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 27-07-2004, 15:34   #16
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Re: DRM filters

Quote:
Originally posted by G8JQW
I may have missed something but I don’t see how having a notch or bandpass filter improves DRM reception.

Removing an interfering AM signal might make a difference for agc but the DRM OFDM carriers on the same frequency as an AM carrier interferer can’t be decoded and filtering it out makes no difference ...

Hi Roger,

I totally agree. I think a lot of people are mislead because they get better results with IF notch filters that have no other effect than to prevent the AGC from turning gain down.

But still, one question remains: Not being able to decode certain carriers because of interference or filtering them out and not decoding them makes no difference, but could it perhaps make a difference in channel estimation?

dk8cb
dk8cb is offline   Reply With Quote
Old 27-07-2004, 16:00   #17
ka2hzo
Registered User
 
Join Date: Jun 2003
Posts: 23
Well something is going on with notch filters that makes the DRM reception better
It may not make sense on paper. But if I am only receiving a DRM signal on 6140 with QRM n 6145 at a SNR at 13 DB and then I add a notch 5 Hz notch filter at 6145 and the signal jumps to 18 db and starts to decode. This filter is doing something.
It may have to do with the AGC. But it works.
So why not use it? Unless something can be done to make the agc work better this is one method to improve reception.
__________________
73
Ron KA2HZO
ka2hzo is offline   Reply With Quote
Old 27-07-2004, 16:11   #18
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Quote:
So why not use it? Unless something can be done to make the agc work better this is one method to improve reception.


It is not a question of using it or not.
Of course one should use it, if it improves reception.

But the question is: Could the same result be achieved by a notch filter in software?
If better reception with the notch filter is only a result of the AGC not turning gain down, then using a software notch filter behind the IF chain (or more correctly: at the last stage of the IF chain) makes no sense as the AGC has then turned down the IF gain anyway. In such a case, a software notch in the DRM decoding software makes no sense and is not worth the effort.

dk8cb
dk8cb is offline   Reply With Quote
Old 27-07-2004, 16:40   #19
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Hi,

another thought:

Could reception with the notch filter be so much better, because some IF chains (including soundcards) are not very linear and thus produce intermodulation products in the presence of interfering carriers?

However, if the soundcard input circuit is responsible, turning sensitivity down at the operating system's audio input mixer should help.

dk8cb
dk8cb is offline   Reply With Quote
Old 27-07-2004, 18:29   #20
carknue
Registered User
 
carknue's Avatar
 
Join Date: Dec 2002
Location: Frankfurt, Germany
Posts: 1,321
So this bandpass filter is only usefull with unmodified reveivers ?!?
__________________
regards, Carsten
AOR 7030 + Wellbrook ALA 1530P & DWT + Degen DE31 + TG33

Free DRM Tools for analysing, controling and automating your DRM reception and logging
carknue is offline   Reply With Quote
Old 28-07-2004, 07:50   #21
corrados
Registered User
 
corrados's Avatar
 
Join Date: Nov 2003
Location: Darmstadt
Posts: 10
Hi all,

let me try to explain the benefit of the additional bandpass filter. In the previous versions of Dream, the unfiltered signal of 20 kHz bandwidth was fed to the OFDM demodulator (which is basically a FFT operation). If only a DRM signal of 10 kHz is present in that bandwidth, the filter cannot improve anything in this case. But if, e.g., a strong signal is close to the border of the actual DRM signal, under some conditions this signal will produce interference in the useful bandwidth of the DRM signal although it not at the same frequency as the DRM signal at all. The reason for that strange behaviour lies in the way the OFDM demodulation is done. Since OFDM demodulation is a block-wise operation, a windowing has to be applied (which is rectangular in case of OFDM). As a result, the spectrum of a signal is convoluted with a sinc function in the frequency domain. Consider now a sinusoidal signal close to the border of the DRM signal, its spectrum will not be a distinct peak but a shifted sinc function -> spectrum is broadened caused by the windowing. Thus, it will reach in the DRM spectrum and acts as an in-band interferer.

There is a special case if the sinusoidal signal is in a distance of a multiple of the carrier spacing of the DRM signal. Since the sinc-function has zeros at certain positions it happens, that in this case the zeros are exactly at the sub-carrier frequencies of the DRM signal. In this case, no interference happens. But this is a special case. If the sinusoidal signal is in a distance of a multiple of the carrier spacing plus half of the carrier spacing away from the DRM signal, the interference reaches its maximum.


