What about the phase noise of my phase locked source ?
I got the phase noise -82dBc/Hz.100KHz.
It is bad or good? :(
What type of xtal oscillator have you got? and the xtal freq?
The phase noise in xtal @100KHz is very poor.
At 2GHz, you can reach better than -100dBc/Hz with any VCO locked to a 10MHz OCXO with PLL (like ADF4113)
MANUEL
Thanks.
So I have to get better Xtal.
One standard trick to clean up crystal oscillators is to follow it by a crystal filter. You can make a bridge circuit with one crystal and a capacitor in the other branch. Look in any radio design book from 1950 or before for the filter circuit.
You should also consider the loop itself. It may introduce some noise or may allow other noise sources to be introduced.
Hello,
Besides the good earlier posts above, there is one factor to be considered. Probably the Quality Factor of your printed microstip line that gives the needed L inductance for the VCO is rather low, say 20-30 or even less.
There is a paper from Rhea on how to design higher Q microstip lines on FR4 material, see http://www.elektroda.pl/eboard/searc...-rhea.html+fr4 here at elektroda).
Or you can use coaxial resonator instead of the microstrip line like in a new product at http://www.synergymwave.com/products...s/SYN2020N.pdf (The coax resonator can be made from a high quality 1/4 wave long semi-rigid coax cable piece.)
There are several VCO manufacturers who produce VCOs for / around 2GHz. Have a look at their data sheets on their phase noise performance to get a better feel on what is achieveable.
Have a look at this url for some insight on low noise VCOs here:
http://www.synergymwave.com/Articles/Articles.htm but of course there are many good papers on this topic.
unkarc
What is the crystal frequency? What frequency are you comparing in the phase detector? Do you have the phase noise of your unlocked VCO at 100kHz? With this information, it will be possible to calculate exactly where the problem lies.
Glad to see many experts here.
The xtal frequency is 18MHz. And the reference frequency about 1.5MHz.When locked, phase noise of 2 GHz output is -82dBc/Hz offset 100KHz.
Mr toonafishy, could you calculate what's wrong? Thanks.
The file in fttp://www.elektroda.pl/eboard/searchtopic41906-rhea.html+fr4 can not download. Anyone can help me? Thanks.
8O :D
Hello,
If you cannot download file from elektroda
http://www.elektroda.pl/eboard/ftopic41906.html , the rhea.pdf file can be downloaded from the eagleware site as #2015: "Designing a Low-Noise VCO on FR4" from
http://www.eagleware.com/apps/ByNumber.htm It is possible that you have to check-in with a correct e-mail to be able to download it from eagleware. It is free.
unkarc
Phase noise of -80dBc at 100KHz is poor. I still consider my PLL design with phase noise of -80dBc at 10kHz is poor as it affects significantly on the quality of modulation at low bit rate (ie. 10kbps-100kbps). To reduce the noise effect, u must identify the noice source. Try to use lower noise VCO, narrower but acceptable loop bandwidth, and as what most ppl suggested: use better Xstal oscillator (phase noise at 10Khz of 140dBc/hz is available in market)
2 activewei
Could you share the name of the suppliers of such oscillators? I am personally looking for 10MHz with -150dBc@10KHz (preferably low-cost one).
Thanks.
activewei
I also want to know this Xtal.
TO: Olxx and microstrip_line: Sorry for the late reply.
I can recommend you Corning product. Check out more at
http://www.corningfrequency.com/tcxo/
SHowing you the phase noise of C2410 TCXO,
-90 dBc/Hz @10Hz
-120 dBc/Hz @100Hz
-140 dBc/Hz @1 Hz
-150 dBc/Hz @10kHz
Pdf file available at
http://www.corningfrequency.com/tcxo/c2410.pdf
Is it great? yes.. it is... I am using it for space application :)
I think the real problem is the large N value. A 1.5MHz compare frequency is too low for a 2GHz oscillator to get good phase noise. The crystal noise is divided by 20*log(R) by the reference divider but is then multiplied by 20*log(N) by the main loop divider. Any noise within the loop will be raised by (in theory) 20*log(N). With certain synthesizer such as charge pumps, its more like 10*log(N).
Your crystal starts out at -100dBc. It is divided to 1.5MHz so now the noise is -100 - 20*(18/1.5) = -121.5dBc but within the loop it is multiplied up, -121.5 + 10*log(2e9/1.5e6) = -90dBc. If your VCO phase noise is -115dBc. then -115dBc + 10*log(2e9/1.5e6) = -82dBc.
i feel the crystal phase noise is quite poor & you can't get better phase noise than that of the crystal that you are using! so it's better to opt for a better crystal ; moreover since the reference freq. is 18 MHz according to you and you are trying to lock a 2 GHz VCO the multiplication /division factor is quite high l; it would be better if you could get a higher frequency reference say around 100 MHz or 50 MHz :)
My friends at Stanford Research Systems make some of the best oscillators I have come across. Not that cheap though. But anyone needing ultra-low phase noise look at this. We used a lot of them in a GPS freq standard product.
http://www.thinksrs.com/products/SC10.htm