Problem with PSU Noise

[alex28]

Hi,

I am working on a project that involves high frequency RF signals. I believe that the RF is maybe travelling on the ground and sending my external external OCOX clock off. The output of my clock is 10MHz and appears to jump about when connected to the same ground as the rest of the equipment. Is it possible to use a De-coupling capacitor to filter out certain frequencies? If so how would i calculate the values for the components needed?

I look forward to your responses.

Many thanks

Alex

 

[shrtrnd]

I'd recheck my test set-up before digging into your circuitry.
Is your 10Mhz clock power source grounded to the same earth ground as your test circuit?
Are there any electrically 'noisy' instruments or electrical devices on the same ground?
If there is any descrepancy in the grounds, it can cause you problems.
It sounds on the face of it, that you're not really grounded at the same point.

 

[CDRIVE]

Is the ext 10MHz clock coupled to your equipment via 50 Ohm coax?

 

[alex28]

Thanks for your quick responses!! The Clock internals are grounded to the rest of the components via a 12v dc to dc. I have also tried using an independant dc to dc for the sole purpose of supplying the clock and still there is noise riding on it. There are multiple pieces of equipment that all transmit rf signals including Power amps and when i use and FSH to measure the clocks output, I can see the various bands being picked up across the range.

The 10MHz clock is linked to all other boards using Coax correct but the SMA connection for the shielding is not linked to the internals of the clock except via the common ground of the chassis which returns to the same DC to DC.

 

[CDRIVE]

<snip>
The 10MHz clock is linked to all other boards using Coax correct but the SMA connection for the shielding is not linked to the internals of the clock except via the common ground of the chassis which returns to the same DC to DC.

Uh, you lost me there. Is the shield ground or not? Is the clock output port a resistive 50 Ohms? Likewise, is the clock input port also 50 Ohms resistive? If transmission line is not properly terminated in its characteristic impedance it becomes inductive, capacitive and possibly an antenna ... among other things. For sure, it no longer possesses the virtues of a transmission line. It can even become resonant.

EDIT: The reason that I specified resistive In/Out ports is because it's a clock signal, that's rich in harmonics. Resistive seems apropos here.

 

[shrtrnd]

We realize you may have a circuit problem. If I was experiencing your problem, the
first thing I would do, would be to try to make SURE nothing outside of your circuit is
causing the problem. If you're in a lab situation, where other RF signals are present,
consider powering-down the test instruments not used at the moment, and recheck your
signals. Hardwire grounds between the clock and the circuit under test, just attempt
to eliminate any other signal problems, before you start tweaking your circuit under test.
We actually had a ground fail to our 10MHz ref standard once, and the grounds between
our instruments started floating. Like CDrive mentioned, I'd want to be sure I wasn't
picking-up a random signal that was screwing-up my 'ground'.

 

[alex28]

Hi sorry for the delay in my reply,

The grounds generally are all common as they all use one 12v Dc to DC module.

The clock im usign for my 10Mhz Ref is an external oven controlled one. The power is supplied via 12v to a DB9 connector. Originally the guys were usign pin 6 for + 8 for the negative for the internals of the clock which is floating in side a metal box and completely isolated from the chassis. Except they bridged the 8th and 9th pin to create a common ground.

I tried to add another dc2dc that isolated its self from the rest of the circuit to see it could could get rid of the noise that is riding on something, but i think there is still some sort of leakage inside this module.

I dont doubt that there is alot of noise inside the lab that im working in but please bear in mind the units i work one themselves actually produce multiple outputs of various frequencies that interfere too.

I really do think it is self generated noise being transmitted on to its self and riding the power wiring. I started to read about bypass and decoupling circuits to remove such noise but its a hard to calculate components and sizes.

 

[alex28]

I also tried using a completely seperated PSU for the clock and that was perfect.
Is there a simple small circuit that could add to the clock power lines to flatten the input voltage across a wide band say 200MHz to 2.6Ghz?

Sorry if these questions are way out there, but i am still a bit of a noob in the world of RF.

Thanks again

 

[CDRIVE]

This thread still has no schematics or photos. Schematic is the language that we speak.

 

[alex28]

Hi CDRIVE,

Its a difficult one for me to take photos of because of the genre of the project to be absolutely honest. Can i fill in any gaps for you? Or literally start again?

I found this and it gives alot of information on what im looking for but i dont really understand the equations and how to calculate a specific band that i want to reduce the noise at.

Filtering Techniques: Isolating Analog and Digital Power
Supplies in TI’s PLL-Based CDC Devices
http://www.ti.com/lit/an/scaa048/scaa048.pdf

Please have a quick glance and see if anything makes more sense of the sort of interference i am dealing with.

 

[(*steve*)]

Its a difficult one for me to take photos of because of the genre of the project to be absolutely honest.

What does this mean?

Are you dishonestly seeking help?

Is this homework?

 

[alex28]

Hi sorry i am being so vague like i say its a difficult one to talk about. But no not homework. And yes honestly seeking help.

 

[alex28]

Some more information i have is that the engineers at my place found a cure using a 10MHz filter on the clock output that attenuates any noise after 20 Mhz, but i still think that this is a temporary solution as the bigger the power amps the more noise will will enter via the power supply.

 

[alex28]

Am I on the right track with the De-coupling caps idea or barking up the wrong tree?