[USRP-users] 10us+ sample delay between daughterboards

Marcus Müller marcus.mueller at ettus.com
Thu Jul 23 06:13:29 EDT 2020


I also just realized that you can be reaaaally silly. If you have a
transistor and two resistors to spare, use the GPIO front panel
connector to generate a faux-PPS pulse, buffer it using the transistor,
and feed it back to the backpanel PPS input... might save you the extra
device, but of course the RPi solution is still elegant, as it allows
for not-quite-but-at-least-close-ish NTP synchronicity.

Cheers,
Marcus

On 23.07.20 11:44, Johannes Demel via USRP-users wrote:
> Hi Marcus,
> 
> I tried to improve the situation. I had another look at [0,1].
> 
> From [1] (N310 manual)
> "[..] which can both be used as time- and clock references. The GPSDO
> will function as a reference even when there is no GPS reception."
> 
> This helps. I set
> Sync: "Unknown PPS"
> Clock/Time source: "O/B GPSDO"
> and all streams are in sync. That's sort of a relief for N310s.
> 
> Since I don't have a GPS signal in the lab, I assume there's no point in
> trying to receive a GPS timestamp and set the device time on next PPS
> [0]. I would prefer to work with timestamps "since epoch" instead of
> "since start of flowgraph".
> 
> I need to figure out how to use GPSDO on those X310s though. Since the
> GPS LED does not light up, I don't think it detects a GPSDO at all. And
> I just hope it can use a GPSDO like an N310.
> 
> Since I work remotely, I can't just manually generate a sync pulse. But
> if I can find a device that works reliably I may switch to that. Thanks
> for the explanation and ASCII schematic.
> 
> Cheers
> Johannes
> 
> [0] https://files.ettus.com/manual/page_gpsdo_x3x0.html
> [1] https://files.ettus.com/manual/page_usrp_n3xx.html#n3xx_synchronization
> 
> On 23.07.20 09:59, Marcus Müller via USRP-users wrote:
>> Hi Johannes,
>>
>> let me increas Marcusness of this by ~3dB.
>>
>> On 23.07.20 09:29, Johannes Demel via USRP-users wrote:
>>> I don't have a PPS signal readily available. Would a 10MHz reference
>>> suffice as well?
>>
>> Nope, that doesn't set a time register. You don't actually need a pulse
>> per second – you need a single pulse.
>>
>> It's a trick I usually tell to impress physicists doing measurements:
>>
>> 1. call set_time_next_pps
>> 2. connect 2× 1.5 V batteries (or similar) to get a voltage between
>> something like 2.5 V and 5V.
>> 3.             |
>>               -----
>>     ----------o   o----------+---------------> SMA inner
>>    |   +                     |
>> ------         push-       | | pull-
>>   ---bat        button      |R| down
>>    |                         |
>>     -------------------------+---------------> SMA outer
>>
>> The pushbutton can effectively be anything that has a nice quick
>> contacting mechanism to yield a nice edge, without ringing dangerously
>> high.
>>
>>
>>> Do I get rid of this issue if I use a GPSDO? We have the hardware
>>> (antennas + modules) available but no GPS signal in our lab.
>>
>> Hmmmm I'd have to think hard here: the GPSDO does offer 10 MHz from its
>> internal oscillator when it's not locked, but I think the PPS is really
>> only present with GPS time
>>
>>> Besides, I was really hoping that one USRP would sync itself.
>>
>> Sounds fair; there's an internal PPS pulse, I've never tried it. Have
>> you tried setting the pps source to "internal" and then setting on PPS?
>>
>> Cheers,
>> Marcus the younger
>>
>> _______________________________________________
>> USRP-users mailing list
>> USRP-users at lists.ettus.com
>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>>
> 
> _______________________________________________
> USRP-users mailing list
> USRP-users at lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com



More information about the USRP-users mailing list