[hpsdr] ADC Question

Wed Jul 25 19:11:49 PDT 2012

With the networked series USRP hardware I've run multiple receivers at 25Msps with Studio1 or single instances with winrad, hdsdr etc...

The sample rates can be pushed to where it appears the gig Ethernet interface becomes the bottleneck, and I imagine the number of receivers would be based on what the PC hardware can handle.

The Hermes hardware looks to be very capable. Hopefully some of the talented developers responding here can help break away from the noted limitations of PowerSDR.

Very interesting project you folks have here. I think I'll buy a board ;-)

On Jul 25, 2012, at 1:36 PM, Larry Gadallah <lgadallah at gmail.com> wrote:

> ***** High Performance Software Defined Radio Discussion List *****
> 
> 
> 
> On 24 July 2012 20:55, Bill Tracey <bill at ewjt.com> wrote:
> Alberto - there's nothing magic in 48/96/192, just happened to be what was there before our hardware.  In a purely technical sense, you can use whatever sampling rate you want.   The constraint is you need to have agreement on those sampling rates at your  interfaces.  In our case, PowerSDR  supported 48, 96, and 192 so that is what we went with at the time.   I could have rewritten the code to support arbitrary sampling rates, but I took the pragmatic approach and went with being compatible with what was at hand.   That also made life easier interfacing to VAC and others.  The problem I ran into in PowerSDR was the CW  generator converted timing data to sample counts assuming  48khz.
> 
> Does anyone actually know how well feeding non sound card based rates into VAC works?   Can Perseus's or QS1R's software feed to/read from VAC?
> 
> Larry -  what sampling rates are you looking for? Would 24/12/8/6 or 384/576/768/1536 khz  - these are all integral decimations from 122.88MHz so are not hard to do in the firmware.  Are you looking for absolutely arbitrary sampling rates, or would integral decimations from 122.88 MHz fit your needs?
> 
> Bill - All I can say is that when I heard that a Gigabit Ethernet interface was going to be provided, I expected that the ADC/CIC decimator spigots could be fully opened, limited only by what the network between the SDR and the PC could handle, and the processing power of the PC. After using the QS1R for some time, and being able to watch 500 kHz or 1 MHz of spectrum at once, being stuck at 96 or 192 kHz makes me feel like I'm looking through a keyhole again :-)
> 
> I understand from the recent discussions on the weekly teamspeak sessions that there are some real and non-trivial reasons why it is not easy to go past 192 kHz sampling rates, and given the relationship of the ADC sampling clock and the CIC decimation, there needs to be an integral relationship between the sample rate to the PC and the ADC clock rate (unless a lot of complexity is introduced into the FPGA firmware). However, I agree with the later comment that the PC-based software should not be dependent on this relation.
> 
> 
> As to where this is baked into the architecture - at the moment the transmit IQ samples are assumed to be 48 khz and the received IQ samples are allowed to be 48/96/192 - this is all the protocol between the PC and the device publishes.  The protocol can certainly be extended, just needs to be done with care as we'd like to avoid breakage to existing programs as much as possible
> 
> Regards,
> 
> Bill (kd5tfd)
> 
> 
> 
> At 04:50 PM 7/24/2012, Larry Gadallah wrote:
> Is there really such a constraint ? Why the software running on the PC must be fed with
> 48/96 or 192 kHz samples ? Aren't those rates an heritage from the sound card era ?
> Take for example Perseus.  Its DDC sends to the PC rates of 125, 250, 500 kHz,
> 1M or 2M samples/sec.  Both the original Perseus software and my Winrad are perfectly
> capable to cope with those rates.  IMHO there is nothing magic with 48/96/192 kHz.
> 
> --
> 73 Alberto I2PHD
> 
> 
> 
> I kind of agree with Alberto here: In addition to the Perseus, Phil Covington's QS1R also has a wide variety of non-integral sampling rates available. In my astonishingly small knowledge of DSP techniques, I know that non-integral resampling is a difficult task, but non insurmountable. Phil and Alex have been working recently on improving the sampling rates, and this happens to be one of my biggest wish list items, so it would be good if we could somehow break away from the soundcard heritage and start to take advantage of the true capabilities of the hardware (i.e. CW Skimmer Server and 5-band WSPR beacon).
> 
> All of this kind of leads to a supplemental question: How much of this "integral of 48 kHz" is baked right into the architecture?
> 
> 
> 
> 73,
> -- 
> Larry Gadallah, VE6VQ/W7                          lgadallah AT gmail DOT com
> PGP Sig: B5F9 C4A8 8517 82AC 16B6  02B6 0645 69F0 1F29 A512
> _______________________________________________
> HPSDR Discussion List
> To post msg: hpsdr at openhpsdr.org
> Subscription help: http://lists.openhpsdr.org/listinfo.cgi/hpsdr-openhpsdr.org
> HPSDR web page: http://openhpsdr.org
> Archives: http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20120725/55e6c483/attachment-0004.htm>