Our Review of the Airspy HF+: Compared against ColibriNANO, Airspy Mini, RSP2

Over the last few months we’ve been posting and getting excited about the Airspy HF+, an upcoming high dynamic range HF/VHF receiver designed for DXing. The Airspy team were kind enough to supply us with an early pre-production unit for review.

Long story short, the Airspy HF+ is probably one of the best low cost SDRs we’ve seen for DXing or weak signal reception out there. So far few details on the availability of the HF+ have been released, but we’re aware that preorders are due to start soon, and the target price is expected to be $149 USD from iTead Studio in China. 

What follows is the full review and comparisons against other similarly priced SDRs. The Airspy team want us and readers to understand that our review unit is a pre-production model, and apparently already the matching and thus SNR has already been improved by about 2-4 dBs, so the sound samples we provide in the review below should sound even better with the newer revision.

Disclaimer: We received the HF+ for free in exchange for an honest review, but are not affiliated with Airspy. We’ve been in contact with the Airspy team who have helped clarify some points about the architecture and technology used in the design.

Introduction

The Airspy HF+ is designed to be a HF/VHF specialist receiver with a frequency range of DC to 31 MHz, and then 60 to 260 MHz. It has a maximum bandwidth of 768 kHz. So the question is then, why would you consider buying this over something like the regular Airspy R2/Mini or an SDRplay RSP2 which both have larger frequency ranges and bandwidths? You would buy the Airspy HF+ because has been designed with DXing and weak signal reception in mind. Basically the main idea behind the HF+ is to design it so that it will never overload when in the presence of really strong signals. Combined with it’s high sensitivity, weak or DX signals should come in much clearer than on the other radios especially if you have strong blocking signals like broadcast AM/FM around.

Aside: What is overloading, intermodulation and dynamic range?

Basically strong signals can cause weak signals to be drowned out, making them not receivable, even though they’re there at your antenna. This is called overloading or saturation. Intermodulation occurs when the SDR overloads and results in images of unwanted signals showing up all over the spectrum.

A simple analogy is to think about what happens when you are trying to drive, but there is sunstrike. The road is very hard to see because the sun is so bright and right in your eyes. The human eye does not have enough “dynamic range” to handle the situation of sunstrike. Dynamic range is a measure of how well a radio (eye) can handle strong (bright) and weak (dark) signals at the same time. The same analogy applies to radios which can struggle to ‘see’ weak signals if there is a very strong signal nearby on the frequency spectrum. There are a few ways to solve this:

  • Filtering: Block the strong signals that you don’t want using LC filters.
    • Eye analogy: using your sun visor to block the sun.
  • Attenuation: Reduce the strength of all signals.
    • Eye analogy: using sunglasses or squint.
  • Increase dynamic range: Get a better SDR with better design/technology and more bits in the ADC.
    • Eye analogy: upgrade your eyes.

Technology and Architecture

The HF+ uses a typical Filter->Tuner ->ADC architecture. So it is not a direct sampling receiver like most of the more expensive SDRs. Direct sampling receivers directly sample the analogue spectrum, without the need for a tuner so they avoid losses and the intermodulation problems that usually come from the mixing stages. But there are some major cutting edge technology differences in the HF+ architecture that should make its performance even better than direct sampling receivers.

Tuner: The tuner on the HF+ is one of the first to use a “Polyphase Harmonic Rejection” architecture. Essentially this means that harmonics produced in the mixing stages are naturally rejected, making the front end filtering requirements much more relaxed. So unlike the tuners used in other SDRs, this one is extremely unlikely overload in the mixing stage.

An additional benefit to this architecture is that the mixer is very low loss, so the LNA in the tuner only needs to use low gain, giving it a very high IIP3 value. So the first LNA which is typically another point of saturation and imermodulation, is very unlikely to saturate in the HF+ design. Most of the amplification only occurs after the mixing stage with the filtered narrowband output of the tuner.

Analogue to Digital Converter (ADC): The ADC is 16-bits and uses a “Sigma Delta” (ΣΔ) design. Basically a Sigma Delta ADC has a natural filtering ability due to its narrowband nature. Instead of seeing say a 30 MHz signal, it only sees 1 – 2 MHz, thus increasing dynamic range and reducing the likelihood of out of band overload.

Digital Down-Converter (DDC): Then after the ADC is a DDC which decimates the output from the ADC, increasing the effective number of bits. The more bits the larger the resolution of the digitized RF signal, so weak signals are less likely to be lost when converted from analogue to digital.

The HF+ Block Diagram
The HF+ Block Diagram

So the block diagram flow goes like this:

A weakly filtered signal enters the tuner, is weakly amplified by the tuner LNA, mixed down to baseband and filtered to 1-2 MHz. It is then amplified and sampled with the sigma delta ADC into 16-bits. The DDC decimates the output into 18-bits which is then sent to the microcontroller and PC via USB.

The Airspy team also compiled this comparison chart for us to understand the differences in architecture between the current SDRs on the market (click to enlarge). This shows that the HF+ is a different type of design compared to other SDRs. Generally the best SDRs out the market right now are direct sampling receivers with many filter banks. The HF+ approaches the problem in a different way, and according to the specs seems to match or better the performance of heavily filtered direct sampling receivers.

Performance from the Airspy HF+ product page is stated as:

  • -141.0 dBm (0.02 µV / 50 ohms) MDS Typ. at 500Hz bandwidth in HF
  • -141.5 dBm MDS Typ. at 500Hz bandwidth in FM Broadcast Band (60 – 108 MHz)
  • -139.5 dBm MDS Typ. at 500Hz bandwidth in VHF Aviation Band (118 – 136 MHz)
  • -139 dBm MDS Typ. at 500Hz bandwidth in VHF Commercial Band (136 – 174 MHz)
  • -138 dBm MDS Typ. at 500Hz bandwidth in the upper VHF Band (> 174 MHz)
  • +26 dBm IIP3 on HF at maximum gain
  • +13 dBm IIP3 on VHF at maximum gain
  • 110 dB blocking dynamic range in HF
  • 95 dB blocking dynamic range in VHF

Software and User Experience

The Airspy HF+ runs on the standard SDRSharp software. The first thing you notice when selecting the HF+ on the SDRSharp menu is how simple the controls are. There is no gain control – the AGC algorithm automatically adjusts the internal gain for maximum SNR, whilst ensuring zero overloading. The only control is the bandwidth selector where you can select from 768 kHz, 384 kHz, 192 kHz, 96 kHz and 48 kHz. Browsing the spectrum without having to adjust the gain slider is quite a liberating experience and the AGC always seemed to optimize the reception nicely.

