XTRX: Soon to be crowdfunding Mini PCIe based TX/RX SDR

Over on the crowd funding site crowdsupply.com there have recently been several updates on the Fairwaves XTRX SDR. The XTRX is an upcoming TX/RX capable SDR in a tiny Mini PCIe form factor. Mini PCIe is the expansion slot system used on some laptops. The SDR itself will be 2 x 2 MIMO, with a tuning range of 10 MHz - 3.7 GHz (down to 100 kHz with some degradation), and have a sample rate of up to 120 MSPS. It uses the LimeSDR RF chipset which provides most of the hardware required.

The XTRX is not yet for sale, and is planned for a crowdfunding run on Crowdsupply 'soon'. You can subscribe to future updates on their page. No word yet on pricing, but according to one of the developers comments on Reddit the price will be somewhere between the LimeSDR ($299 USD) and LimeSDR Mini ($139 USD). Eventually in the future if they can tap into a mass market they hope to get the price down to $50 USD.

Features & Specifications

  • RF Chipset: Lime Microsystems LMS7002M FPRF
  • FPGA Chipset: Xilinx Artix 7 35T
  • Channels: 2 × 2 MIMO
  • Tuning Range: 30 MHz - 3.8 GHz
  • Rx/Tx Range:
    • 10 MHz - 3.7 GHz
    • 100 kHz - 3.8 GHz with signal level degradation
  • PCIe Bandwidth:
    • PCIe x2 Gen 2.0: 8 Gbit/s
    • PCIe x1 Gen 2.0: 4 Gbit/s
    • PCIe x1 Gen 1.0: 2 Gbit/s
  • Sample Rate: ~0.2 MSPS to 120 MSPS
  • Reference clock:
    • Frequency: 26 MHz
    • Stability: <10 ppb stability after GPS/GNSS lock, 500 ppb at start up
  • Form Factor: full-size miniPCIe (30 × 51 mm)
  • Bus Latency: <10 µs, stable over time
  • Synchronization: synchronize multiple XTRX boards for massive MIMO
  • GPIO: 4 lines @ miniPCIe connector, 3 lines @ FPC edge connector
  • Accessories: miniPCIe-USB3 converter, miniPCIe-PCIe converter, etc
XTRX Prototype
XTRX Prototype
Notify of

Inline Feedbacks
View all comments

and to Believe uniden america disclosed that its using rafael micro r836.. Uniden scanner model sds100 and selling for 650.00 usd…
I bought one and its loses receive everytime it has a close by signal and the VHF hi BAND SUCKS.. everyone do a favor DONT BUY THE SDS100 nothing but drama check the forums on radioreference.com.. r836 is 42mhz-1002mhz when scanner is marketed for 25mhz-1300mhz did UNIDEN even read the SPEC SHEET lmao alot of pissed ff customers with this one… a 49db gain on a scanner plus 2 preamps…what a beotch…….

Douglas Goncz A.A.S. M.E.T. NPI1659534493

I build mobile rechargeable supercomputer Klustorouters, a name I made up for the processor core Klust-, the distributed store -stor, and the router component, doh, -router, with 4G connectivity to the router and 3G on each PC in the Klust (ers). Each PC is a Samsun NEOL XE700T1A0-H01US or similar and accepts PCIe.
So I am *interested*.
How can I help?


If you use an ADC with the same performance and the one in the LMS7002M (ENOB 9 bits) the performance would be similar, and if you used a better ADC with the Rafael Micro tuner the performance in most use cases will be better than the LMS7002M. As long as you only need one Rx channel and bandwidth requirement is 10MHz or less, and your frequency is within the specified frequency range of 42MHz to 1002MHz should will run circles around the LMS7002M.

