Page 1 of 1

10MHz / 1pps clock input

Posted: Fri Jul 25, 2014 1:59 pm
by jump
As a disclaimer, sorry if this question seems "weird" or very basic to some fo you but just consider that I pretty new to DSP and radio stuff :)

What kind of application may require to feed a precision clock reference input (Rubidium, GPS) in the bladeRF?
I saw on the forum a (very interesting) subject about ground station and telemetry (DGSN if I remember correctly) and I know that the list may be pretty huge...
I also know that most of the time, for dealing with day to day RF signals from commercial devices I won't need such precision reference and that's why I am interested on the corner cases where it starts to be problematic to only rely on the internal clock. For example, is it required for GSM/GPRS/3G/LTE?

Re: 10MHz / 1pps clock input

Posted: Sat Jul 26, 2014 6:34 am
by bpadalino
That's a good question. Recently someone had wanted to track the beacon from the ISEE-3 satellite. Since it is a beacon, it is a single carrier. Moreover, they wanted to watch the Doppler shift of the satellite. When doing an experiment like this, they wanted to only look at 200Hz worth of spectrum at a time. When you want to look at very narrow bandwidths from very high frequencies, then using a precision LO is basically required such that your signal of interest (way up at RF) doesn't appear out of band.

To have this make a little more sense, lets look at an example. Our VCTCXO is 38.4MHz with a 1.0ppm tolerance over temperature meaning it can be off by as much as 38.4Hz. When using a PLL, it will take this reference clock, and multiply it to get a signal of interest - lets say 2GHz. In this case, it would multiply the output of the VCTCXO by around 58. Up at 2GHz, 1ppm is 2000Hz - around 10x the overall bandwidth of the narrow 200Hz spectrum they wanted to view.

In general, synchronization and frequency offset error is inherent in many systems - but sometimes having a really good oscillator makes life a lot easier.

On a related note, we're still committed to getting a 1pps and 10MHz reference into the board and able to lock onto it.

Brian

Re: 10MHz / 1pps clock input

Posted: Sat Jul 26, 2014 6:54 am
by SDR-Radio.com
bpadalino wrote:On a related note, we're still committed to getting a 1pps and 10MHz reference into the board and able to lock onto it.

Brian
Wow - that's worth paying for. I'm into satellites and stuff like that...

Re: 10MHz / 1pps clock input

Posted: Sun Jul 27, 2014 12:00 pm
by jynik
SDR-Radio.com wrote:Wow - that's worth paying for. I'm into satellites and stuff like that...
I think Brian's wording might have been confusing there...

From my understanding, "getting a reference into the board" implied bringing in the aforementioned signals into the board via added features to the FPGA and having folks connect their input signal to the J71 expansion connector. I assume all of this would be accompanied by some sort of tutorial, perhaps covering the use of some COTS reference device?

So in short -- no new boards or HW to purchase is required, right? Brian, perhaps you could elaborate or correct me if I'm wrong.

Re: 10MHz / 1pps clock input

Posted: Mon Jul 28, 2014 12:18 pm
by jump
Thanks brian for your answer.

Most of the time, that kind of accuracy constraint seems to be related to communication with space (probes, satellites, etc.), right?
Back on earth, is there an application that may require such clock?

Again, sorry if that seems silly. I am basically trying to figure out if it may be relevant, regarding the expected usages I have for my bladeRF, to get a hand into such clock. Because they don't tend to be cheap :)

Re: 10MHz / 1pps clock input

Posted: Mon Nov 10, 2014 12:07 am
by drew23
Hi. The 1pps GPS disciplined TXCO are readily available, new and cheap. They use the GPS on flying drones, the 1pps is just a bonus. See http://vk4zxi.blogspot.com.au/2014/09/c ... lator.html. Regards Drew VK4ZXI

Re: 10MHz / 1pps clock input

Posted: Mon Sep 21, 2015 7:48 am
by scitts2000
Hi, after reading this thread, I would like to ask.
1. What is the way to get bladeRF to be driven by an external clock, bypass the VCTCXO and input the clock via the CLK connector
2. What frequency must be set at the CLK connector if this is possible? I have options to generate 10MHz, 38.4MHz and 40MHz.

Eddie

Re: 10MHz / 1pps clock input

Posted: Mon Sep 21, 2015 10:41 pm
by bpadalino
Drive 38.4MHz into the SMB connector and set the device to MIMO Slave mode.

Re: 10MHz / 1pps clock input

Posted: Sun Sep 27, 2015 5:56 pm
by scitts2000
Hi

The Cyclone IV FPGA has a 1PPS input for time synchronization. Where must I connect the GPSDO/1pps line physically?
Is there a timestamp to mark 1PPS signal?

Assuming bladedRF is running with a host, how can I use libbladeRF to read the timestamp of the 1PPS?

Thank you
Eddie

Re: 10MHz / 1pps clock input

Posted: Mon Sep 28, 2015 7:28 am
by jynik
Hi Eddie,

The short answer is that this functionality isn't currently included in the code base, in the FPGA or in libbladeRF.

However, the other devs have been chatting about the design and implementation of this. I'll ask them to post updates to this thread as things progress here and anything we hear regarding Issue Tracker Item #414.

Sorry, that's definitely not the answer you wanted to hear, but hopefully the guys will have something up in a development branch soon to allow others to help test drive it.

Best regards,
Jon