K33E T/S Parameters

[John Warren]

Someone asked for this information earlier.

This is what I got for the K33E using the mass-shifted Fs method. The values in ( ) are what,

I think, Klipsch provides.

Qms = 9.9 (8.5)

Qes = 0.790 (0.684)

Qts = 0.732 (0.633)

Cms = 388 um/N (305 um/N)

Vas = 368 L (343 L)

L at 1kHz = 1.46 mH

Voice coil DC resistance, Re = 3.4 Ohms (3.2 Ohms)

Resonance frequency, Fs = 35Hz (27 Hz) <<< seems my measurement is a bit high.

[JohnA]

How much of those differences are manufacturing tolerances?

[John Warren]

----------------

On 11/21/2003 9:24:57 PM John Albright wrote:

How much of those differences are manufacturing tolerances?

----------------

Part tolerances, measuring methods, measuring conditions all contribute to differences.

The Klipsch numbers I quoted above are from an unknown source.

[formica]

----------------

On 11/21/2003 9:53:34 PM John Warren wrote:

The Klipsch numbers I quoted above are from an unknown source.

----------------

The K33 T/S if I recall correctly were actually posted by a klipsch employee ... and I've enclosed the original file posted. The values seem to match those in brackets except (Re) which is actually closer to your measured one.

I have no experience with measured versus specified TS... but it would seem that the variances are great enough to noticeably affect the drivers performance in it's cabinet? ... but I haven't taken the time to model it. Maybe if I have some free time this weekend...

Later...

Rob

[John Warren]

Heres the procedure I use.  I suspend the driver from a hook on the ceiling about 3 ft from

any surface. The frequencies are read from a caibrated Fluke 1910-A counter. <?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

1. Measure voice coil resisitance (Re)

2. Measure resonant frequency (Fs)

3. Measure impedance at Fs (Zmax)

4. Calculate ratio of the DC resistance to maximum impedance:

Ro = Zmax / Re

5. Determine two frequencies at some impedance Z above and below the resonance frequency such that:

Z = Re * SQRT (Ro)

and, as a "check" Fs should = SQRT (Fl * Fh)

where Fl = below and Fh = above (I try to get the calculated Fs to within 1Hz).

6. Calculate the mechanical Q at resonance:

Qms = Fs * sqrt(Ro) / (Fh - Fl)

7. Calculate electrical Q at resonance:

Qes = Qms / (Ro - 1)

8. Calculate total Q at resonance:

Qts = Qms * Qes / (Qms + Qes)

(ref: Direct Radiator Loudspeaker System Analysis, R. H. Small, Trans. Audio and Electroacoustics,

vol. AU-19, pp. 269-81, Dec. 1971)

[Marvel]

John,

I emailed Eminence and asked for the T/S parameters for the K33-E, and got this repsonse back from them. Maybe they would send out bogus info, but that wouldn't sem to benefit them very much.

SPEC 15162

PART # K-33

RE OHMS 3.39 FS HZ 34.46

LE MH .96 MMS GMS 78.59

QM 7.39 CMS mm/N .2714

QE .410 RMS NS/M 2.3037

QT .390 VAS LTRS 301.66

XMAX MM 8.20 SD SCM 889.59

BL TM 11.88 EBP 84.4

EFF % 2.91 SPL dB 96.6

Wattage 150rms

Looks a bit different.

Marve

[John Warren]

I agree and I think I see the reason why.

Rob's 1989 data shows a Bl-product of 9.77 T*m and the data you provided is 11.88 T*m.

Qes is proportional to 1/(B*l)² so as B*l increases Qes decreases.

A change from 9.77 T*m to 11.88 T*m would decrease Qes by about 30%.

Assuming that the voice coil length did not change, that would mean that the flux density in the magnet is higher.

The woofer I measured here is a 1991 "vintage" sq. magnet K33E.

[WMcD]

I had noticed this too.

Early K33 are specified to have a Q around 0.3 and later ones are up at 0.7. At first I thought there was a measurement error creeping in, but apparently not. Is this the 33 versus the 33-E ?

This arguably makes for two different versions of the K-Horn, LS, and CW which use them. Further it could start to a suspicion of good versus not so good, or at least classic versus modern. Yikes.

Maybe the change in driver accompanied a change in the x over ?

For example, I suspected but couldn't really prove that a higher Q in the bass horn motor might increase the very low bass at the expense of the high end. Then a second order crossover could peak the high end. Pure guesswork on my part.

Any comments, guys?

Gil

[Marvel]

I know that the specs I got are for the newer version, as it has the 150watt RMS value, where earlier ones were only rated at 100 watts. Perhaps Dennis knows the date of changeover by Klipsch.

Marvel

[djk]

I am totally confused by the high Q figures.

If you take the low Q figures and plug them into a speaker program like BassBox (which I use), then it gives a Cornwall sized box as optimum. Same size box, same F3, same dB/W.

I wonder if the magnet is charged properly?

If a speaker coming off the line is a dud it is re-cycled.

The first thing done to it is de-mag so it doesn't grab all your tools. The speaker still has some residual magnetism at this point.

The cone, coil, etc, are removed and the frame is cleaned up.

The frame is put back in the line and receives a new cone/coil assembly.

The speaker is then re-mag'd.

If the residual field is oriented wrong it may buck the new charge, resulting in a partial or incomplete charge. I usually hit a re-charge twice, without moving the frame, to insure a complete charge.

