Inductor saturation and capacitor differences

[Dave Bullet]

What is the most basic measurement setup required to measure:
a) Saturation point of an inductor
b) Difference in linear or non-linear distortion caused by different capacitor qualities and types

Reason. For my 3 way I want to "cheap out" on large value inductors and capacitors. This means looking at series NP electrolytic capacitors and iron core inductors.

My ears cannot tell the difference between iron core and air core inductors. I was hoping to put some science / objective measurement behind my design choices and was wondering how fancy a measurement setup I might need (or get away with).

Thanks,
David.

[gedlee]

This test could be done with a very simple setup.

An RC or RL circuit ( actually CR and LR, reactive element first) is created and a voltage is applied. A sine wave would be best. Then on a spectrum analyzer you would simply look at the spectrum of the voltage seen across the resistor. Note the harmonics. Spectrum analyzers for sound cards are readily available on the web. You may not get down to .01% resolution, but you should be able to get a good idea.

I have long wanted to do this myself, just don't have the time.

But remember that the NP capacitor will probably show its biggest problem at low voltages. If you see high harmonics at low voltages then you have a very bad situation.

Let me know how this goes. If you have troble I can help.

[Dave Bullet]

Thanks Earl,

I suppose I have to decide on what level of distortion is perceptible. I know his varies with frequency and SPL and various studies have had different results. I remember Jay Butterman published a table in the SW manual which was quite useful.

I suppose something like RMAA might be useful.

I was crudely going to apply a crossover topology to a driver with NP electrolytics then with metal film / foil types and do a "delta" on the fundamental as well as HD spectrum to see variation. The problem is I cannot test for all types of distortion.

Your approach is better and takes away more variables

Cheers,
David.

[xiphmont]

Originally posted by Dave Bullet
Thanks Earl,

I suppose I have to decide on what level of distortion is perceptible. I know his varies with frequency and SPL and various studies have had different results. I remember Jay Butterman published a table in the SW manual which was quite useful.

I suppose something like RMAA might be useful.

I was crudely going to apply a crossover topology to a driver with NP electrolytics then with metal film / foil types and do a "delta" on the fundamental as well as HD spectrum to see variation. The problem is I cannot test for all types of distortion.

Your approach is better and takes away more variables

Cheers,
David.

There are two problems with objectively measuting audible distortion levels-- the levels are not the same for all people, and the line is not hard and fast. There's generally a squishy grey area where distortion is not strictly audible per se, but in an ABX test you begin to see reproducible, statistically significant results where a person can tell the difference, but isn't sure why. So the line of acceptability is more a slippery slope than a line. For both these reasons, designers tend to build in a very large margin of error.

I'm also interested in the distortion effects of inductors, although I'm mlooking at it from the other side: Very large value micro-signal inductors in the input path, where inpedence transformation means you're regularly looking at inductors packing a full Henry or more. And that means super-high passivity ferrite or mumetal cores (u0 of 10,000-25,000)

I've got test equipment nearly set up, I'd love to contribute to the testing.

[Dave Bullet]

Hi xiphmont,

I'm hoping (not a very good way to start a scientific experiment, I know) that the tolerance in components might actually be more significant in altering the sound than the material the component is made out of.

But in measuring distortion of a pure sine wave, then the actual uF or mH value differences between components should not matter, it is their relative introduction of distortion that counts.

What sort of distortion is best to measure? Or do you really need all types to have it covered?

I will post some measurements when I next get "permission" for more speaker related work.

Thanks,
David.

[xiphmont]

Originally posted by Dave Bullet
Hi xiphmont,

I'm hoping (not a very good way to start a scientific experiment, I know) that the tolerance in components might actually be more significant in altering the sound than the material the component is made out of.

I doubt individual variation will affect the type or scope of distortion much. The variation in components usually has to do with rated value. The type and amount of distortion more to do with materials and construction. Eg, I'd expect similar hysteresis in all inductors using the same core material, even if exact inductance value varies 5%.

Originally posted by Dave Bullet
But in measuring distortion of a pure sine wave, then the actual uF or mH value differences between components should not matter, it is their relative introduction of distortion that counts.

What sort of distortion is best to measure? Or do you really need all types to have it covered?

I will post some measurements when I next get "permission" for more speaker related work.[/B]

Well, there's the 'intentional' distortion introduced by the very nature of the filter (that would be there even with ideal components) and then there's the distortion and noise introduced because the components are nonideal.

The ideal distortion can be measured with just about any kind of input; I regularly get my phase/responts plots using pink noise or musical input rather than using pure tones or sweeps.

As far as I know, though, THD measurements have to use either pure tones, pure tone sweeps, or narrowband noise... because you're getting your analysis from what is happening in the 'unused' parts of the spectrum.

[gedlee]

Originally posted by Dave Bullet
Thanks Earl,

I suppose I have to decide on what level of distortion is perceptible. I know his varies with frequency and SPL and various studies have had different results.

I was crudely going to apply a crossover topology to a driver with NP electrolytics then with metal film / foil types and do a "delta" on the fundamental as well as HD spectrum to see variation. The problem is I cannot test for all types of distortion.

Your approach is better and takes away more variables

Cheers,
David.

The higher the order of the harmonic and the lower in level it occurs the more audible it is. This is why I suggested testing the caps at low levels. I doubt that you will find any nonlinearities in the inductors at lower levels - hence the inductor saturation is not likely to be an issue. But poor caps will certainly have zero voltage crossover effects, especially electrolytics.

A complete study would look at several frequencies and several signal levels.

I don't think that you would see much in the test if you used the drivers as they will tend to swamp the distortion. A differencial test will most likely just be looking at test to test variability.

[cabbagerat]

Originally posted by Dave Bullet
Reason. For my 3 way I want to "cheap out" on large value inductors and capacitors. This means looking at series NP electrolytic capacitors and iron core inductors.

Caps I can certainly understand, but air core inductors can be had for cheap.

When I built my first set of speakers, I made my own rig using an electric drill to wind inductors. I then wound 14ga magnet wire around a wood core wrapped in PTFE tape. I then removed the wood core (for re-use) and tied the inductor together with zip ties. This way I got nice air core inductors for a fraction of the price of commercial suppliers.

There are a couple of references to the required equations (and online calculators) around the net. The best thing to do is to wind according to the equations, then measure and add or remove turns to get exactly the right value. With a good measuring setup, you can get within 1% like this.

That said, I am interested to see the outcome of any distortion measurements you do.

[bzfcocon]

Originally posted by cabbagerat
Caps I can certainly understand, but air core inductors can be had for cheap.

When I built my first set of speakers, I made my own rig using an electric drill to wind inductors. I then wound 14ga magnet wire around a wood core wrapped in PTFE tape. I then removed the wood core (for re-use) and tied the inductor together with zip ties. This way I got nice air core inductors for a fraction of the price of commercial suppliers.

Still, a say 8mH, heavy gauge air core inductor will be pretty huge. If also a low DC resistance is required, iron(or other magnetic materials) core inductors might be the better option. Thus, measuring distortions due to saturation is an interesting experiment.