Jump to content


Photo

Ideas for new ways to display data


  • Please log in to reply
65 replies to this topic

#21 lilmike

lilmike

    Super Power Member

  • Members
  • PipPipPipPipPip
  • 157 posts
  • LocationPNW USA

Posted 30 December 2015 - 06:47 PM

Definitely some interesting ideas being kicked around here.

 

I've been thinking about calculating the "area under the curve" through a set bandwidth at a normalized power input as a way to compare different cabinets. I like the idea of considering output and/or extension vs. cabinet volume too. 

 

Not 100% sure how I will go about this yet, but there's got to be a better way to look at the results.

 

How about multiple "scores" plotted on a radar plot? Excel can do these (at least in 2016). Create a consistently calculated metric for extension, efficiency, and output based on the measurements, add volume data, and possibly even driver cost, then plot things on a multi-axis radar plot. Each axis of the plot represents the relative score on each metric, and comparisons between data sets would show the relative strengths and compromises between designs.

 

Interested to see what you come up with here.



#22 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 30 December 2015 - 07:51 PM

Here are a couple of quick example graphs taking a stab at representing acoustic power density. This is my original idea of representing this that I've had for a couple of years. These are using the burst data SPL (dB) which is then converted to Pascal (Pa) and then divided by the volume that the system takes up in liters (L) 

 

This one is shown using the data in Pa. This is the simplest way but I'm leaning away towards converting the data back into (dB) SPL because that's a format more familiar to most who would be looking at it.

 

 

Attached File  PAPD Pascal Graph Example.png   35.96KB   0 downloads

 

 

Here is the same data converted back into SPL decibels. This looks better to me intuitively but it's the same data either way.

 

 

Attached File  PAPD SPL Graph Example.png   43.96KB   30 downloads

 

 

I think something similar would work. Here is what is going on for those who need a bit deeper explanation.

 

The maximum output from a system is being converted into Pa and divided by the total enclosure volume so that we can end up with a base output number either in Pa or dB, per liter of space taken up, for each 1/3rd octave frequency band. The conversion to Pa is needed so that the results are comparable.

 

Here's a simple example.

System A produces 120dB at 25Hz and measures 90x90x60cm (486 liters of space occupied)

System B produces 103dB at 25Hz and measures 43x36.2x39cm ( 60.7 liters of space occupied)

 

System A produced 120 dB which converts to an even 20 Pa

System B produced 103 dB which converts to 2.825 Pa

 

(1 Pa = 94dB / The formula for converting SPL(dB) to Pa is

Pa = [10^(dB/20)]*.00002)

 

We then divide each systems recorded Pa value by the total liters of space it occupies to get the Pa per liter value which we are going to call a acoustic power density rating or APD. Higher numbers are better.

System A =0.0411 Pa per liter. (20Pa / 486L)

System B =0.0465 Pa per liter. (2.825Pa / 60.7L)

 

What does this tell us? System B while smaller offers more potential output per liter of space. It is 1/8th the size of system A and offers far less output one vs one, but a block of 8 takes up the same space and potentially offers a little bit more output than system A.

 

If we convert the results back into SPL it is perhaps easier to digest.

The formula to convert Pa into dB is the reverse of the Pa conversion formula

 

dB = 20*log[Pa/.00002]

 

Applying that we get the following.

System A = 66.3dB per liter

System B =67.3dB per liter

 

To wrap up we have a very large system which offers 17dB more output than a smaller system exactly 1/8th the size. As most know if you double the systems you gain 6dB. Doubling three complete times to get to 8 total of the smaller system B results in an 18dB gain and the same total space being occupied as system A. The result is a 1dB advantage for an octet of system B. This is what the math shows in the final calculated SPL per liter.  

 

One more example of proving out the APD formula. System A was 120dB and 486L. The resulting APD (dB) rating was 66.3dB per liter. At first you may be thinking that can't be correct. Back calculating out how many systems you would need to provide a 53.7dB gain shows that it is in fact 486 total 1 liter systems each with an output of 66.3dB to produce a total of 120dB.

 

 

 

Whether this data is of interest to anyone is the question. It interests me but I cook up weird data like this all the time for my own curiosity. This is clearly a very simple way of looking at it and ignores a lot of issues that would have to be considered. Number one being amplifier power. Are both systems already capable of handling a very large amp? If so, is it really feasible to use 8x the power for the smaller system in this example? At some point efficiency does factor in. (Efficiency actually is tracked in this data for the passive systems above roughly 50 or 63Hz.) However it is ignored in the deep bass and for active, closed loop systems. Perhaps the bigger system A weighs 220lbs and uses a single large driver. What if System B uses a heavy driver itself and already weighs 75lbs despite its small size? 8 of them results in a 600lb system. That may be an issue. Perhaps the system must fire all output from a single enclosure face and this will be impossible with 8 drivers/vents etc... Cost is ignored as well. Perhaps the big system is way too big to consider to begin with? The list of factors for consideration goes on.

 

Thoughts?



#23 3ll3d00d

3ll3d00d

    Ultra Member

  • Members
  • PipPipPipPipPipPip
  • 425 posts

Posted 30 December 2015 - 08:17 PM

My immediate thought is why do you need to jump through the hoops to determine that having 8 of system B would produce similar output in a similar size space to system A? As you said "if you double the systems you gain 6dB. Doubling three complete times to get to 8 total of the smaller system B results in an 18dB gain".

 

Perhaps I'm being dim but it's not obvious to me what knowledge you gain by going under the covers (so to speak)



#24 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 30 December 2015 - 10:18 PM

It seems quick and easy in a very simple comparison like the one I posted and it is, but what about quickly comparing multiple systems at multiple frequencies with dimensions that are not so evenly divisible? Correct there is nothing done that can't be looked at by anyone with the time but isn't a visual aid and having the work already done and condensed into a simple comparable number worth something?

 

This is the kind of feedback I want so thanks.



#25 lilmike

lilmike

    Super Power Member

  • Members
  • PipPipPipPipPip
  • 157 posts
  • LocationPNW USA

Posted 31 December 2015 - 12:26 AM

The main challenge you will face with any single metric is relevance to a viewer. For example, you and I have different criteria that matter "most". You're space constrained, I'm more hamstrung by the budget I've got to work within. Who knows what someone else's main concern may be? That's why the radar plot is kinda cool, lots of unrelated metrics can be displayed in the same plot, which makes it simple to do comparisons once all of the data set is normalized to the same benchmark.

 

I do like your approach because it keeps the frequency portion of the data front and center. Of the two you presented, I prefer the first plot, but I'd suggest that you try using a log scale on the vertical axis. In my opinion, changing the presentation via axis scaling is a LOT easier to do than the math, and that also keeps "dB" out of the plot, which will only lead to confusion. 



#26 3ll3d00d

3ll3d00d

    Ultra Member

  • Members
  • PipPipPipPipPipPip
  • 425 posts

Posted 31 December 2015 - 11:32 AM

It seems quick and easy in a very simple comparison like the one I posted and it is, but what about quickly comparing multiple systems at multiple frequencies with dimensions that are not so evenly divisible? Correct there is nothing done that can't be looked at by anyone with the time but isn't a visual aid and having the work already done and condensed into a simple comparable number worth something?

 

Don't get me wrong, I'm not saying it is worthless but i'm not sure just presenting that data as is would be that helpful, or rather I'm not sure the conclusion that you described (8 of the smaller one is about equal to 1 of the big one) would be that obvious from a simple line graph. It seems like you're getting into modelling territory here and that would benefit from some user input.

 

For example something like the following would be relatively straightforward to code up in javascript

 

Let user pick 2 systems as a baseline, call these A1 and B1

Show a graph that renders it in dB as A1/B1 to show the difference between the two 

Add a control that lets you enter how many instances of each system you have (i.e. to scale up from 1 to 8 in your example), call these A2 and B2

Add another series to the chart rendering A2/B2

Add a control that lets you simulate a difference in power (e.g. perhaps you only have 50% more power)

render A2/B2 again



#27 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 31 December 2015 - 04:25 PM

You know, way back when we started this site we had a pipe dream of having an advanced simulation program built into the site that ran online and would allow people to save and share files online in a common space. Not sure if that will ever happen.  

 

I'm not sure the APD graph idea would be that useful to most people but I might do it anyway. The idea is to show which systems have the most output for their size and how different types of alignments compare to each other in that aspect. That is all. Some of that can be backed out of simulators like HR easily but that's simulated and not based on real data with all of the warts. BTW this info would be put into a big sortable systems chart like the output averages and burst data are now in addition to the graph that would be in the individual systems measurements.  

 

 

 

Mike,

I like the idea of a radar type graph to show several types of data all at once. That might be something to think about. I've not created those before but it can't be that bad. It would probably have to be an image upload rather than a live comparable graph, though. The question is what data gets put on a graph like that? Anybody want to make some suggestions or take a stab at generating one?  

 

 

 

Updated APD graph in Log (Pa) format. Meant to do that yesterday but missed it trying to get things done in a rush.

 

Attached File  PAPD Pascal Graph Example LOG.png   34.72KB   0 downloads



#28 maxmercy

maxmercy

    Super Bass Overlord

  • Moderators
  • 1,851 posts
  • LocationNW FL

Posted 31 December 2015 - 04:49 PM

I would only add an option to do (PAPD)/(Watts to get that Sound Pressure).  While many HT builds 'from scratch' have more than enough current capacity out of the wall, there are many that will use just what is currently available.  People may find out that horns may be better for them if they can spare the space.

