How does an external reference clock affect LimeSDR performance?

Some weeks ago I bought a GPSDO, intending to use it with my LimeSDR 1.4, for experimenting with OAI.

Now that I have some of the tools I need, I realized that I don’t know how an external clock affects the ability of LimeSDR to capture and transmit e.g. GSM frames, or whether it should (or can) noticeably improve the ability to function as an LTE air interface.

Having read the schematic, it appears that I have to feed my external clock directly to the clock buffer CLKIn input, otherwise the phase detector fed by LimeSDR REFIn simply signals the FPGA to adjust the VCXO output, so that frequency stability is still determined by the Rakon clock part.

What effect does feeding a 10MHz GPSDO clock to the Lime have on my GSM, LTE, and WiFi experiments with the radio? I’ve been able to use the LimeSDR with Osmocom’s GSM stack (very cool!), and would like to know whether I even need this GPSDO for LTE and WiFI experiments.

(Yes, I could experiment; but I’d really rather not be spending for additional kit not necessary to use capabilities the LimeSDR provides.)


It’s not the difference between the stock (30.76 I think) & 10 Mhz.
It’s the precision that you are getting. Usually, a GPS 10 Mhz does 10.00000 Mhz minimum. Not 2 decimal points, but 5 or more. Coordination becomes much greater.


Thanks, Ed @AA7QQ. “Better stability”, which I take to mean frequency and phase stability (the clock source’s ability to stay close to it’s specified output, with low jitter shift) sounded like a good thing. What I don’t understand is:

  1. Whether an accurate external clock (with good short- and long-term accuracy + precision) improves the LimeSDR clock precision from ~ 2.5ppm to the stability provided by the external clock. This, in turn, presumably (?) affects both the sampling process and transmit waveform accuracy.

  2. Whether (and, more importantly, how) this affects the capability of LimeSDR to receive and send - e.g. serve as a radio head - in two domains of interest to me a) GSM and LTE, and b) WiFi 802.11b/g.

I saw the demos showing use of the LimeSDR as radio for an LTE eNodeB, with phones at very close range. From that I inferred that RX/TX frame timing was no issue for experimenters like me, operating at low power and negligible distances between mobile and radiohead.

The bottom line question for me is under what conditions it makes sense, for my use case, to spend for improvement in my LimeSDR’s clock precision.

I guess, simply said, 2 clocks on the same clock are going to communicate data better. Especially, if the accuracy is better.
I am no pro at this stuff, but do understand the “Mechanics” of it. Where’s the jetting kit for this Holley?