Date   

Re: Fw: RTLsdr with HDSDR & SDRSHARP

ve1kg
 

Mike you are the "MAN" following your instructions I downloaded the SDRsharp which applies to RTL & got at last the RTL/USB in the configuration Window. I clicked on Configure & a window showed up called RTL settings. I adjusted the RF gain as you suggested up to 40. Now is there anything else I should indicate in that window please. like RTL AGC, TUNER AGC ECT ECT
Thank you so much for the clarity of your instructions.
Serge/VE1KG


Re: [SDRSharp] Re: Computer Requirements to run SDR#

Alberto I2PHD
 

On 9/3/2013 12:07 AM, ian_mm6dos wrote:
�

Well, congratulations Ian (and Youssef and any other involved in the development
of SDR#).

I tested that WAV file against both SDR# and Studio 1 (optimized C++ code),
on a 8 year old AMD Athlon, under Windows XP+SP3.� Athlon, compared to
Intel CPUs, are rather inefficient in their floating point core.

Both programs decoded without problems, with a CPU load of about 40% for both.
This shows that, with the due skills, also using C# and .NET can produce efficient
code, comparable to C++.

73� Alberto� I2PHD

�


Re: Computer Requirements to run SDR#

sdrsharp
 

Let's play a game: Find the Tunisian DX station in that file.

--- In SDRSharp@yahoogroups.co.uk, Alberto I2PHD <i2phd@...> wrote:

On 9/3/2013 12:07 AM, ian_mm6dos wrote:

/Sure.//
/ /
/ /http://www.myradiobase.de/perseus/20101226_FM8850_PaviaItaly.wav//
/ /
//Ian/
Many thanks Ian. I am just now downloading the 1.3 GB (thanks God my ADSL
is rather fast), to perform tests with my programs.


--
/*73 Alberto I2PHD*/


Re: [SDRSharp] Re: Computer Requirements to run SDR#

Alberto I2PHD
 

On 9/3/2013 12:07 AM, ian_mm6dos wrote:

Sure.

http://www.myradiobase.de/perseus/20101226_FM8850_PaviaItaly.wav

Ian

Many thanks Ian.� I am just now downloading the 1.3 GB (thanks God my ADSL
is rather fast), to perform tests with my programs.


--
73 Alberto I2PHD


Re: [SDRSharp] Re: Computer Requirements to run SDR#

Alberto I2PHD
 

On 9/3/2013 12:49 PM, David J Taylor wrote:

Drifting off-topic, but during my first non-amateur experience with
Assembler on an IBM 1130, when you submitted your paper tape for
compilation, all you got back was "OK", or "Error". It was up to you to
find the error! Made you work, I'll tell you!

You make me remembering of when (in my young age...) I was discovered�
correcting errors in a deck of punched cards by reinserting back the
punched chips into the card holes, and hand punching new holes, where necessary...

And it wasn't a deck of source code, but the punched object code output
of a Fortran compilation on an IBM 7090...� A myth was then born...
Today I would be too old to spend my time in such an inefficient way... :-)

--
73 Alberto I2PHD


Re: Scopeview plugin speed control: how it works?

yagiorgio <giorgio.fontana@...>
 

I had some success with:

System.Threading.Thread.Sleep (scopeSpeedTrackBar.Value)

But the thread includes audio and FFT display. They go to sleep too.
Looking for something else.

Giorgio

--- In SDRSharp@yahoogroups.co.uk, "yagiorgio" <giorgio.fontana@...> wrote:

I succeded in recompiling the scopeview plugin.

This is the link to the source code:
https://www.assembla.com/code/sdrsharp/subversion/nodes/1164/branches/bobrich/AutoTunerPlugin/ScopeView/ScopeViewPanel.cs

scopeSpeedTrackBar.Value controls the speed, that I want to reduce.

From:
https://www.assembla.com/code/sdrsharp/subversion/nodes/1164/branches/bobrich/AutoTunerPlugin/ScopeView/ScopeViewPanel.Designer.cs
the minimum value is 1, max is 20

I think that a different algorithm must be used to increase the time between updates of the scopeview screen (to have time to see the trace..)

Any suggestion?
Thank you.


Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

Alan
 

----- Original Message -----
Subject: Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

That assumes that the whole stream is decoded.
Does this happen? 4 HD channels all at the same time?
A multiplex has a band of individual carriers if I remember correctly.
As far as I know, they aren't split down by frequency, but by PIDs on the total stream, so the total stream is required. HD is sent on DVB-T2, which these sticks can't handle. Yes, there are something like 2000 4 KHz carries in an 8 MHz channel (using round numbers). There's an interesting table here:
David,

Yes, I'd forgotten the RTL chip specs say up to 8MHz bandwidth.
Looks like the SDR driver does not allow that, probably for a good reason.

73 Alan G4ZFQ


http://en.wikipedia.org/wiki/Digital_terrestrial_television_in_the_United_Kingdom#The_multiplexes
showing a bit rate of 24 - 27 Mb/s for the basic services, using 64QAM. That suggests a symbol rate of 4+ Ms/s to me (but I'm no expert). The RTL chips include at least partial decoding, so perhaps it's not as much of a strain for the interface and driver software as handling the IQ stream directly.


Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

jdow <jdow@...>
 

On 2013/09/03 02:58, David J Taylor wrote:
From: jdow
[]
Where is the limitation in being unable to run the dongles at 3.2 MS/s?
The
RTL software? So are you running multiple instances of that software on
multiple CPUs?

73,
David GM8ARV
Off hand I'd guess it is inside the dongles somewhere.

{^_^}
===================================

If that were the case, they wouldn't be fast enough for DVB-T, or would
they? It's a channel bandwidth of 8 MHz (at least, in the UK).

David GM8ARV
The DVB-T decoding is done within the dongle. And that different signal
path may feed decoded video out using a different USB format.

{^_^}


Re: [SDRSharp] Re: Computer Requirements to run SDR#

jdow <jdow@...>
 

Leif didn't appear to be running the tests under the same dongle setting
conditions. He tried to work with noise figure as a reference. I figure
that testing is not meaningful in terms of an overload test. So, no, I did
not play with it. I lacked time to do so.

My doctored SDRSharp RTLSDR dll doesn't change anything in the signal path.
It only makes it easy to store and recover settings. I am aware he retired
the one YouTube video. He never seemed to test for the decimator artifacts
I suspect are there. It works well enough I've never tried to hunt for them
in detail, though. A perfectionist would use sharper filters as part of the
decimation process so that aliases have a harder time creeping into the
picture. (You might even have to play with an SDR IQ to see them depending
on the precise shapes you programmed in for the decimator filters.)

To really test it I have to pull my 8640 apart, put a new fan on it, and
fix up some of the other little annoying issues it has. That's a lot of
effort and I'm not sure I have the time at the moment to do it.

{^_^}

On 2013/09/03 02:10, sdrsharp wrote:
(SDRSharp has some performance issues at the digital RF level that show up with careful testing that other software do not exhibit because of basic design differences.
Did you play with Leif's sim1.wav in SDR#? I believe it's a very good testing tool for RF processing correctness. I invite you to play with the downloaded version of SDR# (not the one you modified) and compare to the rest. I have had a talk with Leif about his lab testing environment and the settings he used in SDR#. He performed the test once again and he concluded there are no problems in the processing and he retired his Youtube video.
So, can you please provide a test case (an IQ wave file is fine) which can highlight the problems you're talking about?

Youssef


Re: [SDRSharp] Re: Computer Requirements to run SDR#

Simon Brown
 

You are showing your age! I started with the very modern punched cards, far
more interesting and the cards readers were *loud*. Maybe that's the cause
of my tinnitus?

I remember the first VT 52...

Simon Brown (G4ELI/HB9DRV)
http://v2.sdr-radio.com

-----Original Message-----
From: SDRSharp@yahoogroups.co.uk [mailto:SDRSharp@yahoogroups.co.uk] On
Behalf Of David J Taylor

Drifting off-topic, but during my first non-amateur experience with
Assembler on an IBM 1130, when you submitted your paper tape for
compilation, all you got back was "OK", or "Error". It was up to you to
find the error! Made you work, I'll tell you!


Re: [SDRSharp] Re: Computer Requirements to run SDR#

David J Taylor
 

"The Worst Programming Environment in the World?"

Read and weep: https://github.com/jloughry/BANCStar/blob/master/README.md

Simon Brown (G4ELI/HB9DRV)
====================================

Drifting off-topic, but during my first non-amateur experience with Assembler on an IBM 1130, when you submitted your paper tape for compilation, all you got back was "OK", or "Error". It was up to you to find the error! Made you work, I'll tell you!

