measuring my P/A speakers

[Arbr4]

Hi all,

I'm the sound guy of a 7 piece rock and pop (cover)band, and I tow my system from gig to gig.
I have several sets for smal gigs and bigger gigs.
For smaller stages I use active RCF and HK Audio stuff alongside a digital Soundcraft UI mixer.
The main (FOH-ish) portable system for the smaller gigs consists of two RCF AS8003MK3 subs (18") and two HK Audio Linear 5 MK2 112FA.
The RCF's do a fine job delivering the rock punch and the HK's have a nice long throw (60x40) and have a fine DSP that's fully controllable.
With some tuning this is a good sounding combo!

However:
On the so called HK audio pro users forum I had some discussion on how to setup the system and how the speakers are factory tuned.
For every gig I measure the phase en frequency response (sometimes with ARTA, sometimes REW) and tune it "flat" to make the signal more predictable and more feedback proof.
With most speakers I've used, the standard frequency response is more ore less "flat"... I've used most common active speakers for the smaller gigs: Yamaha DZR, Electrovoice ETX / EKX, Martin, Adamson, Nexo, RCF NX /TT, FBT, JBL etc... If I measure nearfield (1m on axis) the frequency response is mostly flat(ish) with al these speakers.

Only the HK audio boxes (in this case the Linear 5 MK2 112FA) are different; the factory preset they call flat has a very strong " smiley curve".
There is a strong boost towards 10-12khz (+ 8dB) and towards the 100hz region: +8dB. (see attached files). The measurements on the spinorama site give about the same image.

The HK audio representative says that HK tunes the speakers towards the Fletcher-munson curves... They tune their factory presets on "how our hearing works" he says.
The strange thing that occurs is that if I tune the speakers "flat" with the HK DSP control, and compare them to other comparable speakers, high and low seem to be "off" comparing to the nearest competitors; It's like the FIR filter isn't done correctly and delay between high and mid/low isn't correct (high seems to be to far behind the low/mid) and they have tried to compensate by boosting frequencies to get some decent sound image compared to their competitors.
By asking the HK audio guy, he says that my measurements aren't good and that the (HK) have enough experience tuning and measuring speakers.

My questions:
1.How should a factory preset called "flat" should look / sound like? (what should be a good starting point for a P/A speaker out of the box?).
2. Does it seem to you that HK audio has made a design error?

just asking

 

[Sokel]

Hi all,

I'm the sound guy of a 7 piece rock and pop (cover)band, and I tow my system from gig to gig.
I have several sets for smal gigs and bigger gigs.
For smaller stages I use active RCF and HK Audio stuff alongside a digital Soundcraft UI mixer.
The main (FOH-ish) portable system for the smaller gigs consists of two RCF AS8003MK3 subs (18") and two HK Audio Linear 5 MK2 112FA.
The RCF's do a fine job delivering the rock punch and the HK's have a nice long throw (60x40) and have a fine DSP that's fully controllable.
With some tuning this is a good sounding combo!

However:
On the so called HK audio pro users forum I had some discussion on how to setup the system and how the speakers are factory tuned.
For every gig I measure the phase en frequency response (sometimes with ARTA, sometimes REW) and tune it "flat" to make the signal more predictable and more feedback proof.
With most speakers I've used, the standard frequency response is more ore less "flat"... I've used most common active speakers for the smaller gigs: Yamaha DZR, Electrovoice ETX / EKX, Martin, Adamson, Nexo, RCF NX /TT, FBT, JBL etc... If I measure nearfield (1m on axis) the frequency response is mostly flat(ish) with al these speakers.

Only the HK audio boxes (in this case the Linear 5 MK2 112FA) are different; the factory preset they call flat has a very strong " smiley curve".
There is a strong boost towards 10-12khz (+ 8dB) and towards the 100hz region: +8dB. (see attached files). The measurements on the spinorama site give about the same image.

