Orelino / Orelo MKII Fletcher-Munson Curves

[PeterSt]

What you see below first are a few pictures of the switch board at the back of the speaker which allows to set Fletcher-Munson curves. Here you see it in lab setup to determine the proper resistor values etc. before all goes on to a PCB built in the inside of the speaker.

A few examples of the settings (27 in total) :

This is a flat curve and all is "green". On the graph below you can see what it does and it resembles this :

As you can see per "loudness" the curves differ, and we need to look at the 80dBSPL / 90dBSPL curves which assumably resemble our playback levels.

Now, the green-green-green curve is just a flat line and normally loudspeakers will be tuned to such a straight line. The Orelino and Orelo do too, but only if you set it like that.

These are snapshots of the picture above :

and it shows from the shadows on the dip switches how they are set. Also look at the arrow directions next to the switches;

Now think : While the normal loudspeaker will be tuned to green-green-green and the F-M (Fletcher-Munson) curves deviate from the horizontal line partly to under that and partly to above that, the deviation towards that F-M direction will be other than green. Look :

Here you see the switch for the (right) part above the horizontal line switched upwards (higher) and now it is orange.

Here we switched the left and middle part more "into F-M" and both switches are now in their middle position (orange). The right part is green again and the switch is in its most down position.
(the middle led for the right part (normally orange) catches the light of the bottom led in this lab setup).
 

And here you see that we set the middle part all into the red, or IOW into the most extreme F-M setting (again see the switch settings by means of the shadows).

This one again :

but now look at the texts above the switches;
From left to right the switches are named Slope, Dip and High. And with this one repeated you see it happen :

So there's a slope from 600Hz (look at the mouse pointer !) and which "600Hz" is also denoted under the switches :

So see under the left hand switch. But what you also see is that this "Slope" is controlled up to 6800Hz (6K8). See the mouse arrow for that postiion :

Now, the second switch controlls the Dip, and it reaches from 1200Hz (1K2) to 5000Hz (5K). Need to show another F-M plot with arrow pointers ? no ! because surely you got the idea by now.
Notice : The Dip as such is "in" the Slope, so when the slope is set more downwards the Dip automatically goes along with it. However, since the Dip is a separate setting just the same, that Dip setting determines the deepness of it. More below you will understand when we start showing measurements of this.

The rightmost switch controls the "highs" and therefore it is named "High". Its working range is from 6800Hz (6K8) and the 6800Hz is a lever point (you'll again understand this from below measurement plots).
It is no coincidence that the beginning of its working area matches the end of the Slope's working area because this too you see in the F-M curves might you look at the different levels of loudness.

Remark : If you again look at the F-M plots, you see that there's also working area for that under 600Hz. And if you look more close to our listening levels (the 80-90dBSPL) you see that nothing much happens under 600Hz up to 200Hz. From there (towards the left) things start to change again. And it is there where we can adjust all with DSP (Digital Signal Processing) ...


While this may all look nice to you, it obviously has a purpose. And this goes beyond the Fletcher-Munson ideas;
So notice that the idea of these two blokes (back in 1938 or so) was about our hearing system (mechanisms) and how we human are more receptable to certain areas of the frequency spectrum than others. For example, the "Dip" area consists of "danger" noises we humans from origine receive better (louder) than others. A snap of a branche for example. So, while we perceive that more loud, it may jump out in music when equally (for level) presented compared to the other audible frequencies. It may sound nasty ...
(Notice : with not too many words this is not very well solvable if music is to be sounding less "nasty", but still ...)
So what happens with the adjustable F-M curves is that any louder perceived frequency range is attenuated and the result of that as how we perceive it is now a flat response.

What we, as the owner of this loudspeaker can do with it goes beyond what F-M intended with their curves;
It occurred to me that for example the difference between Windows 7 and Windows 8 can very well be attacked with it. So, too much of uncoloured highs ? then ... put in some Dip. So yes, you might expect that some High should be "muffled" out, but it does not work like that. Not for the frequencies what this is all about.