But the R820T2 is an EOL (End Of Life) part, soon you will no longer be able to buy them.
You should be asking about its replacement which will probably the the R836 ( http://www.rafaelmicro.com/product/list/3 – but like all chips from Rafael Micro, it requires an NDA for a datasheet )
But there is an old leaked copy on the Internet which probably has out of date information – http://archive.espec.ws/files/Rafael%20Micro%20R836%20High%20Performance%20DTV%20Silicon%20Tuner%20Datasheet.pdf

One chip costs ~100x times the price of the other, but for the parts that they have in common, the performance is similar.

The R836 has an LNA(gain), mixer(with a gain), fractional PLL VCO, VGA(gain), voltage regulator and a RF tracking filter and IF filters.
One Rx channel in the LMS7002M has a LNA [LNAL/LNAH/LNAW(gain)] , a mixer, fractional-N synthesizer, TIA(gain), two LPF[RXLPFL:0.75 MHz up to 20 MHz OR RXLPFH:20 MHz up to 80 MHz], PGA(gain), 12-bit ADC (ENOB: 9 bits) and then you have a digital TSP(Transceiver Signal Processor)

The TSP can add corrections for gain, corrections for phase, corrections for DC, then you have a digital mixer, decimation, three general purpose FIR stages, a RSSI (Radio Signal Strength Indicator) which can adjust the AGC and scale the digital output.
The LMS7002M has 2 Rx channels, but they share the same LO. It also has two Tx channels, but they share a common LO (different to the Rx LO). So in some cases you need additional external mixer to make use of the second channels.

LMS7002M 850 MHz Phase Noise -96 dBc/Hz @ 1k offset
R836 860 MHz Phase Noise -89 dBc/Hz @ 1k offset
LMS7002M 850 MHz Phase Noise -99 dBc/Hz @ 10k offset
R836 860 MHz Phase Noise -98 dBc/Hz @ 10k offset
LMS7002M 850 MHz Phase Noise -107 dBc/Hz @ 100k offset
R836 860 MHz Phase Noise -110 dBc/Hz @ 100k offset

LMS7002M Noise Figure 2.0 dB @0.95GHz 2.5 dB @2GHz, 3.5 dB @3.5GHz
R836 Noise Figure 2.9 dB


Thank you for the answer and the data. In this case it would be interesting to see a comparison to eg. Airspy mini, which uses similar resultion ADC than LMS7002M.


They are both 12-bit ADC’s but they are not the same.
Airspy R2 ENOB: 10.4 bits
LimeSDR ENOB 9 bits


Sorry the Airspy mini is also ENOB 10.4 bits
But the Airspy R2 and Airspy mini are both no longer being produced, as far as I know this was due to a drop in the quality of the R820T2, since bulk production was moved to a EOL fab. There is nothing wrong with the chip when operated within specification 42 to 1002 MHz defined by Rafael Micro, but every SDR drives them to much lower and higher frequencies than this.

There should be an even higher quality airspy replacement product at some stage (11.6 ENOB – https://airspy.groups.io/g/main/topic/better_rx_for_vhf_uhf/14086303 )


Can anyone point me to a comparison between R820T2 and LMS7002M looking at sensitivity, dynamic range, noise? Of course LMS7002M should perform noticeably better looking at the specification, but I would like to know how it looks in reality.


Let’s say I install one in a x220 Thinkpad or a similar laptop, could existing antennas be used for tinkering in the 300-900mhz range? Antennas are for the WWAN (3G GSM) and WIFI, both have primary and auxilliary conectors, all in all 4 little cables. Are GPS and clock connections mandatory for XTRX?

Alexander Chemeris

I’m not sure those antennas will be optimal for 300-900 MHz band as most of the cellular bands are >850 MHz (for GSM) or >700 MHz (for LTE). That said, while not optimal I think you should be able to receive. You’ll have to experiment, though.

Bendail Vam

I utilized many of these: https://epiqsolutions.com/sidekiq/ for a client who wanted all viable RF recorded at a few of R&D labs/sites for security of product and piece of mind for the investors, I have to tell you, if your shooting for that price model your blowing up the competition. And, you have a larger chunk of spectrum coverage to boot.

