[prev in list] [next in list] [prev in thread] [next in thread]
List: usrp-users
Subject: Re: [USRP-users] Clock Drift Questions
From: tilla--- via USRP-users <usrp-users () lists ! ettus ! com>
Date: 2014-08-21 17:35:40
Message-ID: 2031921285.741425.1408642540109.JavaMail.root () comcast ! net
[Download RAW message or body]
[Attachment #2 (multipart/alternative)]
Dont think I did a reply to all, sorry if duplicated...
Started a short time ago when looking at a Tx signal on a Spectrum Analyer N210/WBX \
combo.
Set to 100.0 MHz Tx and actually reading 100,002,000... Not a huge deal, but was \
causing what looked like an amplitude variation in the TX'd signal.
Kinda lead me to wonder what would make this happen and I think I ended up at \
something related to clock...
Also desiring to Tx at a timed interval as accurate to 10 MSA/sec as possible... So \
10 MSA is 100 ns resolution, figuring 2.5 usec per second of potential drift doesnt \
seem accurate enough...
----- Original Message -----
From: "usrp-users" <usrp-users@lists.ettus.com>
To: "usrp-users" <usrp-users@lists.ettus.com>
Sent: Thursday, August 21, 2014 9:41:00 AM
Subject: Re: [USRP-users] Clock Drift Questions
Hello Tilla,
1) If you use an external reference clock, all clocks in the USRP will be derived \
from that -- the internal clock doesn't influence it anymore. So if you need a \
higher stability than the standard oscillators provide, you need a very good external \
clock.
2) The clock accuracy should be about 2.5ppm, if I remember correctly. Note however \
that this is a value with stochastic characteristics.
3) https://www.ettus.com/content/files/X300_X310_Spec_Sheet.pdf : about 2.5ppm, too.
4)different approaches, dependiing on what your application needs:
a) compensate effect of non-exact clock
- frequency correction on the RX signal, as you'll need it in almost any real-world \
application anyway :) Most common approach for communications; the 2.5ppm frequency \
error is often enough completely irrelevant to actual transmissions. b) share the \
same clock on all used devices to eliminate frequency offsets between \
USRPs
- daisy-chain the clocks of X310
- use a clock distributor like the Octoclock to act as a common external clock
Demands co-localization of all involved devices; doesn't actually "correct" \
frequencies but just avoids having different ones. c) lock to a globally unique \
clock
- use an integrated GSPDO
- use an external GPSDO (e.g. Octoclock-G)
Better clock; most likely the best accuracy-per-money you'll get.
All in all: a 2.5ppm clock is in most cases considered "good". What is your use case? \
Greetings,
Marcus
On 21.08.2014 15:05, tilla--- via USRP-users wrote:
Couple easy ones:
1) Does using an external 10 MHz signal compesate for clock drift vs internal clock?
2) Seen N210 clock drift quoted at 2.5 ppm. Correct?
3) What is the clock drift on X310?
4) Any "simple" ways to compensate for drift?
