Iron-core vs. air-core inductor distortion in speaker crossovers.

[MAB]

There have been some posts recently claiming excessive distortion when using iron-core inductors in speaker crossovers. As with many audio-claims, this legend has a kernel of truth; iron-core inductors do have magnetic and electrical properties that play a significant role at high frequencies. Fortunately for audio, these phenomena typically become a problem at frequencies well-above 100 kHz.

It doesn't help that manufactures claim the same and try to disallow the use of inexpensive components. Nobody ever includes a relevant measurement, or the measurements are of some secondary aspect not related to claimed distortion, or they botch the measurements outright. Instead we have anecdotes about iron-core inductors causing all sorts of bass issues, or that they should never be used in with a midrange or tweeter. Yet some of the world's great speakers use iron-core inductors. Let's see.

Here is a measurement of inductors used in a 6dB/oct filter on a B&C midrange in a sealed cabinet. I have some cheap bobbin-type inductors typically used in basic speaker crossovers, and compare them to air core inductors. For example, two 1.5 mH inductors:

Here is the response in a sealed enclosure with no filter, and with a filter made from each of the inductors:

The small DCR difference between the two inductors causes a slight mismatch, but at high frequency the response of the filter is matched between the two different inductors.

How about distortion? THD is matched:

2nd HD is matched:

3rd HD is matched:

All of the HD components between the two inductors, for instance 5th HD:

Trying another pair of 3.5 mH inductors with similar DCR:

The frequency response of the two filters using the 3.5 mH inductors is more closely matched with similar DCR of ~1 Ohm.

How about distortion?

Matched, 3.5 mH iron-core is no different than equivalent air-core.

Individual HD components for 3.5mH inductors follow.

Iron-core inductors in general are fine for speaker crossovers. In fact, for implementations like some notch filters they have advantages over air-core. However, it's super easy to show that iron-core inductors do not generate distortion as is often claimed. Iron-core inductors do saturate, but measuring the impact of that while dealing with a driver operating at high volume is actually a difficult measurement since the driver tends to dominate. Measuring the supposed effect of hysteresis and eddy-current losses in an inductor used in crossover is very straightforward, and shows that the claims and anecdotes about distortion in iron-core inductors in general are wrong.

edit: I accidentally posted this thread before completion. Added all of the data, and cleaned up some language and typos.

 

[a4eaudio]

There have been a raft of posts recently claiming excessive distortion when using iron-core inductors in speaker crossovers. As with many audio-claims, this legend has a kernel of truth; iron-core inductors do have magnetic and electrical properties that play a significant role at high frequencies. Fortunately for audio, these phenomena typically become a problem at freq

This audio-claim is an old one, and as many of these legends is based on a kernel truth, that iron-core inductors have losses due to the magnetic and electrical properties of the core. In my experience these effects only become significant well-above 100 kHz.

Thank you for this post!!

I have seen lots of "experiments" that show different effects of inductors placed close together, iron-core saturation, etc. from an electrical perspective. I have never seen a single experiment that tested the audibility of such things in a blind-listening test. (I do know of one "test" that only compared a very cheap buy-out inductor that Parts Express sold (this was like 10 years ago) and supposedly it was quite obviously not performing as expected.)

 

[MAB]

Thank you for this post!!

I have seen lots of "experiments" that show different effects of inductors placed close together, iron-core saturation, etc. from an electrical perspective. I have never seen a single experiment that tested the audibility of such things in a blind-listening test. (I do know of one "test" that only compared a very cheap buy-out inductor that Parts Express sold (this was like 10 years ago) and supposedly it was quite obviously not performing as expected.)

I inadvertently posted the thread before I finished it. It now has data and conclusions.

 

[terryforsythe]

There have been a raft of posts recently claiming excessive distortion when using iron-core inductors in speaker crossovers. As with many audio-claims, this legend has a kernel of truth; iron-core inductors do have magnetic and electrical properties that play a significant role at high frequencies. Fortunately for audio, these phenomena typically become a problem at freq

This audio-claim is an old one, and as many of these legends is based on a kernel truth, that iron-core inductors have losses due to the magnetic and electrical properties of the core. In my experience these effects only become significant well-above 100 kHz.

The issue is a dearth of test data for power losses of inductor cores used in crossovers. I have been contemplating running some tests on inductors used for audio, but power analyzers are not cheap and I am now retired. I may pickup a used one and get it calibrated.

Long ago, early in my career, I was tasked with figuring out why inductors were overheating in powerline filters running at 60Hz and 400Hz. They used laminated silicon steel C-core inductors with 12 mil. (0.3mm) and 4 mil. (0.1mm) lamination thicknesses, respectively. I ran numerous tests to isolate eddy current power losses from hysterises losses, and from that data derived constants for use in equations to accurately compute the power losses. I don't have a copy of the report I generated as it was considered by the company to be a trade secret. But, I can say this, the inductor core power losses were significantly higher than those previously computed. Specifically, the hysterises losses were much higher than expected. Previous assumptions that thinner laminations reduced eddy current losses, and that power losses increased significantly with frequency, were confirmed to be correct.

In audio, laminated steel inductors commonly are used. The cost effective cores typically used in mass produced speakers have thicker laminations (maybe somewhere around 0.4mm - 1.0mm, depending on the crossover manufacturer) of what I suspect is cold rolled steel. The more expensive ones use thinner laminations of silicon steel, resulting in lower power losses. Air core inductors have no core, and thus don't have any core losses.

At one point when I was in audio engineering, we conducted a blind test comparing inductor cores. There were three of us, and all of us clearly heard an audible difference in the upper midrange. The difference was not subtle with that inductor value in that two-way speaker.