73,
David GM8ARV
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@blueyonder.co.uk


Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

David J Taylor
 

David

That assumes that the whole stream is decoded.
Does this happen? 4 HD channels all at the same time?
A multiplex has a band of individual carriers if I remember correctly.

73 Alan G4ZFQ
==========================================

As far as I know, they aren't split down by frequency, but by PIDs on the total stream, so the total stream is required. HD is sent on DVB-T2, which these sticks can't handle. Yes, there are something like 2000 4 KHz carries in an 8 MHz channel (using round numbers). There's an interesting table here:

http://en.wikipedia.org/wiki/Digital_terrestrial_television_in_the_United_Kingdom#The_multiplexes

showing a bit rate of 24 - 27 Mb/s for the basic services, using 64QAM. That suggests a symbol rate of 4+ Ms/s to me (but I'm no expert). The RTL chips include at least partial decoding, so perhaps it's not as much of a strain for the interface and driver software as handling the IQ stream directly.

Cheers,
David
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@blueyonder.co.uk


Re: [SDRSharp] Re: Computer Requirements to run SDR#

Simon Brown
 

"The Worst Programming Environment in the World?"

Read and weep: https://github.com/jloughry/BANCStar/blob/master/README.md

Simon Brown (G4ELI/HB9DRV)
http://v2.sdr-radio.com

-----Original Message-----
From: SDRSharp@yahoogroups.co.uk [mailto:SDRSharp@yahoogroups.co.uk] On
Behalf Of jdow

Simon, I mentioned the bus between processor chips as a potential hitch in
your gitalong - if you attempt low latency and heavy multiprocessing.


Scopeview plugin speed control: how it works?

yagiorgio <giorgio.fontana@...>
 

I succeded in recompiling the scopeview plugin.

This is the link to the source code:
https://www.assembla.com/code/sdrsharp/subversion/nodes/1164/branches/bobrich/AutoTunerPlugin/ScopeView/ScopeViewPanel.cs

scopeSpeedTrackBar.Value controls the speed, that I want to reduce.

From:
https://www.assembla.com/code/sdrsharp/subversion/nodes/1164/branches/bobrich/AutoTunerPlugin/ScopeView/ScopeViewPanel.Designer.cs
the minimum value is 1, max is 20

I think that a different algorithm must be used to increase the time between updates of the scopeview screen (to have time to see the trace..)

Any suggestion?
Thank you.


Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

Simon Brown
 

But for DVB-T they use a different driver? I can't say I'm at all up-to-date
with these beasties...

Simon Brown (G4ELI/HB9DRV)
http://v2.sdr-radio.com

-----Original Message-----
From: SDRSharp@yahoogroups.co.uk [mailto:SDRSharp@yahoogroups.co.uk] On
Behalf Of David J Taylor

If that were the case, they wouldn't be fast enough for DVB-T, or would
they? It's a channel bandwidth of 8 MHz (at least, in the UK).


Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

Alan
 

----- Original Message -----
From: "David >
If that were the case, they wouldn't be fast enough for DVB-T, or would they? It's a channel bandwidth of 8 MHz (at least, in the UK).
David

That assumes that the whole stream is decoded.
Does this happen? 4 HD channels all at the same time?
A multiplex has a band of individual carriers if I remember correctly.

73 Alan G4ZFQ


Re: [SDRSharp] Re: USB: 4 ports or external powered port ?

David J Taylor
 

From: jdow
[]
Where is the limitation in being unable to run the dongles at 3.2 MS/s? The
RTL software? So are you running multiple instances of that software on
multiple CPUs?

73,
David GM8ARV
Off hand I'd guess it is inside the dongles somewhere.

{^_^}
===================================

If that were the case, they wouldn't be fast enough for DVB-T, or would they? It's a channel bandwidth of 8 MHz (at least, in the UK).

David GM8ARV
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@blueyonder.co.uk


Re: Computer Requirements to run SDR#

sdrsharp
 

(SDRSharp has some performance issues at the digital RF level that show up with careful testing that other software do not exhibit because of basic design differences.
Did you play with Leif's sim1.wav in SDR#? I believe it's a very good testing tool for RF processing correctness. I invite you to play with the downloaded version of SDR# (not the one you modified) and compare to the rest. I have had a talk with Leif about his lab testing environment and the settings he used in SDR#. He performed the test once again and he concluded there are no problems in the processing and he retired his Youtube video.
So, can you please provide a test case (an IQ wave file is fine) which can highlight the problems you're talking about?