Conclusions:

- If only one DRM signal is present in the 20 kHz bandwidth, filtering has no effect at all

- If the neighbor channel signal is far away from the DRM signal, filtering will not give much improvement since the squared magnitude of the spectrum of the sinc-function is approx -15 dB down at 1 1/2 carrier spacing (approx 70 Hz with DRM mode B) and goes down to approx -30 dB at 10 times the carrier spacing plus 1 / 2 of the carrier spacing (approx 525 Hz with DRM mode B).
-> the improvement of using the filter will be quite small

- special case: a carrier of an in-band AM station (no modulation considered now) lies exactly at the DC frequency of the DRM signal (which is usually the case since the oscillators at regular transmitters are usually very good), no interference to the DRM signal takes place! Only the modulation signal causes the interference -> notch filter might help

- the bandpass filter must have very sharp edges otherwise it is useless -> high CPU power (BTW: the current filter used in the CVS version of Dream is far away from beeing optimized in any way, it is still a test-stadium)



Best Regards,
Volker
corrados is offline   Reply With Quote
Old 28-07-2004, 09:25   #22
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Quote:
Originally posted by corrados
special case: a carrier of an in-band AM station (no modulation considered now) lies exactly at the DC frequency of the DRM signal (which is usually the case since the oscillators at regular transmitters are usually very good), no interference to the DRM signal takes place! Only the modulation signal causes the interference -> notch filter might help

Hi Volker,

how can a notch filter help with the wideband modulation sidebands, when it is only on a single frequency?

DF9RB just wrote in a PM (in german) that he usually gets improved DRM demodulation when the interfering carrier is notched out on 1296 kHz.
Why? AGC effects? Intermodulation in IF stage?

Shouldn't an experiment be carried out with a notch filter that is built into Dream? This would answer a lot of questions.

Roland, dk8cb
dk8cb is offline   Reply With Quote
Old 28-07-2004, 13:54   #23
cesco
Registered User
 
Join Date: May 2004
Posts: 11
> how can a notch filter help with the wideband modulation
> sidebands, when it is only on a single frequency?

just an idea:

if the disturbing carrier is not orthogonal to the drm signal it does affect the surrounding drm-carriers in the way corrados describes.
this means that a bunch of maybe 20 carriers will be corrupted.

with a sharp notch filter the number of corrupted carriers is reduced to the notch filter width.
cesco is offline   Reply With Quote
Old 28-07-2004, 19:09   #24
df9rb
Registered User
 
Join Date: Jun 2004
Location: 150 km north of Munich
Posts: 484
BBC_notch_no notch

I analyzed the DRM signal of BBC on 1296 with speclab (see picture). The lower part shows the waterfall diagramm with notch OFF, in the upper part notch of the receiver is ON. The red line is the strong interfering carrierer. I think it is interesting that it is shown in the lower part that the DRM signal is getting weaker at the right and left side of the DRM signal, with the notch ON the DRM signal has a quite uniform strength up and down 4.5 kHz from the center.
I interprete this as a dynamic problem of the sound card (at least the sound card I use) because reducing the input signal does not
improve anything with notch OFF.
With notch ON reception of BBC is ok, with notch OFF no decoding!
If it is a dynamic problem of the sound card a software notch will not help I presume.

Bernd, DF9RB
df9rb is offline   Reply With Quote
Old 28-07-2004, 19:16   #25
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Suggestion for an experiment

Hi,

let me suggest an experiment using a notch filter.

1. Record an off the air sample of a DRM signal with an interfering centre carrier at 48 kHz samplerate as a wav-file.
In Europe, a BBC WS transmission on 1296 kHz would be a good candidate.

2. Transfer the file to another PC and play it there. Feed the output into your PC with the DRM decoder and log SNR and dropouts.

3. Use a wave-editing software that has got a built-in notch filter to notch out the carrier on the PC where the file is played and either save the new file or play it directly after filtering (depends on the capabilities of the software).

4. Play the modified file or the direct output and feed it into your PC with the DRM decoder and log SNR and dropouts.

Any differences?

