Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Art Welter]

Hi Art,

I'm using two 8 ohm.  An impedance curve is attached, however it's in-room next to other subs....

The "SPL Leq method of finding sensitivity" is very simply.....put processing in place (as per blue trace posted)....simultaneously meter SPL LEQ and average RMS voltage. 
Do the math and bring it to 2.83v at 1m.

An aside.....I really, .....really, want to hear what's wrong, or more fair, than that method for a sub sensitivity spec......

Mark,

Thanks for the explanation, had to look up the meaning of "Leq", so I re-learned an acronym I'd forgot...

Traditionally, a sensitivity rating was considered to be 1 watt at one meter, and the manufacturer would use the appropriate drive voltage for the nominal impedance of the speaker- 2 volts for a nominal 4 ohm, 2.83v for 8, 4v for 16 ohm, etc.
BMS, JBL, EAW still are traditional in that sense, others require math, using 2.83 volts for 4 ohm or even 2 ohm speakers, six dB more "sensitive"!

Without an impedance graph, "sensitivity" is ambiguous- your cabinet could have been anything from a nominal 2 to 16 ohms given it is available in 4 or 8 ohms. If you are going to the effort to "do the math and bring it to 2.83v at 1m" I'd prefer you'd do the math for "1 watt at one meter" so I don't have to do more math converting...

With your explanation and the impedance graph, it appears your sensitivity is more or less "fair".

That said, how "average RMS voltage" is determined- lowest and highest voltage divided by 2, or logarithmic average, the frequency range the average is done over all introduce variability and potentially ambiguous results compared to "put in X volts and graph what happens".

I do appreciate your work!

Thanks,
Art

[Mark Wilkinson]

I feel that loudspeakers should be measured (at least as far as sensitivity and response) with NO processing of any type.

Just the measurement system into the amp, and to the speaker.

It gives a much better indication of the natural raw capability of the loudspeaker.

Anybody can add eq to bump up a part of the response to tame the peaks as needed.

Assuming the generator is flat, and is using a swept sine wave, you can easily measure the applied voltage.  2.83V is suggested.

I never measure at 1m, it is to close to get numbers that translate to other distances.  It is better to get several meters away.  I prefer to increase the drive voltage at the same dB as the distance takes away.  10m=20dB is a easy one to remember and the voltage increase is simply 10x more.

Another advantage to this method is that any power compression will be present, which gives an idea of real world numbers in usage.

Thx for the reply Ivan,
I agree with much of what you say, particularly using voltage as a reference and not power. 
Also measuring at a distance, with more voltage..... 28.3v at 10m makes a ton of sense, for the reasons you give. 

But I feel we need to see both raw and processed curves to know what we can expect. 

I also feel measurements are better than judgement calls for defining specs.

With processing in place we get to see manufacturer claims against intended usage.
Let the HPF and LPF used be stated.  Let any eq be stated.

If they claim an unrealistic low freq response, and use too low a HPF, the efficiency spec will suffer greatly using the method I prescribed.
If they use an unrealistic high response, and use an obviously too high LPF, their game playing sticks out.
If eq is used to raise low areas of response, it will lower measured efficiency, as in real world.
If eq is used to lower peak areas, it will also lower efficiency, as in real world.
(Both of those uses of eq are appropriate processing IMO)

Personally, I try to flatten sub response within its intended passband, and then compare that to other subs similarly set up.
I mean, in the end, I care more about processed performance than raw.
And I have a simple intuitive way of measuring those comparisons....works for me anyway


 

[Mark Wilkinson]

Mark,

Thanks for the explanation, had to look up the meaning of "Leq", so I re-learned an acronym I'd forgot...

Traditionally, a sensitivity rating was considered to be 1 watt at one meter, and the manufacturer would use the appropriate drive voltage for the nominal impedance of the speaker- 2 volts for a nominal 4 ohm, 2.83v for 8, 4v for 16 ohm, etc.
BMS, JBL, EAW still are traditional in that sense, others require math, using 2.83 volts for 4 ohm or even 2 ohm speakers, six dB more "sensitive"!

