High-End Bookshelf speaker

[Zvu]

And how do you proceed to correct for constant time delay? Proceed to angle the baffle?

Regardless of DSP or passive, you have to do physical time alignment. If you do it on axis with DSP, it will not be aligned as soon as you move off axis.

You mention the linearity is great if you use quality parts. The parts you mentioned measure to about 0.2% distortion. This is orders of magnitude larger than the one of a good quality amplifier. Let alone something like the purifi amplifiers...

Any decent amplifier distortion will be burried under loudspeaker driver distortions. I see no data regarding 0.2% distortion. Can you post some measurements to back your claims ?

...Also, even if you use 20W resistors instead of 5W, they still have a thermal drift. For high frequencies this doesn't matter, because the temperature will remain constant. For low frequencies you get thermal modulation effects.

Again, just claims. Post some measurements of effect visible and how it influences the frequency response and distortion.

...I'm not sure how expensive those crossover components are, but you have a Fusion amplifier for about 330 euro for the 2 way version. That also includes amplification and DAC. Seems like a very reasonable price compared to amplifier+DAC+fancy crossover components.

Not fancy crossover components. I said what will give you great results. For two loudspeakers you are looking at 660 euros Fusion amps. High quality DAC's are quite cheap nowadays. Use Khadas DAC, IcePower 125ASX2 and you are left with about 360 euros for crossover parts. That buys crossovers for at least 3 pairs of two way loudspeakers - especially Purifi and Bliesma since those are pretty linear drivers to begin with.

...Besides if OP is developing his own speake as opposed to a DIY kit, I doubt he will get it perfect in the first try. Developing crossovers in digital domain is much easier to adjust.

It is easier with DSP, no question about that. But you still have to make high resolution mesurements (and most people don't) to use full potential of it.

 

[kipman725]

yeah sure its not like you can't get good sound with passive crossovers just that with a DSP or line level analog crossover we can get closer to ideal. I would agree that perhaps at low playback level at home the non ideal behavior of passive crossovers can be reduced to insignificance in the face of driver distortion, but this is meant to be high end, we should aim for the best possible.

The use of passive crossovers is only widespread in home audio, almost all modern studio speakers and prosound speakers are active or run multicore cables to the drivers. Some manufacturers even attempt to compensate for effects like voice coil heating by sensing voice coil resistance. My preferred approach is to build all my speakers with speakon connectors and multicore cables, this way I only need one amp rack for all my speakers so its quite economical.

 

[hidjedewitje]

If you place the drivers within 1/4th wavelength the effectively behave as a point source. They will act as if they are in phase on axis and off axis. The thing is, you want the drivers to be in phase at the crossover frequency. This is much easier done in DSP than analog. It can be done analog, but I prefer to make my life easy.

As for the distortion, I meant 0.02%, it's in the ERSE video you linked. Sure DAC's and amplifiers distortion is significantly lower than the distortion of the driver, but does that mean we should keep using crappy amplifiers? I don't think so.

As for the thermal modulation:
Bruce Hofer - Designing Analog Circuits with Ultra-Low THD - YouTube
Again, it's low, even lower than the linearity measured by your source, but I rather not have it at all.

Microphonics are tricky to measure and specify. Parts behave differently based on how you hit them. As for hysteresis I can only find Bruno Putzeys/Lars Risbo's claim of 0.1% for loudspeaker units. I find them reputable enough that I believe that claim. You can find the article here: Purifi Audio PTT6.5W04-01A midwoofer | HiFiCompass


The Khadas Tone Board is nice, but I wouldn't call the IcePower amps great. The Hypex amplifiers are at least 15dB better in SINAD! Though that's another discussion.

Here's another few pro's of active:
- Going active allows you to choose drivers for their distortion characteristics as opposed to trying to match impedances, FR and sensitivity (you still have to match crossover freq and dispersion, but you lower the requirements).
- You have no tolerances, because the filters are depending on math as opposed to production tolerances. It also allows for use of odd values.
- You can automate the production: Measure, process and done! Though this isn't very usefull for a single pair.
- In case you are interested, you can implement room correction/EQ preferences in the DSP as well.

- You can use feedforward/feedback for distortion reduction. Klippel uses this in their new Klippel Controlled sound technology. For subwoofers servo subs become optional. Current driven amplifiers are also useful sometimes!

 

[celef]

different strokes for different folks