All the controls you get/need for the HF+
All the controls you get/need for the HF+

The HF+ is also compatible with the SpyServer software, which allows you to stream the data radio IQ data over a network. SpyServer saves network bandwidth by sending only the currently actively tuned IQ signal plus the waterfall data. This is in contrast to other SDRs like KiwiSDR which send only compressed audio, or rtl_tcp which sends the full IQ data. Sending the IQ data rather than compressed audio allows you to perform various DSP algorithms to the signal on the host side, such as noise filtering. Sending the IQ data (even if only a slice of it) still uses significantly more bandwidth compared to sending compressed audio however, so internet connections and wideband signals such as BCFM may not work well together over long distances and slow internet connections.

External Design/Photos

Note that our pre-production unit does not have the completed metal finish to it yet. The final version is supposed to have a more aesthetically pleasing metal finish applied to the enclosure.

The HF+ is about the size of a pack of cards, and comes in a 90 x 55 x 3 mm metal enclosure with the Airspy HF+ logo stamped onto the top. This thick enclosure gives the HF+ quite some weight at 190 g and a very sturdy feel to it. There are two SMA ports on the left for HF and VHF antennas, and a USB micro port on the right. Two small status LEDs are placed near the SMA ports.

Inside is the PCB, and the main RF circuitry is shielded with a metal can (ignore the poor soldering on the can as this was removed and replaced by us when performing a small mod to the pre-production unit). This double shielding means that the HF+ is well protected against stray RF and USB noise. Also, one interesting feature is the use of a grounding spring on the bottom plate which ensures that the USB connector is grounded with a low impedance connection to the metal enclosure.

Comparison SDRs

In this review we are doing side by side comparisons of the HF+ against similarly priced SDRs, including the ColibriNANO, Airspy Mini + SpyVerter Upconverter and the SDRplay RSP2.

NameFreq. RageBandwidthTechnologyPrice 
Airspy HF+

DC to 31 MHz

60 to 260 MHz

768 kHz – 18 bit

Polyphase Harmonic Rejection

$TBA – “Under $200”.

Airspy team note “expected to be ~$149 USD at iTead Studio”

 
ColibriNANO

100 kHz – 55 MHz

Up to 500 MHz undersampling.

768 kHz – 24 bit

3 MHz – 16 bit

Direct sampling + LPF Filter$350 USD 
Airspy Mini24 – 1800 MHz
Down to DC with SpyVerter Upconverter
6 MHz – 12 bitUpconverter + Direct Conversion

$99 USD

$148 USD (incl. SpyVerter)

 
SDRplay RSP21 kHz – 2 GHz10 MHz – 12 bitUpconverter + Direct Conversion$169.95 USD 

Comparison Tests

In these tests we compare each SDR on a real world signal. SDRs are cycled through, taking screenshots and recording audio as fast as possible to ensure that conditions don’t change. To verify conditions didn’t change part way through we go through our loop twice to confirm that similar results are recorded.

The HF and below tests use a Wellbrook Loop antenna. VHF Tests use a discone or dipole tuned for the tested band. The RF environment is one with strong broadcast AM and FM stations. The location is 10km away from an AM tower, and LOS to the FM/TV transmitter tower.

In all cases the signal of interest is optimized for best SNR without overloading the SDR. For each SDR we used the officially recommended software package. For the Airspy devices this was SDRSharp, for the ColibriNANO this was ExpertSDR and for the RSP2 this was SDRUno.

With the RSP2 we used the recommended HiZ port for all LF – HF signals and also flipped between then Zero and Low-IF mode choosing the best one. Just to be sure, we tested the A and B ports on the RSP2 as well, but experienced heavy broadcast AM overload (with the filters turned off) and weaker signals than with the HiZ port with the filters on, so did not continue to use these ports.

On the ColibriNANO we used bandwidths at or below 768 kHz to get the 24-bit output.

A modern 2016 Core-i7 laptop run on battery power is used for all tests, but all SDRs were confirmed to run smoothly on an older model Core-i5 desktop PC.

LF (Low Frequency 40 kHz Time Signal)

http://Airspy%20HF+

Airspy HF+

http://ColibriNANO

ColibriNANO

http://Airspy%20Mini%20+%20SV

Airspy Mini + SV

http://SDRplay%20RSP2

SDRplay RSP2

http://RTL-SDR%20+%20SV

RTL-SDR + SV

This signal is a 40 kHz time signal originating from Japan. It is know as the Ohtakadoya-yama LF Standard Time and Frequency Transmission Station (NICT).

From the screenshots we can see that the only SDRs successful at receiving this station where the HF+ and the Airspy Mini + SV.

  • The HF+ comes in with a very clear copy and there is no sign of overloading from broadcast AM.  VLF signals down to 20 kHz are also visible and copyable.
  • The Mini + SV receives the signal too, but there is significant overloading from broadcast AM stations present all around the signal. 
  • The ColibriNANO cannot receive the signal at all. According to the advertised specifications the ColibriNANO starts receiving at around 100 kHz so this is expected. From the screenshot we start to see a response at around 70 kHz.
  • The RSP2 just barely receives the signal (a very faint line is visible in the waterfall), but no audio was copyable. There is some minor signs of overload from broadcast AM as well.

NDB’s (~325 kHz)

NDB’s or Non-Directional Beacons are beacons used to aide with aircraft navigation. In this test all SDRs were able to receive NDBs with good performance and it was difficult to notice a difference between SDRs.

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini%20+%20SV

Airspy Mini + SV

http://ColibriNANO

ColibriNANO

http://SDRplay%20RSP2

SDRplay RSP2

Broadcast AM (Day)

Here we tested broadcast AM during the day. During the day local broadcast AM is generally stronger and more likely to overload a receiver. Distant stations come in weaker. We tuned to a weak station and tested reception.