Alexander Chemeris



While I like the idea of a SDR-TRx for a notebook, I am wondering if the height is equivellant to the max. height and if it will allow insertion in a notebook slot.
While it is necessary to keep the height low enough to all insertion in a miniPCIe slot in a notebook, without some more stable coax connectors at the end of the card like SMA ,I fear the use will be shortlived.
From the slides at https://drive.google.com/file/d/0B-CV_07uSBIuV3pjSVhHdW9RLVE/view the card is only intendended for external use on the shown adapter.

Alexander Chemeris

If your laptop has a full-size miniPCIe, then yes. Most modern laptops have M.2 instead of miniPCIe which is similar, but a different form factor. We’re looking into having an M.2 version of XTRX, but that’ll be later.

XTRX use U.FL connectors just like all modems. These connectors are working well if you don’t disconnect/reconnect them every day. The initial goal for XTRX are embedded devices where this is exactly what you need – you connect cables ones and never touch them again.

For everyday use we’ll have a USB version which will have SMA connectors.

I’m not sure what do you mean by “the card is only intendended for external use on the shown adapter”?

Michael Andreas

If you look midway though the linked slide deck, there’s a slide showing the XTRX mounted either on a USB3 adapter or a PCIe carrier card.

Alexander Chemeris

Those slides show extra features and examples of possible carrier boards and use cases. It was not meant as a comprehensive list of intended use cases. Primary use case is actually an embedded SBC where U.FL makes a lot of sense.

Michael Andreas

I was answering your question in your prior comment on this particular comment branch.

You asked:
“I’m not sure what do you mean by “the card is only intendended for external use on the shown adapter”?”


Also hopefully there will be enough hardware compatibility between this and the LimeSDR that it can still use SoapySDR and help grow the Lime software ecosystem.

My understanding is that the Lime firmware is dependent not just on the LMS7002M but also on the Altera Cyclone FPGA etc. I’m not enough of an expert on exactly how the gateware, firmware, etc. all fit together, but I imagine there might need to be some clever coding to ensure compatibility.


It supports SoapySDR, GNU Radio.

You’re correct the Lime firmware is dependent on Altera FPGA. We’re using our own FPGA implementation that allows us to do many interesting things.

David Gillooly

What advantages will the PCIe bus give?


I think the hope is that a PCIe bus will be considerably faster.
Depending on the revision, mini-PCIe (which is 1x) should be able to handle 250MB/s – 3.9GB/s
Whereas USB 3.0 tops out around 640 MB/s.

Hopefully they will be able to avoid some of the initial LimeSDR design snafus, for example the issue with the antenna matching network that required many early boards to be modified before it could receive effectively on HF (something I am proud to say I did myself with a soldering iron and a toothpick).


Thank you for the point!

PCIe is also more reliable and using DMA and zero copy interface it’s possible to get lower latency and lower CPU utilization. mini-PCIe by standard is x1. We’re using optional pins (that was reserved for second lane initially but didn’t go) and routed them, using special converter board it’s possible to get x2 connectivity. Having PCIe 2.0 x2 you can get up to ~7.2Gbit/s or near 900 MB/s.

Wikipedia says “Accounting for the encoding overhead, the raw data throughput is 4 Gbit/s, and the specification considers it reasonable to achieve 3.2 Gbit/s (0.4 GB/s or 400 MB/s) or more in practice”. In practice I’ve never seen something goes beyond that.

We’re aware of the initial LimeSDR design issue.

Roger T

Low Latency, High speed/bandwidth are main advantage.
But the board looks very tight and I do not know if the XTRX try to perform full potential, it might get super hot and may be unstable. You probably make your own or buy heat sink for sure .


You’re right! We’re working hard to get into 2.5W power budget defined by miniPCIe standard and it’s tough. So far it was 3 revisions of the board and now it works stable! Will post more data about temperature/power consumption when we accurately measure it.