An issue that I commonly see is that people see test results of capacitor comparisons, and just assume that the same results apply to inductors. That is a falacy. Inductor hysterises has a much more significant impact on power losses and time domain response than capacitor equivalent series resistance (ESR) and dielectric absorption (DA).

Another issue I commonly see is that people see a comparison showing a particular result, then assume that result applies to all similar comparisons. That is not always the case. Erin's Audio Corner did a video on that regarding amplifiers.

I spent a good portion of my career designing filters, including designing the inductors and capacitors used in those filters. I am happy to never again use passive filters in my speakers, primarily to avoid use of inductors in my crossovers.

 

[voodooless]

So how about at elevated power levels?

 

[MAB]

So how about at elevated power levels?

I don't listen too often above 110 dB.

Yes saturation can be an issue. I see no evidence of it in this test. I didn't heat up the inductor either. I leave super-high power to someone else.

 

[terryforsythe]

For example, two 1.5 mH inductors:

I suspect the cored inductor probably has an iron powder or ferrite core. It definitely is not laminated steel, which oftentimes is used in mass produced speakers. Iron powder and ferrite cores typically have lower power losses than laminated steel cores.

 

[MAB]

An issue that I commonly see is that people see test results of capacitor comparisons, and just assume that the same results apply to inductors. That is a falacy. Inductor hysterises has a much more significant impact on power losses and time domain response that capacitor equivalent series resistance (ESR) and dielectric absorption (DA).

The inductor's hysteresis is indeed a large effect, but only at really high frequencies, well outside of audio.

 

[terryforsythe]

The inductor's hysteresis is indeed a large effect, but only at really high frequencies, well outside of audio.

Yes it has a large effect, not limited to frequencies well outside of audio. If that were the case, I never would have been tasked at figuring out why inductors operating at 60Hz, and especially at 400Hz, were overheating.

 

[MAB]

I suspect the cored inductor probably has an iron powder core. It definitely is not laminated steel, which oftentimes is used in mass produced speakers.

Uh... These bobbin inductors are the cheapest and suffer hysteresis and especially eddy-current losses more than laminated. Laminated are also more expensive. They came from mass-produced speakers.

I chose the bobbins because they are the most common especially in mass-produced speakers, and have the worst performance (albeit at frequencies above human hearing). The laminate will perform better, but you would need to test at 500 kHz or higher.

 

[Palmspar]

But when a speaker has the same freq curve i dont think it must be sound also the same.
My Revelator/Hypex Diy sub and the SVS SB2000 have also almost the same freq curve, but the Diy sound way better.
But its a nice test!

 

[MAB]

Yes it has a large effect, not limited to frequencies well outside of audio. If that were the case, I never would have been tasked at figuring out why inductors operating at 60Hz, and especially at 400Hz, were overheating.

Please demonstrate then. You are now mixing phenomena.

 

[Nullgator]

I'd also be curious to see data for "signal smearing" or phase shift which is often cited as a reason to use air core inductors. You'd expect this to show up in the distortion measurements of course but I think some people would be but at ease by measurements with an oscilloscope.

 

[MAB]

But when a speaker has the same freq curve i dont think it must be sound also the same.

Oh my, you need help!!!
Please run out and get a copy of Toole's book ASAP!
It's covered in the first chapter!

 

[MAB]

I'd also be curious to see data for "signal smearing" or phase shift which is often cited as a reason to use air core inductors. You'd expect this to show up in the distortion measurements of course but I think some people would be but at ease by measurements with an oscilloscope.

You mean the phase. Like this:

For this comparison, I need to make sure the DCR is matched since that will play a large role compared to the negligible role the inductor plays:

There is no difference in phase. They are line on line matched between air- and iron-core inductor. I was careful to keep mic and speaker position the same run to run. Even a few mm would affect the outcome. Small changes in DCR are way more impactful on the phase, for example the two 1.5 mH inductors have DCR differences that lead to small phase differences:

I hope you appreciate, this time-smearing argument about iron-core inductors is marketing people having their way with customers. It's total BS. A few mm in placement is larger effect, a few mOhm in DCR is larger effect.

 

[Palmspar]

Oh my, you need help!!!
Please run out and get a copy of Toole's book ASAP!
It's covered in the first chapter!

Ha Ha but let say the diference for subwoofers, one with a 13 inch Revelator driver and another with the Peerless stw-350 both 1000w Hypex amps
They cost also the same and its easy with dsp to give them the same freq responce, do you realy think with music or movies they sound the same?
They can have the same freq responce with a sweep, but they will sound night and day different for music/movies.

 

[voodooless]

There is no difference in phase.

Of course there is not. It’s a minimum phase system, so if response is the same, so is phase.

almost the same freq curve

“Almost” is the operative word here… and peeking

 

[Palmspar]

Of course there is not. It’s a minimum phase system, so if response is the same, so is phase.


“Almost” is the operative word here… and peeking

White is the SB 2000 blue line the Revelator Diy
Mic was 1 cm from the drivers.

 

[MAB]

Ha Ha but let say the diference for subwoofers, one with a 13 inch Revelator driver and another with the Peerless stw-350 both 1000w Hypex amps
They cost also the same and its easy with dsp to give them the same freq responce, do you realy think with music or movies they sound the same?
They can have the same freq responce with a sweep, but they will sound night and day different for music/movies.

Please, this thread is about distortion in crossover inductors.

Don't drag us back to the beginning of time and discussions about measurements and audibility.
Do get Toole's book. It addresses your comment from the first paragraph.

 

[Vladimir Filevski]

Here is a measurement of inductors used in a 6dB/oct filter on a B&C midrange in a sealed cabinet. I have some cheap bobbin-type inductors typically used in basic speaker crossovers, and compare them to air core inductors.

Thanks for the measurements.
Max SPL is over 105 dB. What was the measuring distance mic-driver? What was the voltage of the measuring signal?