Nick wrote: ↑Sat Jul 23, 2022 9:34 pm
That being the case, the midrange and HF have to be padded anyway to the level of the bass units
I thought Colin was talking about using a Hypex amp for the LF?
Yes, but for level matching (ignoring everything else for a moment) there are two ways of going about it with a semi-active system. You can raise the output of the bass drivers, or you can pad down the midrange & HF to match them. Both work, & within reason, assuming you don't have a completely limp-wristed amplifier on one leg (most likely the midrange / HF) the difference in that narrow sense is mostly just a question of implementation & method. Whichever you do, your ultimate limit in this sense is always defined by the dynamic headroom and distortion performance of the least efficient element, which is inherent to the driver[s]. Not everybody will agree with this, but for me, overall system level-matching is not the main advantage to a semi-active system; their primary scoring points are greater flexibility for adjusting to room position etc., and avoiding needing some extremely large and expensive components necessary for a passive equivalent -especially inductors, since low DCR is pretty much mandatory to keep the filter functioning as intended & not cripple damping.
Re autoformers -I like the idea, but (correct me if I'm wrong -it's been a long time since I used them) while they are fine for adjusting overall SPL and / or impedance, they'll do just that and nothing more. But since these are drive units, not resistors, their impedance load varies with frequency and some of the impedance peaks need to be damped down to ensure the passive filters function properly. For example, the midrange units for this box will have roughly a 25ohm peak relative to the nominal level at 200Hz, which is right around the (passive) midrange high-pass filter frequency. If that peak isn't flattened out, the drivers will start to roll off, and then either level off, or peak back up again, which will affect power-handling / distortion, and ultimately the overall speaker response. Within reason (it's not always possible / practical) the simplest way of damping the peak down & ensuring the filter behaves as its supposed to is with a resistive shunt. The more complex alternative is to use an LCR Zobel, but that's three components rather than one. Sticking with this example, since the midrange impedance load will be about 12ohms, which is a little higher than ideal for a lot of amplifiers, we could then add an autoformer to address that, and what level adjustment, if any, is desired. But if (if
) you can get the necessary damping of the impedance peak and also drop the level a couple of ohms with a simple shunt resistor -it strikes me as a valid (and cheaper) solution. Since current / power demands tend to drop as frequency rises too, it shouldn't be causing excessive waste, especially if you're starting from a reasonably high efficiency in the first place.
More generally (and this is just a personal thing) -I tend to think that within reason, you can get 'better' results if you make the speaker load as unreactive as possible, even if this means sacrificing a touch of sensitivity or drawing a little more current, than if you maximise sensitivity, but end up with a highly reactive impedance load & electrical phase angles. Context is obviously critical as it's going to depend on a bunch of factors that need to consider in the design stages. Nick -I'd be interested to hear your take on that as an amplifier designer; what sort of load would you prefer? Presumably a flat-line impedance with the highest possible sensitivity, but if we're having to make some compromises -which direction would you take?