Without an impedance graph, "sensitivity" is ambiguous- your cabinet could have been anything from a nominal 2 to 16 ohms given it is available in 4 or 8 ohms. If you are going to the effort to "do the math and bring it to 2.83v at 1m" I'd prefer you'd do the math for "1 watt at one meter" so I don't have to do more math converting...

With your explanation and the impedance graph, it appears your sensitivity is more or less "fair".

That said, how "average RMS voltage" is determined- lowest and highest voltage divided by 2, or logarithmic average, the frequency range the average is done over all introduce variability and potentially ambiguous results compared to "put in X volts and graph what happens".

I do appreciate your work!

Thanks,
Art

Hi Art,  yeah, what a tower of confusion traditional measures of efficiency are, using power against nominal impedance.
I agree, an impedance graph is a must.

I do like sensitivity referenced to voltage not power,
in a way because it almost forces us to do a little math and give impedance  some thought.
Nominal's  meaning is so dang limited....IMHO.

You know, average impedance could be measured though. 
Pretty dang easy really,..... over the same time  period that SPL LEQ and averaged rms voltage are taken, also take an average current reading.

A little Ohm's law and we get measured average impedance, to go with measured SPL sensitivity against voltage.
Heck,  we even get a real 1 watt spec.........

To hell with tradition !!

[Chris Grimshaw]

<snip>

REW has a stepped-sine harmonic distortion routine that is really cool. 
It spends a couple of seconds at each sine step to determine distortion.
You set how many sine steps per octave the neighborhood can stand     I used 12.

<snip>

Last plot is a transfer for the push-push.  Dark green is raw, blue is processed.
My SPL Leq method of finding sensitivity gives 101.3 dB 2.83v at 1m, with the processing in place.
THD hit 10% right at 130dB 1m.

Happy camper!!

Hey Mark,

Just wondering if you can share how to set up REW to do this testing routine. I've gone through the help file and can't find much.

130dB at 10% THD is excellent.

Chris

[Ivan Beaver]

Hey Mark,

Just wondering if you can share how to set up REW to do this testing routine. I've gone through the help file and can't find much.

130dB at 10% THD is excellent.

Chris

Here is a good document on how to do it.

It is real easy and quick.  But be SURE to calibrate the mic to get useful results.

http://www.brentbutterworth.com/cea-2010-measurement-manual.html

[Mark Wilkinson]

Hey Mark,

Just wondering if you can share how to set up REW to do this testing routine. I've gone through the help file and can't find much.

130dB at 10% THD is excellent.

Chris

Hi Chris,

First, as you've probably already discovered, that's a great article Ivan linked,.... full of real world instruction. 
It also made me realize REW is set up to do CEA-2010, in addition to stepped sine.

As far as stepped-sine....I couldn't find much in the REW help file, or on the support forums either.
I spent 2 days wondering how to even get the thing to work, because after making a stepped sweep, the RTA page housing the stepped sine initiation, and the output graph,.... didn't show any harmonics ?
Then, I finally noticed, every time you run a stepped sine, it adds a new measurement on the main REW page, just as if you ran a normal measurement. 
That's all there is to it !!! I was such a dope lol.

Do pay attention to FFT length per the help files if you're trying to measure way low.  And averages if you have a noisy environment.
The only other caveat i can think of right now, is that I kept blowing my circuit breaker trying to test around Fb, at high volume.
I had to move the mic closer to turn down the juice to a PL340 running bridge. 

I'm going to try CEA-2010 next time out....it has to be safer than stepped sine, since it only runs 6.5 cycle bursts. 
Will be interesting to see how they compare

[Mark Wilkinson]

If you are concerned about it, you can apply about 0.6 VDC of the correct polarity to them while stored, and let the voice coil carry the weight.  1/2 lb is about 2.2 newtons.  With BL of 25 T-m, 0.1 A will carry the weight.  Power is trivial at (0.1 A)^2 * 5.6 ohm = about 60 mW, so heating is a non-issue.

