• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Adjusting DSP filter software to compensate for driver sensitivity (for active speakers)

maglev rabbit

Member
Joined
Nov 1, 2022
Messages
5
Likes
1
I am experimenting with building active speakers using FIR filters in a DSP. I need to adjust the level of the tweeter, mid range and woofer audio streams to allow for the different sensitivity of the drivers. I had assumed this would be a simple matter of finding the difference in sensitivity between two drivers, and then adding/subtracting that number of db's from the signal level of the appropriate driver. For example, my tweeter has a 91.5 db sensitivity (2.83 volts / 1 meter) and my mid-woofer has a 87db sensitivity (2.83 volts / 1 meter), a difference of 4.5 db, and I assumed that all I had to do was apply an attenuation of 4.5 db to the tweeter signal within the dsp, likewise a similar appropriate adjustment for the bass driver, send the signals through identical amps with the levels set the same, and i would get at least a reasonable match to start with. However, the high end is way too loud. I know that there are two different db ratings, one for voltage and one for power, and that 3 db increase in power is a doubling of power, whearas 6 db of voltage is a doubling of voltage. Also that the speaker db rating is a power rating. Does this mean that to increase the power of a speaker by 3 db requires a 6 db increase in voltage drive to an amplifier driving that speaker? Is that true? Anyhow, to adjust the signal to compensate for a 4.5 db difference between two drivers, what amount of db (up or down) should i use in the dsp software to make the levels roughly even?
Hope someone can help.
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
7,531
Likes
11,998
Why not measure the frequency response of the speaker and adjust the levels based on that? You don't need fancy equipment for that, just an iPhone is enough.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
I am experimenting with building active speakers using FIR filters in a DSP. I need to adjust the level of the tweeter, mid range and woofer audio streams to allow for the different sensitivity of the drivers. I had assumed this would be a simple matter of finding the difference in sensitivity between two drivers, and then adding/subtracting that number of db's from the signal level of the appropriate driver. For example, my tweeter has a 91.5 db sensitivity (2.83 volts / 1 meter) and my mid-woofer has a 87db sensitivity (2.83 volts / 1 meter), a difference of 4.5 db, and I assumed that all I had to do was apply an attenuation of 4.5 db to the tweeter signal within the dsp, likewise a similar appropriate adjustment for the bass driver, send the signals through identical amps with the levels set the same, and i would get at least a reasonable match to start with. However, the high end is way too loud.

Way too loud how?
Pink noise, white noise, or music?



I know that there are two different db ratings, one for voltage and one for power, and that 3 db increase in power is a doubling of power, whearas 6 db of voltage is a doubling of voltage.

Close… a 3dB increase in voltage is a 6dB increase in power.
Mostly because W = VI and I = V/R.
And W= V^2/R and 6dB is 2^2.

But one also has box affecting things and making more iof an affect on the woofer.
Just measure it like @staticV3 suggests.


You might as well throw out a step function response plot at some point, as that would be interesting to see… and maybe impulse response.
A FIR based approach allows those to be more optimal than IIR.
 
Last edited:

puppet

Senior Member
Joined
Dec 23, 2020
Messages
446
Likes
284
Since you're building this from the ground up you need to invest in some measuring equipment. Forget about the driver sensitivity ratings and measure each drivers output frequency response in their enclosure(s). You can then adjust spl between drivers. I'd run the digital signal levels as hot as you can without output clipping and attenuate at the amplifiers.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
... I'd run the digital signal levels as hot as you can without output clipping and attenuate at the amplifiers.

I think that that assumes that the DSP and amps are at the speaker?

Most people using a FIR would likely have the the DSP between the preamp and amp(s), and signals like for a turntable would get turned into a digital signal.
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
2,916
Likes
3,831
I think you'll just have to measure it. Even then you may want to make some final tweaks by ear. ...There's no point in doing this kind of project if in the end you aren't happy as you could be with some adjustments.

and I assumed that all I had to do was apply an attenuation of 4.5 db to the tweeter signal within the dsp, likewise a similar appropriate adjustment for the bass driver, send the signals through identical amps with the levels set the same, and i would get at least a reasonable match to start with. However, the high end is way too loud.
I would have made the same assumptions!