The dud rate at Eminence is less than 5% so this may not be a good scenario.

Another Eminence OEM woofer I like is the 151269 made for Yamaha. Like the K33E it has an 8mm x-max with only a 56 oz magnet. It has a rubber surround and a heavier cone so the Fs=23.4hz, Vas=16.27, Qts=.34 with the SPL being about 3dB less than the K33E.

[John Warren]

Guys.

I goofed, I'm sorry. When I calculated the side band impedance (Z), I raised Ro to

the .4 power instead of .5. I ended up taking the Fh and Fl at the wrong impedance.

Here is a retest, this time with the right numbers.

Re = 3.39 Ohms

Zmax = 46 Ohms @ Fs = 35 Hz

Ro = 46/3.39 = 13.57

Z = Re*SQRT(Ro) = 3.39 * SQRT(13.57) = 3.39*3.683=12.5 Ohms

Fh=49Hz @ Z=12.5 Ohms

Fl=25Hz @ Z=12.5 Ohms

check >> SQRT (Fh*Fl) = SQRT (49*25) = 35Hz = Fs (good numbers)

Here are the T/S parameter calculations-

Qms = Fs * SQRT(Ro) / (Fh - Fl) = 35* SQRT(3.68) / (49-25) = 5.37

Qes = Qms / (Ro-1) = 5.37 / (13.57 - 1) = .427

Qts = Qms * Qes / (Qms + Qes) = (5.37 * .427) / (5.37 + .427) = .395

This driver is a 1981 driver. I know so because I bought it. The specs that Robs shows (1989) are way off relative to what I am measuring here. They are much closer to Marvels data.

[johnnyp]

John Warren: Thanks for sharing your measurements of your K-33. I've always wondered which set of T/S parameters were correct.

Gil & others: Is the general theory that in a given enclosure, the driver with higher "Q" (other parameters equal) will reproduce lower frequencies more efficiency?

[Marvel]

Dennis,

Do you know if Yamaha made their own on the JA3803 15" Woofer? It has a white cone of strange material.

Marvel

[djk]

"Dennis,

Do you know if Yamaha made their own on the JA3803 15" Woofer?"

Yeah, and it is a non-standard size too. Very hard to find re-cone supplies to fit.

"It has a white cone of strange material."

Very strange, it is paper. Yamaha claims spruce wood. Most woofers with so-called 'paper' cones are only about 20% actual wood fiber content.

Later Yamaha PA speakers use Eminence drivers.

[WMcD]

To answer the Q question from John . . . I'm not an expert on optimization of performance. But I'll give you the big picture.

Generally, speakers with a low Qes (Q of the electrical system) have strong motor systems. This shows up as high Bl product. B being the flux density (magnet strength) and l being the length of wire in the gap. As a result they have high electrical effecencies. A given driving voltage produces a greater amount of force on the diaphragm. There can be high SPL at the mid and upper bass where the diaphragm is a relatively good radiator.

However, this strong motor is also an efficent generator. If you have any sort of motor, if you drive it mechanically, electricity comes out the electrical terminals. There is sort of reciprocity. For example, in many cheap intercoms or hand held radios, the speaker is used as a microphone.

Going back to the generator aspect, if you connect a strong generator to a dead short, it will be more difficult to crank than a weak generator. Going another step, if you short out a speaker's terminals, it is more difficult to physically move the diaphragm by hand, or when driven by its own mass and spring system at resonance.

The result is that speakers with low Qes are overdamped. That means that when the diaphragm tries to move on its own near resonance, the motor system is now a generator. The amp is a short circuit to some extent, and it is hard for the diaphragm to crank the generator. The result is that any resonant movement is reduced, or damped out.

Now you must understand that at least a part of bass response is to exploit the tendancy of the diaphragm to resonate at fs (its resonant frequency). As you go down the frequency spectrum toward fs, the diaphragm is less big in terms of wavelength and a poorer and poorer radiator. However, getting closer to resonance, the amount of excursion is greater. So this is a type of built in equalization.

(If I may be my pedantic self. The above is an important thought experiment. You may have, for example, a 15 inch woofer. A fixed size. However, look at it in terms of the wavelenght of sound being reproduced. As frequency goes down, wavelenght goes up. The 15 inch woofer diaphragm is getting smaller. The equalization is that it is easier to move as we get closer to resonance.)

So now with a low Qes driver, the motor/generator is damping out the resonance. The generator is hard to crank. Therefore there is less bass response than if there was a weak Bl. A weak Bl would create a higher Qes.

Again, sliding down the frequency curve, below the resonance fs, the diaphragm is not moving very much (it is hard to move) and bass out put is really going to heck. Generally, if you look at the bass response of a system, below the resonant frequency, there is a very fast roll off of bass.

You've heard the effect of underdamping. This shows up in cars with high Qes speakers. They are not damped much at resonance, and you get throbbing one-note bass which can be heard at a distance. These are, to some extent, efficent, but only at that note.

There is a lot more to it that what I've explained, you have to look at the Q of the speaker and the box. None the less, to some extent, electrical efficency and extended bass response in a box can't both be accomplished. This is why you can't take a driver with a very, very low Qes (electrically efficent), and get very low bass response at the same time.

Best,

Gil

[Erik Mandaville]

From a few years into the future from the date of this thread:

Thanks for this helpful information.

Erik