People might think that this is all the expensive way of applying equalizers, but that is not so; first off it all happens in the high voltage domain (read : at loudspeaker levels) which in this case is a necessity and is all related to the sensitivity of the speakers (118dB). Meanwhile (with this as a given), the crossovers all stay in-tact. Not an easy task to accomplish but obviously the most crucial.
Foremost though, we don't want to have our sound touched by other means than "normal" (which is passive filtering in this case), assumed no DSP is possible at this moment, but while the speaker fully anticipates on that (future upgrade) and now our "switchboard" is even more complicated than it already was because it anticipates external amplifiers.
Can't follow ? it's not all that important for now ...


So then let's now start measuring this.


All measurements were performed at 88dBSPL. Is this important ? Well, actually Yes *if* we think the Fletcher-Munson curves really mean a thing (they are different for output levels).

What you see here is the Slope with all its three positions. Green, Orange and Red.
The dB scale is not shown but it is 5dB per vertical division.
In communication Bert and me got used to S(lope), D(ip) and H(high) positions. Above you see mixed in one picture :
S=1,D=1,H=1
S=2,D=1,H=1
S=3,D=1,H=1
So each switch has its default at position 1, the middle position at 2 and the extreme position at 3.

S=1,D=1,H=1
S=1,D=2,H=1
S=1,D=3,H=1
Got it ?

Varying the the Dip means varying the "presence" (empirically found).

S=1,D=1,H=1
S=1,D=1,H=2
S=1,D=1,H=3

If you payed attention during the introduction you will have notice with these measurements that the "green" for High does not comply to the promised description;
The reason is that the tweeter really needs burning in and now the fun : no problem during the burn-in period because you can just flip a switch for it. This really matters because the horn for the Orelino speaker I used for the past 7 monts really shows more tweeter output by now, and it is audible too.

Here the both extremes are set against each other;
S=1,D=1,H=1
S=3,D=3,H=3

The red line will make older mor bad recording (could be Rock) sound less grey while still detail is picked up and which makes them sound fresh. It is not the best for good balance, but you may like the detail coming forward which can sound really "on-par".

S=1,D=1,H=1
S=2,D=2,H=2

S=1,D=1,H=1
S=2,D=2,H=2
S=2,D=3,H=2

You can also see this as the difference with the previous picture, where the red curve shows the added Dip.

S=1,D=1,H=1
S=2,D=2,H=2
S=2,D=3,H=3

Same as the previous picture but now also the High is set in extreme position (3).


So these were 7 pictures which showed some of the combinations and you will see (hear) that most of them really make sense when it were about some means of wanted perception. Without telling the combined settings, you can expect this (of course each one mentioned also has its opposite) :

- Dial in presence.
- Dial in spaciousness;
- Dial out harshness;
- Dial in über-detail;
- Dial in Mid-bass;
- Dial in highs-colour.

With my own experience from the past 7 months, a lot of music really can be played with (for these settings) in order to get so much more out of it than "normal". So think of music so well recorded that you can set H=3 and all it does is create a very wide stage and sharpened detail while nothing starts to hiss. Things start to fly in the room. It could be your genre of the night and all it takes is setting some switches.

What crucial about it (to let it fluently work) is the levers you see; one at around 400Hz and one at around 7100Hz. So this keeps on connecting the frequency spectrum in a responsible fashion. And for example, setting up a Dip can create more highs. Yep. So what you will be doing in that case is making the 7KHz region more profound. You do this by making the - actually more profound by nature - 4KHz-5Khz *less* profound, so it allows the 7KHz to be heard. Just look at the last picture to understand this.
Similar can happen with making Mid-bass more profound, assumed this goes from 200Hz to 400Hz or so. Just set S=3 (and see the red line in the last picture).

All this is outside of DSP which operates on the bass (up to 400Hz, but can be more for the real tweakers). This is for another topic.

Peter

[PeterSt]

I remeasured everyhing (don't ask, but I made a stupid mistake as I found out) and I put new screenshots in the original post above. You can now better see how the tweeter extends with the notice that it still needs burning in so it drops off somewhat (~2.5dB at 21.5KHz for the intended straight green) line.
Maybe you need to apply (Ctrl-)F5 to your browser to see the new screenshots (the old screenshots extended to 15.2KHz, the new to 21.5KHz).

