showdown-hackrf-vs-bladerf-vs-usrp.html
Turned up only to find that it was too long. Think so long there is no need to translate, do not meet the habit of fast reading, deep reading should be directly read the original. But now that you've opened your head, you're done.
Here is the original translation:
This year or next, it looks like the golden age of SDR. There will be three new SDR platforms available, with many options for users. This article will compare these three SDR platforms: the HACKRF produced by Great Scott Gadgets, the bladerf produced by Nuand, and Ettus (USRP) b200/210 production.
HACKRF, developed by Michael Ossmann, has also developed Ubertooth, the first and only low-priced Bluetooth sniffer device (Bluetooth sniffer). HACKRF has been in development for several months. Michael has released a free 500 beta version of HACKRF to hackers around the world. Now he is developing the official version and is selling on the crowdfunding website Kickstarter. This will be the cheapest SDR platform on the market, and it can work in a wide frequency range.
Bladerf is also a successful project on the Kickstarter, and the products have been paid to the users. The Bladerf can support a wide carrier range, with a large FPGA, and a high-speed USB3 interface. The developer spends a lot of money on the clock module, providing a VCTCXO that can calibrate the accuracy to the ppb. All modules are designed to be synchronous, so there is no clocktamer. This hardware can now be purchased on the Nuand website.
At first, this article is to take Ettus usrp B100 and N210 with HACKRF, bladerf comparison. But when I sent the first draft to Ettus, they said the new USRP b200/210 was going to be released soon and sent me a development version of B210. So later I revised this article. Now B210 has been released and can be purchased from Ettus. Ettus is probably the oldest SDR hardware producer. They have produced a lot of different usrp. The USRP b200/210 is completely different from the previous old product, which is a "veneer" design and is no longer a combination of "motherboard/sub-board". It also uses the USB3 interface, which supports a wide range of RF bands. The b200/210 has a newly designed GPSDO connector that can be adjusted to several PPB via the GPS module. B210 is also the Ettus's first standalone support for 2x2 MIMO boards. I'll do some tests on the B210 and post the test results on this blog. Here are some high-definition images of B210:)
The following table is a comparison of three hardware! (Alin: Focus on this watch is enough ^_^)
Specs
|
Hackrf |
Bladerf |
USRP |
| x40 |
x115 |
B100 Starter |
B200 |
B210 |
| Radio Spectrum |
Mhz–6 GHz |
mhz–3.8 GHz |
mhz–2.2 GHz [1] |
50mhz–6 GHz |
| Bandwidth |
MHz |
MHz |
-MHz [2] |
61.44 MHz [3] |
| Duplex |
Half |
Full |
Full |
Full |
2x2 MIMO |
| Sample Size (ADC/DAC) |
8 bit |
Bit |
BIT/14 bit |
Bit |
| Sample Rate (ADC/DAC) |
Msps |
+ Msps |
msps/128 Msps |
61.44 Msps |
| Interface (Speed) |
USB 2 HS (480 megabit) |
USB 3 (5 gigabit) |
USB 2 HS (480 megabit) |
USB 3 (5 gigabit) |
| FPGA Logic Elements |
[4] |
40k |
115k |
25k |
75k |
150k |
| Microcontroller |
Lpc43xx |
Cypress FX3 |
Cypress FX2 |
Cypress FX3 |
| Open Source |
Everything |
HDL + Code Schematics |
HDL + codeschematics |
Host Code [5] |
| Availability |
January 2014 |
Now |
Now |
Now |
| Cost |
[6] |
$420 |
$650 |
$675 |
$675 |
$1100 |
[1]–separate Daughterboards is required to receive/transmit. The WBX transceiver is included in this kit[2]–half this if the-samples is used[3]–56 MHz for single half duplex c Hannel, 30.72 MHz per channel full duplex[4]–there are a CPLD on the board, but no Fpga[5]–ettus confirmed that the HDL + Code + Schematics'll be released for the b210/b200[6]-estimated retail price, cheaper though Kickstarter
RF Performance
The RF range of the HACKRF and USRP B210 is wide. HACKRF is 20MHz lower than b200/210, and the highest carrier frequency can reach 6GHz. B210/200 is mainly based on AD9361 chip. This chip is actually working in 70mhz~6ghz, so it looks b210/200 slightly out of use, down to 50MHz. HACKRF uses a different type of RF chip to support wide bands. If you look at its schematic, you will find that it mixes a few chips, each chip is responsible for a spectrum, at least 6 band switching switch. It is hoped that so many devices do not introduce too much noise into the system.
