Non-objective experiment (for self)

[eddantes]

I've said it before, but I am very greatful for this site and all the wonderful people and their tireless labours in contributing to the "objective measurements" discussion we have here. To me it's been an eye opener!

One of the problems that I have is that I have a hard time corelating the spinorama charts we use with a mental picture of they mean soundwise. Looking at two spins - I understand what (more or less) what the charts mean with regards to FR or DI or Distortion... but what does that sound like?

So - I thought I'd run an experiment. Since I don't have any (real) measurement equipment (nor have I ever done anything resembling science), I figured the following might be interesting (to me at least):

Hypothesis: Two speakers with very similar (<1.5db difference, in small dips/paks - high q) ON-AXIS, Directivity, and PIR measurements (and aligned for sensitivity differences) shoud be indistiguishable to my (untrained, somewhat tone-deaf) ears. If there is a detectable difference, then I should not be able to readily articulate what the difference sounds like (that means a statement like "more midbass" or "clearer highs").

This is my current mind-map of spin data.

So... with that in mind, as an owner of Revel M22s, I decided to procure some Revel F35s. Lets look at their spins: (BTW thank you!! @edechamps for your wonderful Loudspeaker Explorer tool, I spend lots of time in it trying to learn. I do hope it can be maintained and kept updated, though I recognize that it's probably too labourious)

If one ignores everything below 80HZ and above 12KHz ( I can't hear much above 14KHz anyways) then the two speakers ought to meet my criterea: <1.5db difference, in small dips/paks - high q.

So here's my proposed experiment to test the hypothesis:

• Using MiniDSP
  • take an incoming stereo signal and make mono;
  • apply any changes to one output to account for sensitivity differences (F35 - 90dB | M22 - 85dB)
  • high pass both outputs at 90Hz (remove the low end) - the area of obvious difference between the two speakers
  • (Optional - low pass at 12KHz - not sure if I need it - I'm unlikely to hear up there)
• Pass the two outputs to my Denon DRH800 (the only thing I have that can A/B speakers with a remote) AUX-in
• Attach one F35 to Speaker A connection | attach one M22 to speaker B connection (keeping in mind which MiniDSP output has the +5DB)
• Set up one of each speaker in the least reflective room of my house, as close to each other as possible and matching tweeter heights
• Test with some pink noise to ensure one isn't noticably louder than the other
• Play a familliar playlist while rapidly A/Bing

Will the two speakers sound the same? Will the differences between them (if noticed) be easily labled, or will I just get stuck at "different"? If the hypethesis holds - then does that mean that speakers that spin largely the same, sound largely the same? Can one (maybe just "can I") look at spin data and make assumptions that reliably hold?

I would apreciate some comments if anyone has decided to read this (thanks in advance!), so that I don't conduct this and fool myself into some falacy. I am hopeful this will help me attach the thoretical knowledge I am getting to realworld, intuitive, understanding.

 

[RayDunzl]

Set up one of each speaker in the least reflective room of my house, as close to each other as possible and matching tweeter heights

I might also place the speakers one on top of the other.

I would think we are less sensitive to small vertical differences than lateral when "comparing".

Reverse the positions if you come up with a preference and try again.

 

[Blumlein 88]

I'd do it both ways. Stacked like Ray says, and separate while listening in stereo and mono. I might try mono pink noise filtered 12/octave below and above 500 hz and 2000 hz for initial adjustments.

 

[eddantes]

I'd do it both ways. Stacked like Ray says, and separate while listening in stereo and mono. I might try mono pink noise filtered 12/octave below and above 500 hz and 2000 hz for initial adjustments.

Thanks for the reply. I can certainly place the m22 upside down on top of the f35, I think that makes sense. One hiccup already is that my 2x4 minidsp apparently has no routing options so mixing to mono is not an option. But I don't think this is a problem, listening to the same stereo channel will still get me to where I want to go. The key to the minidsp is to add the 5db and to strip out the lows. Infact I'll probably set up stereo pairs to follow up the single channel test anyways.