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini%20+%20SV

Airspy Mini + SV

http://ColibriNANO

ColibriNANO

http://SDRplay%20RSP2

SDRplay RSP2

Broadcast AM (Night)

Here we tuned to the broadcast AM band and tested reception with one of the weaker signals.

http://Airspy%20HF+

Airspy HF+

http://ColibriNANO

ColibriNANO

http://SDRplay%20RSP2

SDRplay RSP2

2.6 MHz FAX

Receiving a fax signal about 1 MHz above the broadcast AM band.

http://Airspy%20HF+

Airspy HF+

http://ColibriNANO

ColibriNANO

http://SDRplay%20RSP2

SDRplay RSP2

7 MHz Shortwave

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini%20+%20SV

Airspy Mini + SV

http://ColibriNANO

ColibriNANO

http://SDRplay%20RSP2

SDRplay RSP2

88.9 MHz FM

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini

Airspy Mini

http://SDRplay%20RSP2

SDRplay RSP2

Weakly received from this location, TX 50km away and not designed to cover the RX region.

96.03 MHz FM

This is someones private repeater of a low power FM station which seems to be illegally radiating. Very closely spaced to a powerful station on the frequency spectrum.

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini%20+%20SV

Airspy Mini + SV

http://ColibriNANO

ColibriNANO

http://SDRplay%20RSP2

SDRplay RSP2

96.2 MHz FM

Receiving a regular non-E’s FM station about 120km away.

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini

Airspy Mini

http://SDRplay%20RSP2

SDRplay RSP2

http://ColibriNANO

ColibriNANO

Pagers

In this location we have some very strong pagers at 158 MHz and most SDRs show signs of overloading near to the pager frequency. The HF+ seemed to handle them quite well, however when two transmitted at once there was a about a 100ms period of overload before the AGC kicked in to reduce the gain.

HF+ Receiving strong pagers. Two transmitting at once.

Here we tested a weak signal about 2.5 MHz below the pagers.

http://Airspy%20HF+

Airspy HF+

http://Airspy%20Mini

Airspy Mini

http://SDRplay%20RSP2

SDRplay RSP2

  • The HF+ was able to clearly receive this data station without any sign of overload from the pagers.
  • The Airspy Mini could also receive the station, but much weaker. Turning up the gain any further caused overload, and caused pager + WFM noise to appear over the frequency whenever the pager transmitted. Even at a gain setting of 10 there was some mild interference noticeable in the screenshot when the pager transmitted.
  • The RSP2 could also receive the station with fairly good strength, but intermodulation was severe whenever the pager transmitted, causing a loss of signal. Turning down the gain did not help with the interference, and only reduced the signal of interests’ strength further. Enabling the MW/FM filter did not help as the pager interferer is outside the notch range.
  • The ColibriNANO could not receive this station.

Conclusions

The Airspy HF+ is an exceptional SDR and will truly please any DXers or people wishing to listen to weak stations. It is a relatively narrowband SDR (in comparison to say the Mini/R2 and RSP2) that can only tune up to 260 MHz, so don’t expect to be able to use it as a wideband scanner for trunked radios for example. But on VHF it would perform very well on FM DX, airband voice scanning and for 137 MHz WX satellites.  The reception on the HF+ is almost entirely unaffected by extremely strong pagers in the 157 MHz region.

Below 30 MHz the HF+ also shines. VLF to MW is the best we’ve seen on any sub $300 SDR. Overload is non-existent on broadcast AM, and no effects from the strong AM signals can be seen further up on the spectrum.

The closest competing unit to the HF+ in terms of price and use cases (designed for HF) is probably the ColibriNANO. But the ColibriNANO commands a decently higher price at $350 USD. Performance on HF seems similar, but we do have to give a slight edge to the HF+. The ColibriNANO also has the downside of poor LF/VLF reception (advertised response starts at 100 kHz), and heavily aliased VHF/UHF due to undersampling. A filter is needed for proper operation on VHF/UHF. That said the ColibriNANO itself is a very good SDR, but the HF+ certainly wins out in terms of value and general performance and we can’t see any situation where the ColibriNANO would be a better choice at the moment.

The Airspy Mini/R2 and SDRplay RSP2 also generally perform well for the majority of signals, but will struggle when it comes to really strong signals. Comparing against these SDRs on weak signals near strong blockers really shows where the HF+ shines. But when compared against regular (non-weak) signals or in a tame RF environment without strong signals then it is pretty much impossible to determine which SDR is better.

So there is obviously some brand new cutting edge technology going on in this receiver with the polyphase harmonic rejection mixer and the sigma delta ADC which possibly even puts it on top of the very expensive direct sampling SDRs. On the HF+ weak and DX signals are noticeably more accessible. Performance for the price (expected $149USD) is phenomenal. This is a highly recommended SDR.

Disclaimer: We received the HF+ for free in exchange for an honest review, but are not affiliated with Airspy. We’ve been in contact with the Airspy team who have helped clarify some points about the architecture and technology used in the design.

70 comments

  1. Mike

    What is the sample rate of the Airspy HF+ ?
    (Excuse me if I missed it in the article, but I’m not familiar with all the different terminology – I am new to the radio world.)

  2. rwgast

    Fifi uses a tayloe mixer front end and an audio codec as its interface to the pc, I believe. First thing is first that type of SDR Architecture will never be able to see more bandwidth than 192khz at a time. As far as channel width is concerned that is fine for the most part for the ham interested in DXing, SWLing, CW, Contests, but sometimes when you step out of that part of ham radio the signals are much wider than 192khz but even if you are interested in “normal” ham stuff being able to see 1mhz + of the spectrum is pretty useful. Sec. ond problem is the Tayloe Detector/Mixer, it works and can work well but will always be limited by fast bus switch IC performance, and minute differences in the phase of I and Q not being 100% 90 degrees. Packing the IQ demodulator, preamp, up/down conversion stages in silicon is a much cleaner way to do things with a better NF and image rejection and laser trimmed and matched components. Not to mention it eliminates any parasitic capacitance, inductance, and impedance variation caused by the pcb traces linking the parts of a QSD/Tayloe Detector.