In addition, HACKRF bundled an up-converter "Ham It up", if through the Kickstarter two things to buy, just add a $. This board can reduce the carrier frequency of the HACKRF to 300KHz. This "Ham It up" can also be purchased separately, about $43. I think USRP b210/200 and WBX can also use this "Ham It up". For older USRP B100, it can cover different bands with different sub-plates. The WBX board can cover 50mhz~2.2ghz. The new CBX board can be covered to 6GHz. But the disadvantage of this single-purchase sub-board scenario is that the cost is higher than HACKRF and Bladerf. The Bladerf can support 300MHz to 3.8GHz, and it uses a chip that is lms6002d. This chip offers the most RF functions, including all mixers, ADC,DAC and other features. This chip is similar to the AD9361. However, it can only support up to 3.8GHz. This means that it is impossible to use Bladerf to achieve 802.11n in the 5GHz band. Now, Bladerf plans to release an expansion board, allowing the carrier frequency to be lowered to 10MHz, but the plan is still underway, not yet done.
Duplex Performance
It is important to note that HACKRF differs from the other two hardware, which does not support full duplex. This means that to switch between receiving and sending, you must send a command to the controller each time. It may take a microsecond of time for the microcontroller to process the switchover. If you want to count the time the signal arrives at the computer, the switchover time will be longer.
Both Bladerf and USRP b210/200 can support full duplex. USRP B100 also supports full duplex. There are some older sub-boards that are not supported for full duplex when used on B100. But most of the sub-boards can be supported.
According to the documentation on GitHub, the B200 has a full-duplex channel. The B210 has two receivers and two transmitters, which are designed to support 2x2 MIMO. Two receivers can be transferred to the same frequency point, the transmitter is the same (can be different from the receiving frequency point). This can take advantage of the spatial diversity of the wireless channel to transmit higher-rate data. MIMO technology has been implemented in 4G LTE and 802.11n systems.
It is important to note that if you are receiving and sending at the same time. The transmitter may produce some noise to the receiver, because the transmitter is close to the receiver after all.
communication with the host
For SDR, communication with the host is important because it determines the bandwidth and reliability of the signal.
USRP B100 and Hackrf both use the USB2.0 interface. This determines that the highest data transfer rate is 35mb/s. However, since we often insert other devices on multiple USB interfaces, they are shared bandwidth, so the actual data rate is lower than this.
USRP b210/200 and Bladerf use the USB3.0 interface. It can support the transfer rate of the 400mb/s. This is enough bandwidth for most SDR applications. Similar to USB2.0, multiple USB interfaces share bandwidth.
One potential problem is that USB3.0 may be disturbed. Intel warns that signals in the 2.4GHz band may cause interference with USB3.0, suggesting some shielding measures. One simple way is that you can wrap a sheet of foil in a card. Bladerf and b210/200 developers have done some testing and think it's not a big problem. The Bladerf RF module is covered with a shell outside. The b210/200 also makes it easy to add copper shells. In addition, B210/200 's circuit board also has a large area of "paving", which can also play a very good shielding role.
After my test, USRP B210 with my Asmedia controller is still problematic. It can only work in USB2. In fact, before I received B210, Ettus's engineers reminded me that Asmedia USB3 chip was not fully compliant with the USB3 standard. Finally, I bought a PCIe USB3 interface card, it uses the VL805 chip, the price is about $ $, now I can work normally in USB3 mode. Since Bladerf also uses a FX3 chip, I think it may not match asmedia.
