Coil resistance increases from 54 ohms to 8k at 220kHz, why?

[jrive]

edit: The key thing I am after is why the ferrite might be responsible for the increase in resistance I measure on the coil.

Hello,

i have a coil designed to be 2.2mH and54 ohms. When an MN60 ferrite (T-ish core) is used with it, the inductance rises to 6.1mH as expected, and the resistance remains the same up to about 10kHz. Above 10kHz it begins to rise, and at the desired operating frequency, it rises to 8Kohms! (measured with LRC meter). The resistance of the coil by itself (without the ferrite) rises slightly, but is still below 100ohms at 220kHz.

So, we surmised the issue has to do with the losses in the ferrite, but we cannot pinpoint exactly what. We simulated this in Flux 3D, including the ferrite per the material datasheet, and the simulation does show a rise in ac resistance (not reactance) due to proximity effect as expected, but it is only a hundred ohms or so (clearly, we modeled some o the parameters but not the "right" one to demonstrate the behavior we see with the real part).

Does anyone have any experience with this that can shed some light as to what may be happening? Or can explain some basics physics of ferrite materials and how the losses manifest themselves (is this what I'm seeing?).

Also, I just started looking into the Loss Factor, since I noticed from the datasheet that it goes up significantly for this material at 300Khz (it goes from 12 at 100kHz to 55 at 300kHz), but I am having trouble understanding what this really means. Any help here is also appreciated..

Thank you
Jorge

 

[sophiecentaur]

i have a coil designed to be 2.2mH and54 ohms.

Silly question but is that 54 ohms in series? Does your RLC meter tell you series or parallel values?
Play around with http://www.daycounter.com/Calculators/Parallel-Series-Imedance-Conversion-Calculator.phtml and you may find an answer. The equivalent parallel resistance is a lot higher if you convert your original coil values.

 

[jrive]

Silly question but is that 54 ohms in series? Does your RLC meter tell you series or parallel values?
Play around with http://www.daycounter.com/Calculators/Parallel-Series-Imedance-Conversion-Calculator.phtml and you may find an answer. The equivalent parallel resistance is a lot higher if you convert your original coil values.

Yes., that's 54ohms in series (Ls-Rs setting)...

 

[nasu]

The inductive reactance of the coil with inductance of 6.1 mH, at 220 KHz will be
## X_L=\omega L \approx 8.4 k\Omega ##
So your measurement seem to be OK. Why do you think you have a problem?
The effect of the ferrite is to increase the inductance of the coil.
The resistance of the coil may increase due to skin effect.

 

[jrive]

Thanks

The inductive reactance of the coil with inductance of 6.1 mH, at 220 KHz will be
## X_L=\omega L \approx 8.4 k\Omega ##
So your measurement seem to be OK. Why do you think you have a problem?
The effect of the ferrite is to increase the inductance of the coil.
The resistance of the coil may increase due to skin effect.

Thanks for your response. The resistance (real part) is measuring 8.4kohms, not the reactance. The meter measures Ls-Rs (L and R in series model). The rise in R is kills the Q that I need (and have at lower frequencies).
Yes, I believe the resistance is increasing due to the skin and proximity effects --but it is increasing much more so than I expected. So I was wondering what aspect of the core could be causing this. It is much more than just the resistivity and the permeability, because with those in my FLux 3D simulation, I don't get anywhere near the 8kohms that I'm measuring.

 

[sophiecentaur]

Yes., that's 54ohms in series (Ls-Rs setting)...

OK. Only, if you do a series parallel conversion, the R_p is well over 1k at your frequency, which is in the direction of your problem.

Thanks

Thanks for your response. The resistance (real part) is measuring 8.4kohms, not the reactance. The meter measures Ls-Rs (L and R in series model). The rise in R is kills the Q that I need (and have at lower frequencies).
Yes, I believe the resistance is increasing due to the skin and proximity effects --but it is increasing much more so than I expected. So I was wondering what aspect of the core could be causing this. It is much more than just the resistivity and the permeability, because with those in my FLux 3D simulation, I don't get anywhere near the 8kohms that I'm measuring.

Have you done the series parallel conversion? Do you know what I am talking about? You have to eliminate that before looking for losses in the core. If there are losses then you can find out from the spec sheet.