Is anyone around who has got all the necessary tools available?

dk8cb
dk8cb is offline   Reply With Quote
Old 28-07-2004, 19:31   #26
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
Re: BBC_notch_no notch

Quote:
Originally posted by df9rb
I think it is interesting that it is shown in the lower part that the DRM signal is getting weaker at the right and left side of the DRM signal, with the notch ON the DRM signal has a quite uniform strength up and down 4.5 kHz from the center.
I interprete this as a dynamic problem of the sound card (at least the sound card I use) because reducing the input signal does not
improve anything with notch OFF.

I think the more constant shape (i.e. flat top) of the DRM signal with the Notch on is just a result of the filter's gain characteristics, with the gain increasing still a little towards the edges of the DRM signal.

But it is also clearly visible, when looking at frequencies far off the DRM signal, that overall gain is much higher with the Notch filter on.
This is exactly, what I always suspected: The carrier forces the AGC to turn down overall gain!

BTW: If it were a problem of the insufficient dynamic range of your soundcard, reducing the input level should improve reception.

I think, with the carrier not filtered out, the DRM to noise ratio is already too low at the receiver's first or second IF.

dk8cb

Last edited by dk8cb : 28-07-2004 at 19:54.
dk8cb is offline   Reply With Quote
Old 28-07-2004, 20:25   #27
simone
simone
 
simone's Avatar
 
Join Date: Jan 2003
Location: Bad Salzschlirf, Hessen, Germany
Posts: 10,144
Re: Re: BBC_notch_no notch

Quote:
Originally posted by dk8cb



This is exactly, what I always suspected: The carrier forces the AGC to turn down overall gain!

Hi Roland,
thatīs the problem with my configuration too, I mentioned that a long time ago.

btw, interesting experiment that you suggested! but I donīt have the chance to do it the next days.

73, Simone
simone is offline   Reply With Quote
Old 29-07-2004, 01:14   #28
ka2hzo
Registered User
 
Join Date: Jun 2003
Posts: 23
Re: Re: Re: BBC_notch_no notch

Quote:
Originally posted by simone


Hi Roland,
thatīs the problem with my configuration too, I mentioned that a long time ago.

btw, interesting experiment that you suggested! but I donīt have the chance to do it the next days.

73, Simone


Yes I agee.

This I have posted many many times in the past.
If you go back and look at some of my many Sackville results you can see. I have great copy as long as there are no AM carrier's above and below 9800 khz ie. 9795 and 9805 kHz.

How can we fix this problem without using a notch filter?
__________________
73
Ron KA2HZO
ka2hzo is offline   Reply With Quote
Old 29-07-2004, 07:23   #29
dk8cb
Roland
 
Join Date: Mar 2004
Location: Munich, Germany
Posts: 2,748
DRM and AGC

Quote:
Originally posted by ka2hzo
How can we fix this problem without using a notch filter?

By using a receiver that satisfies the following two conditions:

(1) It has a large dynamic range (a lot of headroom) in the IF chain, i.e. it is not easily overdriven by a carrier inside or next to the bandwidth of the DRM signal
(2) It uses an AGC, that derives its control voltage from a part of the spectrum that is not affected by the carrier.

Condition (2) requires, that the AGC voltage is either derived by the demodulating software itself (intelligent AGC) or by an envelope detector and by filtering the IF signal ahead of this detector, such that it is not affected by a carrier.
This could be accomplished by using an already existing built-in narrow filter to derive the AGC envelope and by adjusting the receiver in such a way, that an interfering carrier does not fall into the bandwidth of this narrow filter.
The DRM signal itself would have to be run through a separate IF chain, but this is already often the case, when a receiver is modified by tapping the IF signal ahead of the too narrow second IF filter and feeding it to an external mixer circuit such as an SE612 or NE612 mixer.

Of course, you could as well just use manual gain control. ;-)

dk8cb

Last edited by dk8cb : 29-07-2004 at 07:47.
dk8cb is offline   Reply With Quote
Old 29-07-2004, 07:47   #30
G8JQW
Registered User
 
Join Date: Dec 2003
Location: Birmingham, UK
Posts: 12
Hi All

A DRM signal is just wideband noise to the receiver. It is the amplitude of the AM interferer that will influence the agc more than the DRM signal and this can have a detrimental effect on DRM reception.


73s, Roger
G8JQW 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 19:41.


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