    As far as bits go, when the SDR can sample more than a 192khz at a time it can make use of all the bandwidth it sampling and turn it in to dynamic range. So your FiFi will always sample 192khz or 48khz of spectrum at the codec rated 24 bits (and only 18 to 20 of those bits are probably usable). While and RF Space can sample a whole 60mhz of bandwidth at once @16bits and then through decimation it will digitally scale that 60mhz down to the 1mhz or 500khz you want to see and this effectively gives you an 18 to 24 bit sample instead of 16bit. An AirSpy samples 10mhz at a time and I see 80db of dynamic range, when I decimate the input 64 times I end up with I thin 120khz of bandwidth at -100dm effectively my 12 bit sdr is no a 14 or 16 bit device. When sensitivity and dynamic range isn’t needed I can run at default 10mhz sample rate and decode a 6mhz analog tv stream.

    QSD’s like FiFi, SoftRock, on and on had there time, they introduced us normal folk to what was only military technology at the time and inspired technical people to push SDR performance to the next step(s). That being silicon tuners/transceiver like the AirSpy, RTL, Ettus, LimeSDR, BladeRF, Analog Devices Pluto and they also inspired the Digitally Down Converted “Cream of the Crop” HF SDRs like Flex,RF-Space,WinRadio,HPSDR, and anything else with a 1000 dollar price tag.

    If your happy with your FiFi and it works for you keep it and use it no reason to upgrade…. But when you want to listen to trunked signals and decode wifi or spread spectrum it will be time to think past the 24 bit audio codec SDR.

    • Bernie

      If the ADC was the only parameters in the affair, everybody would be using SoftRocks by now. The Tayloe Mixer itself generates harmonics that harm its linearity and the linearity of the stages behind it. Interestingly the sound card based designs can be seen as a less sophisticated version of the HF+ architecture. They also use a “high linearity” mixer and a sigma delta ADC (audio codec).
      In terms of raw number of bits, the 18bits of the HF+ at 768 kHz can be decimated to audio level bandwidth of 48 kHz and give 20 effective number of bits (ENOB). And given the level of integration, better linearity if the harmonic rejection mixer and probably some other fancy filtering, the HF+ could still take the edge on 24bit narrow band receivers. The best and most expensive audio codecs nowadays hardly achieve 20 ENOB.
      I hope someone can compare these radios.

    • Buzz

      It’s always fascinating to see how people cannot grasp how modern SDR technology surpassed softrock and other QSD derivatives like the fifi, foofoo etc. HAM radio folks are probably living under a rock.

      • Bernie

        Interesting. So, Fifi has 107 dB at 192 kHz (17.5 ENOB) vs. HF+ with 110 dB at 768 kHz which translates to 116 dB at 192 kHz (19 ENOB).

        • Beer It

          The Fifi has a low jitter 61.5 ppm (3.3V) Si570 oscillator with 0.3 Jitter (ps RMS) vs 0.5 ppm high precision, low phase noise clock in the HF+ which I would guess would be X2 on the back side of the PCB.

  3. admin

    Raised the FM dipole and tested BCAM during the day. Stronger signals on both accounts.

    Added two new results at 88.9 MHz and BCAM (Day) to the post.

  4. rwgast

    The AirSpy people do deserve props for making SDR’s with great specs for a good price. That doesn’t mean people like Flex and RFSpace skipped out on engineering though…. There radios are designed very well using tons of DSP engineering in software. The guys at AirSpy are smaller so it is a lot easier for them to test out parts and ideas considered to be on the bleeding edge in order to drive prices down. The bigger guys have a rep to keep and doing things like using the same tuner in there radios the is used in 5 dollar tv dongle doesn’t help keep that quality part high engineering standard rep. RFSpace NETSDR has an optional upgrade board using an r820t2 but the only way I figured it out was to google image the pcb and blow up the pics, it is no where in there material they use it

    I think in a few years the standard tech used in SDR’s will be these new RF ADC chips. They are able to sample up to [email protected] right now and output a digitally downconverted signal. So basically you will have Antenna->RF ADC->USB3/Ethernet transceiver and that is it. There will be no technical reason do use any more parts than that for a standard SDR that just pipe IQ to the PC like an AirSpy. The problem is the ADC chips are brand new and cost 500-2000 dollars direct from Analog Devices and TI. Give it a few years though and the chips will go way down and you will have a dongle that can do direct sampling from DC-6Ghz with 60mhz instant bandwidth @ 16 bits, and 24 bits when decimated down to 1 or 2mhz. The beauty is a dongle can be made out of just two chips and some passives plus an SMA connector for the antenna. The original Lime Transceiver chip used on the bladeRF was $100+ just 3 years ago and now it is only $30. Once these ADC chips are at $100 dollars im pretty sure almost all hi end ham gear will be SDR whether it has a waterfall display or not.

    .s it true that an old version of the Coolibrinano was used in this review as another commenter stated? As I said I want a hi dynamic range, low noise receiver to mix microwaves down too. Due to the AGC the Airspy HF isn’t usable for me. The Coolibrinano, was a big consideration until I read this review, but if the review used an older model than what they sell than im not sure how valid the comparison is unless the reviewer can tell us what has been updated since he received this calibiri unit.

    • admin

      I still can’t find any mention of a newer ColibriNANO, and the product page still refers to 100 kHz as the lower limit. Though even on that page they sometimes specify 0.01 MHz or 0.1 MHz as the lower limit. The ColibriNANO used in this review was purchased 3 months ago.

    • Bernie

      I fully agree. These guys tend to squeeze every bit of performance from general purpose consumer chips – which is amazing.
      The Airspy HF seems to have no problems with overload from AM BC while the ColibriNano had hard time keeping up. I doubt the Airspy is using any advanced filtering due to the limited PCB surface under the metal can, so the dynamic range must be really huge – or there’s some other trickery we are not aware about.
      The AM BC test demonstrated very good in-band performance and the 41m Short Waves test has shown that out-of-band interferers have virtually no effects at all, especially when you consider the time of the test with the strong nightly propagation of AM BC signals.
      In the pager test, it looks like the AGC doesn’t mess with the signal level when reducing the gain. This is the first time I see this behavior in a SDR. When using standard SDR’s as we know them, you will either have the noise floor jumping or will reduce the gain manually and include your own human response time in the loop.
      When tuning 2.5 MHz below the pager, the AGC doesn’t seem to trigger at all despite the pager being probably > 120 dBc higher. This is another mystery to figure out.