The HK audio representative says that HK tunes the speakers towards the Fletcher-munson curves... They tune their factory presets on "how our hearing works" he says.
The strange thing that occurs is that if I tune the speakers "flat" with the HK DSP control, and compare them to other comparable speakers, high and low seem to be "off" comparing to the nearest competitors; It's like the FIR filter isn't done correctly and delay between high and mid/low isn't correct (high seems to be to far behind the low/mid) and they have tried to compensate by boosting frequencies to get some decent sound image compared to their competitors.
By asking the HK audio guy, he says that my measurements aren't good and that the (HK) have enough experience tuning and measuring speakers.

My questions:
1.How should a factory preset called "flat" should look / sound like? (what should be a good starting point for a P/A speaker out of the box?).
2. Does it seem to you that HK audio has made a design error?

just asking

In hi-fi terms neither is flat (except the measurement is at one meter, then the corrected on could be flat)

At some distance though (which I presume would be far-field) the stock response could end up being flat.

 

[dasdoing]

The HK audio representative says that HK tunes the speakers towards the Fletcher-munson curves... They tune their factory presets on "how our hearing works" he says.

that's obviously BS, but

If I measure nearfield (1m on axis

this might be too near for this speaker.....it MIGHT be the reason for the smiley curve, which makes no sense for a PA speaker

 

[Sokel]

this might be too near for this speaker.....it MIGHT be the reason for the smiley curve, which makes no sense for a PA speaker

I thought so too but it's a very small speaker after I checked.
Maybe...

 

[ThatSoundsGood]

Where did you measure from? How loud did you measure? What are you measuring with? PA speakers don't generally measure flat from 1 meter on-axis. They're intended for far field listening which is a different animal.
My hunch is that they change as they get louder......they will get flatter as they get louder. So when you measure them, you see the smiley face/fletcher Munson curve but as they get louder, the mid range increases.

 

[DVDdoug]

Yeah... Are you measuring/listening at the main or "average" listening position? If the frequency response is EQ'd to be similar the sound should be similar but it's going to sound different in different venues and at different listening positions... "similar" not identical...

Are you measuring at every venue? I don't think that's often done but I could be wrong.

If it's a "typical" band that's not playing huge venues there is probably a LOT of sound coming directly from the stage (drums & amps, etc.) and you don't have control over that.

P.A. speakers usually aren't going to sound like hi-fi speakers or studio monitors. They shouldn't sound bad but there are different compromises, specifically sensitivity/efficiency. They are also usually made to be lightweight and rugged. Usually the subs roll off higher because they can be made to go louder to to fill a large venue with bass strong enough to feel in your body.

But overall a good PA in a good large space (or outdoors) is going to sound better than the typical home system. Loudness and the natural reverb (not outdoors) adds a lot! And you don't get the "small room" problems.

 

[staticV3]

1.How should a factory preset called "flat" should look / sound like?

Signal in = signal out.

PA engineers should be expected to know about Fletcher munson, indoor vs outdoor house curves, and frequency response in general and should be expected to be capable of choosing and implementing an appropriate house curve.

A speaker whose stock response is anything but signal in=signal out, adds uncertainty and impedes that process.

 

[ThatSoundsGood]

Signal in = signal out.

PA engineers should be expected to know about frequency response and to be capable of choosing and implementing an appropriate house curve.

Agreed

A speaker whose stock response is anything but signal in=signal out, adds confusion and impedes that process.

How would you define this though? Most of these speakers are active and have a curve on them that doesn't measure flat from 1 meter on-axis because they're meant for far field listening.

 

[staticV3]

How would you define this though? Most of these speakers are active and have a curve on them that doesn't measure flat from 1 meter on-axis because they're meant for far field listening.

The designers ought to know from which distance their speaker has the intended driver summing, so just adhere to the specified reference axis and minimum listening distance when measuring and all's good.

Worst case: simply measure at increasing distances until the readings stabilize

 

[ThatSoundsGood]

The designers ought to know from which distance their speaker has the intended driver summing, so just adhere to the specified reference axis and minimum listening distance and all's good.

That's fair. But that's why I want to know how he was measuring these. I also would want to know what's going on in the processing as they get louder. Does compression (driver or DSP based) set in and flatten the curve as it gets louder? Sure, we could say that the engineer should be doing that but maybe that's their niche as a speaker manufacturer. Just an idea on that. I'm not sure.