Usually it's 1 meter and 1W but that's based on the nominal impedance. The actual impedance varies with frequency and the "average" may not really be or 4 or 8-Ohms. So it's usually 2.83V. I think I've seen some 4-Ohm speaker systems rated at 2.83V, but I'd expect an individual driver to be measured with the voltage corresponding to it's nominal rated impedance.

Amplifiers are "constant voltage" and speakers (and headphones) are spec'd and tested with "constant voltage". So if the driver has higher impedance at certain frequencies, the speaker won't be getting as much current or the expected-calculated wattage. But that's not a problem because speakers aren't designed & tested with "constant power".

I know that there are two different db ratings, one for voltage and one for power, and that 3 db increase in power is a doubling of power, whearas 6 db of voltage is a doubling of voltage. Also that the speaker db rating is a power rating. Does this mean that to increase the power of a speaker by 3 db requires a 6 db increase in voltage drive to an amplifier driving that speaker? Is that true?]/quote]Doubling of the voltage (+6dB) results in doubling the current (Ohm's Law). Power is calculated as Voltage X Current so +6dB is 4 times the power.

The dB calculation is different when you are measuring volage or wattage, but in both cases +3dB is twice the power.

For DSP the digital value corresponds the voltage out of the DAC so you should be using the voltage {amplitude} dB formula. Or if your DSP software works in dB and you're not using ratios or amplification factors, the software will do the right thing with a dB adjustment.
 
OP
M

maglev rabbit

Member
Joined
Nov 1, 2022
Messages
5
Likes
1
I would really like to understand the relationship between signal level decibels and speaker spl decibels, regardless of whether it is necessary. No explanation I have heard so far makes any sense to me (my fault). What I really want to know is this. Suppose I am sending a constant digital signal sine wave of 1Kz to an amplifier connected to a speaker, and let's suppose i measure the speaker spl at a particular distance, somehow, and it is reading 80db. If I then make no other adjustments except to increase the signal level in the DSP by 3 db, does the speaker level then go up to 83db? Or do I have to increase the signal level by 6db to make the speaker spl go up to 83db? Can anyone answer that question?
By the way, I am running brutefir FIR software on a headless raspberry pi 400, driving an Asus Xonar U7 MKII 8 channel sound card, connected to three class-d amps driving a ScanSpeak SSD2608 tweeter, a 5 inch SB Acoustics MFC poly mid-woofer, and a single 15 inch bass woofer. It's an entirely experimental set up.
When I say that the tweeter is way too loud, it's that the sound overall is very thin and top heavy. I have allowed the audio streams to go through unaltered in level and adjusted the levels at the amps, by ear, and can get a reasonable balanced sound. I have used TrueRTA with a frequency sweep to look at the frequency response of the overall speaker with all drivers operating, but the room acoustics make a nonsense of the readings. It's not really practical to do the experiments outside in the garden.
Also I am wary of making adjustments by "ear", since the ear/brain is heavily influenced by psychological expectations - you make a small change, hoping to hear a difference, and you hear one, even if it's not really there. Also, your ear has been trained by previous exposure to other systems as to what is "correct".
As I said in the beginning of this post - I would really like to understand what I am doing at this point, and avoid trial and error until the later stages.
Many thanks for your posts so far. Any help much appreciated.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
I would really like to understand the relationship between signal level decibels and speaker spl decibels, regardless of whether it is necessary. No explanation I have heard so far makes any sense to me (my fault). What I really want to know is this. Suppose I am sending a constant digital signal sine wave of 1Kz to an amplifier connected to a speaker, and let's suppose i measure the speaker spl at a particular distance, somehow, and it is reading 80db. If I then make no other adjustments except to increase the signal level in the DSP by 3 db, does the speaker level then go up to 83db? Or do I have to increase the signal level by 6db to make the speaker spl go up to 83db? Can anyone answer that question?

Well look at the resultant level in the time domain.
It will either be twice as big or 1.414x bigger.

Back to the ohms law equation… W = V*I
The 1.414x bigger voltage makes the current 1.414x bigger as well.
Square root of 2… squared… is 2.
So you do 3dB in voltage, and you get 6 dB in power.