 

[nasu]

Thanks

Thanks for your response. The resistance (real part) is measuring 8.4kohms, not the reactance. The meter measures Ls-Rs (L and R in series model). The rise in R is kills the Q that I need (and have at lower frequencies).
Yes, I believe the resistance is increasing due to the skin and proximity effects --but it is increasing much more so than I expected. So I was wondering what aspect of the core could be causing this. It is much more than just the resistivity and the permeability, because with those in my FLux 3D simulation, I don't get anywhere near the 8kohms that I'm measuring.

Can you indicate the name of the instrument you use for the measurement? I am not familiar with this type of instrument.

 

[jrive]

Can you indicate the name of the instrument you use for the measurement? I am not familiar with this type of instrument.

The RLC meter? It's a Keysight e4980a Precision LCR meter, 20-2MHz range.
If it is Flux 3D, it is a 3 dimensional finite element analysis tool to model and simulate electromagnetics. Similar to Ansys.

 

[sophiecentaur]

Self resonance can produce very high impedance effects. Can you measure what itself resonant frequency is?

 

[jrive]

OK. Only, if you do a series parallel conversion, the R_p is well over 1k at your frequency, which is in the direction of your problem.

Have you done the series parallel conversion? Do you know what I am talking about? You have to eliminate that before looking for losses in the core. If there are losses then you can find out from the spec sheet.

Thanks for engaging in the discussion...I find it useful to discuss these things since almost always one of these questions leads to an answer (or at least deeper understanding on my part!).
I don't think I need to do the Rs to Rp conversion. The meter is giving me Rs, and that is indeed how it is modeled in my circuit. But, what do you mean by "you have to eliminate that before looking for losses in the core?"
If I measure the inductor without the ferrite, the measured Rs is in the 54ohm range. Same measurement, but with the coil "in" the ferrite core (it surrounds the coil and the center portion of ferrtie goes through the center of coil), the series resistance measured by LRC is 8k-ish.

 

[jrive]

Self resonance can produce very high impedance effects. Can you measure what itself resonant frequency is?

I have an image I can share, but don't know how (the image icon wants a URL, so I can't navigate to my folder to get image)...It shows the Ls-Q of the inductor with and without the ferrite...the self resonance of the coils is around 840kHz.

 

[sophiecentaur]

But, what do you mean by "you have to eliminate that before looking for losses in the core?"

I was anxious for you to accept the possibility that you could be measuring the parallel L and R and not the series L and R because the R_p is significantly higher. We never know the level of competence of people who post questions. Hence it is all too easy to 'insult' someone by talking down. This series / parallel thing with components is very important to get right. Sorry if you already knew all this but there is less potential harm in telling someone what they already know than not telling them something they didn't know. Self resonating coils can measure as nothing like what's written on the side. The parallel parasitic C can give you a very high impedance.
`PS if you want to post an image then clicking the 'UPLOAD' button bottom right of your posting window and select the file with your GUI as normal.

 

[jrive]

No, @sophicentaur...I completely understand and appreciate all the questions and inputs.-- thank you.
Here are two of the plots I generated with data from the LRC meter, same coil, with and without the ferrite. (Ls-Rs, Ls-Q). I'm also including the coil and the ferrite for reference...

 

[jrive]

I think the issue is with the material of the core. I did a quick calculation for the Resr using the following relation: $$\frac{u''}{u'}=\frac{Resr}{2*\pi*f*L}$$ using data from the MN60 ferrite material datasheet, u''=1800, u'=6000, f=200kHz, and the corresponding L as shown in the Ls-Rs plot @ 200kHz (previous post), and got around 1.4kohms which is in the same order of magnitude as what the LRC is meausring.
I then used the data from an N48 ferrite material: u''=35, u'=2500, and the Resr is around 68ohms which is reasonable. Next step is to see if I can have someone fabricate some of these same ferrites in this material to prove out the theory.

A plot of the complex permeability of the MN60 material is attached, as is the one for the N48. Notice how the inductance is flat and the losses (u'') much lower at 200kHz than the MN60 material. It is almost I'm using the MN60 material above the frequency range for which it is intended. I think this is the reason for what I see...