Software Defined Radios are Flaky

[righthalfplane]

Software Defined Radios are Flaky and they are much more flaky than people think. As the author of SdrGlut, I though that people might want a peek behind the curtain at some of the things that are being hidden from them by the software programs that control the hardware. This video points out a few of the problems with the hardware that the software is hiding.

Software Defined Radios are Flaky

SdrGlut

Release SdrGlut V1.24 Adds a Help Window and Some Bug Fixes · righthalfplane/SdrGlut

This release is obsolete - please use the latest release - https://github.com/righthalfplane/SdrGlut/releases/tag/v1.29 If you are new to SdrGlut, You should watch the video "An Introduction ...

github.com

 

[ArkTex]

Software Defined Radios are Flaky and they are much more flaky than people think. As the author of SdrGlut, I though that people might want a peek behind the curtain at some of the things that are being hidden from them by the software programs that control the hardware. This video points out a few of the problems with the hardware that the software is hiding.

Software Defined Radios are Flaky

SdrGlut

Release SdrGlut V1.24 Adds a Help Window and Some Bug Fixes · righthalfplane/SdrGlut

This release is obsolete - please use the latest release - https://github.com/righthalfplane/SdrGlut/releases/tag/v1.29 If you are new to SdrGlut, You should watch the video "An Introduction ...

github.com

The Heat output by the SDRs is the reason for the variations, right?

I know it takes awhile before the dongles reach peak efficiency.

 

[PDXh0b0]

Lol, it's like finding mustard on your Wendy's jr. bacon cheeseburger at one location, pickles on it at another location

 

[prcguy]

What do you expect from a $20 dongle receiver? Is it supposed to compete with a $2,000 on up commercial radio? These dongles were designed primarily for wide band TV reception and over the years people have written software that allows them to receive various narrow band signals but at a huge compromise.

The purpose built SDRs are much better like the Airspy and SDRplay types. The Icom R8600 is an SDR and probably the best performing HF/VHF/UHF receiver of any kind ever marketed for consumers. All of the newer military radios like the Thales MBITR, Harris PRC-152 and others are SDRs and are top notch radios, so the general statement of "Software Defined Radios are flakey" is not true.

 

[KA1RBI]

This is an interesting vid but there are some problems with the analysis......

1) The spike ("hump") in the exact middle of the spectrum in the HackRF is present in all proper SDR (i.e., zero-IF) receivers. It appears in the Ettus SDR series also. SDRs (such as the RTL) that utilize a non-zero final-IF frequency do not exhibit this artifact. The spike at the center of the band may also be referred to as the "DC offset".

2) Before reaching a final diagnosis of the dropped frame issue, a separate test should have been shown with each of the four SDRs running separately, and the remaining three SDRs disabled for each test. Placing all four SDRs in operation concurrently would be a severe exercise for any USB bus hardware.

3) It may not be correct to say that the software drivers cover up for dropped frames. In GNU Radio for example, "O" (Overrun) messages are issued to the console in the event of dropped frames. Typically these overruns become more likely as sampling rates are increased and/or additional bus devices are activated. Overruns are most often caused either by insufficient bus bandwidth or by insufficient CPU horsepower to keep up with the selected data rate. The former issue cannot be cured by throwing more CPU at the problem...

 

[dlwtrunked]

This is an interesting vid but there are some problems with the analysis......

1) The spike ("hump") in the exact middle of the spectrum in the HackRF is present in all proper SDR (i.e., zero-IF) receivers. It appears in the Ettus SDR series also. SDRs (such as the RTL) that utilize a non-zero final-IF frequency do not exhibit this artifact. The spike at the center of the band may also be referred to as the "DC offset".

2) Before reaching a final diagnosis of the dropped frame issue, a separate test should have been shown with each of the four SDRs running separately, and the remaining three SDRs disabled for each test. Placing all four SDRs in operation concurrently would be a severe exercise for any USB bus hardware.

3) It may not be correct to say that the software drivers cover up for dropped frames. In GNU Radio for example, "O" (Overrun) messages are issued to the console in the event of dropped frames. Typically these overruns become more likely as sampling rates are increased and/or additional bus devices are activated. Overruns are most often caused either by insufficient bus bandwidth or by insufficient CPU horsepower to keep up with the selected data rate. The former issue cannot be cured by throwing more CPU at the problem...

All true as well as the most important which I will add as an item 4.
4) None of the things he mentions makes a performance difference if one know what they are doing and using an PC with enough capability.

I have an R9500, R8600, various SDRPlay, and various AirSpy models. I mostly use the Airspy's by a wide margin even with the Icoms sitting right next to it not in use..