 

[Arbr4]

That's fair. But that's why I want to know how he was measuring these. I also would want to know what's going on in the processing as they get louder. Does compression (driver or DSP based) set in and flatten the curve as it gets louder? Sure, we could say that the engineer should be doing that but maybe that's their niche as a speaker manufacturer. Just an idea on that. I'm not sure.

I've measured these on-axis (between mid en high driver, aimed with my laser to be shure the mic was on the correct hight!). Measured at 1m, 2m and 5m distance (the REW screenshots are from the 1 meter measurement). I've used my calibrated Isemcon on REW on my MacBook, nothing in between. The curve didn't flatten out while going from 1 tot 5 meters, the accuracy of the measurement did however... Compression / limiting of the DSP was off while measuring, and when pushed towards 90% level only the distortion level in low mid en mid went sky-high..... the frequency curve did not change. Off course I realize that P/A speakers don't measure ruler flat like Hi-Fi or studio monitors. but most of my other P/A speakers measure within reasonable margin " flattish"... my point is that there is a factory preset on these Speakers (called Linear 5!!!) that's named "flat" and it's not! the discussion of how a P/A speakers should measure for a good reference is very interesting! IMHO, a P/A speaker should measure "flat" from 1 to 5m on axis out of the box (near/midfield) so it's predictable and reliable and easy to setup. Maybe not important, but when mixing a band (random band on a random stage) you LPF/HPF, EQ, gate, compress, limit, FX etc... each channel individually; there you make the mix. Setting up the system, before the soundchecks, starts with Arta / pink noise en some sweeps for time delay/phase... also you want to disable the feedback: if a speaker has it;s own Curve / boosted highs and mids, then Feedback suppression is a big pain in the ass! if you mixing several bands in one night on the same stage, and someone brings on their own soundguy and plug-in the Yamaha DM7 or DiGiCo mixer, the whole preset is instant unusable due to the curved FR!

 

[ThatSoundsGood]

I've measured these on-axis (between mid en high driver, aimed with my laser to be shure the mic was on the correct hight!). Measured at 1m, 2m and 5m distance (the REW screenshots are from the 1 meter measurement). I've used my calibrated Isemcon on REW on my MacBook, nothing in between. The curve didn't flatten out while going from 1 tot 5 meters, the accuracy of the measurement did however... Compression / limiting of the DSP was off while measuring, and when pushed towards 90% level only the distortion level in low mid en mid went sky-high..... the frequency curve did not change. Off course I realize that P/A speakers don't measure ruler flat like Hi-Fi or studio monitors. but most of my other P/A speakers measure within reasonable margin " flattish"... my point is that there is a factory preset on these Speakers (called Linear 5!!!) that's named "flat" and it's not! the discussion of how a P/A speakers should measure for a good reference is very interesting! IMHO, a P/A speaker should measure "flat" from 1 to 5m on axis out of the box (near/midfield) so it's predictable and reliable and easy to setup. Maybe not important, but when mixing a band (random band on a random stage) you LPF/HPF, EQ, gate, compress, limit, FX etc... each channel individually; there you make the mix. Setting up the system, before the soundchecks, starts with Arta / pink noise en some sweeps for time delay/phase... also you want to disable the feedback: if a speaker has it;s own Curve / boosted highs and mids, then Feedback suppression is a big pain in the ass! if you mixing several bands in one night on the same stage, and someone brings on their own soundguy and plug-in the Yamaha DM7 or DiGiCo mixer, the whole preset is instant unusable due to the curved FR!

I apologize if I sounded like I was questioning your methodology or your knowledge on this. I really wasn't. I just wanted the details and you seem to already have a comprehensive answer, which is that the those speakers are not flat when they say that they are, especially if they measure like that at every distance. I agree that this is unsatisfying and I completely agree with the problems you have pointed out with this. It's kinda silly that they're doing this. I guess they could say that those frequencies are less sensitive to the ears due to the Fletcher Munson Curve (or whatever the name of it is). I do know a well seasoned FOH guy who would EQ his PA with a dynamic EQ to control this as there PA gets louder.
So my question is this:
How do they sound when they're quiet vs. when they're loud? How are they to mix into? Are you flattening their weird curve or attempting to mix into it?