Adc/dac
The quantization accuracy of the ADC and DAC is important, and adding a bit can double the accuracy. Therefore, the USRP B100 using the 14-bit DAC is 64 times times better than HACKRF accuracy using the 8-bit DAC. While it is said that a cheap 8-bit precision RTL-SDR can receive NOAA's Meteorological satellite imagery, the higher accuracy is obviously more useful. Of course, better antennas and gain settings can also be helpful.
Another indicator is the conversion speed of the ADC and DAC. Higher sample rates require greater processing bandwidth. Many older communication systems can use very low-speed ADCs or DACS, but newer communication systems, such as WiFi a/b/g, require a minimum of 20MSps of ADC/DAC. Of these three hardware, only USRP b210/b200 is capable of handling 40MHz of 802.11n signals. However, even if USRP can handle it, it is still a huge challenge for the computer to handle such high-speed data. In fact, even if you just want to store such high-speed data, it's a hassle.
BandwidthUploading all baseband data to the computer is the main bottleneck for all SDR hardware because the amount of data is too large. For USRP B100 and HACKRF using USB2.0, this bottleneck is obvious. and other hardware that uses USB3.0, this bottleneck is relatively relaxed a bit. Although the Bladerf also uses the USB3.0 interface, it cannot achieve the same sampling bandwidth as the USRP b210/b200. Because the Bladerf uses a bandpass filter in the lms6002d. In the Bladerf forum, there is discussion on how to turn off the filter so that an external filter can be used to increase the sampling bandwidth. With regard to the USRP b210/b200, its filter can allow a signal with a bandwidth of up to 56MHz to pass.
FPGAHow do you use these baseband signals? It is either uploaded to the computer or processed on the board. Bladerf and USRP b210/200 have more powerful FPGAs, as well as FX3 microcontrollers. The B210 uses the Spartan 6 LX150 FPGA, which has 150k logic units, B200 uses LX75 FPGAs, and has 75k logic units. The BLADERF uses Cyclone 4 fpga,x40 with 40k logic units, and x115 has 115k logic units. USRP B100 is a small FPGA with 25k logic units. While HACKRF is using CPLD, the signal processing relies on the microcontroller on the board.
The number of logical units determines the processing power of the FPGA, obviously the larger the better. The advantage of FPGA is parallel processing, the disadvantage is that the frequency is generally lower than the microcontroller. If the developer is not very good at HDL language, the processing efficiency may be relatively low.
The USRP FAQ page has an FPGA usage note that shows how much of the resources the FPGA has left to use. For B100, the space left for the user to develop is very small, while the b210/b200 has a larger amount of space available for users to use. With regard to BLADERF, it is said that the x40 FPGA is currently using about 15%, so there is plenty of space left. In addition to being a bridge between Adc/dac and FX3, FPGAs can also perform signal processing tasks such as digital filters. The USRP includes digital frequency conversion, pumping value and interpolation module and so on. I did not see the Bladerf function, probably similar to USRP. One difference to note is that Ettus uses Xilinx chips, and Nuand uses Altera's chips, so it's slightly different. There are more DSP modules in the FPGA than Altera,xilinx, including pre-adder, multipliers and accumulators, while Altera FPGAs have only multipliers in the DSP module. This means that the addition needs to be implemented with a logical array, so the same functionality as the Altera FPGA requires more logical units. Also, Altera has less ram than Xilinx. But for Bladerf, the RAM on the chip might be enough. It is also important to note that B210 's LX150 does not support the free Xilinx ISE, while the LX75 and Altera FPGAs are free to use for development software.
Finally, the price of FPGA is emphasized. The x40 Cyclone IV Price is about $100, x115 Cyclone IV is about $315. This is an offer on Digi-Key, and may not be the cost price of the manufacturer. But this chip selection, at least, shows that Nuand is not a high-end hardware.
