[time-nuts] Noise and non-linear behaviour of ferrite transformers

Richard (Rick) Karlquist richard at karlquist.com
Sat Jul 19 21:38:17 EDT 2014

I did some tests of residual phase noise using an
Agilent E5505A and found that air coil inductors
did not add noise (at least down to my noise threshold)
but that ferrite core inductors had easily seen noise.
It was on the order of ADEV = 1E-10 close to the
carrier.  I would describe this as noise rather
than non-linearity.

In my case, the inductors were used to make tuned
circuits.  This is different than your case where
the ferrite is being used to make a transformer.
So as they say "your milage may vary".  And you
probably don't have the option of an air core.

After obtaining the E5505A and using it for a while,
it became clear that anyone who is serious about
component noise has to measure the components of
interest to them in the application of interest
to them on the E5505A or equivalent.  (And the
Agilent E5202A Signal Source Analyzer isn't
equivalent.)  Unfortunately, the E5505A is quite
expensive and requires a highly skilled user.

You mentioned "theory" several times.  This sort
of thing is 1% theory and 99% experiment.  Or
another way of putting it is you do a bunch of
measurements and then construct a theory to
explain what you already know experimentally.

Rick Karlquist N6RK

On 7/19/2014 3:09 PM, Attila Kinali wrote:
> Hi,
> I'm currently looking at some way of breaking the ground loop between
> several systems. The obvious idea would be to use transformers. I would
> like to have some kind of rule of thumb to guess how much noise such
> a transformer would add. But unfortunately i cannot find any theory
> or measurements of this. Does anyone have some pointers to documents
> on what kind of noise i could expect (type, and strength) and
> what/how strong the non-linear behaviour of transformers would be?
> Thanks in advance
> 			Attila Kinali
> PS: although this started as something with a real application in mind,
> i'm now interested in this as an endavour of its own. So all and any data,
> theory or rule of thumb would be appreciated

More information about the time-nuts mailing list