 

[iMONITOR]

What do you expect from a $20 dongle receiver? Is it supposed to compete with a $2,000 on up commercial radio? These dongles were designed primarily for wide band TV reception and over the years people have written software that allows them to receive various narrow band signals but at a huge compromise.

The purpose built SDRs are much better like the Airspy and SDRplay types. The Icom R8600 is an SDR and probably the best performing HF/VHF/UHF receiver of any kind ever marketed for consumers. All of the newer military radios like the Thales MBITR, Harris PRC-152 and others are SDRs and are top notch radios, so the general statement of "Software Defined Radios are flakey" is not true.

I agree Mike. My first experience with SDR's was a RTL-SDR I bought from Amazon for around $14.00, and totally disappointed.


Recently I squired a SDRplay RSPdx and the difference was like night & day. Granted it retails for $199.95. My next SDR will be and SDRplay RSPduo which runs $279.95. More expensive than an RTL-SDR for sure, but just as much fun (maybe not the performance) of an IC-R8600. It all depends on what you want and what your expectations are.



Check out SDRplay's SDRuno software:

SDRplay SDRuno Software User Manual

https://www.sdrplay.com/docs/SDRplay_SDRuno_User_Manual.pdf

 

[vagrant]

Overtasking A Computer May Produce Unwanted Results is a better title. The OP’s expectations seem inappropriate. Still, we live we learn.

I use those $20 dongles with their built in TCXO on 3B+ Raspberry Pi’s. One dongle on each Pi and they work very well. One handles APRS and the other ADS-B data. I do not task the Pi‘s to do much else in order to handle their intended function correctly. Oh wait, the APRS one also handles Pihole. I’m living on the edge!

 

[majoco]

I graduated from an early RTL-type SDR when they first came out - complete with the TV software on a mini CD. The radio software had to be downloaded and installed which wasn't as easy as it is now. Having been a radio user both as an amateur and a professional, it became pretty obvious that the SDR had plenty of shortcomings so I invested in Winradio G303 and Icom R7000 receivers for HF and VHF - much better - the Winrad software was a bit clunky but the performance was excellent. Impressed, I was offered a Winradio G33DDC - a deal I couldn't afford to miss and it's proved to be really excellent - more bells'n'whistles than you know what to do with. Does the old RTL SDR have it's uses? Certainly, I use it as a bit of test gear - but it's still a TV tuner, so don't expect the Holy Grail for twenty bucks!

 

[R8000]

What do you expect from a $20 dongle receiver? Is it supposed to compete with a $2,000 on up commercial radio? These dongles were designed primarily for wide band TV reception and over the years people have written software that allows them to receive various narrow band signals but at a huge compromise.

The purpose built SDRs are much better like the Airspy and SDRplay types. The Icom R8600 is an SDR and probably the best performing HF/VHF/UHF receiver of any kind ever marketed for consumers. All of the newer military radios like the Thales MBITR, Harris PRC-152 and others are SDRs and are top notch radios, so the general statement of "Software Defined Radios are flakey" is not true.

100% correct.

 

[merlin]

While they do work, they are designed around a TV tuner that has poor gain, no filtering at all and the perverbial cheap Chinese interface to the USB. Takes a really good driver with good bandwidth for the I/Q
These tuner chips are vintage clones going back to the 80s.
ALL the processing and demodulating is done in the software for SDR, some good, some, not so good.
This FFT processing takes a computer with loads of processing power, speed, and memory.
Over that, they add radio lab displays like a spectrum analyzer, waterfall, and IF scope.
Fine if you are a radio engineer. Handy, but a learning curve for the layman.
Sorry, none of these devices will never match or compare to radios dedicated for the task.
You get into the high end SDR devices, they do better but prices generally start at $300 and up.
73s
PS: Yeasu FTDX10 is $1600---the FTDX101MP around $4200

 

[righthalfplane]

The SDRplay RSPDuo was about $300 and the HackRF One was about $400 and the NetSDR+ was $1400 - I had only one cheep RTL stick.

 

[dlwtrunked]

...
Sorry, none of these devices will never match or compare to radios dedicated for the task.
You get into the high end SDR devices, they do better but prices generally start at $300 and up.
...

I am afraid this is not true unless you define "dedicated to the task" to make it seem otherwise or you need to define "these devices". A AirSpyHF+ Discovery (Around $200) on a typical PC (often I used an old Win 7) is on par with an ICOM R9500 and ICOM R8600 on HF. Yes, I own all of these. In discussing this subject, one should start by reading actual careful comparisons such as in the 2021 WRTH (World Radio Television Handbook) and taking a look at the table on pg. 24. The problem with most cases of people not liking SDRs is that they are not using the right settings in the software; unfortunately it is true that some people will not take the time to figure those out . As someone who has helped them when this happens, I have seen that many times. Cost does not determine performance here; else the $12,500 ICOM 9500 would significantly outperform the $200 (which it does not). Additionally, when you say " Over that, they add radio lab displays like a spectrum analyzer, waterfall, and IF scope", in the circles I move in, those are not "lab displays" but strongly desired of any serious receiver and regularly used. (I am a 50 year SWL also. Started with an E.F. Scott 21 tube receiver for HF that I bought for $20 at an Amish flea market of all places that looked like the one at E.H. Scott allwave 15 radio chassis | eBay .)

