DRM Software Radio Forums  

Go Back   DRM Software Radio Forums > DRM Software Radio - User Forums > General Topics
User Name
Password
FAQ Members List Calendar Search Today's Posts Mark Forums Read

Reply
 
Thread Tools Search this Thread Display Modes
Old 22-08-2004, 16:33   #46
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
A further note on AGC

Hi,

one of the purposes of the AGC is to prevent distortion in the IF amplifier by decreasing IF levels in the amplifier to a safe value.
If you have a strong interfering carrier that triggers the AGC to turn down gain then distortion in the IF amplifier is prevented. If however, the AGC voltage is derived in such a way that the carrier will not turn down gain, distortion may happen and the quality of the DRM signal will suffer.

Thus it is of prime importance to have an IF amplifier that exhibits enough headroom (> 30 dB) when using this kind of AGC approach.
Also, if done properly, the AGC voltage derived from the DRM signal should not be derived from the average but rather from the peak level of the signal.

dk8cb

Last edited by dk8cb : 22-08-2004 at 16:46.
dk8cb is offline   Reply With Quote
Old 22-08-2004, 16:46   #47
PY2PLL
Registered User
 
Join Date: Aug 2004
Location: S. B. Campo, SP (Brazil)
Posts: 3
Hi Roland et al.

> 2. Run this wav-file through DL4YHF's Spectrum Lab programm ...

It's a good idea. I'm already assembling another setup:

Since my filter is an external unit, I want to loopback Dream (audio muted) in TX module => RX module trhu this filter and measure:

1)S/N with / without filter. OK, probabily lots of dBs S/N

2) Mix at the filter input a 12KHz center freq spectrum, tuning my HF receiver to any crowded SW-AM band sort of have 3 stations 5KHz appart.

So this will reproduce a DRM signal co-chanelized with an AM transmition and another two on the 10KHz BW edges Then I'll test the notches, etc

3) Using the same config, I'll change to MW-AM band and place 2 stations 10KHz appart (here ITU reg 2) at the 10KHz edges. Test the notches etc ...

On this way, the SW signals will be fading in and out but the DRM will be stable in amplitude. This can reproduce the behaviour of a modified receiver where AGC os driven by the DRM signal only (and amplifiers have room enough to handle all this signals).

Yesterday I took a snapshot of DW signal on 3995 and another from CBC on 11955 I guess.

May be I got why Dream doesn't lock on 3995 but it does on 11955, where the CBC signal was S0 (zero) and DW was S9+20: looking at the signals with Spectran, with resolution enough to "see" the OFMD carriers (and the missing one at the center) I could observe that CBC was much more consistent and DW have a vy fast selective fading, seems multipath with several signal arrival time on a "rate" that keeps the software crazy.

The only way to make it lock on DW was choose Time Interpolation = linear and Time sync tracking = First Peak.

But anyway, no audio at all.

I'm trying to modify the external DSP filter code to have left channel as DRM 10KHz bandpass + notches and the right channel a 2KHz wide filter, at 9.5KHz. This energy will be then rectified and used to control receiver AGC.
__________________
73
Marcus
PY2PLL/PY3CRX
http://py.qsl.br
PY2PLL is offline   Reply With Quote
Old 22-08-2004, 17:54   #48
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
> I could observe that CBC was much more consistent and DW have a vy fast selective fading, seems multipath with several signal arrival time on a "rate" that keeps the software crazy.

Have you inspected the estimated impulse response? It should give an indication on the number of paths. Is the guard interval violated?

Interesting to read that DW's signal is so strong in Brazil, I still think they are using an antenna with a mostly vertical radiation angle in order to be able to serve Europe well. But still, quite a considerable amount of power seems to be radiated at such a low angle that it can reach you.

Perhaps time constants for the channel estimation are set a bit too high so that the estimation cannot track the real channel fast enough. I wonder if there is a way to adapt the various time constants of the channel estimation to the kind of channel encountered or to provide a few sets of constants for special types of channels.

dk8cb

Last edited by dk8cb : 22-08-2004 at 18:09.
dk8cb is offline   Reply With Quote
Old 22-08-2004, 19:51   #49
PY2PLL
Registered User
 
Join Date: Aug 2004
Location: S. B. Campo, SP (Brazil)
Posts: 3
Hi ...

>Have you inspected the estimated impulse response?

nop ... I'll do it tonight ...

I'm setting up another DSP card to run Chirp (from G3PLX). There's a chirp sounder in Svalbard and may be the path trhu here can be considered almost the same to DW.

I'll left it running on 3992KHz, measuring the propagation delay from Europe to PY2LAND. There's a spectrogram too, which shows me the main chirp signal and , if any, the delayed signals (even can show RTW one, delayed by 138ms).

BTW, Peter (G3PLX => PSK31) probabily will implement a DRM decoder on such DSP cards.