Micro Controller
In addition to the older USRP B100, several other boards have very powerful microcontrollers. B100 uses FX2 to provide USB2.0 connection, only 16KB of RAM. Both Bladerf and b210/200 use FX3, which provides a USB3.0 connection. HACKRF is a dual-core lpc43xx chip that handles the USB2.0 interface and controls the RF chip.
The HACKRF microcontroller operates at a 204MHz frequency and is manufactured by NXP, with an arm M4 core and a M0 coprocessor. Contains 64KB of ROM and 264KB SRAM. This chip is responsible for a lot of work: send and receive data on the USB link, control all RF chips on the board. There are plans to add a suction and interpolation module to it. Microcontrollers are chosen rather than FPGAs because they want users to be able to use the C language for faster code changes rather than using HDL language. If the microcontroller connected to a PORTAPACK,HACKRF can not connect to the computer, it becomes a spectrum analyzer directly.
The FX3 microcontroller operates at 200MHz and has a arm926ej-s core. The chip has a GPIF that, once turned on, allows the arm core to be idle. It has 512KB SRAM and no ROM. This chip has several start-up methods, including USB boot mode, which firmware often load in this way. The Bladerf has a 4MB SPI Flash, which contains the microcontroller and FPGA code to support Board operation when offline. B210 only 32KB EEPROM, used to store some basic configuration programs, no flash. Because there is no flash, so b210/200 want to run offline, it is very difficult. Nuand developers want to be able to run Openbts and Openlte offline on Bladerf. There are already examples of LTE systems running on B210, but in the case of connecting the computer (Core i7 processor). It is doubtful whether the LTE system can be run on the FX3.
Developer Community
Software Radio is a big concept that has existed for more than 10 of years. The user forum or community of a hardware platform is very important. These people can provide each other with technical support and share new ideas. They drive the wheels of innovation to keep on moving forward. Therefore, for a company that produces SDR platforms, the developer community is very important and deserves strong support. One of the easiest ways to do this is open source, open hardware. Let's talk about how the three companies are doing it.
Source Code
Software is very important for SDR hardware. Fortunately, the three hardware supports the GNU Radio, which contains a lot of free and open source code. It also has a very good graphical interface for rapid development and testing. Hackrf and Bladerf in the GNU Radio, in the GR-OSMOSDR project, with theRTL-SDR dongle is the same as the driver package. Bladerf drivers are added in the last few weeks (Alin: The author's writing Time is August 2013), so if you want to use it, please update the code in time. HACKRF's drive has been released for some time. For USRP,GR-UHD, which is part of the GNU radio, you need to install the UHD Drive library, which can be obtained from the Ettus website.
These three hardware boards, all the code, HDL files and circuit schematics are available for free, except USRP b210/b200 has not yet been released. HACKRF is more open, and it even publishes all the KICAD board files, including the original circuit schematics (not pdf) and PCB cabling diagrams. The schematic diagram of USRP B100 and Bladerf is in PDF format. I hope b210/200 can also open the schematics as soon as possible. The openness of the HACKRF allows others to continue to improve the design of HACKRF, and is a very good learning material for readers of the developer community. I think other people can easily reuse some of these designs.
USRP has a unique advantage that its application class code is very rich. Because USRP has a long history, since 2006, many people have used USRP. A number of academic papers have been experimented with USRP and GNU Radio. There are a lot of new applications and code emerging. UHD, which is the universal interface for all USRP boards, has been very stable and mature after so many years of optimization and improvement. A GNU radio application is easy to use on a variety of USRP boards as long as the hardware meets the requirements. In other words, USRP's support for the GNU Radio is the best. Ettus also released a free Linux image that includes the GNU radio and other tools to build a GNU radio development environment as quickly as possible. Although this image is currently only available in UHD, it should be easy to add GR-OSMOSDR.
HACKRF is much shorter than USRP's, and has just begun to develop a set of code (Alin: What the development environment drives), but it is progressing quickly. It should be compatible with the GNU radio and is being tested. The advantage of HACKRF is that it has a lot of hacker-type fans. At least 500 free HACKRF have been delivered, and 1100 hackrf have been booked on Kickstarter until the time of writing. These users can make a great contribution to software development. HACKRF has greater influence in the hacker world than USRP in academia. Of course, there is a part of the overlap between the two groups. But I think hackers are more capable of writing more good code.
