Re: [Discuss-gnuradio] Universal device API, parameter standardization?

[Moeller]

 On 05.08.2010 19:38, Marcus D. Leech wrote:
> On 08/05/2010 01:10 PM, Philip Balister wrote:
>> Non-Ettus products can also support UHD.
> Indeed, all the code is freely available, and it would seem like a
> company like, for example,
>   ThinkRF (who already have Gnu Radio support) could head in the UHD
> direction.  Nothing
>   stopping them--no "secret handshakes" required, etc

But is the UHD really device-independent ?
In the sense that you compile a program and later you don't have
to change the source to support a different device?
That could be possible with a device selector, configuration files
and a certain device-independent capabilities data structure.

My impression from the source is different.
In rx_timed_samples.cpp
the URSP-device is hard-wired, as a uhd-derived class.
"uhd::usrp::simple_usrp::sptr sdev"
So, once compiled, it can only support the USRP-type.

I would prefer to leave this dynamic.
So, e.g. you open the spectrum analyzer and you have a button
to choose a device. And if there is only one available, this one
is chosen automatically as default.
So, all example programs, spectrum analyzers etc. work for all
supported devices, without source patching.

Is that possible with UHD, or does it need a different design.
I did not find a universal API, a dynamic data structure of device
capabilities that can be changed in run-time.

ThinkRF solutions would cost around $10000. That's not on my radar.
I think USRP2 is already a good start for doing serious experiments.

Additionally I wish I had a small SDR device for private amusement,
as a spectrum analyzer and oscilloscope. Also for the kids to play.
A smaller frequency range and bandwidth would be accepted, for
a price around $100-200, RX-only. I think that should be realistic.
The ADC are not that expensive. There are some amateur solutions
around 100$, but they use the soundcard as ADC. For an oscilloscope
that's not enough bandwidth.

Moeller