In the HackRF Pro specifications we’ve mentioned improved RF performance compared to HackRF One. Now that we’ve finalized any last tweaks to the RF front-end and production is well underway, we’d like to share some more details about the improvements.

One key metric for radio systems is sensitivity, which measures the minimum signal strength that a receiver can detect. However, for software defined radio hardware, sensitivity is difficult to define in a useful way as it changes depending on the modulation scheme in use and even the software implementation.

Instead we can look at noise figure: this measures the degradation in signal-to-noise ratio caused by components in the RF signal chain, so it gives a good representation of the hardware’s contribution to achievable sensitivity. The higher the noise figure of the hardware, the more noise and/or loss it contributes, so we would like the noise figure to be as low as possible.

We’ve been using noise figure measurements throughout HackRF Pro development to optimize the RF front-end and the tuning algorithm that picks LO/IF frequencies. We do this by using a switchable wide-band noise source that generates a known level of noise (HP 346B) along with SDRAngel’s noise figure measurement plugin.

Below is a comparison between HackRF One and a final prototype of HackRF Pro:

This shows typical expected values, measured from single HackRF One and HackRF Pro prototype units; there may be small changes from unit to unit and also if we make further changes to the tuning algorithm. The measurement is also quite susceptible to external interference, so the result can be affected by ambient RF signals in the lab.

The plot shows a solid improvement across almost the whole tuning range, with significant improvement at higher frequencies. In particular the extra tuning range above 6 GHz is now more useful. The plot is also much smoother, thanks to better PCB layout and signal integrity and an improved tuning algorithm.

I wanted to do some extra testing to demonstrate how this plays out in a real-world example, so I set up a head-to-head comparison between HackRF One and HackRF Pro receiving ADS-B location data from planes. I set both of them up in a window in a fairly challenging location with limited sky view and no view of the horizon, so most signals would be reflected off buildings and be pretty weak. They each used a simple dipole tuned for 1090 MHz attached directly to the HackRF, so no extra amplification or filtering.

The plot below shows maximum observed coverage after collecting data for a few hours, with HackRF One in red and HackRF Pro in blue:

In this test the HackRF Pro generally got around 15 to 50 km extra maximum range, and also received double the number of valid messages.

Then to see how well it could do in good conditions, I took the HackRF Pro up a hill with almost 360° horizon view, and received some positions out to almost 400 km!

I was really impressed to see that sort of reception range with just an antenna directly attached. These ADS-B observations have made it clear that our hard work to reduce HackRF Pro’s noise figure resulted in improved receive sensitivity in the real world, and we’re excited to see what interesting applications people have for it.