I use an helix vertical antenna (40m of wire) with some buried radials in a lake margin. With this setup was easy to listen to DW on 3995 (when in AM some weeks ago) or Sondergrese 3320 (S.Africa).
__________________
73
Marcus
PY2PLL/PY3CRX
http://py.qsl.br
PY2PLL is offline   Reply With Quote
Old 31-08-2004, 21:24   #50
DigiBC
Registered User
 
Join Date: May 2003
Location: near Cologne, Germany (51°00'N / 6°49'E)
Posts: 36
Exclamation Dream 1.1 released

A week ago the "Dream Team" has officially released Dream version 1.1.

Compared to previous CVS versions there are no major changes, but Hamlib is fully integrated now (e.g. a S-Meter was added to the "Stations Dialog"). So Dream is able to control a wide range of communication receivers (see: http://hamlib.sourceforge.net/support.html).

Unfortunately the new Dream version won't start anymore if there's no "libhamlib-1-2-2-2.dll" installed. That's a bit annoying if you want to use Dream simply as a DRM decoder.

Now Dream offers 3 color "themes" for the "Main Plot" display. They can be chosen by the following start parameters:
-y 0 Blue-White (default)
-y 1 Green-Black (looks familiar somehow...)
-y 2 Black-Grey

After first start the color setting is written to the "Dream.ini" file.

If you compile the latest CVS source you will already get version 1.1.1cvs.

Regards - DigiBC
DigiBC is offline   Reply With Quote
Old 01-09-2004, 00:57   #51
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Hi,

but it is very easy to inhibit the hamlib functionality if you don't need it. Just edit \common\GlobalDefinitions.h by removing the comment (//) in front of
# undef HAVE_LIBHAMLIB
and adding a comment (//) in front of
# define HAVE_RIG_PARSE_MODE 1
if you do not need the library because you don't have a suitable rig. I did so without any problems.

Of course, you won't need the libhamlib dll then.

Roland

Last edited by dk8cb : 01-09-2004 at 01:10.
dk8cb is offline   Reply With Quote
Old 01-09-2004, 18:34   #52
DigiBC
Registered User
 
Join Date: May 2003
Location: near Cologne, Germany (51°00'N / 6°49'E)
Posts: 36
Hi Roland,

you're right of course, but I'd prefer a universal and more elegant solution: The standard Dream version should work even without that DLL and simply disable all Hamlib features if the required DLL isn't found.

According to Volker (corrados) that's not a matter of Dream but of Hamlib because the DLL is demanded by the "libhamlib.lib". So maybe I should contact the Hamlib developers...

Regards - DigiBC
DigiBC is offline   Reply With Quote
Old 01-09-2004, 20:11   #53
corrados
Registered User
 
corrados's Avatar
 
Join Date: Nov 2003
Location: Darmstadt
Posts: 10
Hi,

what's the problem with having hamlib included? The only dll you really need is the main one which is not very big. If the default setting in Dream is used (which is "none"), hamlib is not used (not even initialized at all). Maybe sometime we get it managed that hamlib is compiled in a static library so no dll is needed at all. I guess this would be a resonable solution for you :-)

Best Regards,
Volker
corrados is offline   Reply With Quote
Old 15-09-2004, 10:02   #54
GuidoS
Registered User
 
GuidoS's Avatar
 
Join Date: May 2004
Location: Antwerp-Belgium
Posts: 14
1.11 supports P-Stereo now.
__________________
Kind regards,

Guido Schotmans.
GuidoS is offline   Reply With Quote
Old 15-09-2004, 11:01   #55
DRM-Fan
Registered User
 
Join Date: Jun 2003
Posts: 209
Quote:
Originally posted by GuidoS
1.11 supports P-Stereo now.

It sure does! I find I'm getting a good 1 or 2 db more than using the DRM s/w radio also. Both can be run at the same time of course
DRM-Fan is offline   Reply With Quote
Old 15-09-2004, 21:29   #56
midre
Michael
 
Join Date: Jan 2003
Location: Schöneck, Hessen,Germany
Posts: 46
Thumbs down

ups,

annother >>forbidden<< post got lost...........

Michael
midre is offline   Reply With Quote
Old 23-09-2004, 23:35   #57
VE3MEO
Registered User
 
Join Date: Oct 2002
Location: Cobourg, Canada
Posts: 289
Group Delay Plot & Higher Reported SNR in latest version

Attached is a screen shot showing an earlier version of 1.1.1cvs on the left and a current one on the right (numbered the same but different!). The one on the right shows the new group delay plot in the Transfer Function screen - it's the black line while the blue line is the amplitude response. Notably, this version reports the SNR to be about 0.6dB higher than the earlier one and it's my impression that it uses 5-10% lower average CPU on a 2 GHz. The SNR difference with a 1.0.8cvs is more variable, from 0.7 to 1.2 dB higher.

I'll post some more shots about the Group Delay plot and refer you to SourceForge forum for more review.

73, Tom
Attached Images
File Type: jpg 9800gd-9800.jpg (96.3 KB, 379 views)
VE3MEO is offline   Reply With Quote
Old 24-09-2004, 01:53   #58
VE3MEO
Registered User
 
Join Date: Oct 2002
Location: Cobourg, Canada
Posts: 289
Re: Group Delay Plots - Off-Air Signal

Quote:
Originally posted by VE3MEO
I'll post some more shots about the Group Delay plot and refer you to SourceForge forum for more review.

