So, I wrote the following script for testing this morning:
http://www.ccera.ca/files/tuning_test.grc
It does random frequency hopping with 1MHz offsets, with a base
frequency of 2.4GHz.
It runs for 60 seconds and then just dumps complex I/Q data to a file.
I then wrote a quick C program to do run-length analysis, and indeed,
there are long runs of exact zeros after a tuning event, averaging about
4.2ms with the above parameters.
I made no attempt to use any of the fancy-tuning options available via a
"tune_request_t".
What I do know is that the AD9361 isn't ideal for frequency-hopping
applications. A lot of "stuff" needs to happen every time it is tuned.
My recollection is that tune times used to be in the range of 50-100msec
for this part, but that an optimization was put in place to reduce this
for "short hops".
On 2017-10-27 01:23, Marcus D. Leech via USRP-users wrote:
> On 10/27/2017 12:44 AM, Gilad Beeri (ApolloShield) wrote:
>
>> Thanks Marcus.
>>
>> I was told by Ettus in the past that nearby hops (X00's of MHz) should take
>> some tens of microseconds to settle (for the AD936x chip), while I see 3
>> milliseconds.
>> Can you reconfirm the former saying, or confirm that you also experience ~3
>> milliseconds of 0's when you change frequency with a delta of tens of MHz?
>>
>> In addition, I would expect that during the settling time, I will get
>> garbage data / noise, and not 0 samples. Do you know the reason I get 0's?
> My *guess* is that this is the AD9361 doing a "mute during tuning" thing. I
> don't know whether that's a feature that can be disabled, and whether at
> some point in the code history, it was enabled. A cursory inspection of the
> code doesn't reveal anything obvious.
>
> When you experience your zero-valued samples, is it a sharp cutoff, or is
> there a ramp into, and out-of the zero-valued-samples state?
>
> Are you using manual or automatic gain control?
>
> On Thu, Oct 26, 2017 at 5:27 PM Marcus D. Leech via USRP-users
> <usrp-users@lists.ettus.com> wrote:
>
> On 10/26/2017 06:43 AM, Gilad Beeri (ApolloShield) via USRP-users wrote:
> Hi,
> Whenever I switch center frequency with a USRP B205mini (changing to a nearby
> frequency), I get 3 milliseconds of 0 samples.
>
> * What's the reason behind that?
> * Can I make some changes that will reduce that timespan without data?
>
> The analog hardware takes a finite amount of time to switch frequencies.
> During that time, there will inevitably be *some* type of glitch.
>
> A PLL synthesizer will take some amount of time to change frequency. Every
> time you change frequency, you take it out of its converged state, and
> it has to re-converge.
>
> https://electronics.stackexchange.com/questions/76197/pll-loop-bandwidth-lock-time-and-jitter
>
> _______________________________________________
> USRP-users mailing list
> USRP-users@lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
