How to depress OCXO very low freq spur?

[tony_lth]

My OCXO has a very low spurs, that is -100dBc@135Hz offset, -101dBc@270Hz offset.
I have two questions:
1. Why OCXO can stimulus so low freq spurs?
2. Have any way to depress the spurs by me, not by vendor?
Thank you very much.

 

[jiripolivka]

What you call "spurs" in fact is phase noise specification. OCXOs are designed to have a low phase noise close to carrier as they are often used as a reference source in frequency multipliers and synthesizers.
To my knowledge, an user cannot affect or improve OCXO by filtering or other means. A careful installation of an OCXO using foam cushions can help to reduce microphonic effects- vibrations do affect the mechanically oscillating quartz resonator inside.
One method does exist to improve phase noise of an OCXO: locking it to a lower-frequency reference which is another OCXO that can have a better phase noise.

 

[vfone]

Inside of an OCXO usually the thermal control-loop (thermistor, amplifier, heather, etc) is not connected electrically to the crystal oscillator, and use its own power regulator.
The thermal control-loop generally is stable due to relative slow temperature response of the thermistor, so there are low chances to oscillate and producing spurious, or to affect the oscillator phase noise (by this reason).
I found some OCXOs that behave funny in the warm-up period, but they were fine after that.
Try to supply the circuit from a battery.

 

[biff44]

Question #1: Are you experienced in making such measurements? It is really easy to mistake a harmonic of 60 Hz AC frequency to cause such a spur, either on the ocxo, or a phantom reading where the measurement system is totally at fault.

I would carefully set up the test with the ocxo and an rf amplifier both running off a battery--no external connections to anything. Then I would use an inside/outside DC block to connect up to the measurment system. Rerun the test and see if the "spur" went away.

Another trick would be to have the OCXO drive a low phase noise frequency multiplier, and run the test at a higher frequency. If it is a real spur on the oscillator output, at the higher frequency it will be the same offset frequency but 20 Log N higher in dBc. If it is an AM spur, the multiplier will probably strip it off.

Rich

 

[tony_lth]

Thank all of you.

Another trick would be to have the OCXO drive a low phase noise frequency multiplier,

In my system, the spur become higher after multiplier and met the law.
And in China, AC current is 50Hz.
Do you think it is ecessary to test with battery?

 

[biff44]

Thank all of you.

In my system, the spur become higher after multiplier and met the law.
And in China, AC current is 50Hz.
Do you think it is ecessary to test with battery?

5 x 50 = 250, which can look a lot like 270!
Try the battery test and see. Spectrum analyzers have a lot of ac chassis leakage, and the battery will tend to fix that. You can get a ground loop between your power supply feeding the ocxo and your spectrum analyzer. If you have a vcxo, then you especially have to worry about such small currents modulating the phase/frequency

I like these things, they will not pass 50 hz:
**broken link removed**

At microwaves you can take two waveguide to coax transitons, and butt them together with a thin sheet of plastic in between and plastic screws--excellent ground loop block.

 

[RCinFLA]

Outside of a problem with the OCXO itself, making sure the power supply is clean, and making sure you are not picking up something from your circuit that is coupling into the OCXO output line, there is not much else.

There is one other thing, even with the OCXO having a one stage output buffer, if there is something modulating the output load impedance there might be an injected shift this way. Strobing the circuit on and off that OCXO is connected to can cause this. I once had this kind of issue with a TCXO feeding separate WCDMA and GSM transceiver chips. When the GSM strobed on and off it changed the load on TCXO that was connected in parallel with WCDMA chip. It screwed up the WCDMA chip operation.

 

[Dr.Drew]

Another problem is EMI from ground or industrial EMI from AC line. I often see an industrial EMI at 1.5 kHz going from near factory. And i can't supress them.

 

[tony_lth]

It is from AC power supply. Now the 50Hz spur is very strong, and yesterday the 250Hz is very strong. Several months ago, when power supply have good grouned, there is no spurs at all.

Thank all of you.

 

[biff44]

Yep, there is all sorts of whacky low frequency noise floating around. Power supplies, switching regulators, leakage out of spectrum analyzer, building AC ground floating above earth ground, stray industrial equipment noise (like induction heaters), ground loops, high impedance pickup on Vtune lines, wireless transmitters, etc etc.

A good tool to have is a battery box, where you have a big battery, and a 10 turn potentiometer with a 10 uf ceramic cap across the output, and use that to be your Vtune supply whenever you think you have a real spur and want to check it.

And don't throw away those old HP analog power supplies--you might need them someday to finish up a test (just check them before using to make sure their old capacitors did not fail and cause unstable output voltages!).

Rich

 

[tony_lth]

I used one HP old power supply, and it is really good to use.

---------- Post added at 02:45 ---------- Previous post was at 02:43 ----------

Now I am using Agilent power analyser as power supply, it is also very good.