My favourite treble tends to come from a low treble tweeter crosssed to a semi -super tweeter at around 12000-1300hz [in the BBC tradition] or a dome midrange running to around 7 -8000Hz crossing over to a small tweeter.The old Celestion HF2000 as used in classic speakers like Gale 401s and Spendor BC3 sounds beautiful when used above about 7000Hz.
I suspect a mid tweeter with a natural frequency response of say 700Hz to 12-13000Hz supported by a semi -super tweeter would be ideal.
That way you avoid problem crossover points .
I suspect a mid tweeter with a natural frequency response of say 700Hz to 12-13000Hz supported by a semi -super tweeter would be ideal.
That way you avoid problem crossover points .
Last edited:
Will that make them the weaker link, in this context?
Post one:
50Hz & 95dB requires about 4mm of excursion from a 8" driver.
If you run a typical woofer much lower or louder than this, it will also become non-linear.
...so it seems to me that a 2" and 8" will run out of steam at approximately the same SPL.
Except you don't need a $5000 mic or an anechoic chamber to get accurate measurements of loudspeakers.
I know they are popular now, but I experimented with digital crossovers in the late 1990s, and I *never* managed to get good imaging with them.
Digital crossovers usually involve many milliseconds of latency, which is enough time for sound waves to travel several feet. Unless you are very precise about exactly matching the latency to each driver, you can end up with the equivalent of a tweeter that's mounted four or five feet ahead - or behind - the midrange.
Computing power has changed significantly since the 90s. Hardware DSPs are now available that are synchronous to the incoming datastream and do all of their processing within one sample period. This is obviously for IIR filters. FIR are a different kettle of fish, but there is no need to move to those unless you want linear phase, which there isn't really any point in anyway. FIR will introduce delay, but it is still of a known quantity and could easily be accounted for.
Modern DSPs are powerful. The one I've designed does a pair of 3 way active mains, an active 3 way centre channel, EQ for 4 surround channels and complex EQ for 2 subwoofer channels. That's 15 channels running at 192kHz/24bit with all processing being done within one sample period plus any time delays I've deliberately added in.
If somebody gave you free and convenient access to an excellent anechoic chamber, are you saying you would not use it?
Modern pulse measurement techniques let you get reasonable results without an anechoic chamber, but getting good results still requires some ingenuity, and gating away the nearest-time reflections reduces the frequency resolution of your measurements.
You can close-mic a woofer to get a decent low frequency response measurement, but only for frequencies where the sound wavelength is very much larger than the woofer diameter.
You can bury a speaker box in a hole in the (flat) ground, front baffle flush with the ground, and make half-space / infinite baffle measurements with a hanging mic, if you have a suitable large flat space with little ambient noise.
You can hang a speaker box from a cable suspended between two tall pylons (microphone suspended on a boom 1 m ahead of the speaker), and if you get it high enough off the ground, and far enough from reflective objects, that works as a substitute for an anechoic chamber, too.
But each of these clever tricks has its limitations. It is hard to beat having access to the real solution - an anechoic chamber designed for making loudspeaker frequency measurements.
As for the measurement mic, yeah, there are cheap electret mics available from a few sources, but they don't compete on equal grounds with the $5000 Bruel and Kjaer when it comes to precise measurements. The one we used was virtually ruler-flat from 10 Hz to 20 kHz (see attached image.) I mean flat within less than plus/minus one-tenth of a decibel.
Beyond the audio band and still within less than +/- 3 dB, this mic can measure down to 2Hz, or up to about 85 kHz. The 2Hz end turned out to be useful when developing a servo-feedback woofer. Measurements out to 50 kHz or so sometimes gave useful additional information about tweeters or crossover networks.
As an example of what this mic can do, we found a steep, narrow, tall, resonant spike ("oil-can" resonance) in the titanium dome tweeter used in the original Mackie HR824. The spike was, from memory, at 40 kHz.
We never saw that spike in anybody else's published frequency response measurements of the same speakers, most probably because they were using cheaper measurement microphones that simply didn't have the ability to measure such a narrow, high-frequency feature.
-Gnobuddy
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.