Using resistors or L-pad for SPL matching of multiple speakers

[R-Carpenter]

The argument came from this thread:
http://forums.audioholics.com/forums/showthread.php?t=57469
I don't want to keep going there so I am starting a new one. If moderators would be so kind as to move the last bunch of “back and forth” from there to here.
The argument is as follows.
Can resistors be used to adjust for a sensitivity and SPL matching of a different loudspeakers.
Basically, can a reviewer use a simple switch box with variable L-pads or a bunch of resistors between the amplifier and a loudspeaker?
I say that the differences in frequency response of a loudspeaker are so minuscule that it is quite possible and very cheap indeed.
Measurements were taken with calibrated Behringer ECM800 microphone, using Behringer Mic preamp, Behringer 500 amp and Briston preamplifier.

 

[R-Carpenter]

Here's a first one. Impedance taken with and without an 6 ohm resistor in series with the loudspeaker. The resistor is placed between the amplifier and the speaker.
Sand cast 10 watt resistor was used.
The impedance remains identical except it's 6ohm higher across the board.

 

[R-Carpenter]

Will the Frequency response of the speaker change? Let's see.
Measurements were taken 20cm from the woofer, 0.7m mic/ground.
No resistor, 3.5ohm, 8ohm and 15ohm sand cast resistors were used.
Crossover was connected at all times.

 

[R-Carpenter]

cschang was also arguing as to how strongly frequency response of a loudspeaker changes with the SPL increase.
I've stated that based on my own measurements and experience, FR changes are very minor given the fact that the driver or a loudspeaker is used within it's intended limits.
Here's a random volume increase in the test signal from approximately 70 to 85 db.
I do not have a calibrated SPL meter so this is just a relative comparison.

 

[R-Carpenter]

In short, I believe that impedance matching speaker switch boxes can and should be used for an SPL matching for a multiple speaker listening tests.
As you can see it is only one driver/speaker combo so it is only so valid.

 

[cschang]

Just a clarification of what I was arguing.

I said you run the risk of changing the characteristics of the crossover, and in turn, the characteristics of the speaker. I did not say it would happen every time. The fact that it can happen is the reason not to do it.

Also, in the links that I posted, it stated that if the speaker doesn't have wide impedance swings, the effect of inline resistance is minimal. Also, I believe using an L-pad in this situation will yield less deviations than simple series resistance.

One test, on a speaker with a relatively flat impedance, does not really prove anything. Again, in the previous thread, I sited some information, and I think there is more out there. Also, maybe a speaker designer or EE can chime in here.

I am glad you were able to prove to yourself your own point, but I still think it is a bad idea.

 

[cschang]

Here's another site for you:

http://wolfstone.halloweenhost.com/Audio/spkimp_SpeakerImpedanceMatching.html

Add a resistor in series with the speaker. A four ohm speaker in series with a four ohm resistor = 8 ohm load.
This is simple in theory, but problematic in practice. You need to use a high power resistor (in the 4+4=8 example, the resistor will dissipate as much power as the speaker). Also, sound quality will suffer, because the speaker impedance changes with frequency, but the same is not true for the resistor.

 

[TLS Guy]

Here's another site for you:

http://wolfstone.halloweenhost.com/Audio/spkimp_SpeakerImpedanceMatching.html

You are correct. Auditioning speakers using series resistors to balance spl levels would not be a fair test. Qt of the system will be changed, the effect of the spl reduction can not be constant with frequency if their are large swings of impedance. The only way to carry out the test is with separate amplifiers with trimmer pots at the inputs.

 

[highfigh]

The argument came from this thread:
http://forums.audioholics.com/forums/showthread.php?t=57469
I don't want to keep going there so I am starting a new one. If moderators would be so kind as to move the last bunch of “back and forth” from there to here.
The argument is as follows.
Can resistors be used to adjust for a sensitivity and SPL matching of a different loudspeakers.
Basically, can a reviewer use a simple switch box with variable L-pads or a bunch of resistors between the amplifier and a loudspeaker?
I say that the differences in frequency response of a loudspeaker are so minuscule that it is quite possible and very cheap indeed.
Measurements were taken with calibrated Behringer ECM800 microphone, using Behringer Mic preamp, Behringer 500 amp and Briston preamplifier.

As posted, you should match the levels before the amp and listening to speakers at different levels makes a difference on several levels. First, human hearing acuity isn't the same at all SPL. Also, the room's acoustical properties dictate the way the sound energy continues after being produced.

The speaker's response will change less than the perceived response. The only way to verify the speaker's response at various levels is in an anechoic chamber because testing this in a reverberant field will be affected by the room's acoustics.

