Magnetic Isolation Between Transformers

[Warpdrive]

Greetings, everyone.

Usually I give many answers in this forum; now I have a question a magnetics designer would enjoy.

I have two transformers that must mount in close proximity to each other, with their respective B-fields dictating an unacceptable amount of mutual inductance between them -- enough to disrupt signal integrity to say nothing of the impedance, since each core operates at a different frequency. They have a ferrite spacer between them, in physical contact, that tends to bridge the cores together instead of isolating them per the designed intent.

Using the existing spacer, how would one modify it to improve its isolation performance?

Thanks in advance!

Cheers,

Warpdrive

 

[nbucska]

What is their freq. ranges and core shapes ? What kind of circuit uses them ?


<nbucska@pcperipherals.com>

 

[Warpdrive]

>> What is their freq. ranges and core shapes ?
>> What kind of circuit uses them ?

Power operates in the tens of KHz.
Data operates in the hundreds of KHz.
Both signals are differential.
Core shapes are cylindrical, non-toroid -- like a choke. The power is larger than the data.
Spacer is about 6 mm thick.

 

[Highspeed]

I have been through a similar situation before
ferrite spacer is useless, an air gap is better
Every thing depends on the boundary conditions close to the gap.
Before trying to find a magnetic model of your system you could :
Try to add a millimeter or 2 (or more) between your transformers .
If this not going to work due to your layout, then 2 iron shields separated by an air gap, can do the job as long as your induction is reasonnable.

 

[nbucska]

It looks like the circuit was design to maximize the noise.
To minimize it you should use closed cores (toroid or EI )
with minimal or no gap, oriented 90 deg.

The poser should be the higher ferquency OR mush lower
frequency, sinusoid.


<nbucska@pcperipherals.com>

 

[Warpdrive]

Nbucska,

I concur with your design assessment. I've been trying to teach the core designers (not me!!) and their contracting agents the virtues of optical data links for that reason, but they don't want to listen. They even claim to have their design (as is) working.
Closed cores and toroids are mechanically impossible in my configuration. Power requirements also preclude using a sine wave without upping the amplitude (which will translate into a higher B-field, since I=V/R coil).

I'll go with Highspeed's recommendation.

Thanks!!

Warpdrive (signing off)

 

[nbucska]

Can you rotate one xformer ?

<nbucska@pcperipherals.com>

 

[Heydave]

I have delt with the same problem and solved it by rotating adjacent transformers 90 degrees. Adjacent boards in a cage must also reorient transformers which unfortunatly results in two board layouts for the same function. If all else fails, shield with mu-metal.