Print

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

[Matthias McCready]

I have been embarking upon a rabbit hole, which is testing speakers.

More to the point I have been tasked at work with going through roughly 8 installed PA systems and verifying that things are working as they should. I know some of the systems are not working as they should, so I wanted to create a more comprehensive way to examine the loudspeakers and to provide data up the chain of command, "hey these need to be taken down, and parts need to be replaced."

I ended doing a lot of research and creating a 17-page procedure manual for myself, which being a week or two into the project already needs a few revisions! 

---

I have 3 ways I am trying to test each loudspeaker:

Procedures 1 & 2, are both conducted at 3x the baffle distance for flown systems, with 1:1 resolution, via unplugging other boxes if needed. Procedure 1 is done at least 20dB above ambient SPL, and Procedure 2 is done at about 40dB less than the rated output of the box, at the measurement mic.

1) A Smaart TF
2) a REW THD measurement
3) An amp impedance measurement, I have been able to utilize this on my L'Acoustics Rig (Enclosure Check), but I have not been able to do this with my D&B systems, via the "Load Monitoring." The D&B systems do not have 1:1 resolution, which I am guessing has a lot to do with that.


----

Anyways getting on to my question:

What level of THD is acceptable?

I know this is probably a, "it depends" situation," as I presume THD is dependent upon the drivers, the design used, and volume tested; these are PA speakers after all!

Going into the project I had presumed from my reading that above 1% would be bad.

Once I began I realized just about everything I tested has above 1% above 8K or so, even new L'Acoustics X12's, which sound great!

It has been recommended to me to compare THD between models of the same type of speaker. So that is what I am trying to do, parsing through all of the measurements that I have taken so far.

---

This has been helpful, as in comparing D&B Q-Subs of three in one particular sound system, one of them sounded a little funky and did not have the output of the others. When I looked at that particular sub it had 126% distortion, compared to the 30% of the other two Q-Subs. So I believe that data there would be showing me the driver (or something internal) is ready for replacement.

When it comes to top boxes though things haven't been quite so clear for me, and I feel uncertain about what to make of my data. Some boxes have 10-20% in the HF and others have 30% or 70%. For those would all the HF drivers that are not matching the lowest, say 10% need replacement? I mean it is an OLD rig so this is conceivable to me.

----

Typically most boxes seem to have below 1% in the 200hz-5k range, although some models I have measure have a stereotypical 1-2% thing going on that is consistent box to box in the low mids (L'Acoustics A10). I did have a few boxes on my older systems where there was a spike of 4-6% in a low-mid frequency, which I would presume denotes a bad woofer, being that comparatively everything is under 1% in the measured range, for other loudspeakers of the same model.

----

Sorry for the long post, I am hoping one of you audio wizards can help me figure this data out.

Thanks!

Matthias

[John Roberts {JR}]

Not sure what your real task is, but a good idea for installed system maintenance to benchmark nominal (working) speaker measurements to compare to later.

Some smart install system amps can measure speaker current draw that can indicate gross failure, and sometimes "fixin to break".

Have fun...

JR 

[Riley Casey]

REW will make impedance measurements although you'd need to create a jig that interfaced with your amp rack outputs. For tracking speaker performance from a maintenance POV impedance covers a lot of bases.

[David Sturzenbecher]

I rely heavily on complex impedance for “speaker health” as quite lot can be gleaned. For this I ether use a Dayton Audio DATS system or an IAudioInterface.   This unfortunately will not alert you to a polarity issue in a biamped box (that is what I use a Smaart TF for). It is also largely a low voltage test and some issues won’t manifest itself until driven harder. For that I use Systune Pro with a LinearX VIbox, however they are no longer made.

Here are some sweeps from a recent system check.  It’s pretty obvious to spot major problems.  What is not obvious is what to do about smaller differences. These boxes were all in the air with no lift available so the issues were documented and that is about it.


Sent from my iPhone using Tapatalk Pro

[Matthias McCready]

Not sure what your real task is, but a good idea for installed system maintenance to benchmark nominal (working) speaker measurements to compare to later.

