Re: Direct Sampling frequency limit in r 1628 onwards #sdrsharp #spyserver

Kenneth Sejkora

Based on the revision log, it looks as though build 1627 was the last version of SDRSharp available before the 14.4 MHz ceiling was implemented in direct sampling mode.  As Martin points out, there are situations where it might be advantageous to at least have the option to run a dongle in direct sampling mode in the 14.4 to 28.8 MHz range as opposed to being forced to use quadrature sampling in this frequency range.  I totally respect Youssef's desire to implement the 14.4 MHz ceiling for the technical reasons he stated, but for an end user the option to exceed the ceiling should also be respected and left up to the user.  The user can 'assume' the negative consequences of operating direct sampling in the 14.4 to 28.8 MHz frequency range. However, at least for now, SDRSharp is freeware and Youssef has every right to call the shots.

Although it circumvents any improvements that might be afforded by new revisions to SDRSharp, it might behoove anyone that wants the option of using direct sampling in the 14.4 to 28.8 MHz range to obtain a copy of build 1627 to at least have the option of using that range of frequencies in either direct sampling or quadrature mode.  Not the best "fix", but a workable one.  Fortunately, I still have the Zip file for build 1627, and can maintain that version as a separate installation for whenever I need to use direct sampling above 14.4 MHz.

Ken, WBØOCV Norton, MA USA
41.959546N, 71.163996W  FN41kx

From: Martin via Groups.Io <martin_ehrenfried@...>
Sent: Friday, November 24, 2017 9:36 AM
Subject: Re: [airspy] Direct Sampling frequency limit in r 1628 onwards #sdrsharp #spyserver

Hi Joanne,

Yesterday I tried to setup three concurrent FT8 decode sessions on 24MHz in order to test the receive performance of an Airspy R1, generic 820T2 dongle and RTL SDR V3 in direct sampling mode.

Unfortunately the propagation gods were not favourable and I discovered a few other issues, so today I connected them up to my tes tkit to try and figure out what was going on.

First the maximum sensitivity I could obtain for each type. These were measured using plain carrier and USB in a 2400KHz BW. The generator output was adjusted to obtain 10dB SINAD at 27MHz. Note my generator and attenuators may not be fully in calibration, but I used the same setup for each test, so I'm comparing like with like.

Airspy R1 = -122dBm (Gain setting in free mode 10/10/15)

Airspy Mini = -117dBm (Gain setting in free mode 10/10/15)

Generic 820T2 = -119dBm (Gain setting 49.6)

RTL SDR V3 Quadrature = -118dBm (Gain setting 49.6)

RTL SDR V3 Direct Sampling = -107 (No gain adjustment)

So in theory the Quadrature mode seems to offer about 10dB better sensitivity than when used in Direct Sampling mode.

However when I connected an antenna to the various SDR's using the maximum gain settings they all showed signs of overload and IMD products, apart from the V3 in Direct Sampling mode.

In each case I adjusted the gain settings until the overload and IMD products just disappeared and then remeasured the sensitivity with the new gain settings.

Airspy R1 = -116dBm (Gain setting in free mode 10/10/6)

Airspy Mini = -115dBm (Gain setting in free mode 10/10/9)

Generic 820T2 = -112dBm (Gain setting 29.7)

RTL SDR V3 Quadrature = -111dBm (Gain setting 29.7)

RTL SDR V3 Direct Sampling = -107 (No gain adjustment)

So now the difference is only 8dB best case (Airspy Mini) and is typically only 4dB (Generic 820T2 and RTL SDR V3)

In practice these figures could be improved by placing bandpass filters ahead of the SDR's but I just wanted to test their basic out of the box performance.

Other observations:-

All the dongle sensitivities remained fairly consistent across their frequency ranges.

When using Quadrature mode non of the dongles locked consistently below 24MHz. The 820T2  would not go below 24MHz, the Airspy R1 occasionally manged it down to 19MHz and the Airspy Mini would normally only lock down to about 25MHz. I could sometimes see a signal but it was not on the correct frequency (usually 1-1.5MHz low) So if you have it connected to an antenna you may think you are seeing signals below 24MHz, but they may not be on the correct frequency.

There is no mechanism to receive signals between 14.4MHz and 24MHz using Quadrature mode. Direct Sampling does work up to about 27MHz but above this frequency the 28.8MHz clock oscillator causes very high level unwanted noise spikes.

When using the RTL SDR V3 in Direct Sampling mode the sensitivity is typically -107dBm from 2MHz to 25MHz. It falls outside these limits as the Bias tee inductor and input diplexer add additional losses. This permits good reception on most bands up to about 20MHz (with a suitable antenna) and is only limited by the natural atmospheric noise floor. Reception is still adequate on higher frequencies up to about 27MHz, especially in urban areas where the noise floor is likely to be higher.

The RTL SDR V3 in Direct Sampling mode does have Nyquist Alias signals wrapped around 14.4MHz (1/2 of 28.8MHz sample rate). This results in signals appearing twice (one the correct frequency and one unwanted on the Nyquist image). Most of these can be removed by adding a simple high / low / Diplexing filter or if only certain frequency ranges are of interest a band pass filter.

This is why like to have the option of being able to use the Direct Sampling mode between 14.4MHz and 27MHz.


Martin - G8JNJ

Join to automatically receive all group messages.