Also, the plots with the resistors shows a difference at both extremes- a divergence of output levels at low frequencies and convergence at high frequencies, although the convergence may be due to the source or software. Any component placed in series with a speaker system will affect the signal. If you want to make the resistance less detrimental, look into fixed L-pads.

Variable L-pads aren't generally used as a volume control, they're made for adjusting the level of mid-ranges and tweeters.

 

[R-Carpenter]

As posted, you should match the levels before the amp and listening to speakers at different levels makes a difference on several levels. First, human hearing acuity isn't the same at all SPL. Also, the room's acoustical properties dictate the way the sound energy continues after being produced.

I don't see how hearing relates to this. Fletcher Munsen tables relate more to the human hearing sensitivity FR/SPL not to speaker vs speaker comparison.

The speaker's response will change less than the perceived response. The only way to verify the speaker's response at various levels is in an anechoic chamber because testing this in a reverberant field will be affected by the room's acoustics.

Digital MLS will allow you to get near identical data in comparison to anechoic chamber. Most of the companies do not have anechoic chamber handy and use software packages like LEAP, Praxis or in my case Sound Easy. Besides, post processing is used even in the data acquired in the anechoic chamber.
For example, how do you measure a loudspeaker with rear firing port? It needs to be summed up later on. The easiest way to SPL match is with SPL meter at a 1000hz or pink noise.

Also, the plots with the resistors shows a difference at both extremes- a divergence of output levels at low frequencies and convergence at high frequencies, although the convergence may be due to the source or software. Any component placed in series with a speaker system will affect the signal. If you want to make the resistance less detrimental, look into fixed L-pads.

The data below a 100hz is not valid as it's shown on the graph. I'd like to debate where a minor SPL differences in the -30db range of the woofer could be audible unless, like you said it wasn't caused by software or source/voltage variations. Based on your statement we can go back to debating speaker cables and such.

The point was very simple. If you can't match through the preamp or a receiver and you are comparing 2 pairs of loudspeakers, they should be played at the same SPL and if one is couple of db more efficient then an additional resistive load should not cause any changes in FR or over all impedance.

As Mark mentioned the over all Q of speaker/amp will change and the differences have been shown with loudspeakers that have difficult impedance loads. Is it humpy impedance or Zin/phaze that have to swing up and down?

I am willing to provide room and measuring equipment for DBT of the same loudspeaker (not necessarily mine, any loudspeaker) with the resistor inline and higher voltage from the amp and without a resistor SPL matched. I am located in Brooklyn if anyone is interested of doing it.

 

[cschang]

RC...not sure why you are so bent on this issue of using resistor or not. TLS pretty much summed up the reasons. Have you read any of the links I provided. It just isn't good practice as it introduces the possibility of affecting the sound of the speaker. If it was good practice, the industry would be doing it.

The easiest way to SPL match is with SPL meter at a 1000hz or pink noise.

Matching at 1000hz is not the wisest thing to do. Two speakers can vary greatly at this one point....more so than their differences in sensitivity. I believe the industry norm is to average and normalize for sensitivity between 1khz-3khz. Pink noise limited to this bandwidth would be great.

 

[highfigh]

I don't see how hearing relates to this. Fletcher Munsen tables relate more to the human hearing sensitivity FR/SPL not to speaker vs speaker comparison.

Digital MLS will allow you to get near identical data in comparison to anechoic chamber. Most of the companies do not have anechoic chamber handy and use software packages like LEAP, Praxis or in my case Sound Easy. Besides, post processing is used even in the data acquired in the anechoic chamber.
For example, how do you measure a loudspeaker with rear firing port? It needs to be summed up later on. The easiest way to SPL match is with SPL meter at a 1000hz or pink noise.

The data below a 100hz is not valid as it's shown on the graph. I'd like to debate where a minor SPL differences in the -30db range of the woofer could be audible unless, like you said it wasn't caused by software or source/voltage variations. Based on your statement we can go back to debating speaker cables and such.

The point was very simple. If you can't match through the preamp or a receiver and you are comparing 2 pairs of loudspeakers, they should be played at the same SPL and if one is couple of db more efficient then an additional resistive load should not cause any changes in FR or over all impedance.

As Mark mentioned the over all Q of speaker/amp will change and the differences have been shown with loudspeakers that have difficult impedance loads. Is it humpy impedance or Zin/phaze that have to swing up and down?

I am willing to provide room and measuring equipment for DBT of the same loudspeaker (not necessarily mine, any loudspeaker) with the resistor inline and higher voltage from the amp and without a resistor SPL matched. I am located in Brooklyn if anyone is interested of doing it.

Will you be doing listening comparisons or sitting there, looking at graphs and meters? If listening, How does human hearing not enter into it?

Are you asking this just for the sake of argument or is there really a point? No valid speaker comparison can be done if the SPL is drastically different. If you have ever auditioned speakers, you should know this.