My goal is exactly as you describe, to have an annual process for going through every rig completely, and get data on every box. A health report if you will. As new systems are installed (we have 3 new PA's going up in the next year), I will be taking baseline measurements before boxes go up, which also helps to make verify that the boxes are wired in phase from the factory. 

Some smart install system amps can measure speaker current draw that can indicate gross failure, and sometimes "fixin to break".

All the systems are either D&B or L'Acoustics, both have this feature, however I believe 1:1 resolution is needed, which some of the systems do not have. So I will probably need to figure out something there (maybe using a single amp and channel and bringing that to each box).

REW will make impedance measurements although you'd need to create a jig that interfaced with your amp rack outputs. For tracking speaker performance from a maintenance POV impedance covers a lot of bases.

In your mind would these be superior to measuring distortion, and would getting some data here mean that distortion is not valuable? Or would you consider distortion and impedance to be two distinctive points of Data that could give a better picture?

I rely heavily on complex impedance for “speaker health” as quite lot can be gleaned. For this I ether use a Dayton Audio DATS system or an IAudioInterface.   This unfortunately will not alert you to a polarity issue in a biamped box (that is what I use a Smaart TF for). It is also largely a low voltage test and some issues won’t manifest itself until driven harder. For that I use Systune Pro with a LinearX VIbox, however they are no longer made.

Here are some sweeps from a recent system check.  It’s pretty obvious to spot major problems.  What is not obvious is what to do about smaller differences. These boxes were all in the air with no lift available so the issues were documented and that is about it.

This is helpful, those are some tools I don't have in the kit as of yet.

Thanks!

[Matthias McCready]

I am also still quite curious about measuring THD in speakers for several reasons:

1) I had a D&B C4 rig I mixed on for a little over a year. It was a fine rig, but it was designed for the room it was in, and the deployment was rubbish. I got a new rig, and the D&B C4 was fortunate enough to go to a better home, where its coverage pattern is close to perfect. When I went a listened, the system sounded just phenomenal, there was clarity I simply never had (at a similar listening distance). When I asked the installer what they done, he mentioned that all of the HF drivers were quite old and tired, and needed replacement.

2) I know a sine Wave can audibly reveal problems with a driver, so THD should provide a data point for this.

----

So for those of you measuring THD, what are you looking for in the data?

----

An additional but related question I would have, is should I be measuring my amp channels?

With modern amps (presuming the right DSP is put in) are they are either working or not? Or is this something that I should be measuring as a potential factor?

---

Thanks again for help with all of the questions!

Matthias

[Chris Grimshaw]

I'd say that modern amps are smart enough to know if they're working or not.

Old iron amps might have their class AB bias drift around a bit over the years, possibly resulting in crossover distortion (NB - nothing to do with speaker crossovers, it's a slightly "sticky" zero crossing on the amp output) if the bias runs too low and it ends up class B.


With regards to speaker health, I'd run a sine sweep at -10dB from maximum output, and see what you get. I'd look for:

- Frequency response within 3dB of original measurement across the band
- Harmonic distortion within certain limits. Maybe +/-50% of the original measurement. ie, original = 10% THD, current = max.15%.

I haven't done much/any research into how thoroughly-broken-in drivers behave compared to new, but I think the above guidelines are sensible with regards to deciding whether something needs replacement.

NB - Mic position is flexible, but must be consistent. Putting the mic up against the grille, for example, is okay, so long as you put it in the same place next time you do the measurements. As you move the mic further from the cabinet, other acoustic variables can contaminate the measurement.

Chris

PS - I'd also do a measurement at any bass reflex ports, which might help you spot internal rattles.

[Matthias McCready]

I'd say that modern amps are smart enough to know if they're working or not.

Old iron amps might have their class AB bias drift around a bit over the years, possibly resulting in crossover distortion (NB - nothing to do with speaker crossovers, it's a slightly "sticky" zero crossing on the amp output) if the bias runs too low and it ends up class B.

Point in case I had an L'Acoustics LA12X die on me two-weeks back, it let me know that it had "an internal short-circuit and would be powering off." 