 

[YSC]

Actually that’s easier to just use two of the same speakers, in mono, just make sure level match the two. For example I remember unit to unit variation of JBL J305P have wider unit to unit variation, whereas higher end stuffs like Genelecs and Neumann are having 0.5db deviation tolerance, just buy a pair of entry level jbl and level match at say 1khz and you will be done

 

[eddantes]

Actually that’s easier to just use two of the same speakers, in mono, just make sure level match the two. For example I remember unit to unit variation of JBL J305P have wider unit to unit variation, whereas higher end stuffs like Genelecs and Neumann are having 0.5db deviation tolerance, just buy a pair of entry level jbl and level match at say 1khz and you will be done

Thanks for the reply. The idea I'm trying to firm up in my head is that two "different" speakers that spin alike, sound alike. While I see your point about unit variation, I don't feel it satisfies the curiosity I have.

 

[YSC]

Thanks for the reply. The idea I'm trying to firm up in my head is that two "different" speakers that spin alike, sound alike. While I see your point about unit variation, I don't feel it satisfies the curiosity I have.

well, if one speaker unit differ from another speaker with 2db, then it's not across the frequency 2db, the peaks and dips would be just like what you want to achieve, and that the directivity will be exactly the same as they are the same, spin very close to on another and the physical limit of their extension will be similar, thus achieving what you try to achieve, if you inisit on two different brands or models, then we know that the directivity will differs quite a lot, so that's the room modes contributing mostly and I can gurantee it will sound different

 

[eddantes]

Agreed! This is why I want to "experience" it. But as to my two candidate speakers directivity - it's fairly close, with F35 being the better. So whatever differences I hear - I now may be able to ascribe to directivity - and that would be a successful exercise.

 

[eddantes]

Experiment will need to wait, while I get a remote controlled A/B switch. DRA800 speaker switching goes through A>B>A+B and A+B puts into protection with both M22s and F35s attached... So more equipment needed. Will update once resolved.

 

[napilopez]

I think this is a very interesting experiment and I'm curios to see what you find. My own experience is that even small differences in directivity are quite audible and these speakers have rather significantly different directivity.

In this case, although the overall DI curves are similar, the speakers have quite different horizontal directivity, so it'll be interesting to see if and how that manifests. My prediction is that you will hear a difference; I'd imagine the M22 would sound a little brighter and perhaps more mids recessed, while also sounding a little wider.

 

[Dennis Murphy]

I think this is a very interesting experiment and I'm curios to see what you find. My own experience is that even small differences in directivity are quite audible and these speakers have rather significantly different directivity.

In this case, although the overall DI curves are similar, the speakers have quite different horizontal directivity, so it'll be interesting to see if and how that manifests. My prediction is that you will hear a difference; I'd imagine the M22 would sound a little brighter and perhaps more mids recessed, while also sounding a little wider.

It should be possible to judge the impact of differing horizontal directivity using just the M22's. Place one upright, and the other on its side sitting on some books or whatever to get the tweeter at the same height as the vertical sample. Then switch back and forth and see whether you hear a difference in tonality. That would be a more precise test than comparing two different model speakers, and they will be level matched with no guess work. I've done this using a speaker with the worst-case MTM configuration, and the results were pretty unambiguous. This assumes you're just testing for differences apparent from a listening position directly in front of the speaker, not for differences apparent from an off-axis position.

 

[preload]

Interesting experiment. Taking a look at the two on-axis curves that you provided, I would offer the prediction that you should be able to hear at least subtle differences. The degree of FR deviation (~2dB over >1oct) should be audible. People make even smaller corrections when EQ'ing their systems and can differentiate.

What might spice up your experiment a little more is that if you do detect differences, to then EQ the curves to match using measurements, to see if you can still reliably differentiate.

