Hello all - in response to some questions from some of you who are upgrading your crossovers, I’d like to post some of my reflections. I have uploaded a CAD drawing of my modular outboard crossover cabinet to my virtual system. That drawing is empty of components, which can be added as appropriate. The cabinet system holds individual boards for each driver and can be size-scaled as needed. It is simply form follows function as were the speaker cabinet designs.
First some general background.
Upgrading a crossover is fraught with uncertainty. All components interact with all other components, and all components have some amount of other electrical elements. In other words, a cap doesn’t just have capacitance, but also has series resistance, which varies with load, etc. Foil coils present differing resistance and capacitance than wire coils, etc. Re-balancing those elements after any change is highly demanding, and beyond my knowledge. I try to stay out of trouble by being cautious and self-critical of sonic changes. These considerations are why Rob is ’unenthusiastic’ about cap changes, or any changes. He has made his own determinations whereby he wants to stay with Jim’s results. I am cautiously venturing further and value all feedback from you who brave the territory.
As a general comment, I will be repeating some things I’ve said over the past few years here, on the chance that repetition might be better than omission for some.
Why consider outboard? Every performance parameter is improved by getting the crossover away from the electromagnetic, vibratory, hot stew inside the cabinet. The crossover is built-in for cost and marketing considerations. I have developed three solution styles to address the problems.
1: Hang the crossover on the outside/back of the cabinet (with a covering grille)
2: Locate it in a ventilated and isolated sub-base
3: Locate it in a separate enclosure a couple feet behind the cabinet.
Let’s look at some problems.
A: Temperature fluctuations are destructive. Components’ electrical values change with temperature, more than a little - enough to make a notch filter not match its target driver resonance, etc. A hot resister can influence or even melt a cap. Driver coils can burn out or come loose. Note that all elements generate heat, not just resistors. Jim used small-gauge coils where appropriate because their dissipation factor is better than a small resistor. Heat dissipates by convection into a cooler ambient environment, but about 80% of the heat is dissipated by direct radiation. So, keep components away from other components and away from elements that reduce direct radiation. Note, we can increase dissipation by raising components off the board for 360° radiation and convection. This improvement is only possible when we isolate the xo from the vibratory environment in the cabinet where all components must be locked down.
A special case is temperature inside the sealed cabinet, which can reach about 200°F. I am venting the enclosure via small diameter air inlet on the bottom and outlet on the back near the top for convection air-flow. Long-term sessions remain more sonically consistent via more consistent ambient temperature.
A difficult case is cooling the voice coil. An aluminum VC former connected to aluminum diaphragm works great - the cone can get hot enough to sizzle spit. But resultant electrical eddy currents are sonically degrading. The bullet phase plug (such as the 3.6 midrange) allows airflow to the coil. I have developed a heat sink on the back of the cabinet for each feed wire to draw heat directly from the wire while electrically isolating them to avoid eddy currents. The motor structure gets surprisingly warm in heavy use. I’m hooking a drain wire from the back plate to a heat sink on the cabinet back to dissipate heat as well as drain the electrical charge which accumulates from the relative motion of the parts.
Thiel speakers have always been considered best as small signal systems most suitable for simple music rather than rock or Wagner. Taken together these heat management methods move Thiels toward more robust use.
That’s all for now. Next session I’ll speak to EMF and other fields.