Read the AES/EBU reports- this has all been covered.

I don't know your views on speaker cabling but I have to conclude that you aren't a proponent of high end cables because if you were, adding resistors wouldn't be considered a good thing in light of the claims made by people and cable manufacturers and dealers.

 

[R-Carpenter]

RC...not sure why you are so bent on this issue of using resistor or not. TLS pretty much summed up the reasons. Have you read any of the links I provided. It just isn't good practice as it introduces the possibility of affecting the sound of the speaker. If it was good practice, the industry would be doing it.


Matching at 1000hz is not the wisest thing to do. Two speakers can vary greatly at this one point....more so than their differences in sensitivity. I believe the industry norm is to average and normalize for sensitivity between 1khz-3khz. Pink noise limited to this bandwidth would be great.

I am not hell bent on it. It's just an issue and I am offering healthy arguments. If nobody did, we'd be comparing the sound of cables.

A 1000hz is where most inexpensive SPL meters are less flaky. Industry uses $600 calibrated SPL meters that most don't have. Also a 1000hz region is extremely critical to the character of the speaker in relation to the rest of the FR so I would think that any loudspeakers that are made with significant deeps or humps there will be highly colored.

 

[R-Carpenter]

Will you be doing listening comparisons or sitting there, looking at graphs and meters? If listening, How does human hearing not enter into it?

Both. Mic, RLC meters and MLS software are much more sensitive then even the best ear. The purpose is to figure out threshold and what changes affect the perception, how.

If you are comparing speaker A to speaker B, you remain the same. Your hearing and all it's fluctuation remain the same. You are the constant in this case and your hearing isn't under test, the speakers are.

Are you asking this just for the sake of argument or is there really a point? No valid speaker comparison can be done if the SPL is drastically different. If you have ever auditioned speakers, you should know this.

As a matter of fact, I have stated earlier that speaker comparison should be done at the same SPL. I am arguing because I think it's valid argument and don't like then things are taken for granted.

Read the AES/EBU reports- this has all been covered.

If you have a direct link to this issue I'd appreciate it. But again, I'd like to see some SPL and impedance curves for particular cases not just a bold statement.

I don't know your views on speaker cabling but I have to conclude that you aren't a proponent of high end cables because if you were, adding resistors wouldn't be considered a good thing in light of the claims made by people and cable manufacturers and dealers.

On a contrary, I don't place any value on a marketing techniques of cabling industry. My opinion on the cables is purely aesthetics. I think they look GREAT!. But interestingly enough while we are are at the cable issue, R,L,C of a cable is there and can be measured but in DBT can't be heard. In the graphs that I provided FR difference can't even be measured.
So again, can we use resistive spl matching? I have listened myself but it was not done blind and only with one loudspeaker so it does not qualify as any type of test.

 

[cschang]

I am not hell bent on it. It's just an issue and I am offering healthy arguments. If nobody did, we'd be comparing the sound of cables.

A 1000hz is where most inexpensive SPL meters are less flaky. Industry uses $600 calibrated SPL meters that most don't have. Also a 1000hz region is extremely critical to the character of the speaker in relation to the rest of the FR so I would think that any loudspeakers that are made with significant deeps or humps there will be highly colored.

Your arguments go against basic electrical theory. That does not make for a healthy argument. Again, did you read any of the links I provided?

But you agree that averaging and normalizing for sensitivity between 1khz-3khz is a better idea, right? If you wanted to just suggest a single a point, why did you also mention pink noise?

 

[R-Carpenter]

I suggested pink noise because it contains a spectrum of acoustic energy and will give a better chance to SPL matching. If you don't have a pink noise handy a 1000hz would be ok or if you want to be precise, use a few tones. As far as I remember, Rat Shack SPL meter has a deep around 3k and a peak at 6k. Most of the people will not run multiple tone signals. It requires time and everyone usually runs to CD player to listen to their favorite CD. The industry standard and a consumer testing loudspeaker are two different things.
In the set of measurements I presented, there was no measurable difference with or without different resisters in series. I'd like to try a loudspeaker with a difficult impedance and look at the results.
Basic electric theory tells you that a cable should sound different. LCR differences can be easily measured with a good DMM/LCR meter. Do you hear the difference in the cables? Nore do I.
Put it this way, if you can measure the difference, you cannot necessarily hear it. If you cant even measure it, oh well.
I'll find a volunteer and run an ABX. I'll need to design a particularly unfriendly crossover or just find a crappy speaker.

 

[cschang]

Put it this way, if you can measure the difference, you cannot necessarily hear it. If you cant even measure it, oh well.

I agree, but the whole purpose is to eliminate the possibilities out of the test. What is "good enough" for one person, may not be for another.