[Premium-Rx] The new R&S EM510 Direct-Sampling HF Receiver

[Pieter Ibelings]

Francis,

The signal that the fast ADC sees is the composite of all the HF signals. 
This waveform looks like noise, but in reality it contains all of the HF 
information. When using ADCs, it is very important that you do not exceed 
the ADC clipping voltage. If you do that, the noise and distortions get 
abruptly worse. You can limit the level into the ADC by using filters or 
attenuators. The filters reduce the composite level by not allowing bands of 
frequencies into the ADC. The attenuators limit the signal by reducing the 
level across the whole band. This impacts noise figure and sensitivity.
After the signal is converted to digital, there are many methods that can be 
used to separate the one very complex waveform into its different frequency 
components. This methods include the FFT, FIR filters etc. You can, for 
example, have a very strong sinewave near full scale on the ADC plus a small 
signal at another frequency and still be able to separate them. As long as 
the peak voltage of all the signals does not clip the ADC, everything is 
fine. You also want to have enough bits of resolution to be able to see the 
small signal while not clipping on the big one.

There are a couple of different radios that are direct sampling 
(USRP,SDR-14,SDR-IQ,HPSDR,QS1R etc)

Pieter



>From: FRANCIS CARCIA <carcia at sbcglobal.net>
>To: Pieter Ibelings <sietetrescincoprimo at hotmail.com>
>Subject: Re: [Premium-Rx] The new R&S EM510 Direct-Sampling HF Receiver
>Date: Fri, 16 Feb 2007 19:40:53 -0800 (PST)
>MIME-Version: 1.0
>Received: from web82313.mail.mud.yahoo.com ([209.191.85.233]) by 
>bay0-mc12-f12.bay0.hotmail.com with Microsoft SMTPSVC(6.0.3790.2444); Fri, 
>16 Feb 2007 19:40:54 -0800
>Received: (qmail 42301 invoked by uid 60001); 17 Feb 2007 03:40:53 -0000
>Received: from [69.0.83.251] by web82313.mail.mud.yahoo.com via HTTP; Fri, 
>16 Feb 2007 19:40:53 PST
>X-Message-Info: LsUYwwHHNt3660MmjhEvYg2f34OAemlK3oXsmRrh6gU=
>DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;  s=s1024; d=sbcglobal.net; 
>  
>h=X-YMail-OSG:Received:Date:From:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding:Message-ID; 
>  
>b=QmnFlizMGFA8Gq3cv3ZNNG4HS3fL66xPvzfUAzwNzWGUt1OpbFEqnfDxlEHjZlKtCiM+aJmP9eoplX+SFrMVblRuUcJquHlBwZWPJ/ycH3VQnCtaO5A7CF2ffMNtjMy18nd5eASZuKkhDXv5bL8Qz30ZQ0BwiJia/laG4fBM6cs=;
>X-YMail-OSG: 
>7wY.1PgVM1lgSflJtK8_OjZoAVIROGfGrEaAay8MEsXXdS6Mt221aHewkvO5LqmhOGgpPAUUhXTrk7_4mpc_XYh.hr6FXB_yAt.qBIEx5E1N5TNVebYSdQ--
>Return-Path: carcia at sbcglobal.net
>X-OriginalArrivalTime: 17 Feb 2007 03:40:54.0144 (UTC) 
>FILETIME=[6D302000:01C75245]
>
>SOOOO you do only have a few bits actually 0 through 15 (4 bits) sounds 
>like enough to create good information but the increased sample rate is 
>where I was in the dark.I'm confusing a sound card performance with a much 
>higher sample rate.
>   This also tells me I don't want to mess with down conversion to a low IF 
>but wait for the direct connection like the board N8VB has coming out.
>   The 2050 has a roofing filter to limit bandwidth so tha system can 
>operate at a lower sample rate and decreased number of bits.
>   Now that you have broken through my thick skull tell me how the A/D 
>handles strong signals near the signals you want to demodulate. The  A/D 
>input only measures a voltage at each sample and has no idea what is going 
>on in terms of frequency. Say the A/D is near saturation for a period of 
>time due to a low frequency signal on top of the high frequency you want to 
>demodulate?
>
>
>Pieter Ibelings <sietetrescincoprimo at hotmail.com> wrote:
>   Francis,
>
>That's another misconception. I am going to give you the example of an
>SDR-14. There are times when there are only 4 bits wiggling and I can see
>the whole band and demodulate SSB, CW and all signals with no problem. You
>need to remember that the few bits are being sampled at 67 MHz in this 
>case.
>They have an EXTREMELLY large amount of information. This is not 4 bits
>wiggling after a 2.4 KHz bandpass filter being sampled at 8 KHz like in a
>soundcard.
>
>Let's do the numbers. Let's say you have a 2 bit radio and that you are
>sampling at 100 MHz. The S/N in the nyquist band is 2*6.02+1.76 or roughly
>14 dB. The nyquist band is 50 MHz wide so you get a per Hz improvement in
>C/N of 10*log (50MHz)= 77dB. that means that if I used a digital filter 
>that
>is 100Hz wide (a 20 dB hit from 1Hz) you could still get 70dN of C/N. This 
>I
>would say is more than sufficient to demodulate CW, FSK, PSK etc.
>
>This technology is right around the corner with better performance and 
>lower
>cost. I hate to say it since I am a die hard RF guy, but the more I play
>with this technology the more potential I see.
>
>Regards,
>
>Pieter, N4IP
>
>
> >
> >Yes but how many bits above the noise floor are required to demodulate 
>real
> >information.
> > Even cw is going to take a few bits to decode. This will cut into the
> >noise floor of the A/D.
> > The DSP needs more than 1 bit to determine if there is a carrier or
> >noise. I got tricked into looking at displays rather than listening to 
>the
> >audio coming out of the DSP.
> > Yes this stuff is here to stay but it will be a while until we can
> >afford real performance. gfz
>
>_________________________________________________________________
>Find what you need at prices you’ll love. Compare products and save at MSN®
>Shopping.
>http://shopping.msn.com/default/shp/?ptnrid=37,ptnrdata=24102&tcode=T001MSN20A0701
>
>

_________________________________________________________________
Find what you need at prices you’ll love. Compare products and save at MSN® 
Shopping. 
http://shopping.msn.com/default/shp/?ptnrid=37,ptnrdata=24102&tcode=T001MSN20A0701