Thread: Voicing/tuning a 3-way home brew speaker cab

So what is a good strategy for creating the preset for a pa cab? Is there a best practice for this sort of thing or is it just what works for you.

I've got a couple of trap cabs I'd like to use for a general rock-n-roll beater pa. 2 JBL 2226H 15"s and 1 RCF N850 2" driver on some mystery 90x40 horn. The 15s are separated - the top one sealed and the bottom one ported. Cabs would sit on top of some dual 18's - so 100hz on up should work. Boxes will be powered by 2 Xti 2ks and 1 Xti 4k.

I have some measuring tools - REW and HolmImpulse. So far I came up with some presets that take into account the frequency response, roll offs, and sensitivities of the drivers. Played around with driver delays and lining up the impulse response peaks - but results are disappointing and inconsistent. I suspect I do not have the drivers playing well together yet.

I think where I'm lost is what order to make settings and take measurements, measurement mic placement, and what to shoot for in a voice? Flat on the display? Sounds neutral with mic/voice? Indoors or out? Near field or far? Low volume or high? Ignore ground bounce or use ground plane measurements? Average of all?

I suppose I could grab a 58 (or even a flatter mic) and talk and tweak til it sounds decent/neutral across the listening area and call it a day but with perhaps some pointers and a fresh start............

"To tune one loudspeaker cabinet, first you must tune 1,000."
-- Confucius

Or something like that. Seriously, though, unless you're using a tool that lets you look at phase you're going to have a lot of confusing results. There could be 10 things wrong with your tuning that are simple and easy to fix that an RTA will not show you.

For the voicing part, my advice is to not try and do it all in one shot. I find when I spend too much time listening to the same tracks on the same speakers for a marathon session my ears adjust to the sound of the cabinet. When I come back later it might sound awful once I've had time to clear my head. Do it a little at a time, and it helps to have a known "good" speaker or set of headphones so you can have a reference and keep your head straight.

Originally Posted by Bennett Prescott

"To tune one loudspeaker cabinet, first you must tune 1,000."
-- Confucius

Or something like that. Seriously, though, unless you're using a tool that lets you look at phase you're going to have a lot of confusing results. There could be 10 things wrong with your tuning that are simple and easy to fix that an RTA will not show you.

LOL.

R.E.W. and HolmImpulse are both programs that do let me see phase, frequency response, and impulse response of whatever I'm measuring - along with other parameters. It's probably my methods and reading of the data that's messing me up.

Hi John-

I'm not familiar with REW, and I downloaded (but never installed) Holm Impulse. I've been a Smaart user for a few years and I've made hundreds of bad measurements. I still do, but I can usually spot them now. At any rate, I think Bennett assumed that Holm Impulse only did "impulse alignment".

What all the cool kids do these days is attempt to get consistent summation, both in amplitude and pattern, throughout the acoustic crossover region.

That said, all your LF measurements should be ground plane. That might mean laying a trap box on it's side and blocking it up so the front is 90° to the "ground." Likewise, your measurement mic should be as close to the ground plane as practical. Your measuring distance should be at least 3x the longest (diagonal) dimension of the enclosure.

The different loading of the 2 15" speakers will create an interesting acoustic crossover between them and I suspect this is part of what doesn't work well with an impulse response alignment. I'm also presuming you are driving both 15" from the same amp channel, which may (or may not) be possible depending on what you figure out about their behavior.

I suggest you drive and measure each 15" separately, using the same stimulus. Each speaker gets its own amp channel (at least for now). If Holm will let you save traces to the screen, you'll want to save both the magnitude and phase traces AND the impulse response. The reason for the impulse response is to look at group delay; it will be different for each loudspeaker loading. Next power the 15"s from separate amp channels. Bring up the first channel and look at the magnitude and phase, then bring up the other amp channel and see what happens and at what relative levels. Now experiment with delaying the earliest-arriving 15" to match the later speaker. What happens to the phase trace where the energy contribution of both speakers is within 6dB of equal? What do your ears tell you as you try these various evaluations?

Once you have the 15s playing nice with each other, we can deal with getting the uppermost 15 to play nice with Mr. Horn.

One thing you HAVE to ask yourself "Do I want the loudspeaker to be accurate or to "sound good"?"

Very often these can be completely different.

If you try to voice it to "sound good"-then be sure to use a WIDE RANGE of different types of music. You will hear all sorts of different production "sounds"-specifically on the low end.

Which one is "right"? Hard to say.

Personally I go for accuracy-letting the user "voice" it to their tastes. I try to provide as "clean" a pallett as I can upon which to "paint".

You also need to look at different mic positions in different parts of the coverage patterns.

You also have to consider things like "sounds neutral with a mic".. What mic? What mic position-is proximity considered? How do you know what "neutral" sounds like?

Maybe it would help if you would post some of the measurements you have made-specifically showing phase-and of course amplitude.

This NOT a simple process. Some designs are a bit easier than others.

Thanks Tim and Ivan for tips and input................I posted an update reply with more info.......but it seems to have vanished????????? Will retype later.

The best thing to do, is to simply use it for shows and continually make small improvements. Over time you will work out the bugs. Software works great IF you know how to use it, and IF you have the time and space to set up the test.

Just using these things in anger will allow you to tweak them until they work for you, whether you use software or not.

Originally Posted by Tim Weaver

The best thing to do, is to simply use it for shows and continually make small improvements. Over time you will work out the bugs. Software works great IF you know how to use it, and IF you have the time and space to set up the test.

Just using these things in anger will allow you to tweak them until they work for you, whether you use software or not.

I find it doubtful that anyone is going to be fixing things like crossover slopes, phase alignments, polar patterns and overlaps, or anything else other than basic EQ on-the-fly at a show.

As I've mentioned in other threads, and how Bennett so nicely put it, building a tuning for a speaker cab is no easy task. Think about how many manufacturers get it wrong (Vertec is up to V5!!!!).

Don't forget the obvious task of reading what the manufacturer of the driver(s) has to say about cabinet tuning and operating ranges / crossovers for their drivers. It's usually a pretty good starting point!

Originally Posted by Silas Pradetto

I find it doubtful that anyone is going to be fixing things like crossover slopes, phase alignments, polar patterns and overlaps, or anything else other than basic EQ on-the-fly at a show.

As I've mentioned in other threads, and how Bennett so nicely put it, building a tuning for a speaker cab is no easy task. Think about how many manufacturers get it wrong (Vertec is up to V5!!!!).

I don't do it DURING the show, but before while setting up the PA I'll play with crossovers and inter-driver delay times to hone in on a setup. This is on low budget stuff, like the OP has. We're not talking any type of big show here.

I'll tweak one thing and test it for that show. If it's better I'll leave it, but if it's worse I'll change it. This tweak/test/tweak cycle will get you into a pretty good setup within a dozen or so shows.

Originally Posted by Tim Weaver

The best thing to do, is to simply use it for shows and continually make small improvements. Over time you will work out the bugs. Software works great IF you know how to use it, and IF you have the time and space to set up the test.

Just using these things in anger will allow you to tweak them until they work for you, whether you use software or not.

That is about what I have been doing up til now - just got tired of the cab sounding whacked each time I moved to a different room. Decided to try and use my fledgling knowledge of tweekistry to come up with a decent tuning and reach out here for a few pointers.

Originally Posted by john lutz

That is about what I have been doing up til now - just got tired of the cab sounding whacked each time I moved to a different room. Decided to try and use my fledgling knowledge of tweekistry to come up with a decent tuning and reach out here for a few pointers.

Then I would suggest you start recording your mixes and listening to them later on good headphones or a good stereo. This will tell you what was missing, or too hot in the PA because the recording will have the inverse thing happening.

Too much midrange on the recording? Then you didn't have enough mids in the house PA causing you to mix things mid-heavy to compensate. When your recordings start sounding like what the house sounded like, you know that your rig is getting pretty well optimized.*


*Please don't talk to me about phase-response. Right now we are trying to get the rig to be semi-consistant. Uber-tweaks can come later....

Originally Posted by Tim McCulloch

...........I suggest you drive and measure each 15" separately, using the same stimulus. Each speaker gets its own amp channel (at least for now). If Holm will let you save traces to the screen, you'll want to save both the magnitude and phase traces AND the impulse response. The reason for the impulse response is to look at group delay; it will be different for each loudspeaker loading. Next power the 15"s from separate amp channels. Bring up the first channel and look at the magnitude and phase, then bring up the other amp channel and see what happens and at what relative levels. Now experiment with delaying the earliest-arriving 15" to match the later speaker. What happens to the phase trace where the energy contribution of both speakers is within 6dB of equal? What do your ears tell you as you try these various evaluations?

Once you have the 15s playing nice with each other, we can deal with getting the uppermost 15 to play nice with Mr. Horn.

This little project is for my fun and I know it will take some time and effort.

The cab is wired a 3-way. 2" horn isolated + 15" mid in a small sealed box + 15" low in remainder ported cab. I have WT3 so I have the impedance data and port tuning freq. I have data sheets on all the drivers. The horn is a mystery for now - possibly an old Community. I power it with 3 amp channels, Xti2k + Xti2k + xti4k crossed at LR24 1k, 250hz and 100hz.

FWIW it would be easy to convert this to a simpler dual 15" + 2" horn cab if that configuration makes more sense.

For starters I am going do as you suggest. This will take me a while to set up and test.

