microphonics and ferrite inductors

[Colon]

Hi everyone,

I have a board working from 10kHz upwards and have been seei9ng microphonics whenever the chassis/box is vibrated/shocked. Working to get rid or at least seriously improve this I have so far changed all my capacitors in the signal path from ceramic X7Rs to tantalums (COGs not available in 10uF). This has made it much better, apart from a single key component, a transformer.

The transformer is an RM10 bobbin with two 3C90 ferrite halves clipped together. A test board with just the transformer on will product noise spikes on a spectrum analyser whenever it is shocked. I am trying gluing various parts of the structure and also considering mounting the PCB differently. One or all of these might fix the problem.

What I would like to know is what the mechanism is for the generated noise. The test board is not powered. I didn't think that ferrite was magnetic in itself, so that would rule out the ferrite moving relative to the coil. I suppose there is always the earths magnetic field...Can a ferrite generate a magnetic field under vibration?

Also, any tips generally would be welcome.

Thanks
James

 

[D.A.(Tony)Stewart]

most ceramics and ferrites are microphonic. even if epoxy encapsulated, but that may help. For example Touch screens always use plastic caps in critical areas of detection amplifiers.

 

[FvM]

most ceramics and ferrites are microphonic.

For ferroelectric media, it should be true in presence of residual magnetization. I'm not sure if it also applies to a perfectly demagnetized ferrite. Do you have any references?

 

[sky_123]

I also remember seeing some article that mentioned positioning inductors in certain positions on the
board, to minimise flex on them - and for example to cut slots in the board so that the inductors
experience less flex once the board is mounted.

 

[D.A.(Tony)Stewart]

https://www.youtube.com/watch?v=vkDLQAvrEQQ watch at 1:30 into demo.... This unit has a ferrite bobbin core.
Suggestions:
Consult with Ferroxcube
Find a more suitable solid torroidal core.
Experiment with gap bonding agents prior to clamping and measure effect on d(inductance)/dt microphonics.

V= L dI/dt + I dL/dt

EPoxy coat outside cores.

Consult with Ferroxcube
good luck
( did I say consult with supplier tech support {)

 

[mtwieg]

There are a variety of ways in which a ferrite transformer can transduce shock into electrical interference. Magnetostriction is one of them, and is difficult to deal with. You can only reduce shock to the cores by so much, and one of the only thing that can further reduce the effects of magnetostriction is to change the air gap, but that's not always feasible.

Another issue that hasn't been mentioned is motion of the windings. If your windings are allowed to move within the magnetic field, then they will generate EMF (basically lorentz forces in reverse). Rigidly fixing the windings' position relative to the core will reduce this.

You didn't mention where the microphonic signal is actually being observed, and how that system is coupled to the transformer. We should first try and figure out whether the interference from the transformer is conducted or radiated.

Edit: also I wonder if powdered iron cores are less susceptible to magnetostriction?

 

[Colon]

Ok, thanks for the advice guys.

Tests this afternoon have shown that 'potting' the gap between the core and the bobbin has had a significant effect. I would therefore say that the majority of the effect was caused by relative movement between the core and bobbin.

If a magnetic field were present I would be happy that I knew the mechanism by which the noise was being generated. But, unless the ferrite core is magnetic (is it? even slightly?) then the magnetic field must be from elsewhere. Remember this is an unpowered device. The input to the transformer is terminated and the output is connected to a high gain amp which then feeds a spectrum analyser. This is how I have observed the generated noise.

 

[D.A.(Tony)Stewart]

piezo-electric effect works on many ceramic compounds including ceramic materials with metallic oxides as well Also dL/dt (f) should be considered if there is a frictional air gap.

There is a real impedance change from pressure on the ferrite gap. Possible due to misaligned crystaline centro-Asymmetry and motion current flowing thru the device to the 50Ω network analyzer and gap modulation from pressure impulse.

Shock isolation and rigidity help reduce this effect.