Peter

[CoenP]

Impressive!

Am I the only one thinking of three new buttons in XXHigh end?

 

regards,
Coen

[Leo]

Exactlly what I was thinking Coen
And have you looked at the curves for the bass part. About 34 db higher at 16.5 hz compared to 1000 hz. That will be tough for your amp...
Leo

[CoenP]

And have you looked at the curves for the bass part. About 34 db higher at 16.5 hz compared to 1000 hz. That will be tough for your amp...
Leo

Hi Leo,

The BD design eq is within the horizontal 5dB line spacing of Peter's graphs. I see no eq below 600 Hz.... The only tough one would be the high, since that is a 'gain' and we need (imho) to avoid clipping.

Regards,  coen

[PeterSt]

Hi there Leo,

About 34 db higher at 16.5 hz compared to 1000 hz.

Coen is just ahead of me, but even after reading this 10 times I can't understand what you really meant to say !

Peter

[PeterSt]

The only tough one would be the high, since that is a 'gain' and we need (imho) to avoid clipping.

Coen, surprisingly it is more smart than that, and it is not extra gained as you suggest. That is, not within the one powered frequency range (like 400Hz up to 22KHz could be one frequency range to power and one band in there is boosted).

I already see it is hard to explain without explaining all (I tried hard on the latter ), because some things sure will be kept in secrecy.


Peter

[Leo]

Probably my mistake, but:
If I read it correctly you take the FM curves for the correction for perceived loudness. The original FM curves and the present form ISO203-2013  go way up for lower frequencies. About 6db up starting from 1000 hz up to +34dB for 16,5 Hz.

[CoenP]

Gain, attenuation whatever you use, I still see only moderate corrections that can be done digitally without a big loss of bits.

If the passive components are more 'transparent' than the 1 bit digital eq loss (for selected frequencies) the hassle would make sense. And of course for the speaker users that have no interest in computer audio/XXHE/dsp.

Anyway it is a perfect candidate for another unique feature of the XX player!

regards, Coen 

[PeterSt]

Oh, Coen, maybe only now I understand what you were saying. Including the clipping.

Well Yes. And so this is how the speaker contains that other dimension (in electronics) : how to switch off all the passive filtering when done from DSP. But this too is for an upcoming topic.

Peter

[PeterSt]

I prepared this somewhat longer post longer ago, but never got time to finish it until today.

Probably my mistake, but:
If I read it correctly you take the FM curves for the correction for perceived loudness. The original FM curves and the present form ISO203-2013  go way up for lower frequencies. About 6db up starting from 1000 hz up to +34dB for 16,5 Hz.

Aha ...

I have talked about this previously (in the Ambience topic I think) where I said that first of all the modern version of the F-M curves you refer to are not correct. Well, they just can-not-be;

Maybe this gives me the opportunity to explain how the F-M curves for the Orelino/Orelo MKII emerged in the first place :
When I was finished with my "tuning" of the Orelino and which was based on my own filter I used for many years with the same top horn (above 270Hz there) it appeared I was using the F-M curves. I didn't know that, and while I just told one of the due customers of the MKII some numbers, he came back with "but that's exactly Fletcher-Munson !".
Coincidence ?
Can't be eh ?

One thing : For the bass this is not true. No-way.
But why ?

I tried to sort that out with the same guy who came up with the "That's F-M !" and we need to be speculative ...
Notice : I forgot the sequence of things somewhat, but this is not so important;

Say that back in the days (1938) headphones where used. That's cool, but I don't see the really low frequencies come out of there while at the same time the official SPL (Sound Pressure Level) is to be mapped on to that. Maybe it can be done, but I'm sure we all feel that it won't be reality much for a one square inch driver to move air like in real open air (with the notice that headphones are closed and which does a few things too).

Say that in 2013 or whatever recent, normal loudspeakers were used, then I want to see which and how the real lower frequencies were squeezed out because already something will have gone wrong on the distortion levels. On this part : Try to create a threshold of pain (see official F-M curves) for 20Hz without not first implying 40Hz for a 2nd harmonic. It is to-tal-ly impossible - I tell you. Now I'd say that whoever applied that test(s) will have known about 20Hz not being allowed to turn into 40Hz because it is so obviously audible but *if* it did, it would blow out the windows because of that 40Hz, while 20Hz only will shake the room.
So what the heck did they really do ?

I know, and I guess I am amongst the very very few, what happens when a fair amount of SPL is put out distortionless. So, this is about 20Hz not being audible at all, while the meter still reads (the reference) 88dBSPL. Same for 23Hz (also not audible), same for 19Hz. Also same for 18Hz and beyond (I stopped at 8Hz), but from there (18Hz) the output first has to be squeezed or otherwise the 18Hz (and beyond) becomes audible (thus too much distortion). Now :

What is not obvious at all is how a normal speaker will just not put out anything much onder 27Hz because of the, say, acoustical roll off (mechanical also OK). Notice that in normal practice the limit is 100Hz for a fine woofer.
Now some smart guys thought that pressure chambers and ports and what not could help that. Well, let's say that is correct. The same smart guys thought that rooms would help too. This is already more doubtful for me, but alas.
But it does not work. It does not work because you can easily measure it is not true if you only pay attention. Actually it is quite simple : Grab such a fine speaker with specs like "straight to 27Hz !" and measure. It just won't happen. And oh, you will perceive a nice bass alright, but it is at way lower output than e.g. 100Hz (say 10dB for the discussion).

Now we buy speakers according those specs. We also *use* them according those specs. And if you now don't pay enough attention, before you know it you use them to apply such tests. And what happens ? you need to apply 10dB more "power" to them in order to hear the same level (level of pain or whatever level).

So what am I saying ? well, that this speaker 100% really goes straight to 19Hz and if I now apply 10dB more to 27Hz it already goes very very bad and if I'd apply 34dB or whatever (F-M) more to 20Hz, well, ... do we know how much 34dB more is in that low regions ? So, all tuned nicely in balance and then add 34dB there ? I'd say you would die.

Even if I am half-correct only, it is completely clear that those tests for the lower regions must have totally failed. So I am 50% correct but test still failed 100%. It is just too obvious.

Here is another kind of proof, but from another angle :
So we have subwoofers (like I had them). 700W Hypex in there and such, and the 700W still is critical when fed with line level (same line level the MKII etc. is fed with). So how is it possible that e.g. the MKII need something like 40 Watts only ?
This seems directly related to efficiency, but this really can not be all. I mean, there's one tiny 12" woofer in that sub, and although the driver may not be as efficient it just can not be so that it needs to wildly excurt and vibrate the cabinet while our 3x15" hardly move. Right ?
So in my view, all that cabinetted speakers have just fighting-itself drivers for the chamber pressure they create and which works counter productive. So much so that not even real waves come out (distorted ones yes). From there can be reasoned that the frequency gets lower (a sort of sub-harmonics) but not at the same SPL as a "full wave" (sine) would do it nicely.
How this measures in comparison I don't know (I never tried) but I do know that standing waves form easily and nothing will be good.

There is some more to it, like lower frequency waves spreading more easily (are not directional) from which can be calculated that what you perceive from it should be 1/4 easily compared to when it would have been directional. A microphone would not capture that differently. But outside (anechoically) would be different from inside, because inside reflections play a role and add again.
Now how come that when I tune the lot with the microphone from a fair distance (few meters) the bass will blast the windows out again when tuned so. This just won't go. So it has to be done the other way around : from way close. And now all is in balance.
For net result at listening distance the bass rolls off heavily while I think that everybody would expect it the other way around.

And so it has to be my conclusion that the F-M curves for the low spectrum are the exact other way around; we humans should be very much more sensitive to the low frequencies than e.g. 1KHz. But I think I can tell you that all is related to how well these frequencies are formed and how an e.g. 32Hz sine is really that instead of a wobbling thing with spread energy (!) to lower and higher frequencies than the 32Hz intended.

Sorry to be so fuzzy but where science is lacking somewhat we need to start expressing empirical finding. That can easily fail and it is only my thinking as of now. Those receiving the speaker soon : you will see. It is a whole new world ...