You could make up a storage harness hooked to a wall wart.  You can run it all year for a pair of PPSL subs (4 drivers) for about 30 cents of electricity.  It is cheap insurance for $2800 worth of drivers and you don't have to scrape up the top of your cabinets.

Marc

P.S.  Never mind.  Seemed like a good idea until I thought about the internal wiring.  The upper driver needs to be pulled into the cabinet and the lower needs to be pushed out of the cabinet.  That is the opposite of how they are wired for audio output.

I've ended up settling on a horizontal push-push box (clamshell) as per pict below,
and I think Marc's idea to stop cone sag has become very easily doable.

If I haven't screwed up, a single 4-pole double-throw switch, can allow speakon connection on one side of the switch, and low voltage DC connection on the other. 
I'm going to use a mono TRS jack to feed 1.6vdc to the 2 drivers in series.
Parts Express has a nifty cheap 12vdc to 1.6v converter, that will allow plug into a power pack or trailer harness. 
Circuit is below....if anyone spots an error or 'what if' problem, thanks for pointing it out.

I measured cone sag, horizontal vs no-sag vertical, and it's about 1mm. 
I put the 1.6vdc on the two drivers (0.8v each) it looks like the 1mm is fairly exactly offset, but I can't measure as close as I'd like.
Marc, you recommended 0.6v....I'm hoping 0.8 is close enough that I'm not making a small 'reverse sag problem' .  The mass is supposed to be 267 grams.....thx.

[Mark Wilkinson]

Here is a good document on how to do it.

It is real easy and quick.  But be SURE to calibrate the mic to get useful results.

http://www.brentbutterworth.com/cea-2010-measurement-manual.html

Thanks again for this link.... yes, it really is easy and quick !!
REW makes both CEA-2010 and stepped sine measurements so simple. 

The CEA-2010 test, which uses only 6.5 cycle bursts at 1/3 octave points,  seems to me to be about measuring distortion on peak transients.....
whereas stepped sine (as posted earlier) seems more about measuring distortion at high continuous SPL levels.

Anyway, I hauled one of the push-push double 18"s out into the grass today.
Below are the SPLs @ 2m that passed CEA.  These SPLs are as read by REW, and are not adjusted to 1m.
 
Clearly, if +6db is added to adjust those 2m SPLs to 1m, the levels are peak, plain and simple.
I'd call them peak RMS I guess...


The high CEA numbers don't surprise me a bit. 
The pair I've built so far, are a real joy to listen to.

The "anti-sag" hook up appears to be working great, so no concerns there.... 
A standard powerpack with a 12v output looks like it would hold both subs against gravity for about a month before needing recharging.

I've added a pict below of a finished sub, with the modular ground stack top boxes been making.

[Don T. Williams]

Thanks for sharing.  It looks great!

[Mark Wilkinson]

Thanks for sharing.  It looks great!

Thanks for the kind words Don

One thing I'd like to add about the CEA test, is that I think it might be a good help in sizing an amp to a sub.
I first ran the test off a single channel of a PL380.  When I hit the end of the amps output at 80Hz and up, CEA was still showing pass.
So I switched the amp to bridge and ran again.  Fail levels came within trying for only another dB on all three 1/3 oct centers, 80, 100, and 125Hz.....

The 6.5 cycle burst is modulated as below, so I'm thinking it probably represents the shortest peak worth trying to have enough amp for..??
I can't see why I would want a bigger amp than a single channel PL380, even for transients.  (Drivers are 8 ohm 1500w AES in parallel.)

Now I have no idea what the weighting formula is for pass/fail, or how applicable it is for what we hear at high sustained levels,
but it was easy to see that pushing the slider up was not increasing SPL linearly...so compression and distortion were appearing in sync.

What I really like about the short pulses, is they are much less scary to push to the edge, to listen for distortion and compression, than any test signal I've found. And you can test a sub at any specific freq.  REW can generate the bursts at any freq, not just 1/3 octave centers.
I can see using them just for listening tests...dialing in HPF filters..setting peak limiters...etc.

[ProSoundWeb Community]