A noob's question regarding MFB with a DSP (ADAU1772/1777)

Hey!

So i am kinda interested in MFB but too cheap to purchase a ready-made solution from company like Velodyne. I am also not really good in electronics so making an analog version of an MFB woofer is too much for me.

I found this board based on ADAU1772/1777 on AliExpress.

attachment.php


So it looks like it has line level inputs and line level output.

Can I properly place an accelerometer on a woofer, feed the signal from it to the DSP, get the line level subwoofer audio signal and feed it to another input of this board, then mix the woofer signal with an inverted signal from an accelerometer and basically have an MFB?

I believe, this DSP should be programmed in Sigma Studio. I saw examples of different filter implemented with ADAU1701 and I don't perceive it as something very difficult.

What else is needed for a proper MFB to work? Should signal from accelerometer be amplified before it is fed into the dsp? Any specific kind of amplifier to be used with subwoofer?

I understand that this is extremely noobish, but someone had to ask this kind of question :)
 

Attachments

  • adau.jpg
    adau.jpg
    99.1 KB · Views: 356
There's a chance that board could be helpful to accelerometer MFB using the mic input to amplify to sensor's signal. But you'd still have to power the accelerometer and shield the wires and tricky to mount.

Easiest way to start is with a series resistor in the ground-end of the driver, like .5 Ohm. You'd take the signal and mix it into the signal from the pre-amp which goes into the amp. An accelerometer and current feedback do the same thing: ensure the motion of the cone is true to the signal.

Be sure to use a driver than won't be destroyed by the max signal the amp might perversely throw at it one bad day.

MFB is great but you need to understand the technology for DIYers.
 
There is this document: Bachelor Thesis. Digital implementation of motional feedback.

It is a theoretical implementation of ADAU1777/1772 based MFB. In this, authors obtain an error signal by subtracting feedback signal from source signal in a differential or an instrumentational amplifier and then feed it into DSP along with input signal (i believe) and implement MFB correction.

I wonder, if, for a first experiment sake, i put a microphone from a preamp in the front of a loudspeaker and then feed this signal in the DSP along with source signal, can't i then apply correction within a DSP?

If this works out (a proof of very simple concept) I may then buy a real ACH-01 and use OPamps as for offered in the aforementioned thesis.

Anyway, I'll order this development board to play with.
 
StabMe - thanks for link.

Yes, there are no mystery parts. You simply amplify the sensor signal and mix it into the signal path upstream. But certain important EQ and band-limits are essential because woofer speakers have atrocious phase movements which makes simple direct feedback impossible.

Yes, the first thing people think of is using a mic. Not feasible, again, because of phase gyrations all around.

Yes, the ACH-01 us the common approach. But the current sensor is simple and cheap and works as well, even if there are minor shortcomings.

B.
 
Thank you for spoonfeeding that simple info, bentoronto.

Does one need to somehow modify the amp before it is ready to be used with a MFB sub or is it enough to provide the sub with DSP corrected signal?

I hear that current drive is somehow more appropriate for this kind of application.
 
I admire your intrepid courage to go forward. But it is my custom not to give detailed cookbook-like directions on any little topic I have a grasp of so anybody can swim in water over their heads. You can easily destroy an amp and drivers unless they are very robust.

Nothing special about the amp - but good to down almost to DC is helpful. Nothing special about DSP except it lets your dial-in changes that a hard-wired accessory board can't do.

My initial efforts to use the familiar (and wonderful) Behringer DCX2496 came to nought because the processing in that elderly design was far too slow.

The accessory board has to handle a few chores: condition the sensor signal, power an accelerometer (if one is used), ensure right polarity, feed right amount of feedback into amp mixed with the original signal source, and control the bandwidth of the feedback so no squealing slips in. EQ the feedback so that the driver is voltage driven (acceleration, I think). Which might not be the whole list.

B.