Best way to balance speaker for a flat response

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Sort of. Can you measure the drivers in the box/baffle, both on axis and off axis, without any xover filters? You could feed these data into a speaker CAD software, draw the xover schematic and simulate the system result. Vituix CAD comes to mind.

Ok cool. I got the FRD and ZMA files from the manufacturer websites and fed them into xsim when I designed my crossover. I guess that’s the next best thing next to actually spending the money and testing physically.
 
Yes sal, obviously you should listen to the drivers 'raw' before purchasing, each type has its own coloration.

If you go with drivers a good philosophy is to have similar materials/manufacturer drivers. however there are good speakers with dissimilar materials, like plastics, metal, others have kevlar and berylium, but I would use similar materials at least for the sub/bass/mid, this ensures consistency. Remember, the sound is a whole, it is the total of all the parts

the simulator is a good way to know how complex the xo will be and the box volumes.

the simulator gives an idea, it is far from what will actually be.

I would not use music to tweak the xo, it is a common mistake, use music to tweak the response curve! then you design the xo accordingly.

pink noise? well it should measure without spikes and as close as possible to your desired response curve. I would use sweep tones with holm impulse or other very good software, trueRMA also helps finding peaks and resonances in real time.
 
If the flat response is the goal, the best (and possibly the only) way to correct the frequency response of the new unknown speaker in the room is, to compare it with a reference speaker which is proven to be flat. All the other method will end up the speaker with the response that is based on your subjective preference, which is not bad at all, though.
 
Yes sal, obviously you should listen to the drivers 'raw' before purchasing, each type has its own coloration.

If you go with drivers a good philosophy is to have similar materials/manufacturer drivers. however there are good speakers with dissimilar materials, like plastics, metal, others have kevlar and berylium, but I would use similar materials at least for the sub/bass/mid, this ensures consistency. Remember, the sound is a whole, it is the total of all the parts

the simulator is a good way to know how complex the xo will be and the box volumes.

the simulator gives an idea, it is far from what will actually be.

I would not use music to tweak the xo, it is a common mistake, use music to tweak the response curve! then you design the xo accordingly.

pink noise? well it should measure without spikes and as close as possible to your desired response curve. I would use sweep tones with holm impulse or other very good software, trueRMA also helps finding peaks and resonances in real time.

Ok thanks man!

So with the pink noise should the chart look flat on the RTA? I’ve always been confused with how it should actually look on screen. Or look like a downward slope from left to right?
 
If the flat response is the goal, the best (and possibly the only) way to correct the frequency response of the new unknown speaker in the room is, to compare it with a reference speaker which is proven to be flat. All the other method will end up the speaker with the response that is based on your subjective preference, which is not bad at all, though.

I agree, it’s not bad but I would certainly prefer the best response from a technical point of view and then maybe colour it a little according to my band’s needs.

It’ll never be perfect, but then what is!
 
I agree, it’s not bad but I would certainly prefer the best response from a technical point of view and then maybe colour it a little according to my band’s needs.

It’ll never be perfect, but then what is!

I see. If the speaker has to be flat for that reason, you might have to buy or rent a proven pro monitor speaker and check your room (and cheap measurement mic) response before starting correcting your speakers. Without doing it, I'm pretty sure that you will just confused. I don't think you can build a reference speaker without the other reference speaker, unfortunately... :eek:
 
Hi sal,

the slope could be anything that you want. A flat curve is alright, for rock etc. For old classical recordings a little bass boost is welcome.

Some rooms have lot of reverberation of hi frequency so the sloped curve is a solution to reduce the high frequency energy due to reverberation.

In bigger rooms it is not necessary to have a downward curve!!!!

The bass boost at 100hz is a solution to a closed or radiator bass, it makes the small speakers sound more whole and not tin sounding.

If you use big woofers and big bass you can have a sloped bass, the perceived bass will be very clean and deep, the room boosts the bass...

The small room cancels deep bass and acts as a filter, to counter this effect you boost bass.

The big rooms boost bass and enable the long waves to fully expand, so a decreasing bass is preferred. A ruler flat bass in a big room can sound too much bass.

for example in a bedroom a boost at 100hz and nothing at 50 hz is perfect, you will not notice.

In a big room you can have a -3db point at 30 hz and no boost at 100 at all, it will sound nice.
 
Last edited:
For the DIYer, there is only one way to balance a speaker to flat response that i know of...

Learn to measure, and find as reflection-free way as possible.
End of story....forget reference speakers..forget speaker driver matching etc...

Learn to measure, and find as reflection-free a way as possible...and then you can make almost anything sound like anything....
with capable processing, as it makes makes so many of these discussions seem plain silly.

That is, until you move into a reflection environment, and hear how everything changes,
and realize how much polar measurements matter...
And realize that all the eight billion recommended tweaks, important details, are pretty much pee in the wind...

What is flat? On axis ? Listening spot? Where? Power response? SPL? :)
 
Here we go again. Somebody ardently posts, "There's SOMETHING amiss with my sound" and everybody runs off in all directions to fix "it".... instead of replying, "If you don't mind, how about telling us what seems to be your problem?"

