I'm working on a circuit that will attenuate the signal if the sub is reaching its max excursion. Can I use the output from the amp and translate that into excursion (for example: if the output is 65v, then the sub is at max excursion), or do I have to use some kind of a pressure sensor inside the box (sealed) and calculate it from that? Is there any other way to do it?
Excursion is frequency dependant, and will have different values with similar voltages at different frequencys. The feedback devices that I'm familiar with use a small accelerometer attached to the cone of the speaker.
OK, so you can estimate the cone amplitude by building a 2nd order lowpass filter with the same Q and f0 as the box (fc and Qtc). If you feed that filter with the music signal, the output from the filter will be directly proportional to the cone amplitude in the low-frequency region. A bass-reflex box is far more complicated.
If max excursion itself is a meashure, two photedetectors can bee used. Bouth of/dark at inner position, and bouth on/light at outer. Simple logic and dampening of signal can bee triggered.
Svante said:
More complicated? The excursion of a driver in a vented box can be modeled by parking a notch filter after the 2nd order lowpass mentioned above. The fc would be the box resonant frequency, naturally, and I'm guessing Q would be the box alignment Q. Not quite trivial, but not horrendous either.
DSP_Geek said:
Ok, so let's take that path too. You are right in that it is doable. The second order filter should now have a Q and F0 corresponding to the driver in an infinite baffle (ie not in the box). The notch filter should have a F0 corresponding to the fh of the box. The depth of the notch is not terribly important, since it corresponds to a low cone amplitude, and we don't need the protection circuitry to be activated here anyway.
I fail to see right now if there is a third parameter, namely the width of the notch, or if that is just a result of the depth of the notch.
As you say, "not quite trivial, but not horrendous either". 🙂
xplod1236 said:
At the same output excursion in a sealed box quadruples (increases 12dB) for each halving of frequency.
Below the pass-band you get to a 12dB/octave roll-off so excursion remains constant for a given input voltage. I don't understand what happens when you get close to DC although with a sub-sonic filter on the input you won't have to worry about it.
Within the sub-woofer's pass-band excursion for a given input voltage is increasing 12dB for each octave lower you drop.
As you go beyond the pass-band output decreases (12dB/octave?), you have an even bigger (24dB/octave?) decrease in excursion, and can assume that with a music signal you're going to be limited by Pmax.
Feed the input signal to a 12dB/octave low pass filter where the low frequency pole matches the sub-woofer's. Additional filters will be needed to adjust for any boost in the plate amp. You probably need to use all-pass filters to get a delay between that and the amp input.
Run the output of this into a peak detector.
Feed that output to the negative input of a differential amplifier with the positive input tied to a higher voltage reference equal to what your VCA expects to provide unity gain.
Feed that op-amp output to a VCA on the sub-amp input.
Attempts to exceed the sub-woofer's mechanical limits will lower the volume.
Would it be possible to do this with a microcontroller? I would use a low pass filter with the same q and f0 as the box to estimate the excursion, and I would run that into a peak detector, connected to the micro. The micro would sense that the signal is too high, and attenuate it with a pga2310. That's the simplified version. For the real version, I would need to set up 2 or 3 peak detectors each activating at a different amplitude to figure out how much I need to attenuate the signal. I would also need to add a delay before the amp so that the attenuation can take place before the signal gets to the amp.
Would a microcontroller (pic16f84a @ 20MHz) and a pga2310 be fast enough for this?
Would a microcontroller (pic16f84a @ 20MHz) and a pga2310 be fast enough for this?
Probably YES.
As you are know running into the gospel T-party playing your faworite loud, you are not going to hear the cresendo anyway or maybee, it either breaks speaker if not implemented or malfunktioning. If it works there will bee no cresendo anyway. where is common sense (?) when to turn down ? distorted or before distorted, or before a distorted cresendo playing musik not test tones.giving a loud signal for a period... would you calculate elements compression and altered Q due to voice coil heat up.
Some bassrefleks cabinets build up internal pressure as airflow is easier one way compared to the other, in out of cabinet. How much displacement limit do you consider signifikant before turning musik down as you would like to ceep playing, ie the show must go on as the next tone to play is unknown ?
Or mikrophone level or currentflow, estimate from power consumtion, vibration sensor ( spesial calibration prosedyre ).
first order filtering (integration) from f3 with level detection will bee just as good afterall.
shaken not stired or stired not shaken like to bee or not to play
As you are know running into the gospel T-party playing your faworite loud, you are not going to hear the cresendo anyway or maybee, it either breaks speaker if not implemented or malfunktioning. If it works there will bee no cresendo anyway. where is common sense (?) when to turn down ? distorted or before distorted, or before a distorted cresendo playing musik not test tones.giving a loud signal for a period... would you calculate elements compression and altered Q due to voice coil heat up.
Some bassrefleks cabinets build up internal pressure as airflow is easier one way compared to the other, in out of cabinet. How much displacement limit do you consider signifikant before turning musik down as you would like to ceep playing, ie the show must go on as the next tone to play is unknown ?
Or mikrophone level or currentflow, estimate from power consumtion, vibration sensor ( spesial calibration prosedyre ).
first order filtering (integration) from f3 with level detection will bee just as good afterall.
shaken not stired or stired not shaken like to bee or not to play
Can I run an opamp from 0 and +5v supply and feed it a +/- signal, or do I have to use a diode to remove the - signal?
I didn't mention that I only need the signal from 0 to +5v. Would it kill the opamp if I fed it -V signals when its lowest supply is 0v?
It might do. Some op-amps don't like inputs outside the rails, some are more tolerant. It will be specified in the datasheet.
There were once some speakers by KLH that used a tuneable highpass (with a peak) that was put into a control loop in order to achieve the momentarily best compromise between LF extension and max SPL.
Some modern P.A. systems use the same thing BTW.
I have the original AES article describing the KLH system if anyone is interested.
I also once developed a tuneable highpass using an OTA for that purpose.
The description of an excursion detection circuit using a lowpass (closed box) or a lowpass and notch (reflex) followed by a peak detector is correct.
Regards
Charles
Some modern P.A. systems use the same thing BTW.
I have the original AES article describing the KLH system if anyone is interested.
I also once developed a tuneable highpass using an OTA for that purpose.
The description of an excursion detection circuit using a lowpass (closed box) or a lowpass and notch (reflex) followed by a peak detector is correct.
Regards
Charles
- Status
- Not open for further replies.