Youssef


Re: [SDRSharp] Re: Computer Requirements to run SDR#

jdow <jdow@...>
 

Simon, I mentioned the bus between processor chips as a potential hitch
in your gitalong - if you attempt low latency and heavy multiprocessing.

{^_^}

On 2013/09/03 00:53, Simon Brown wrote:
David,

In the main I agree with you, however one extra option is parallel
processing with OpenMP. I don't use it a lot but when I do it has an amazing
effect as expected. It's such a shame to have all those cores idling away...

A very CPU intensive program I want to write sooner or later is satellite
crash detection - detect possible satellite collisions for the next day /
week / month using all available TLE data. This will involve Intel's IPP,
OpenMP and my nice new Haswell 4770 :) .

FWIW VS2010 has excellent profiling.

Simon Brown (G4ELI/HB9DRV)
http://v2.sdr-radio.com


-----Original Message-----
From: SDRSharp@yahoogroups.co.uk [mailto:SDRSharp@yahoogroups.co.uk] On
Behalf Of David J Taylor

As there are so many different Intel and AMD CPUs out there - all with
different optimisations - I am unconvinced that spending a lot of time on
"bit-level", CPU-specific optimisations is time well spent. On the other
hand, using the vector instructions available, and perhaps now using the GPU
to its maximum extent, may provide the best reward. But do get your
algorithms optimised first.



------------------------------------

Yahoo! Groups Links



Re: [SDRSharp] Re: Computer Requirements to run SDR#

jdow <jdow@...>
 

On 2013/09/03 00:13, David J Taylor wrote:
From: jdow

Loren and his coworkers are working with VERY heavy optimization levels
on a mainframe emulator running on Windows architectures. Out of order
execution and a host of other CPU features make true optimization a sort
of black art. They concluded that attempting clever optimization tricks
in C code is close to hopeless. They DO have some assembly routines
embedded in their emulator. Mostly it's straight C++ without using a lot
of the C++ frills. 4096 BYTEs is a magic size to consider. So is the
total size of the cache. It's all PFM, really.

{^_^}
=======================================

Joanne,

That sort of optimisation is rather different, though, to running x86 code
on an x86 CPU, if I understand you correctly, it's running foreign CPU code
on x86. Agreed on the memory issues - I do wonder how much there might be
to gain by running some of the SDR stuff in 64-bit mode - perhaps wider
memory bandwidth, faster memory moves etc. I expect the guys have already
tested this. I wonder whether profiling has a role here - to highlight
performance hotspots, or whether it's sufficient obvious that the graphics
or the basic signal processing is the limiting factor?

By the way, Embarcadero's Delphi (Pascal-like language) can produce quite
efficient code, and I expect their C/C++ would be similar, and now it's
starting to be available for iOS and Android as well. I'm with Alberto on
Embarcadero vs. Visual Studio. There have been a number of community
enhancements - noticeably much faster memory allocation etc. routines -
which have been incorporated into the commercial product. I've also been
playing with Lazarus, which is a cross-platform version of Delphi (and it's
free) supporting Linux and the Mac. I've not done a lot of speed tests, but
it's not too slow even on the Raspberry Pi.

73,
David GM8ARV
Between his work and his side efforts my partner's discovered such odd
things as the CPU to CPU bus characteristics for dual processor motherboards
as implemented by the motherboard designers can mess you up materially. The
bus between processors runs in bursts controlled by the motherboard and to
some degree the OS. This REALLY messes up very low latency audio processing
for a 512 channel audio tool - matrix mixer, equalizers, delays, etc.

Other compilers have a long way to go before they get up to Intel's
compiler's performance on Intel processors. (I can't speak for the AMD side
of things.) We're talking about finer grained effects than memory
allocations and the like. Yes, those can be improved. And simple things
like memory pools can sometimes make huge differences, especially when
memory fragmentation may become a problem. Often those effects are program
specific rather than generic.

This is wandering a tad off topic which started with two programs showing
different performance based on code that has different algorithms and
design goals. (SDRSharp has some performance issues at the digital RF level
that show up with careful testing that other software do not exhibit because
of basic design differences. The other software uses more CPU. What is the
proper CPU vs RF performance tradeoff? I must observe that SDRSharp's
tradeoff is pretty good for most uses and does seem to require less CPU
to run, as Ian's test shows.)

{^_^}