Thanks,
_______________________________________________
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
[Attachment #5 (text/html)]
<html><body><div style="font-family: Arial; font-size: 12pt; color: \
#000000"><div>Dont think I did a reply to all, sorry if \
duplicated...</div><div><br></div><div><div style="background-color: #fdfdfd;" \
data-mce-style="background-color: #fdfdfd;">Started a short time ago when looking at \
a Tx signal on a Spectrum Analyer N210/WBX combo.</div><div style="background-color: \
#fdfdfd;" data-mce-style="background-color: #fdfdfd;"><br></div><div \
style="background-color: #fdfdfd;" data-mce-style="background-color: #fdfdfd;">Set to \
100.0 MHz Tx and actually reading 100,002,000... Not a huge deal, but was \
causing what looked like an amplitude variation in the TX'd signal.</div><div \
style="background-color: #fdfdfd;" data-mce-style="background-color: \
#fdfdfd;"><br></div><div style="background-color: #fdfdfd;" \
data-mce-style="background-color: #fdfdfd;">Kinda lead me to wonder what would make \
this happen and I think I ended up at something related to clock...</div><div \
style="background-color: #fdfdfd;" data-mce-style="background-color: \
#fdfdfd;"><br></div><div style="background-color: #fdfdfd;" \
data-mce-style="background-color: #fdfdfd;">Also desiring to Tx at a timed interval \
as accurate to 10 MSA/sec as possible... So 10 MSA is 100 ns resolution, \
figuring 2.5 usec per second of potential drift doesnt seem accurate \
enough...</div></div><div><br></div><hr id="zwchr"><div \
style="color:#000;font-weight:normal;font-style:normal;text-decoration:none;font-family:Helvetica,Arial,sans-serif;font-size:12pt;"><b>From: \
</b>"usrp-users" <usrp-users@lists.ettus.com><br><b>To: </b>"usrp-users" \
<usrp-users@lists.ettus.com><br><b>Sent: </b>Thursday, August 21, 2014 9:41:00 \
AM<br><b>Subject: </b>Re: [USRP-users] Clock Drift Questions<br><div><br></div>
Hello Tilla,<br>
<br>
1) If you use an external reference clock, all clocks in the USRP
will be derived from that -- the internal clock doesn't influence it
anymore.<br>
So if you need a higher stability than the standard oscillators
provide, you need a very good external clock.<br>
<br>
2) The clock accuracy should be about 2.5ppm, if I remember
correctly. Note however that this is a value with stochastic
characteristics.<br>
<br>
3) <a class="moz-txt-link-freetext" \
href="https://www.ettus.com/content/files/X300_X310_Spec_Sheet.pdf" \
target="_blank">https://www.ettus.com/content/files/X300_X310_Spec_Sheet.pdf</a> : \
about 2.5ppm, too.<br> <br>
4)different approaches, dependiing on what your application needs:<br>
<br>
a) compensate effect of non-exact clock<br>
- frequency correction on the RX signal, as you'll need it in almost
any real-world application anyway :)<br>
Most common approach for communications; the 2.5ppm frequency error
is often enough completely irrelevant to actual transmissions.<br>
b) share the same clock on all used devices to eliminate frequency
offsets between USRPs<br>
- daisy-chain the clocks of X310<br>
- use a clock distributor like the Octoclock to act as a common
external clock<br>
Demands co-localization of all involved devices; doesn't actually
"correct" frequencies but just avoids having different ones.<br>
c) lock to a globally unique clock<br>
- use an integrated GSPDO<br>
- use an external GPSDO (e.g. Octoclock-G)<br>
Better clock; most likely the best accuracy-per-money you'll get. <br>
<br>
All in all: a 2.5ppm clock is in most cases considered "good". What
is your use case?<br>
<br>
Greetings,<br>
Marcus<br>
<br>
<div class="moz-cite-prefix">On 21.08.2014 15:05, tilla--- via
USRP-users wrote:<br>
</div>
<blockquote cite="mid:229812056.697245.1408626347061.JavaMail.root@comcast.net">
<pre>Couple easy ones:
1) Does using an external 10 MHz signal compesate for clock drift vs internal clock?
2) Seen N210 clock drift quoted at 2.5 ppm. Correct?
3) What is the clock drift on X310?
4) Any "simple" ways to compensate for drift?
Thanks,
</pre>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre>_______________________________________________
USRP-users mailing list
<a class="moz-txt-link-abbreviated" href="mailto:USRP-users@lists.ettus.com" \
target="_blank">USRP-users@lists.ettus.com</a> <a class="moz-txt-link-freetext" \
href="http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com" \
target="_blank">http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com</a>
</pre>
</blockquote>
<br>
<br>_______________________________________________<br>USRP-users mailing \
list<br>USRP-users@lists.ettus.com<br>http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com<br></div><div><br></div></div></body></html>
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic