[rythmik]

brucek wrote:

Basically the series resistor modifies the output impedance of a solid state amplifier and changes its reasonable approximation of an ideal voltage source into a modified current source. This mod (if my memory serves), was actually incorporated into some Sunfire amps years ago by Bob Carver in his attempt to allow switching of his solid state amps into a tube sound. If anyone wants their system to enjoy the same distortion and high output impedance of a single-ended triode, install a one or two ohm power resistor on the output of your amplifier.

Since you know the transfer function experiments, it is very easy to communicate the ideas with you as you understand more than just basic circuit theory. Most of stuff I will say next has been posted at the other forum.

When I first got into this hobby, I had thought low Qts, that is, higher (BL)^2/Re is a universally better driver, even though there are conventional wisdom everywhere that would tell you a low Qts drivers are better for vents and Mid Qts drivers are better for sealed. My rationale of favoring a low Qts driver is simple (probably very similar to your concern right now), that is the speaker is basically an ideal motor with a serial voice coil resistor. When everything connected together, we essentially see a lumped resistance of amp resistance, speaker cable resistance, plus the voice coil resistance, driving a back EMF which equals to B*l*v, where B is the average magnetic strength, l is the voice coil wire length, and v is the cone velocity. This lumped resistor is the barrier for us to completely control this back EMF, which completely control the cone velocity. It is like flying a kite. If the lumped resistance goes to superconductor, the amp output directly drives the back EMF which means we would have completely control of the cone. Therefore my rationale is low Qts driver (which means BL is very high) is a better driver (which is in line this quest of always higher (BL)^2/Re). But when I tested actual subs, I also observed one thing, at resonance frequency of a sealed box, a low Qts driver actually produces less output. This can be explained by B*l*v because at the resonance peak, almost all voltage from amp output goes to back EMF, or B*l*v. However, higher B*l leads to lower v, this is what people refer to as stall effect. And that has completely changed my mind about low Qts. Even though low Qts gives us lower distortion, but the price to pay is very high. A system with Q=0.4 actually needs 4 times of power (2x the voltage) to drive the driver to the same output of a Q=0.8. Let me also give you another actual example that NCA told me. One time their customer used a very large motor to design a driver, after it is done, it was found the Qts is too low, not good for sealed box sub. So this customer widened the gap to weaken the motor. Poor customers !!! They could have paid less if the solution had been keeping the gap the same, but used a smaller magnet. But if the magnet becomes smaller, they may not able to sell as many of them because everyone wants to see big motors.. Should the customers to be blamed for this insanity? Now this story also indicates that, unlike most other engineering optimization problem, there is no so called optimal solution for Qts selection. It all depends on application. For the same reason, there is no optimal (BL)^2/Re. The added resistor has better tempco than copper and that means lower thermal compression.

I don't think it's arguable that a large portion of the amplifiers power is lost in the series resistor. This power loss alone would be unacceptable by most people.
The resistor would need to withstand considerable power when the driver was not in its resonance curve. For safety sake, probably a 200% over spec would be prudent. I would think a resistor of this rating would be wire wound and subject to inductance of its own. That wouldn't be good.

Just think of this resistor as part of the voice coil resistance. Actually it may be better because the heat can be dissipated in open air, instead of in that small gap.

I would also have a concern over the drivers damping. A standard solid state amplifier with a typical output impedance of 0.01 ohms driving a 4 ohm load would create a damping factor of ~400 (ignoring wire impedance). The series resistor (of 4 ohms, for example) would create a modified output impedance of slightly greater than ~4 ohms, and would create a damping factor of slightly less than 1. What would stop the cone movement? It has no low impedance path to ground. I can imagine the bass would sound muddy as a result.

Damping factor is a notion of the past. If you look at the lumped resistance I mentioned above, which one is the major contributor? It is not the amplifier output resistance. It is the voice coil resistance itself which has a temp. coefficient of 0.3%/Celsius. 20 degrees C will give you a variation of say 3ohms x 0.3% x 20=0.18ohms. As you can imagine that temperature can get a lot higher. So in the big pic, the 0.1 ohms amp output resistance is not significant at all. If we look further, there are production unit to unit variations, a 10% reduction of BL value is equivalent to 19% increase of Re value. That would have even more impact. Not to mention we would have learnt from history that exact ideal transfer function does not give us the best sound either. I am wondering if you still remember shortly after Bob Carver's transfer function work, there were a frenzy of ideas derived from that, such as Hafler which has a tweaking control to exactly match the amp out to amp input under load. That idea does not go anywhere. The transfer function is useful only to "copy" the sound. It tells nothing how to come up with the first "best" sound. So we still have no clue which contributes good sound or not. Also Bob Carver's transfer function will not work for speakers either. It is funny because speaker has the least number of components, but each one of them is so difficult to model.

BTW, I still like the low Qts and that is how we implement our servo subwoofer. We use a driver with proper Q that will give us best physical output (without stalling), which normally means the driver that will achieve Q=0.8 when not equalized. And then use servo to get us the lowest Qts value we can possibly achieve so-called "fastest" sound possible. That is how we get the best of both worlds without being pull on both ends.


Jman,

BTW, the exact cause of bottoming is not as simple as what I have described. A lot of that also has to do with the spider roll thickness and surround width. In other words, compliance of a speaker is one of the least linear components and linear compliance normally work against us when it comes to bottoming. It is hard to say you didn't get bottoming because the spider system prevent you from getting it.