If you're then not able to differentiate, it might be interesting if others can repeat the experiment in different rooms and (of interest to me), with speakers from different manufacturers (using different driver technology and design goals). That's where I bet it'll fall apart.

 

[eddantes]

It should be possible to judge the impact of differing horizontal directivity using just the M22's. Place one upright, and the other on its side sitting on some books or whatever to get the tweeter at the same height as the vertical sample. Then switch back and forth and see whether you hear a difference in tonality. That would be a more precise test than comparing two different model speakers, and they will be level matched with no guess work. I've done this using a speaker with the worst-case MTM configuration, and the results were pretty unambiguous. This assumes you're just testing for differences apparent from a listening position directly in front of the speaker, not for differences apparent from an off-axis position.

Great suggestion, I'll add this to the exercise! Although, the main thrust is to map my mental image of spin data to physical reality, as I am curious if two different speakers with similar spins sound alike.

 

[eddantes]

Interesting experiment. Taking a look at the two on-axis curves that you provided, I would offer the prediction that you should be able to hear at least subtle differences. The degree of FR deviation (~2dB over >1oct) should be audible. People make even smaller corrections when EQ'ing their systems and can differentiate.

What might spice up your experiment a little more is that if you do detect differences, to then EQ the curves to match using measurements, to see if you can still reliably differentiate.

If you're then not able to differentiate, it might be interesting if others can repeat the experiment in different rooms and (of interest to me), with speakers from different manufacturers (using different driver technology and design goals). That's where I bet it'll fall apart.

Unfortunately no measurement equipment (other than the phone, which might help ensure level matching w pink noise). I agree - I expect to hear differences, but not to readily articulate them.

 

[dasdoing]

a wider dispersion speaker will sound brighter since the meassurement will have more reflected sound included....which our brains seperate in the high frequencies.
easy to understand this way: a flat meassuring speaker in a anaechoic chamber = nice. a a flat meassuring speaker in a reverbant room = bright. the more reverbant the more tilt you need

 

[Dennis Murphy]

Unfortunately no measurement equipment (other than the phone, which might help ensure level matching w pink noise). I agree - I expect to hear differences, but not to readily articulate them.

Where are you located?

 

[eddantes]

a wider dispersion speaker will sound brighter since the meassurement will have more reflected sound included....which our brains seperate in the high frequencies.
easy to understand this way: a flat meassuring speaker in a anaechoic chamber = nice. a a flat meassuring speaker in a reverbant room = bright. the more reverbant the more tilt you need

Interesting. I did equate flat with bright already, but, again - I'd love to have the confirmation of sensory experience. The two speakers here look to have very similar tilt:

I just visualy placed that. I think, if one would calculate the trendline, there might be a material slope difference... but I'm not sure if it would be audibly (to my ears at least) significant.

 

[dasdoing]

Interesting. I did equate flat with bright already, but, again - I'd love to have the confirmation of sensory experience. The two speakers here look to have very similar tilt:

View attachment 141832
I just visualy placed that. I think, if one would calculate the trendline, there might be a material slope difference... but I'm not sure if it would be audibly (to my ears at least) significant.

my point was that you can't see how much reflected sound is included in those meassurements. if the blue line had more reflected sound in it than it's direct sound will be less tilted than the orange one's

 

[eddantes]

Experiment will need to wait, while I get a remote controlled A/B switch. DRA800 speaker switching goes through A>B>A+B and A+B puts into protection with both M22s and F35s attached... So more equipment needed. Will update once resolved.

I'm an idiot. I didn't detach the 2nd M22, hence protection. With only one of each - no protection, experiment can proceed.

 

[eddantes]

Grrr... something is worng. I have a highpass on all channels at 80hz:

But F35 is putting out noticably more low end... This doesn't jive at all with the charts...

I must've screwed something up. I'm going to recheck everything and move to a different space. Will restart this tomorrow.

(JIC anyone asks, I'm pretty sure they're volume matched.

Checked with db meter app on phone and pink noise.

Not exactly lab quality - but at least it's something.)