… If I then make no other adjustments except to increase the signal level in the DSP by 3 db, does the speaker level then go up to 83db? Or do I have to increase the signal level by 6db to make the speaker spl go up to 83db? Can anyone answer that question?

Yes 83dB


… Or do I have to increase the signal level by 6db to make the speaker spl go up to 83db? Can anyone answer that question?

If you went 6dB in voltage then you would get 12dB in power, not 3dB.
I am pretty sure your DSP would be working in power, but it is possible it could be voltage.

So 3dB could give 6dB in power, but it 6dB will never give you 3dB… It will either be 6dB or 12dB, depending on whether it was voltage or power.\

Measure the AC voltage on a tone, and then increase the DSP by 3dB… see if the new measurement is double the old one or just 1.4x the first measurement.
Might as well take out the SPL app on the phone and also measure the level.
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
I would really like to understand the relationship between signal level decibels and speaker spl decibels, regardless of whether it is necessary.
There is a simple way to see this. Sengpiel Audio has a bunch of great calculators for anything audio and the page linked below is helpful for this task.

http://www.sengpielaudio.com/calculatorVoltagePower.htm

You can then use the voltages that you find for a given dB increase and put these into a speaker box/enclosure simulator like there is in VituixCAD, Unibox etc.

As an example put 1V in and see the flat SPL level, change it to 1.41 V for 3dB voltage gain, and then 1.99dB for 6dB voltage gain.

The SPL in dB will increase in line with the voltage gain in dB.

While this can help you understand the relationships, there can be much more going on in a real speaker that will influence the SPL that each driver produces such as the effects of baffle diffraction and overall directivity. Measurements are much better, even using an app on a smartphone would be good enough to get an idea of where the issue is.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
There is a simple way to see this. Sengpiel Audio has a bunch of great calculators for anything audio and the page linked below is helpful for this task.


As an example put 1V in and see the flat SPL level, change it to 1.41 V for 3dB voltage gain, and then 1.99dB for 6dB voltage gain.

Maybe try it in real life And show the volt meter and SPL meter.
(And avoid playing with the calculator)

The units do not even look right… 1.4V does not equal 2V…
(and later 1.99dB /= 6dB?)
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
Maybe try it in real life And show the volt meter and SPL meter.
(And avoid playing with the calculator)
That isn't simple for me to do right now, but I have done it enough times before to be confident in the validity of the simulator.

Here is an example from Unibox

Option one has 1.0W power which is a voltage of 2.36V and gives 91.2dB SPL maximum output

Single Voltage.PNG


If I double the voltage to 4.72V an increase of ~6dB of voltage this results in 4W of power and 97.2dB max SPL, the SPL has increased in dB's by the same amount the voltage increased in dB's, which is also the same as 6dB's of power gain. 4 times power equals 6dB

Double Voltage.PNG

The units do not even look right… 1.4V does not equal 2V…
(and later 1.99dB /= 6dB?)
I was using 1V as the base reference voltage for simplicity

3dB of voltage gain applied to 1V results in 1.414V

6dB of voltage gain applied to 1V results in 1.99V (GV = 20×log (V2/V1) It is not 2V as 20log (2/1) = 6.02dB

The difference in SPL output from a speaker when applying the three voltages will increase in line with the dB of voltage gain.
1.414 V applied will result in 3dB SPL more than when 1V is applied.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia

3dB of voltage gain applied to 1V results in 1.414V.

No… you have 1.5dB of voltage gain, which gives 3dB of power gain.
power is proportional to voltage squared, because current it proportional to voltage,


6dB of voltage gain applied to 1V results in 1.99V (GV = 20×log (V2/V1) It is not 2V as 20log (2/1) = 6.02dB

2V is 3dB more than 1V.
2V is never 6dB of voltage gain, it is 3dB of voltage (dBv)
… It might 6dB (dBw), but it is not 6dBv.

If 1V gives Y Watts, then 2V gives 4*Y watts.

6dB of power gain is 3dB of voltage gain, or double the voltage, which is 4x the power… (6dB)



The difference in SPL output from a speaker when applying the three voltages will increase in line with the dB of voltage gain.
1.414 V applied will result in 3dB SPL more than when 1V is applied.

Correct.
You put in 1.414x more voltage and you get 2x the power, because power is proportional to voltage squared.

