Difference between two halves and monolith ferrite cores?

[arvinfx]

I am designing a SMPS but found out that monolith ferrites have more stable curves on oscilloscope and not make any hearable noise! but about EI or EE cores everything is awful! to much audio noise you can hear from core and voltage and current curves in oscilloscope have too much oscillation and spikes. what is my problem? and why open side (monolith ) ferrites code are like this?


monolith ( open side ):

two halves ( close side core):

 

[shortbus=]

Read this for why they use gaps - https://www.theferritegateway.com/TDK-Distributed-Air-Gaps-in-Ferrite-Cores-__n-8.aspx

 

[arvinfx]

Read this for why they use gaps - https://www.theferritegateway.com/TDK-Distributed-Air-Gaps-in-Ferrite-Cores-__n-8.aspx

So you say only air gap is different in this two type of core is Air-Gap?

 

[unclejed613]

so, the air gap in a core needs special treatment. if you look at a commercially made inductor or transformer using split cores, they put plastic spacers in the gaps, and then once the construction of it is completed, they soak the whole thing in varnish, which fills any remaining air spaces, and "glues" the whole thing together. this is how they eliminate the noise that they would normally produce because of magnetically induced motion. an old trick was to find the part of he core or winding that was noisy, and stick a wooden toothpick into the noisy area to dampen the noise. there is actually more than one source of noise possible in an inductor or transformer. the pole pieces slapping against each other (if the gap is not filled properly between the pole pieces), the wires in the winding slapping together (yes the magnetic fields around the wires attract and repel each other, or the wires are attracted or repelled by the ferrite core). if the inductor or transformer uses mounting hardware, the hardware can get loose and make noise. i have seen a lot of different method of eliminating noises, the first, as i mentioned being the toothpick trick, or removing the whole device and dipping in varnish. the device could be potted in silicone potting compound. i used to use a compromise method on CRT monitor flyback transformers, and that was to inject silicone glue into the areas most prone to cause noise (one way that worked well was to disassemble the core, and put silicone glue on the pole pieces and spacers, reassemble the core, and use silicone glue to dampen mechanical contact points between the mounting hardware and the core.

 

[ronsimpson]

oscillation and spikes

I can not see what you are doing. It is common for people that have never used a "EE" core to have no gap. This causes very high inductance and the core will probably saturate. The high inductance will cause a large delay through the PWM and the error amplifier will become unstable and oscillate in the audio range. OR The core will saturate and drive the current up very high. The supply will current limit and that can easy happen in the audio range. Under saturation it is possible for the transformer to act like a speaker. The wires and the core move at extreme current.
I use tape or plastic to gap the core and that keeps the two pieced of core from hitting each other and making a sound.
Also: I have seen PWM on the edge of current limit/core saturation to get into a 1/2 frequency mode. (1/3 or 1/4) Say you should be running 50% but the supply gets into a strange mode where it does this: 4%, 95%, 5%, 95%..... (or 0%, 100%, 0%, 100%) This sub frequency can easy be in the audio range.

Not enough information. There are too many things that cause audio noise in a PWM.

 

[dr pepper]

Barrel type chokes as you pictured dont have a magnetic path like an E core, this makes them behave differently, and probably quieter.
The feedback system in a smps is usually compensated to prevent oscillations, and audible sounds from the tx.
A poorly designed smps, or one used in a manner that wasnt intended can make audible sounds.

 

[kubeek]

I can not see what you are doing. It is common for people that have never used a "EE" core to have no gap. This causes very high inductance and the core will probably saturate.

Could you elaborate a bit on gapped vs ungapped? I was under the impression that with a gap having lower inductance results in more turns in order to get the same flux as without the gap, so what am i missing?

 

[ronsimpson]

I think we agree.
The drum core from post #1 has a very large gap. It takes many turns to get inductance.
The EE core from post #1, with out a gap, has almost no gap and it takes a small number of turns to get the same inductance.
The EE core has the option to set the gap to what you want.

 

[gophert]

The problem may be magnetic saturation. What does the core manufacturer say about current and frequency? Are you within the range. Are you using correct wire diameter to minimize dc resistance?

What specific ferrite material are you using?

Also, the open side (can shaped/barrel shaped) cores do not have a closed return path for the magnetic circuit so the open magnetic field return path is not defined in space on your board. The fields can be several inches in size and overlap with other inductors to cause unexpected spikes and noise from the magnetic cross-talk. That is why almost every high amperage smps uses toroidal cores - usually iron powder cores for high current below 1MHz. Ferrite are generally used at switching speeds above 1MHz (be careful selecting mosfet at these speeds).