  5. KD0CQ

    Wow, what a cluster—- this thread turned into. Regardless of all the noise below I’m still interested to get one in hand and see for myself. Airspy did a fantastic job with the Mini and R2, so I expect the same or better from this offering. If it is even close to the specs provided then it is well worth the price in my opinion. Putting it on my shortlist.

    • LemeurSabre

      My take home message from all the noise here is a receiver that is holding up against hardware costing more than double the price. It looks like the bang for buck is going to be very high.

      As some others have commented, one path is to skip any engineering, throw the costly hardware components at it and produce unaffordable hardware for the majority of us. On the another hand Airspy have been shown to take low cost parts and squeeze every last dB out of them. THIS is what I respect the most.

    • admin

      The ColibriNANO used in the review was purchased about 3 months ago. I’m not sure if it’s already an older model or not. But their official page at https://sunsdr.eu/product/colibrinano/ states the following:

      “ColibriNANO USB SDR Receiver 0.1 to 55 MHz

      325.00 € 299.00 €

      The New ColibriNano takes USB dongle SDR receivers to a new level. This is a direct sampling, high quality SDR receiver covering 100 kHz to 55 MHz with up to 3 MHz sample rate in a sturdy aluminium USB stick.”

      So it’s 100 kHz. But then again lower down on the page the receiving bandwidth is stated at “Receiving bandwidth: 0.01 – 55 MHz”, so it doesn’t quite match up, although they might be referring to the fact that under 100 kHz is filtered for some reason.

      I did see Mile’s videos before I wondered how he got performance that low, but came to the conclusion that it’s possible that his Miniwhip provides high gain at VLF to completely overcome the weaker response under 100 kHz. If there are two versions of the ColibriNANO then this might also explain it.

    • admin

      It’s definitely a real signal, but it is a bit of an odd one. I actually think it’s someones home repeater of a LPFM station which is spewing out long distance. Might try and update for another BCFM test today at a different location.

  6. rwgast

    The few and far between documentation on these mixers makes it hard to understand much… But generally I would say you could use a chip like the ADF4153 to make a 4ghz oscillator, as far as phase noise of the chip im not sure how that would factor in.

    Judging by the size of the AirSpy HF I think they are using single purpose chips for IQ modulation and DDC (Although can someone clear up if DDC in this case means Digital Down Conversion or Direct Down Conversion), BUT I have been searching for polyphase based IQ demodulators and TV tuners all day along with single function DIGITAL Down Conversion chips and have found nothing.

    I was told on twitter that rtl-sdr.com will release pictures of what is under the RF can as soon as the AirSpy HF+ is officially on sale, out of respect for AirSpy. So we will just have to wait and see unless someone finds a PolyPhase based chip that fits the bill here. They could be doing the DDC on a dsp chip or small FPGA but I would think that would drive the cost up.

    BTW if the polyphase IQ demodulator is rolled from discretes I have to give these guys mad props for creating such a crazy good receiver using discretes

    • WTH is Polyphase Harmonic Rejection

      If you look at the patent for “Polyphase harmonic rejection mixer”, it started to be filed nearly 10 years ago, and was finally granted on the 10th of December 2013. It can typically take two to four years from concept to a new chip being produced, and few companies would spend money on production without exclusive rights, so the wheels would have only really started spinning once the patent was secured. In my mind it has to be new chip(s), which probably come with a stack full of NDA’s.

  7. WTH is Polyphase Harmonic Rejection

    After reading more about “Polyphase Harmonic Rejection” I am now wondering how many phases are being used 4, 8, 16, 32, 64, 128, 256. Because lets say that they are mixing at 250 MHz and were using 8 phases, does that mean that a clock rate of 2 GHz is being used ? And if it was 16 phases, would that be a 4 GHz clock rate ? Does that sound right or am I missing something. I will freely admit, that I’m still trying to get my head around this.

  8. Michael

    The HF+ was touted to be as good as or if not better than high priced SDRs such as Perseus and RFSpace so until there’s a comparison against these on HF frequencies I may just hold off buying one, there’s not much difference I can see from the review between the HF+ and the Airspy R2 and SpyVerter. That’s not to say there is not difference I just think the review wasn’t particular usable as a sales tool. Not convinced yet.

      • Michael

        What I think I meant to say was until there’s a side by side comparison of the HF+ against a Perseus or SDR-IQ (or similar RFSpace product) then I’m not convinced the HF+ would be any better than an AirspyR2 plus Spyverter, so why would I sacrifice bandwidth and not buy anR2 and Spyverter. I think RTL-SDR.com should have compared the HF+ against the receivers it was touted to be as good as. So, will I pre-order, no unless there’s another review in quick succession that shows the HF+ is as good as the high end receivers.

  9. rwgast

    Im going to guess no one is going to open up about details of this special mixer, whether its of the shelf and what part it is, let alone take pics with the case removed?

    One thing im curious to see…. is for rtl-sdr.com to add a quick set of pictures using an H mode mixer like the spyverter plugged in to a DDC receiver. Obviously mixing HF to an HF receiver is pointless but it should give us an idea on just how well this polyphase mixer work vs a well respected h-mode. Also all the googling im doing on poly phase mixers seems to be examples and lectures about the mixer being used in microwave and higher applications.

    • admin

      Once they’re shipping we’ll release the pics of inside the case and the chip part #’s etc. It’s a bit too early to release that as they could be copied and they’re not even shipping yet.

    • Adam

      Have you ever see the Airspy H-mode mixer up-converter lab reports? Does anybody measure the real performance of so called H-mode mixer? The H-mode mixer technology they are using, we had tested a few years back and the mixer struggling already above 50MHz. At the 150MHz the H-mode mixer was very difficult to tune for the descent performance. First of all, the FST switches were not designed for that type of operation so fiddling with the currents and getting the proper signal shape after mixing was a real challenge. H-mode mixer does work on the HF but the up-converter use the 120MHz LO for driving the switches and this is a high frequency for that technology.