Quite mixing volts and watts.
You can talk about voltage gain, so use dBv.
Or talk about power gain and use dBw.

If you do the math in volts then you use a factor of 20Log10 versus 10Log10 for getting power.
You need to be consistent in volts or consistent in watts.



If I double the voltage to 4.72V an increase of ~6dB of voltage this results in 4W of power and 97.2dB max SPL, the SPL has increased in dB's by the same amount the voltage increased in dB's, which is also the same as 6dB's of power gain. 4 times power equals 6dB

No…
- Doubling the voltage increases voltage by 3dB. (dBv)
- Doubling the voltage increases the wattage by 6dB. (dBw)
- 3dB /= 6dB… but 3 dBv does equal 6 dBw.
 
Last edited:

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
No… you have 1.5dB of voltage gain, which gives 3dB of power gain.
power is proportional to voltage squared, because current it proportional to voltage,
The formula that I know for voltage gain is
Voltage gain (dB) = 20×log (Audio output voltage / Audio input voltage).
Voltage gain is defined as the ratio of the output voltage to the input voltage in dB.

20log(1.414/1)=3
2V is 3dB more than 1V.
2V is never 6dB of voltage gain, it is 3dB of voltage (dBv)
… It might 6dB (dBw), but it is not 6dBv.
I didn't use dBu or dBV, but if you want to then 1V RMS is 0dBV, Analog Devices also agrees that 3dB of Voltage gain applied to 1V will result in 1.413V

https://www.analog.com/en/design-center/interactive-design-tools/dbconvert.html

Analog Devices.png


1.413V RMS is 3dBV, so yes the increase is 3dBV

Analog Devices 1.413.png



You need to be consistent in volts or consistent in watts.
I have only referenced voltage gain and how many actual volts RMS it would take with the same speaker to get a specific SPL increase.
We seem to be speaking past each other.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,223
Likes
17,799
Location
Netherlands
Sounds like the tweeter may not be the only problem. If you used sensitivities for the woofer level matching as well, you may be off by quite a bit. Most large woofers droop heavily down from the midrange. And there is also the bafflestep te account for.

Try making your bass 3 to 9dB hotter, see if that brings more balance.

In the end, you’ll need to measure it to really make it function properly.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia

I didn't use dBu or dBV, but if you want to then 1V RMS is 0dBV, Analog Devices also agrees that 3dB of Voltage gain applied to 1V will result in 1.413V

Let’s call it 3 dB of gain, or 3 dB of power gain.

Which is only 1.5dB of voltage gain, which is why we are speaking posts each other.
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
In my examples the impedance of the input and output is considered to be the same which is why the dB numbers for power and voltage end up being the same.

"I’m hearing more and more customers ask ”how does gain through a signal chain change with different load impedances?" And, "when does voltage gain and power gain coincide when measured in dB?” I wanted to share the answers with the Analog Wire audience in case any of you have the same quesitons. So, here we go…..

In a single-ended signal path with 50 Ohms termination, the gain calculations are very easy because the voltage gain (20* log (Vout/Vin)) is equal to the power gain (10* log (Pout/Pin)). Things become a bit more complicated though when the impedance of the load, or source, changes."


https://e2e.ti.com/blogs_/b/analogw...ologies, voltage,power gain changes with load.
 

tinnitus

Active Member
Joined
Jan 26, 2021
Messages
144
Likes
199
Location
Germany
You need the gernan acoustic knife accurate from audio vero. You can do do driver linerasation, corssovers, roomcorrection with messurements at yourr listening position. Build your target curve and get your FIR filter which you cyn integrate in CammilaDSP.
 

fineMen

Major Contributor
Joined
Oct 31, 2021
Messages
1,504
Likes
679
I am experimenting with building active speakers using FIR filters in a DSP. I need to adjust the level of the tweeter, mid range and woofer audio streams to allow for the different sensitivity of the drivers. ...
Please consider to pre-adjust the relative SPL of the drivers using a passive resistor attenuator on line (not power) level. That attenuator goes in front of the amplifier. To equalise SPL digitally may cost some bits of resolution. E/g: a bass & horn combo might need up to -25dB for the horn when low bass e/q is taken into consideration also.
 
Top Bottom