Is it dangerous for me to assume that any differences in frequency and distortion (presuming everything is the same on the front end) is due to measurement position (room interaction) and the loudspeaker?

With regards to speaker health, I'd run a sine sweep at -10dB from maximum output, and see what you get. I'd look for:

- Frequency response within 3dB of original measurement across the band
- Harmonic distortion within certain limits. Maybe +/-50% of the original measurement. ie, original = 10% THD, current = max.15%.

NB - Mic position is flexible, but must be consistent. Putting the mic up against the grille, for example, is okay, so long as you put it in the same place next time you do the measurements. As you move the mic further from the cabinet, other acoustic variables can contaminate the measurement.

Chris

PS - I'd also do a measurement at any bass reflex ports, which might help you spot internal rattles.
[/quote]

So far I have been using the same mic positioning I have been for Smaart Traces; which is based on the baffle size of the speaker, and taken on axis of the loudspeaker.

----

So far it has been an interesting project.

1) In one system I found what appear to be many drivers that need replacement, in addition to boxes that were not designed to be arrayed together.
2) In another system, I found a few of the boxes had the HF horns that were out of polarity, and I don't think those boxes are biamped. This goes to show, how important it is, even for high-end brands, to test boxes BEFORE they get installed.

[Chris Grimshaw]

Is it dangerous for me to assume that any differences in frequency and distortion (presuming everything is the same on the front end) is due to measurement position (room interaction) and the loudspeaker?

I'd say that's a sensible assumption with modern amps.

So far I have been using the same mic positioning I have been for Smaart Traces; which is based on the baffle size of the speaker, and taken on axis of the loudspeaker.

Fair enough.

IMO, for health checks, you could put the mic to the grille and eliminate pretty much all acoustic effects. Sure, the frequency response curve will be weird, but as long as you stay consistent, it'll be easier to spot changes that are due to driver age, rather than someone-has-moved-a-nearby-flight-case.

Chris

[John Roberts {JR}]

2) In another system, I found a few of the boxes had the HF horns that were out of polarity, and I don't think those boxes are biamped. This goes to show, how important it is, even for high-end brands, to test boxes BEFORE they get installed.

You might want to check with the mfr to confirm the reversed polarity is in error. It might improve behavior in the crossover transition region.

JR

[Matthias McCready]

IMO, for health checks, you could put the mic to the grille and eliminate pretty much all acoustic effects. Sure, the frequency response curve will be weird, but as long as you stay consistent, it'll be easier to spot changes that are due to driver age, rather than someone-has-moved-a-nearby-flight-case.

Chris

These are good points. My measurement distance is loosely based of the Tracebook Project (Merlijn Van Veen & Nathan Lively). Their project gives a recommended distance, so that frequency response can even out. However the difference is that the loudspeakers I am measuring are staying rigged; I know this is not ideal and was not sure how well this would work, but for the most part I have not had too much ripple variance and the differences tend to be speaker differences, rather than the room from what I can tell.

You might want to check with the mfr to confirm the reversed polarity is in error. It might improve behavior in the crossover transition region.

JR

For that situation, there were 9 D&B 24S boxes. 7 of them had an identical phase trace, and 2 had an identical, but deviated phase trace that started around 1k, and then went about 180 degrees out of phase, comparatively. So two boxes are different, in the exact same way, than 7 other boxes.

[John Roberts {JR}]

For that situation, there were 9 D&B 24S boxes. 7 of them had an identical phase trace, and 2 had an identical, but deviated phase trace that started around 1k, and then went about 180 degrees out of phase, comparatively. So two boxes are different, in the exact same way, than 7 other boxes.

sounds like a mistake to me...

JR

[Riley Casey]

Measuring at the grill as already suggested is the correct solution for knowing what the speaker system but at the grill you no longer have a speaker system you have several speaker components that all need to be measured separately for useful results. Identify a location on axis with each driver exit and take a separate measurement for each driver. It doesn't matter that you are getting overlap from other drivers, what matters is that each set of measurements remains the same across all the speaker systems. In my experience its unlikely that mechanical speaker driver failures produce relatively low levels of harmonic distortion. A torn cone driver or diaphragm produces high levels of distortion very quickly. For isolating that kind of issue in a multiway flown speaker system simply running a sweep tone thru the box and listening to a mic on a fish pole on headphones would probably be most effective .  Its probably not terribly useful information to know that speaker seven in an array has THD of 20% when the others are at 5%. You want to know if its a woofer or a mid driver.

