Help with "tweeter protection"

[Pafi]

The capacitor therefore does not have to be large (a common rule of thumb is to size it so that its roll off starts two octaves below the crossover frequency), but sized to roll off at the anticipated crossover frequency.

So NOT 2 octave below...

 

[rfj1862]

Whatever is going on that is a very poorly written sentence!

Just for the record I spoke with Alan Langford, and he said:

"the rule of thumb is the value of the cap should be 2 octaves below the crossover point (also taking into account the impedence of the tweeter). The values don't need to be precise. Note that you don't need this cap in place as its only for DC protection from the amplifier or some upstream clicks that could damage a tweeter."

So ~40 uF at ~500 Hz with an 8 ohm driver.

 

[Andersonix]

35uF for $2.50 from Apex Jr:
Axial Capacitors

 

[AllenB]

"the rule of thumb is the value of the cap should be 2 octaves below the crossover point

Indeed that makes sense when trying to go close but not too close. Is there a reason not to size it for the crossover frequency?

 

[Pafi]

I can't believe that only Richidoo managed to understand the manual and the situation...

 

[picowallspeaker]

I can't even imagine how a device such a DEQX or XXXX could COMPENSATE

 

[wintermute]

@Pafi, Steten and Allen did as well

@Pico, why is it hard to imagine? even a rudimentary adjustment could be made based on the cap introducing a 6db rolloff and 90 deg phase shift. If the system is smarter, then a measurement of the response with the capacitor in place before creating the filter would enable the creation of a filter which gives the desired acoustic slope (which is what you really want).

Tony.

 

[phase_accurate]

I can't even imagine how a device such a DEQX or XXXX could COMPENSATE

The "trick" that is used is the reduction of the order of the highpass within the electronic crossover (be it digital or analog) by one and to add some EQing to it. The firmware of the DEQX seems to be able to do this automatically.
The only problem (or better: challenge) is to accurately dimension the series capacitor such that the desired cutoff frequency is achieved (given that the load called tweeter is not frequency independant and resistive).
But if executed well it looks like an elegant and effective solution.

And it is not just restricted to crossovers like the DEQX. It can also be done with "hand-knitted" active crossover topologies. For the proper execution of the latter some knowledge about defining target transfer functions is usually needed.
Regards

Charles

 

[AndrewT]

I would use the smallest capacitor value I reasonably could, which means putting its rolloff right at the crossover. While this offers better protection, smaller capacitors are normally less expensive and better performing. Additionally this makes the capacitor a part of the crossover rather than a redundant part.

The DC blocking capacitor can be used to only block DC and not interfere with the crossover.
Or the capacitor can be a dual purpose DC block and single pole passive filter. It's actually half of an LR2

eg. crossover set to 2.24kHz.
set the DC protection capacitor to have F-3db = 560Hz, or lower. That 2octaves below avoids much of the phase effects of the added capacitor.
or
Use the capacitor scaled to match the crossover of 2.24kHz and give the first pole of a multi-pole filter.
The remaining electrical filter requires one less pole to achieve the desired roll-off.

I would be very tempted to go this second route. By adding a speaker Zobel to the treble driver you can allow the passive filter to actually work @ the desired crossover frequency. An LR4 would now require an electrical 3pole filter which combined with the single pole capacitor+Zobel gives the required Q for an LR.

 

[picowallspeaker]

The "trick" that is used is the reduction of the order of the highpass within the electronic crossover (be it digital or analog) by one and to add some EQing to it. The firmware of the DEQX seems to be able to do this automatically.
The only problem (or better: challenge) is to accurately dimension the series capacitor such that the desired cutoff frequency is achieved (given that the load called tweeter is not frequency independant and resistive).
But if executed well it looks like an elegant and effective solution.

And it is not just restricted to crossovers like the DEQX. It can also be done with "hand-knitted" active crossover topologies. For the proper execution of the latter some knowledge about defining target transfer functions is usually needed.
Regards

Charles

Hi, probably all *this stuff* is needed in order to get a good response from a CD tweeter, which often needs dual-knee/slope to match with the DI of the horn.

 

[rfj1862]

I can't believe that only Richidoo managed to understand the manual and the situation...

Probably my failure to explain properly due to my limited knowledge. But I'm catching up slowly. If anyone has molecular biology questions please feel free to PM me but for now I'm going to be all take and little give

 

[rfj1862]

Also, I am going to hunt around here, and this should probably be a different thread, but can anyone recommend a good book on this stuff? I am electronics naive but, at least I hope, good at math and science, so it should start from the basics but no need to be too simple.

 

[AndrewT]

Linkwitz site is very comprehensive. But too much mathematics for a raw Beginner.
ESP's site is more descriptive and less maths.

 

[picowallspeaker]

The Education part of the Lenard Audio site is a big rock to climb

 

[Andersonix]

Linkwitz site is very comprehensive. But too much mathematics for a raw Beginner.

Ja, but SL shows the right kind of math, the kind with graphs and figures to help you visualize the magic....