      Going back to your question, as you spent some time on reverse engineering and found nothing. It is because most probably the “Polyphase Harmonic Rejection” is a fancy technology term that makes us hard in the morning but this beast is not sitting below the can tuner protection that you are so anxious to see. When you do the reverse engineering you have to consider the cost and the technology. If you have a brand new patented technology, this is hard to cost a $20 or so… Secondly, you need a mass production for such a low prices. Take the example of the DVB-T dongle. When you combine the cheap price, mass production and add the frequency range of the receiver you can estimate the device. In this case, the HF+ is using nothing more than a simple Automotive technology car stereo tuner. And yes, any radio tuner has the LW, MW, SW in one band, then the VHF band and the latest DAB/DAB+ band tuning up to 240Mhz. So this is what you have under the can shielding. As there is some writing that this is NXP proprietary technology, check the NXP/Siemens latest Automotive car radio tuners and pick up the one that suits the datasheet. So both Adam Farson and Rob Sherwood are right, first, there is a heterodyne receiver and mixing to the IF and then an SDR sampling.

      Of course, we all know that half of the web writing is a crap, so you may include my post in that crap too. If you find my post interesting, let me know.

      • Paul

        Interesting Adam.

        I’ve seen something like a R820T perform significantly better in design A compared to design B.

        Frankly I don’t care what chips are used, as long as somebody out there puts enormous efforts into squeezing the most performance out of it for us hobbyists. It enables cheaper technology combined with higher specs than we’re used to at that price level.

        I respect that much more than some people just burning and complaining about new devices that generously are made available for very reasonable prices.

        But continue your cruisade if that is more valuable to you.

      • Tech Guy

        So both Adam Farson and Rob Sherwood are right, first, there is a heterodyne receiver and mixing to the IF and then an SDR sampling.

        Do you have any evidence of this assertion?

  10. Larry Dighera

    Here are comments from Adam Farson and Rob Sherwood:
    —–Original Message—–
    From: Adam Farson [mailto:[email protected]]
    Sent: 19 July, 2017 13:45
    To: ‘Rob Sherwood.’
    Subject: RE: Airspy HF+

    Rob, Larry,

    Just looked over the Airspy HF+ web page. Firstly, a look at the block
    diagram tells me that this is a superhet receiver, with analogue mixers and
    LO’s for HF and the two upper bands feeding IF’s into an ADC. Hence the
    “image rejection” term is relevant. (As an aside, the non-native English
    writing style of the website is a little weird.)

    No RMDR value is stated, but the specified 100 MHz phase noise value does
    not look too good when expressed as RMDR.

    “Very low phase noise PLL (-110 dBc/Hz @ 1kHz separation @ 100 MHz)”.
    Assuming B = 500 Hz, this resolves to 83 dB RMDR – less than stellar by any
    account.

    The device appears to have a varactor-tuned tracking preselector. Due to
    parametric effects in the varactors under strong-signal conditions, this can
    be a horrendous IMD generator, as we saw with the early JRC receivers and
    transceivers.

    Rob, I agree completely with your comments regarding receivers vs.
    transceivers and the uses to which dongles are put. For HF and 6m, I believe
    that the Russian Colibri Nano is a much more interesting device than the
    Airspy. It is a 14-bit direct-sampler. Here is my Nano test report:

    http://www.ab4oj.com/sdr/sunsdr/nano_notes.pdf

    I am more than willing to test serious contenders in the receiver world. For
    example, Icom Canada will loan me an R-8600 for testing, probably next
    month. I am no longer motivated to test any purely legacy architecture;
    although I am retired and have more available time, I do have a life outside
    my little 6.5m2 lab.

    73, Adam VA7OJ/AB4OJ

    —–Original Message—–
    From: Rob Sherwood. [mailto:[email protected]]
    Sent: 19 July, 2017 12:28
    To: [email protected]; [email protected]
    Cc: [email protected]
    Subject: RE: Airspy HF+

    Hi Larry,

    Just a couple of comments. IP3 is a meaningless number for a direct
    sampling radio. I also don’t understand the meaning of “image rejection” as
    applied to a DS SDR.

    There must be quite a bit of preamp and/or ADC driver gain to push the noise
    floor down to the levels quoted. What is the ADC overload value in dBm for
    any preamp settings?

    What is the phase noise at 10, 20 and 100 kHz on 20 meters? That is likely
    more important than at 1 kHz offset where in reality if listening to a weak
    CW signal next to a very strong CW signal, the adjacent signal’s key clicks
    will dominate before phase noise.

    Is there an RMDR spec for the receiver at 2 and 20 kHz, as is now usually
    quoted in any review in QST?
    One would expect the RMDR value to be HF ham band independent.

    When I was writing for Passport to Worldband Radio for its 25 years of its
    existence as their laboratory, receivers as opposed to transceivers would
    have been of prime importance. Today with a significant drop in shortwave
    broadcasting, I don’t think the shortwave market is a shadow of its former
    self.

    On the other hand, SDR dongles are being used for spectrum scope for
    transceivers missing this feature, and for CW Skimmer and reverse beacon
    reception.

    At present with new product announced at the Dayton Hamvention, I am only
    planning to test DS SDR transceivers this year. Examples: Icom IC-7610,
    Flex 6600M, Apache ANAN 8000DLE, and a second and possibly a third sample of
    the Elecraft KX2. I have the second KX2 here now in my possession as of
    last evening.

    Adam may have more time than I for testing receiver only products, as I am
    still working part time with a sub set of my IT client base, at least for a
    few more years.