Point in case I had an L'Acoustics LA12X die on me two-weeks back, it let me know that it had "an internal short-circuit and would be powering off." 

Is it dangerous for me to assume that any differences in frequency and distortion (presuming everything is the same on the front end) is due to measurement position (room interaction) and the loudspeaker?

...

[Chris Grimshaw]

You want to know if its a woofer or a mid driver.

Riley,

See attached. REW gives harmonic distortion by component vs frequency. The two graphs are the same box at different power levels. SPLs are somewhere near correct.

Chris

[Matthias McCready]

Measuring at the grill as already suggested is the correct solution for knowing what the speaker system but at the grill you no longer have a speaker system you have several speaker components that all need to be measured separately for useful results. Identify a location on axis with each driver exit and take a separate measurement for each driver. It doesn't matter that you are getting overlap from other drivers, what matters is that each set of measurements remains the same across all the speaker systems. In my experience its unlikely that mechanical speaker driver failures produce relatively low levels of harmonic distortion. A torn cone driver or diaphragm produces high levels of distortion very quickly. For isolating that kind of issue in a multiway flown speaker system simply running a sweep tone thru the box and listening to a mic on a fish pole on headphones would probably be most effective .  Its probably not terribly useful information to know that speaker seven in an array has THD of 20% when the others are at 5%. You want to know if its a woofer or a mid driver.

These are great points.

In an ideal world, THD, probably should be done per driver, and field magnitude phase traces should at minimum be taken on the ground plane in the same exact spot, to lessen the impact of the room on the measurements (meaning the position of the mic/speaker in the room).

---

My specific goal for this project, is to perform annual inspections on multiple installed PA systems; we are adding two more rigs within the coming 6 months or so. So in this situation I am willing to forgo some accuracy for the sake of efficiency. If something shows up as funky, then I can investigate further (measure individual drivers, pull the box down, go through the box etc). So as long as my THD data is good enough to show me that a driver in the array box 7 is bad (which THD by frequency would give a good hint), I don't necessarily need to immediately know which one is bad.

[Jim McKeveny]

I've been aware for years of how tolerable percents of distortion are at low frequencies. What I wasn't aware of was the increase the THD on HF drivers - all compression drivers apparently. I have often half-jested ' I never met a compression driver I liked". This spittiness at HF/VHF is an ubiquitous characteristic of so much "PA sound". I would gladly trade off some of the extended high-end frequency response chase for lower distortion overall (if that is part of the compromise).

[Tim McCulloch]

I've been aware for years of how tolerable percents of distortion are at low frequencies. What I wasn't aware of was the increase the THD on HF drivers - all compression drivers apparently. I have often half-jested ' I never met a compression driver I liked". This spittiness at HF/VHF is an ubiquitous characteristic of so much "PA sound". I would gladly trade off some of the extended high-end frequency response chase for lower distortion overall (if that is part of the compromise).

I'd give up 8k-16k if it made 2k-8k sound better.

[Brian Jojade]

The actual numbers you come up with don't really matter much.  What you're testing against is change since the last test.

Testing with the measurement mic in front of each component of the speaker should give similar results across speakers and across tests.  If you find one speaker that's giving severely outlier results, that speaker would be suspect during the initial testing.

Now, next time you go to test, just compare against the saved data from the exact test from the exact speaker.  Is it the same?  If so, then you're fine.  If it has changed, determine if the change is from something you did different in testing, or if the change is minor enough that it's still within the range of the other tests you've made with the rest of the speakers.  Anything outside that range now is suspect for further inspection.

Remember that the goal of this test is to just find faults, so you don't need to measure every possible spec and aspect, since faults likely would disrupt many different things all at once.

[ProSoundWeb Community]