 

[righthalfplane]

I did run all of the devices separately and they got the same errors. GNUradio reports audio overruns with an O. CubicSDR, GQRX, and GNUradio just ignore the bad records - if they didn't, you would see the constant glitches in the waterfall and spectrum displays.

 

[righthalfplane]

As for the ICOM-ICR8600, it does not create any bad records, but it has a few more problems. The software only runs on windows - they have patched a USB driver to work. On linux, the ICR8600 acts like a USB device, but when you try to read the data it errors. When I run SdrGlut using the ICR8600 on Windows 10 with the current SoapySDR and ICR8600 software versions it runs very nicely, but crashes if you try go at more than 960000 SPS. On Window 7, with old versions of SoapySDR and ICR8600 SdrGlut would run at the full 5 MSPS, but now it just crashed. I likely could find the problems and get everything going on Windows, but there dos not seem to be much interest. SDR Console does not work with the ICOM-ICR8600 on Windows 7 and SDR Console is only partly working with the ICOM-ICR8600 on Windows 10.

 

[JerryX]

You get into the high end SDR devices, they do better but prices generally start at $300 and up.
73s
PS: Yeasu FTDX10 is $1600---the FTDX101MP around $4200

The FTDX101MP isn't a direct sampling SDR rig. It has a conventional superhet front end and only uses DSP at the last IF stage, similar to how rigs have been doing it for the last 25 years. Rigs such as the IC-7610, on the other hand, are true direct sampling rigs.

"SDR" can have a wide range of definitions, but most people here would probably say that direct sampling rigs are true SDR while superhet rigs that use DSP in the final IF are not.

 

[maan118]

While they do work, they are designed around a TV tuner that has poor gain, no filtering at all and the perverbial cheap Chinese interface to the USB. Takes a really good driver with good bandwidth for the I/Q
These tuner chips are vintage clones going back to the 80s.
ALL the processing and demodulating is done in the software for SDR, some good, some, not so good.
This FFT processing takes a computer with loads of processing power, speed, and memory.
Over that, they add radio lab displays like a spectrum analyzer, waterfall, and IF scope.
Fine if you are a radio engineer. Handy, but a learning curve for the layman.
Sorry, none of these devices will never match or compare to radios dedicated for the task.
You get into the high end SDR devices, they do better but prices generally start at $300 and up.
73s
PS: Yeasu FTDX10 is $1600---the FTDX101MP around $4200

I do not think this is entirely accurate. The USB interface is made by Realtek, a Taiwanese company (there is a very good chance your PC has a Realtek network card, by the way). It also includes an 8-bit ADC, DVB-T demodulator and FM and DAB(+) radio support. (https://www.realtek.com/en/products/communications-network-ics/item/rtl2832u)

The most popular (and best) tuner chip for RTL-SDRs is the Rafael Micro (also a Taiwanese company) R820T, which was specifically designed to receive DVB-T, a European digital television standard that is used in many other parts of the world as well (some countries have switched to its successor, DVB-T2), which was certainly not around in the 1980s.

 

[merlin]

The FTDX101MP isn't a direct sampling SDR rig. It has a conventional superhet front end and only uses DSP at the last IF stage, similar to how rigs have been doing it for the last 25 years. Rigs such as the IC-7610, on the other hand, are true direct sampling rigs.

"SDR" can have a wide range of definitions, but most people here would probably say that direct sampling rigs are true SDR while superhet rigs that use DSP in the final IF are not.

While that is true, it is also a dedicated HF+ radio, and newer models are no different when they take the IF into digital. A great big difference when talking SDR and yes, the 'back end is no different from a dongle and software DSP, just dedicated to the task. You add that front end even to a cheap SDR you vastly improve it.

 

[merlin]

but crashes if you try go at more than 960000 SPS.

Most will crash if you go beyond some sampling level, My RX888 won't even run on SDR# with sampling over 64000kbps, but why the need for such rate anyway.

 

[merlin]

A AirSpyHF+ Discovery (Around $200) on a typical PC (often I used an old Win 7) is on par with an ICOM R9500 and ICOM R8600 on HF

Give me some of what you been taking. Or is this comedy relief.