    73,
    Rob Sherwood
    NC0B

    —–Original Message—–
    From: Larry Dighera [mailto:[email protected]]
    Sent: Wednesday, July 19, 2017 9:55 AM
    To: Rob Sherwood. ; [email protected]
    Cc: [email protected]
    Subject: Airspy HF+

    Gentlemen,

    I’d be interested in your opinions on this new entry into the SDR market, as
    I’m sure many others would be also. You may be able to request a sample for
    testing from: [email protected]

    Technical specifications

    HF coverage between DC .. 31 MHz
    VHF coverage between 60 .. 260 MHz
    -138 dBm MDS at 500Hz bandwidth in HF
    -142 dBm MDS at 500Hz bandwidth in VHF
    +26 dBm IIP3 on HF at maximum gain
    +13 dBm IIP3 on VHF at maximum gain
    110 dB dynamic range in HF
    95 dB dynamic range in VHF
    120 dB Image Rejection
    0.1 dB Spectrum Flatness
    0.5 ppm high precision, low phase noise clock
    660 kHz alias and image free output
    18 bit Digital Down Converter (DDC)
    +10 dBm Maximum RF input
    Very low phase noise PLL (-110 dBc/Hz @ 1kHz separation @ 100 MHz)
    2 x High Dynamic Range Sigma Delta ADCs @ up to 36 MSPS No Silicon RF switch
    to introduce IMD in the HF path Routable RF inputs Wide Band RF filter bank
    Tracking RF filters Sharp IF filters Smart AGC with real time optimization
    of the gain distribution All RF inputs are matched to 50 ohms
    4 x Programmable GPIO’s
    No drivers required! 100% Plug-and-play on Windows Vista, Seven, 8, 8.1 and
    10
    Industrial Operating Temperature: -45°C to 85°C

    Best regards,
    Larry
    WB6BBB

    • Bernie

      Larry, do you really, honestly think Adam Farson has a clue what’s going on there?
      You are shaming yourself by posting such noise in a product review that takes a good radio like the Colibri as a reference.

    • Tech Guy

      So for Rob Sherwood this new radio is a Direct Sampling design, and for Adam Farson this is a legacy heterodyne receiver, and obviously both are wrong here.
      I think they just skipped your long and annoying email and didn’t read the specs.

    • Paul

      Both gentlemen didn’r read very well.

      Adam is only interested in “legacy” designs carrying the Icom brand,
      a company which only just started to deliver SDR based radiis with less than optimal dynamic behavior.

      Rob is generally unbiased but just not into SDR.

      Did they agree to forwarding their private mail btw?
      It may not look too professional dropping some loose first impressions, for them.

  11. Drone

    Where are the DETAILS??

    Schematic and BOM, NO (A BIG PROBLEM)
    L.O. phase-noise, NO (only at 1KHz offset)
    External L.O. ref. input, NO (A BIG PROBLEM)
    Rx noise figure vs. freq., NO
    Rx ret. loss vs. freq., NO
    AGC range details vs freq., NO (A BIG PROBLEM)

    The worst thing about this “new” design is a lack of DETAILS about this HF “polyphase harmonic rejection (HR) mixer”. MORE DETAILS PLEASE!

    This SDR seems to target the HF band. But yet there is NO external reference connector. HF work today is VERY centered on small-signal work, especially when it comes to very narrow band digital modes. By not providing an external reference clock input, this design is an EPIC FAIL when it comes to small signal work.

    Without more DISCLOSURE about this “new” product, I would NOT recommend purchase. Web “reviews” are one thing. KNOWING how a product actually works and why is FAR more important.

    • Paul

      You see a lot of problems for $149 ! I see a lot of great DX opportunites 🙂

      (hope your shift key still works btw)

    • Useful Gain

      Adam Farson probably never heard the word Polyphase of this entire life. 😂
      Yet, we can clearly see (and hear) how this architecture beats everything known to human kind.
      I’m sold.

    • KWRXLB

      You must be coming from Communist North Korea. People don’t work for years to share schematics and the fruit of their thinking with communists like you.
      In ‘Murica nothing is free, and this radio looks like too cheap and too good to be true.

    • calamityjohn

      BAAAAHAHAHAHA An anonymous blog commenter not recommending a random product in the Internet is claiming his RIGHT to get the SCHEMATICS. THE Internet IS doomed!

  12. rwgast

    I would really like to see what is under the RF Shield, is there any possibility you can desolder and take some pictures for us?

    I currently have a few LTC2217 105msps/16bit ADC’s I planed on building a DDC receiver with and I would like to see what the HF+ is doing for DDC without using an FPGA. Im also interested to see if its mixer is handmade or an off the shelf product. My interests are decimating large ADC samples all the way down to very narrow band channels to get the best SNR possible.

    Im in to Radio Astronomy, and weak interstellar signals. I’ve been looking in to getting a DDC receiver like the Net-RF line or even a HPSDR stuff, so I would have a very good platform to mix 407mhz/1420mhz/8ghz down too. The problem is they are so much more expensive than my airspy, and the performance edge a DDC receiver gives isn’t worth an extra $900 dollars, at least to me! Im really impressed with this HF+ and I thought it may be the ticket (especially since im use to the airspy api) but if you cant disable AGC I have to pass, unless a firmware mod adds manual gain down the line. If I cared about receiving HF for HF radio sake though by far this is one of the best HF SDRs besides maybe FLEX I have seen. Im surprised it did so much better than the Calibri, which I have also considered as an off the shelf solution for myself. These guys just keep pumping out great hardware for the price if they stay on this path they will be a huge player when SDR is the main stream. I just wish there core software was a little more open or Linux friendly.

    • admin

      They don’t want us to release whats under the hood until they are actually shipping. That’s understandable as full images would let people copy the design before they could even get it out.

  13. rlwsdr

    Great review! Looking forward to cracking open my wallet and getting one. Back calculating from the MDS specs, it appears that the noise figure on HF is around 6 dB. Very decent for such an inexpensive receiver and sensitivity is more than enough for HF.

    Implied from both the review and the data sheet on the Airspy HF+ website, there does not appear to be any to control the AGC for the HF tuner. The ability to adjust or at least turn the AGC off is important when decoding HF digital transmissions. If the AGC can’t be turned on and off, it will be interesting to see how well the Airspy HF+ performs particularly when decoding some of the weaker digital transmissions such as JT65-HF and JT9.

    • Paul

      I think in this case the mentioned AGC is pre-ADC, adjusting the HW gain stages in a smart/selective way. Don’t think DSP AGC will be in the same loop, so you can still adjust and disable your AGC like you are used to.