 

[Arbr4]

I apologize if I sounded like I was questioning your methodology or your knowledge on this. I really wasn't. I just wanted the details and you seem to already have a comprehensive answer, which is that the those speakers are not flat when they say that they are, especially if they measure like that at every distance. I agree that this is unsatisfying and I completely agree with the problems you have pointed out with this. It's kinda silly that they're doing this. I guess they could say that those frequencies are less sensitive to the ears due to the Fletcher Munson Curve (or whatever the name of it is). I do know a well seasoned FOH guy who would EQ his PA with a dynamic EQ to control this as there PA gets louder.
So my question is this:
How do they sound when they're quiet vs. when they're loud? How are they to mix into? Are you flattening their weird curve or attempting to mix into it?

Didn't mean to be "pedantic" (English is not my native language)... , I really appreciate all the good tips, tricks and feedback. The human hearing curve is the most unreliable reference there is! If HK audio did, by purpose, design this Frequency Response to that Fletcher/Munson Curve, they should fire their R&D department! Anyways, till the time HK Audio does make some good reference presets, everyone is stuck with these ones. What's really good on the Linear 5 MK2's is their DSP control software and their long throw architecture... the 60x40 degrees dispersion on the HF compression driver really gets your sound there where you want it, and the Sica low/mid driver gets along with this in audio quality. These are definitely some strong points for my purposes. What I did is tune the signal from the basic flat prest to "really flat" with the DSP control, with some good measurement tools it's easy to do. When tuned "flat" (near field 1 - 2 m on axis)... The time alignment doesn't really fit... it's like the HF is behind het mids and articulation and audio around X-over (1600hz) isn't on spot... That's why it seems the FIR filtering on the HF isn't done correctly. BTW I use the main EQ / PEQ on my mixer to tune the sound on locations... but that's only finetuning and Feedback elimination.

 

[Salt]

If I understand correctly, there are different presets. How do these perform?

 

[Arbr4]

If I understand correctly, there are different presets. How do these perform?

There are 4 presets. 1. Flat, 2. Boost, 3.Low cut, 4. Custom. these are selectable with a button on the back of the speakers on the plate amp. attached all the REW measurements including the overlays. The custom preset can be build with the HK Audio DSP control software. Flat is the out of the box preset, Boost gives a big boost in the 60-100hz area, low-cut gives the opposite from boost. the Boost preset isn't Usable at all given the very high distortion it delivers... the Sica 12" mid/low driver has just some 3-4mm Xmax. mind u: these are €1200 speakers! (about 1400 US dollars I guess)... So these are definitely not some cheap DJ speakers IMHO; I'd expect something more sophisticated tuning and better design for use in the Pro- audio range.

 

[Arbr4]

The only thing I could imagine for this standard smiley curve is a better speech intelligibility.... D50 is the only really straight line I measure (REW - clarity graph)... on the other hand the other clarity graphs give an overly bright picture.

 

[dasdoing]

if we are talking about longer distances, the HF boost helps, cause it will attenuate stronger over distance.
The LF boost makes zero sense in a PA speaker

 

[Salt]

Have you measured at different volumes?
The smiley curve smells a bit like loudness curve and should then flatten with increasing volume. That's where Fletcher Munson comes into play.

 

[Arbr4]

Hi Salt, I did measure at different volumes. FR curve doesn't change. To me it seems a little strange that by increasing volume, the FR should change. if this was the case (but it isn't) it should be done by the DSP measuring Volume and comparing it to FR; The DSP doesn't active measure real life volume nor does it measure FR or distortion. Otherwise it should be by compression and distortion; if this was the case the summed signal in mid (clean signal + distortion) should even out the HF and LF boost... I would say that's not a good thing...

 

[dasdoing]

I never heard of speakers changing the FR when changing volume. I think people here fell into a mandala effect, confusing it with the loudness compensation feature present in some special DACs and software.