Would my first measurements be without x-over filters applied? Do I need to see what the drivers do naturally? Or am I only interested in what I'm probably going to feed them and start with best guess generic points Like LR24 1K and 250hz ? Would a different filter be a better start point? BW18? LR48?

Originally Posted by Silas Pradetto

As I've mentioned in other threads, and how Bennett so nicely put it, building a tuning for a speaker cab is no easy task. Think about how many manufacturers get it wrong (Vertec is up to V5!!!!).

I know where your coming from, and V5 does sound better than V4, but mainly from hardware limitations of the dsp moving from non HD ITechs and dsp in the Harmon family (previous versions I agree with your above statement).

Originally Posted by john lutz

Would my first measurements be without x-over filters applied? Do I need to see what the drivers do naturally? Or am I only interested in what I'm probably going to feed them and start with best guess generic points Like LR24 1K and 250hz ? Would a different filter be a better start point? BW18? LR48?

John,

The first 90% of tuning is the easiest, as you can mostly see and hear the problems quickly and correct them quickly. The remaining 10% is the stuff that isn't clear on the measurement, and maybe you can only tame but not fix with EQ, and getting the bandpasses to line up off axis.

I always start with raw driver response to take a look at what I've got, as well as sensitivity differences between different passbands. Seeing how well behaved the raw driver is gives me an idea of what crossover points I can use, and whether I'm going to have to really pack them in or whether I've got some breathing room. Then I go for the nastiest out of band problems with EQ.

I normally use BW or BS 12 or 18dB/octave slopes. Maybe steeper for the sub low pass. I have never used LR48 except as a demonstration of how to make subs sound "slow". I only use LR24 if I need to work in a tight frequency space with rough measurements.

Thank you for that distillation, Bennett. My reply was much longer and probably was less useful.

John, what I described was "raw driver response - as loaded". I think it's important for a couple of reasons, but the first reason is what you'll learn from it. Watching and listening to some truly gifted designers over the years, the most important thing I learned is that optimal performance starts with transducer selection based on how they interact with each other, more than how they work (or sound) individually. You'll never know those things unless you look at each pass band transducer by itself and continue your measurements as you combine them.

And I'll say again that the odd loading (reflex and sealed) is a way of creating an acoustic crossover between the mid-LF and LF, but you need to verify how it works with the speakers you've chosen to load into the boxes. You might be able to run them in parallel or they may well need a separate amp channel and DSP (and my money is on the latter), but it depends on what you find as you work with them.

Or fire up some arbitrary crossover points, set levels by ear to get a general "voice" to the PA, and go make some noise. Do the "lather, rinse, repeat" that Tim Weaver suggests. It's a valid concept, too... but since you have some measurement tools I suggest you take the investigatory approach (particularly combined with the "use it at the gig" that Weaver mentions) until you strike gold or get tired of it. This kind of project can take you into all sorts of "what if" situations where you can play with DSP and measure/listen to the outcome. Consider it part of your education.

Thanks all for the input - your suggestions, based as they are on lots of experience, are very helpful. I know polishing an old rock box is pretty far below what most of you do.

I already see I'm going to have to access the user forums on the software I'm using to try to figure out the phase trace functions. Very hard to get understandable results ! Moving the mic or changing the delay / xover causes displays I can't interpret. The frequency response and impulse response functions are pretty straight forward though.

Some questions:
As I measure the 3 sections I see the delay of each in impulse graph. What am I looking at? Is that group delay? Or physical offset? Or both?

If I then apply xovers and delays to each section and line up all the impulses, what have I really accomplished? It appears to smooth out summation - sort of - sometimes............ this is fun.

John,

Don't knock it, I used to make good money polishing up "old rock boxes". Compared to the cost of a new PA, fly hardware, and amps a new DSP and a day's worth of my time and travel was cheap, and usually good for a PA that sounds modern and maybe even has more output.

If you line up impulse responses with delay, ask yourself: At what frequency am I aligning these impulse responses? Also, getting an impulse out of your lowest passband after applying a low pass filter is kind of a joke. Getting it before applying your low pass filter is useless, as you have seen adding filters changes the timing relationship between the passbands.

This is a great project to embark on, and it looks like you're really doing it right by starting with the basics and asking the right questions.

Originally Posted by Tim Weaver

Then I would suggest you start recording your mixes and listening to them later on good headphones or a good stereo. This will tell you what was missing, or too hot in the PA because the recording will have the inverse thing happening.

Too much midrange on the recording? Then you didn't have enough mids in the house PA causing you to mix things mid-heavy to compensate. When your recordings start sounding like what the house sounded like, you know that your rig is getting pretty well optimized.................

Wait, I think I just got what you mean here........compare a direct outa the board tape? If that's it then I'm doing something right because the vocals at least sound close. There's never much snare or guitar though. ;-) Seriously though, I'm going to go back and relisten to a few. I never thought of it that way before.