In the attached zip file are 4 screen shots of reception of CBC Sackville 9800 kHz:

9800GD-1.png - shallow fade around carrier 60 with corresponding bump in group delay.

9800GD-2.png - almost exactly the receiver response with no selective fading.

9800GD-4.png - severe selective fading, like a comb filter, showing corresponding spikes in group delay, sometimes erroneously negative, as Volker advised.

9800GD-5.png - rapid ripple on both amplitude and group delay.

73, Tom
Attached Files
File Type: zip 9800gd.zip (82.6 KB, 327 views)
VE3MEO is offline   Reply With Quote
Old 24-09-2004, 03:19   #59
VE3MEO
Registered User
 
Join Date: Oct 2002
Location: Cobourg, Canada
Posts: 289
Re: Group Delay Plots - 455kHz Downconverter

Quote:
Originally posted by VE3MEO
SourceForge forum for more review.
Attached are plots from the latest DReaM 1.1.1cvs of the estimated transfer characteristics for amplitude and group delay versus carrier index for a loopback test from a DReaM transmitter to the DReaM receiver through my protoboard upconverter and 'dead-bug ugly' downconverter:

455GD-10kHz.png - 10kHz bandwidth DRM signal centred at 14550 Hz results in a centre of 455kHz with my converter; I captured the group delay plot at a moment when it was flat and average at -19ms. It is typically bouncing around between -12 and -24 ms with +/-1 pixel 'noise' - I don't know why nor what the meaning of negative delay is (relative to what?).

455GD-10kHz-2.png - 10kHz bandwidth DRM signal centred at 12000 Hz results in a centre of 457.45 kHz with my converter and seems to fit the filter better, yielding a little higher reported SNR. The little step in the group delay plot is the transient 'noise' in the display.

455GD-20kHz.png - 20kHz bandwidth DRM signal with DC frequency of 8000 Hz is actually centred at 13000 Hz and just fits within the sound card Nyquist limit of 24kHz, i.e., from 3kHz to 23kHz. At the output of the upconverter, the centre frequency is 456.45kHz, close to the filter's optimum (8500Hz is a 1/2 dB better in reported SNR). As the DRM signal is wider than the filter's 6dB bandwidth, the extreme carriers are severely attenuated. One would expect to see significant group delay distortion at these extremes - it is just evident with the coarse resolution of this scale near the carrier index of 10.

455GD-20kHz-2.png - the DC frequency was shifted down to 7000 Hz from 8000 Hz, thus shifting the spectrum to 2-22kHz and the upconverted centre frequency to 457.45 kHz. There is more severe attenuation of the low index carriers and the screenshot was grabbed when the time-varying group delay plot showed the spike near carrier 30. At other times the GD plot was less severely distorted. Note also the difference in average GD between these two shots, -10 ms for this one vs -24 ms for the previous - both jumping around over time; this is unexpected and unexplained behaviour - they should be static and some carrier should be at reference 0 ms.

73, Tom
Attached Files
File Type: zip 455gd.zip (72.4 KB, 255 views)
VE3MEO is offline   Reply With Quote
Old 25-09-2004, 22:36   #60
VE3MEO
Registered User
 
Join Date: Oct 2002
Location: Cobourg, Canada
Posts: 289
Selectable Input supports two receivers with one sound card

The latest 1.1.1cvs now has a command line parameter with which you can select the input channel to be decoded:
-c , where = 0, left channel
1, right channel
2, mix of both channels (the default).

You can run two instances of DReaM, one on the left channel, the other on the right channel and feed each input with a different DRM stream or the same stream from two different receivers. With just a two-channel sound card, it is now possible to:

- listen to one program while logging another
- record two different programs simultaneously (the WAV outputs are direct from the streams, not via the sound card, and are kept separate)
- compare the performance of two different receivers with the same DRM signal (take advantage of DRMCalc's compare feature)

With Windows, I installed the application software in one folder and created two sub-folders: \left and \right. I then created 3 shortcuts to the application:

1. Normal shortcut with the "Start in" parameter set to the same folder as the application.

2. A shortcut for left channel decoding:
Target = "C:\Program Files\DReaM\DReaM1.1.1dual\Dream.exe" -c 0
Start in = "C:\Program Files\DReaM\DReaM1.1.1dual\left"

3. A shortcut for right channel decoding:
Target = "C:\Program Files\DReaM\DReaM1.1.1dual\Dream.exe" -c 1
Start in = "C:\Program Files\DReaM\DReaM1.1.1dual\right"

The "Start in" parameter sets where the ini, schedule, log and wav files are written. With a unique "Start in" folder for each, you can run two instances of DReaM completely independently. Unfortunately, the audio output from each is fed to both output channels so you cannot use cocktail party effect to discriminate between their respective programs. It would be nice to have control over output routing, too.

73, Tom
VE3MEO is offline   Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is Off
HTML code is On
Forum Jump


All times are GMT. The time now is 02:22.


Powered by vBulletin Version 3.5.3
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.