      • rlwsdr

        Thanks Paul. I think we’re talking about the same AGC. The purpose of the AGC, I’m concerned about, is to level-off or reduce or, better yet, optimize the input RF power of the transmission prior to it hitting the ADC to avoid overload. If that occurs automatically, weaker digital signals may be more difficult to successfully decode in presence of a nearby stronger signal regardless of the ability to adjust the audio gain. The proof though is in real world testing.

  14. DK8OK

    Thanks for what I consider the very first hands-on test of a practitioner after we had read some pro and con, just referring to the paperwork. Both sounds profund – the receiver, and the test report given. Nice and convincing examples of daily cases. 149 US$ really is a surprise for such a high quality.
    BTW: I really miss an output for steaming sunspots up into today’s ionosphere …
    73 Nils, DK8OK

    • Paul

      “BTW: I really miss an output for steaming sunspots up into today’s ionosphere …” In that case a 10 to 20dB better MDS compared to most Direct Sampling SDRs certainly helps, Nils! Unfortunately in most examples the Signal+(galactic/band)Noise combined are wel above the RX self noise making this (vast!) sensitivity diffence mostly unnoticable.

  15. Adam

    Let say good value for the money, but not the serious dx-er first choice. If this receiver will have the continuous coverage up to 2GHz and a BW of 5MHz than this will be interesting radio for $149. I see that gain of the product is on HF. Well, only a 31 Mhz lowpass filter on the HF is not enough for the serious performance. When it comes to the HF performance you should not forget the reciprocal mixing problem and there is no mention about that. Your oscillator reference should be high quality to meet that demands and just from the size of it we can see that this may be a big problem. The quality reference oscillator is bulky and expensive, at least the one we can find in the latest HAM HF radios. I understand that the portable size of presented receiver can not accommodate such an oscillator and the price will be at least double but an external 10MHz reference input should be a wise option.

          • Adam

            So, Bernie explain to me what is the influence of the harmonic rejection mixer on the reciprocal mixing?
            How can harmonic rejection mixer cure the phase noise problem of the local oscillator, the main cause of the reciprocal mixing? At the end, you still have to use the rusted 70’s RF oscillator technology.

            • Bernie

              Why would you expect a “Polyphase Harmonic Rejection Mixer” to behave like a standard mixer that shits its pants and suffers from its own products? The patent seems to say that the unwanted products are suppressed by the mixer architecture itself, just like Zero-IF and Low-IF achieve image rejection with proper phasing. They seem to have cracked the linearity and noise problems of the LNA as well by using multi-stage “polyphase” amplification.

              Your statement remind me of RF engineers from the 70’s era who thought Zero-IF was unpractical because no one can filter images that are this close to the LO. Yet it’s a proven technology at the heart of everything we use today from TVs, smartphones to cable modems, etc.

              Take a breath and read carefully like we are all doing.

              • Adam

                An oscillator that is used by the mixer has nothing to do with the mixer own products. The mixer is just a device the mix the input signals, garbage in – garbage out. If your oscillator has a bad Phase noise, the output result from the mixer will be affected by the phase noise of the bad oscillator and no mixer can cure that problem. The mixer is not an intelligent device that can mix only the certain frequency signal. It will mix all you bring to the input, in this case all the oscillator signal skirt and mixed with the inband or the out-band signals create problems that may mask the weak signals. In that case, the dynamic range will not be 110dB as advertised but much lower.

                OK, let’s wait for the lab reports and measurements, but from the video presented I do not see such an impressive advantage over the other SDR radios from the test using “crapy RF technology from 70′”

                • Paul

                  “OK, let’s wait for the lab reports and measurements” Exactly. It would be professional holding assumptions untill after that in the future. It seems to me you don’t like any product competing with your own. In that case: start making SDRs yourself and prove you can do much better. Like the 2GHz/5MHz DDC SDR you mentioned below $149. Let us all know when it’s ready.

                • bloody paywalls

                  It is a real pity that most of the documents I can find are behind pay walls, but the abstract for at least one of them (“Distortion cancellation by polyphase multipath circuits”) does sound interesting:
                  “It is well known that in balanced (or differential) circuits, all even harmonics are canceled. This cancellation is achieved by using two paths and exploiting phase differences of 180/spl deg/ between the paths. The question addressed in this paper is: what distortion products (harmonics and intermodulation products) are canceled if more than two paths (and phases) are used? These circuits are called polyphase multipath circuits. It turns out that the more paths (and phases) are used, the more distortion products are canceled. Unfortunately, some intermodulation products cannot be canceled without also canceling the desired signal. An analysis of the impact of mismatch between the paths shows that the suppression of distortion products will be larger if more paths are used. As an application example, the design of an upconversion mixer with a clean output spectrum is presented.”

    • WTH is Polyphase Harmonic Rejection

      I suspect that you totally missed the new technology used in this, created by “University of Twente” and patented by NXP, B.V., from what I can find online. I’m still trying to digest the patent, but I strongly suspect your comment may be obsolete and may need to be upgraded.

      • Tech Guy

        From what I understand, this technique improves the linearity of the mixer and LNA while reducing the filtering requirements using some obscure phasing techniques. Exactly what we need in SDR’s. Looking forward to getting one of these little gems.

        Good job boys!

    • admin

      The design is supposed to reduce the phase noise requirements naturally. I have a decent low phase noise siggen on the way in a few months, so should be able to test this then…

  16. snn47

    Thank you for the comparisson. I looked closer ont the two FM signals, assuming they were the most stable and continiouly received signals and noticed Spikes. Those spike can be seen in the HF+ and Aispy pictues on 96.03 MHz + 0.2 MHz, and 96,2 MHz + 0.45 MHz? Is this a real (local) signal overlapping, or something the two SDR produce? Did you choose a wider span for the aispy mini on purpose?

    • admin

      That spike is a real signal, some local interference. You can see it in the RSP2 as well.

      The airspy (and rsp2) can’t match the bandwidth of the HF+ exactly so it could either decimate by 4 to get 1.2 MHz or 8 to get 0.6 MHz. Rather than going smaller and loosing half the signals showing the nearby blocker I went larger. Audio result is exactly the same either way.

Post a comment

You may use the following HTML:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>