Peter

[Leo]

Hello Peter,

That makes a lot of sense (instead of nonsense science with weak bass speakers)!

Thank you!

Leo

[Stanray]

Interesting the old Fletcher-Munson curves.

For those interested in some background information, read http://www.lindos.co.uk/cgi-bin/FlexiData.cgi?SOURCE=Articles&VIEW=full&id=2.

The tests were repeated by Robinson-Dadson in the fifties and more recent for the ISO-226 standard. The discrepancies indeed appear in the low frequencies.

Regards,
Stanley

Edit: also see http://www.lindos.co.uk/test_and_measurement/SOURCE=Articles/SOURCE=Articles|VIEW=full|id=10

and

http://www.lindos.co.uk/test_and_measurement/SOURCE=Articles/SOURCE=Articles|VIEW=full|id=17

[vrao]

So,
I've been playing with these curves for a while, and they are indeed very very interesting! Some of the design concepts Peter/Bert came up with make this design a landmark in speaker technology, and that I would not say lightly.

Since these speakers play "realistically" one can just use real music as a guide. My experience so far is that these curves help compensating for all the small changes from recording process. One can see how a poor recording was done, or how a good recording/mixing artist shines.

This is quite apparent, when plays a particular genre of music for a while and adjusts the FM curves, forgets about the curves. Let's say play electronic music, FM curves are adjusted for spaciousness, and then switch over to a live concert, suddenly things sound not quite right, image density very poor. Tighten up the curves, and things fall into place.
Play classical music for a while, then switch to a the girl with the guitar, suddenly girl seems out of size (not that it's whacked out or anything, but not real), switch curves to change image size, back to authentic proportion. There are many smaller paths one can take, right size, push the image a bit back, for those too forward recordings or vice versa.

In many ways one can get an idea of what went on with the recording/mixing. Maybe even comment on the speaker he/she used for playback, because certain things were not solved or heard, and let thro?

Many of the things Peter talks about "dialing in" I'm yet to experience with.

It even might be used to compensate for some hearing loss.

All in all, it's is refreshing Peter and Bert did not just with the "flat" response, but gave an option to manipulate the sound for ones own benefit.

 

[PeterSt]

Hey VJ, nice that it works out !

Ok, let's unveil a small secret;

VJ (as how Mr Rao prefers to be called) knows a bit more about our auditorial system than the average us. It is actually his profession.

It was him whom I invited to do a nice seminar in the US about these things, mainly knowing about how my own perception of this all (think "Phasure" name for now) would nicely match the human anatomy VJ knows all about. Envision an email from me to MD VJ just asking (and I did not know him other than from being a NOS1 customer).
We prepared for the seminar totally indepently from each other. And it was fun (we recorded it all and should upload to YouTube).

At some stage VJ ordered the MKII's.

After that I pointed out to VJ what my personal filter settings were. So, on the Orelino of back then (say October 2013) I had my own longer ago "set" filter response and I explained it by words.
VJ came back to me and said "but that is Fletcher-Munson curves !"

Here "MD's theories" started to match practice beyond ideas put forward on a nice seminar and what I did was making that F-M curves really practice. Well, Bert had to do it of course and I can guarantee you that this is NOT an easy thing for a passive filter (because this all happens in that passive domain).
VJ did not even know this ...
Or not at first, because it first had to work out and that took maybe 6 months apart from quite some hairs.

Actually all is the other way around; say that I am responsible for the filter response with the notice that this is all about one pair of ears plus maybe another few pair in the house over here, but still; once you are used to XXHighEnd "dials" you know that it can go all sorts of directions, including not yours (read : findings by others for the better). So it is quite a responsibiity to apply a not-flat response curve in a not-so-cheap speaker only because *you* like it that way.
So this is how it was made set-able and now I could not be blamed for applying some strange curve ...

That curve though appears to be "science" (I knew about it but never realized I had been doing *that*) and diving into it more it was obvious that this is not the same for everybody. Now the settings became a fairly explicit thing or better, a goodie which is not to be underestimated.

This long lead-in actually leads to one small message I wanted to put out and the reason of this post :
It is caused by VJ.

Thanks man,
Peter