The Nuand team is the only recently released GNU Radio driver for Bladerf. I estimate that about 400 users have got the Bladerf board from Kickstarter. In addition, a significant percentage of users have ordered Bladerf directly from Nuand's website. All of these users are able to contribute to Bladerf code development. Because Bladerf and USRP B210, using FX3, large-capacity FPGA and single-chip RF transceiver chip, I estimate two of the hardware driver code, there are a considerable portion of it can be shared. While the difference between Xilinx and Altera is a bit of a hassle for code reuse, I think that as long as hackers make enough effort, Bladerf can be compatible with UHD interfaces and thus be compatible with the vast number of applications USRP already have.
Hardware
In terms of the openness of the hardware, HACKRF is the best. and USRP, b210/b200 use of the AD9361 chip may be an open barrier. Because the Analog Devices website only provides 1 pages of datasheet. It is therefore difficult for others to obtain more detailed information than the Ettus company, unless you have signed an NDA with the ad company. But Ettus Company's people promised that they would open the source code for the relevant driver. In addition to the AD9361 chip, other chips on the b210/b200 have more detailed information. The bladerf,lms6002d chip has up to 15 pages of datasheet, and there are 45 pages of programming and calibration guides that are readily available to other developers.
the last reviews
The HACKRF is an SDR board with the widest frequency coverage and the cheapest price. Almost all of its information is open source, even including KiCAD files. The disadvantage is that it has no FPGA, uses the low-speed USB2 interface, the ADC/DAC accuracy is lower. Overall, HACKRF is ideal for hackers with high openness requirements, and those who are price-sensitive.
Bladerf, its highlight lies in the large-capacity FPGA and high-speed USB3 interface. It can support a wide band, but it's not as good as the other two. It also has a good ADC/DAC accuracy. I suggest that those who want to run programs offline and shoot frequent points don't need to be too high, consider choosing this hardware,
USRP B100, this is an older card, can not support high-bandwidth applications. It changes the RF band by replacing the sub-plates, which can support up to 6GHz. It supports UHD interfaces. B100 Price is the same as B200, but the ability is much worse than B200. So I suggest that you consider B100 only if you have some very special applications, or if you want to use your own development of a sub-board.
The USRP b210/b200 can support a wide band and also support high-speed interface bandwidth. They have large-capacity FPGAs and fast USB3 interfaces. But the AD9361 of the chip is slightly less open. B210/B200 is the most expensive of the three hardware. But many of their indicators are already comparable to Ettus's other high-end N210 board. Furthermore, B210 is the only one by one board that supports 2x2 MIMO directly. I believe b210/b200 will be the most powerful SDR platform on the market recently, and will be supported by Ettus company. I recommend this hardware platform for applications that require high bandwidth, wide band, and do not need to be used offline.
Finally, I want people to be able to usethese three very good software radio platforms have developed more innovative applications. I spent a few weeks collecting information about these three platforms, some of which are not available in the article, and if you have any questions, please make them in the comments. Thank you for reading, and you are welcome to come back to see the results of the test I will be publishing about B210 (Alin: This test result has not been published until November 27, 2013).
Alin Comment:
Two hardware are USB3.0 interfaces, indicating that USB3.0 will become the mainstream interface of the SDR platform. This is consistent with what we thought two years ago. The advantage of USB3 is that the number of USB ports, but also popular, fast enough. But the disadvantage is that the transmission distance may not be far enough, so I personally think that the 10G or higher speed Ethernet interface will be another mainstream interface.
In addition, I found that the three hardware of the circuit boards are printed on the black-bottom white, more cool than the original green. It seems that hardware engineers have also begun to notice that "the board on the back of the cabinet is also beautiful".
[Reprint] Three SDR platform comparison: Hackrf,bladerf and USRP