BTW, sad to see nobody can think about anything besides speaker near-field FR.

B.
 
Is your idea of reflection-free near field?

Mine isn't, at all.
Had your wise post #28 appeared before I wrote mine, I would have added, "mark100 excepted".

But still, you are in the FR universe, perhaps led astray by the title of this thread. Really the inquiry (after hearing what the problem seems to be) is what sources? Could it be distortion or TH character? How "loud" and what is the tonal balance of the room reverberation? Where do you sit? And so on.

B.
 
Last edited:
well, then one good idea is to take a proper individual measurement.

Place your loudspeaker outside in a non-windy and quiet day, you can leave the city and go in a park idk,

Place it on top of a stepladder or something high with no nearby objects, the microphone and the speaker distance is measured. Everything should be far from that distance of both the microphone and the loudspeakers, that is why you place it on top of the ladder.

You take individual measurements without XO and with XO

you bring this home, it should pin point what the problem is, it could be time delay, phase, response, null points, crossing slopes, beaming, dips, peaks etc....
 
No doubt, I got **** loads to learn and your replies are helping massively.

My main problem with my current speaker is the balance between all 3 drivers but more importantly between the mids and highs. I tend to go to studios with with pretty non reflective rooms but i was clearly doing things in the wrong way when it came to “tweaking”.
Listening to different types of music and then tweaking between songs which, now looking back, was the wrong way of doing things. I also think the inherent design of my current box is adding to the issues.

I never really thought about off axis responses or other parameters until you guys brought it up!



If people think an active solution is easy to implement, they’re in for a rude awakening (my opinion)
 
Just another Moderator
Joined 2003
Paid Member
I was just about to say what Allen did :) ultimately you are trying to get an acoustic target where the drivers blend well, without any nasty off axis anomalies.

Finding the crossover points and slopes where everything just comes together is the tricky bit. You will likely end up with differing electrical slopes on your filters for each driver to get the target acoustic slopes. Paying attention to phase matching at the crossover point and also for a reasonable distance either side of it, IME pays dividends with driver integration.

A flattish looking response can be achieved with poor phase matching, but will sound very different to a very similar response that has been achieved with good phase matching.

There are many aspects to consider when picking a crossover point, and some tradeoffs inevitably need to be made. In the end it comes down to your design goals, and what is physically possible with the drivers in question.

Tony.
 
A flattish looking response can be achieved with poor phase matching, but will sound very different to a very similar response that has been achieved with good phase matching.

This seems to imply that one could achieve a flattish amplitude response with different phase response. That's not really possible since the two are intimately linked in many ways. The axial response might be similar, but the off-axis response will then likely differ considerable. The "correct" phase match is the one that yield both on and off axis optimization.

But you are very correct that one has to "tweak" all the variable at ones disposal to optimize a crossover. It's not a simple thing to do. Especially with waveguides.
 
A flattish looking response can be achieved with poor phase matching, but will sound very different to a very similar response that has been achieved with good phase matching.

Odd to my ears to hear folks using crude cone drivers sweating over phase angles. It's hard enough to document good sound at one spot in the room without also worrying about phase angles and polar response.

Phase coherence is a meaningless concept at my chair. Relative phase from two speakers (and also "within" one Rice-Kellogg-type driver) changes and scrambles millimetre by millimetre as you breathe* and as the sound originates from anything but a tiny point source (not to mention the gargantuan tricks played on the sound from the piano in the process of sending it to the speakers).

B.
* that point does not apply to listeners who hold their breath during test auditions, assuming they are using a bite-board to immobilize their skull
 
Little-known tweak trick ...

Is to literally level your speakers. You may use shims or other methods to make sure top of speaker is as level as possible. This is an "insider" trick to not only getting flat (level) frequency response, but it also helps insure that the sound waves will reach your ears at the proper 0 or 90 degree angle and not tilted a few degrees. :D


Actually I have a non-smart-*** response too! Since you are dissatisfied with tonal balance, perhaps the problem is that you have not determined what your preferred listening curve is? None of my ideas are original but this is what worked for me. I determined my own personal house curve by listening on headphones to white noise bands (1/3 octave) and adjusting using a graphic EQ until repeated listening tests give me similar subjectively equal loudness levels at more or less my preferred listening dB (due to Fletcher-Munson curve). Take your time determining this curve, because it will be personalized to your alone, and will be used for the downstream tuning.


After your have learned your preferred EQ, then you are to make your speakers sound as close as reasonably possible to it. How this is done will depend on your EQ available. The procedure is the same if you will EQ to a "flat" frequency response (which is seldom recommended), or (what I advocate here) is to EQ to match your "house" curve.



I have done this years ago and am so happy that I rarely feel a need to bugger my system's frequency response, as you Brits would say :)
 

Attachments

  • On the level.jpg
    On the level.jpg
    75.8 KB · Views: 231
Last edited:
Just another Moderator
Joined 2003
Paid Member
Earl I should have qualified that as a "flattish looking on axis response" which is what I meant :) my metric for good phase matching is the deepness and symmetry of the reverse null. Obviously for even order acoustic targets.

Tony.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.