Scan-Speak 6600 Le

[DSP_Geek]

I might have found something with the 6600: the voice coil inductance (Le) specification seems way off. The sheet has it at 0.03 mH, but actual measurements on my end, along with reading the impedance curve on the sheet, seem to indicate Le is closer to 0.01 mH.

This has obvious implications for crossovers: building a Zobel with R=3.3 ohms and C=3 uF, as theory would have it from the spec sheet parameters, drops the impedance down to 2 ohms at 16 kHz. Replacing it by a Zobel with R=3.3 ohms and C = 1 uF seems to keep the impedance at a constant 3.3 ohms across the passband.

Has anyone else noticed this?

[dlr]

Originally posted by DSP_Geek
I might have found something with the 6600: the voice coil inductance (Le) specification seems way off. The sheet has it at 0.03 mH, but actual measurements on my end, along with reading the impedance curve on the sheet, seem to indicate Le is closer to 0.01 mH.

This has obvious implications for crossovers: building a Zobel with R=3.3 ohms and C=3 uF, as theory would have it from the spec sheet parameters, drops the impedance down to 2 ohms at 16 kHz. Replacing it by a Zobel with R=3.3 ohms and C = 1 uF seems to keep the impedance at a constant 3.3 ohms across the passband.

Has anyone else noticed this?

I'd be cautious about comparing your measurement with theirs. For one, Le isn't constant in pretty much any dynamic driver, including tweeters. I don't know at what frequency it's referenced, but the motor non-linearity that creates the rising impedance makes Le in a tweeter generally not important for most usage.

In addition, in a tweeter it's not much of an issue insofar as the crossover work. The Le usually isn't a concern and if a zobel is used, a manual calculation is a start, but in the CAD software with actual impedance measurements, tweaking of the zobel is a simple matter. In cases where I've used a tweeter zobel I doubt that any have been a "classic" RC combo. I usually only use it for the case where I want to add a small rolloff to the tweeter's top end.

Dave

[Svante]

Agree, Le is not an ordinary inductor. The inductance varies with frequency. This can be modelled by a network of more than one inductance and resistor, sometimes simulation softwares and datasheets allow for entry of these.

Another way of looking at it is like Leach did, where the impedance is proportional to w^n, where n is typically 0.6-0.7. An ideal inductor has n=1. Have a look in the tech docs and user's guide of Basta! in my signature.

So, it is not strange that the oversimplification of assuming one value for the inductance goes wrong.

[DSP_Geek]

Yeah, I gathered that Le wasn't a true inductor. I would expect it to be somewhat closer than previously mentioned, however. For the inductance to be off by a factor of 3 with an w^.65 factor means I'd be measuring about 5.5 times away from the nominal frequency for Le tests, which for most tweeters is about 10 kHz. Since I was measuring at 16 kHz I'd expect the number to be low by 25%, not 65%.

DLR: the equivalent inductance would mess up my circuit, unfortunately, as I have a fairly complex network around the tweeter.

[dlr]

Originally posted by DSP_Geek

DLR: the equivalent inductance would mess up my circuit, unfortunately, as I have a fairly complex network around the tweeter.

I don't follow that. The inductance of a driver is what it is. The crossover has to be designed to interact with the actual driver impedance and acoustic response.

It's also not necessary to perfectly flatten the driver impedance for most situations. A good zobel will turn down the rising response to an approximately flat response above some frequency. There will be something of a plateau or the rising response can simply be tailored, a lot or a little.

But as I use them, it is not for flattening the impedance, I usually use them to tailor the acoustic response when tweaking by ear.

Dave