 

JSS



#29 SME

SME

    Super Bass Overlord

  • Members
  • PipPipPipPipPipPipPipPip
  • 1,002 posts

Posted 02 January 2016 - 08:33 AM

I was thinking you wanted a metric that would be relatively independent of the particular box chosen.  It looks like instead you want a metric that applies to systems in general.  One fault I see right away is that the same driver put in a smaller box will (almost?) always exhibit more burst pressure per liter.  A HIL-limited sealed sub will drop 3 dB or 0.707 times with a halving of box size, and that will make its apparent output per liter go up.  Of course, what's really happening is that you're using more drivers per liter of space to achieve this, which only supports the old adage of "there's no replacement for displacement".

 

In reality, efficiency does matter, and for a driver in a very small (what I'm calling HIL-limited) sealed box one can expect the acoustic power per liter per electric watt to be fairly similar between boxes of different (very small) sizes.  Note that acoustic power and acoustic pressure are different quantities.  Acoustic power scales with the square of acoustic pressure.  So all else the same, every doubling of SPL is a quadrupling of acoustic power.



#30 Contrasseur

Contrasseur

    Power Member

  • Members
  • PipPipPipPip
  • 71 posts

Posted 02 January 2016 - 03:34 PM


 

 

attachicon.gifPAPD SPL Graph Example.png

 


Whether this data is of interest to anyone is the question. It interests me but I cook up weird data like this all the time for my own curiosity. This is clearly a very simple way of looking at it and ignores a lot of issues that would have to be considered. Number one being amplifier power. Are both systems already capable of handling a very large amp? If so, is it really feasible to use 8x the power for the smaller system in this example? At some point efficiency does factor in. (Efficiency actually is tracked in this data for the passive systems above roughly 50 or 63Hz.) However it is ignored in the deep bass and for active, closed loop systems. Perhaps the bigger system A weighs 220lbs and uses a single large driver. What if System B uses a heavy driver itself and already weighs 75lbs despite its small size? 8 of them results in a 600lb system. That may be an issue. Perhaps the system must fire all output from a single enclosure face and this will be impossible with 8 drivers/vents etc... Cost is ignored as well. Perhaps the big system is way too big to consider to begin with? The list of factors for consideration goes on.

 

Thoughts?

Well you have at least my vote for this. I agree that SPL is a lot easier to read, especially in the deep bass. This is of great interest to me because it very effectively demonstrates output per liter.



#31 Contrasseur

Contrasseur

    Power Member

  • Members
  • PipPipPipPip
  • 71 posts

Posted 02 January 2016 - 03:57 PM

 

 

The metric I was wanting to show is a little different than just showing the true efficiency. Yes HIL governs the deep bass efficiency of a system but that in turn also limits the theoretical maximum output. In the deep bass the maximum output is usually limited by the driver or vent displacement capabilities these days, rather than the system efficiency since huge amplifier power is cheap. I'm more interested in which systems have more output potential per unit of volume than which ones have more efficiency per unit of volume. Both are important and are linked but perhaps require separate metrics.

For a lot of people, huge amplifier power is not cheap. Consider that your average person spends probably $2-300 on a Best Buy sound system for their home theater. Especially if you're limited to a 15A circuit, whatever you have left over dedicated to the subwoofers won't be able to run most of these systems to their very limit. Dividing that remainder between 3-4 subs for a Geddes setup, and now the small-signal behavior of these systems becomes very relevant. I think it would be extremely useful to have both efficiency/volume and max output/volume.



#32 lilmike

lilmike

    Super Power Member

  • Members
  • PipPipPipPipPip
  • 157 posts
  • LocationPNW USA

Posted 05 January 2016 - 06:35 AM

Radar plots are easy, you just need to normalize the data relative to each other so the scales are the same. Excel can handle them, and the math isn't bad.

 

Metrics? Pick any of them. Compare systems, drivers, whatever. Cost, power handling, extension, size, 1W efficiency, take your pick. Just normalize things in a consistent manner. Make the "good" things normalize to a big score when more is better, but invert things when a lower score represents better. Power handling and efficiency? More is better so higher numbers are good. Extension? Lower numbers are better, so 1/extension might be a better approach, DC would be an unobtainable result, just like the real world.... Size? Smaller = better, so a 1/volume might be appropriate too. Cost is easy, bigger cost is something that's universally understood. Still, it has to be normalized, so the ceiling value is simply the highest one to date.

 

Lots of ways to do this. Shoot me an email if you want to see some examples.

 

If I could attach pictures, I would.... I can't seen to manage that here, at least not consistently.



#33 LTD02

LTD02

    Advanced Member

  • Members
  • PipPipPip
  • 38 posts

Posted 08 January 2016 - 06:10 AM

are you looking to determine output density for the subs that were measured or determining the output density in an ideally sized cab for each driver?



#34 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 08 January 2016 - 05:35 PM

are you looking to determine output density for the subs that were measured or determining the output density in an ideally sized cab for each driver?

 

Potential output density assuming unlimited amplifier or close. Important for working within defined space or shape limits. Sort of a way to investigate the effective utilization of minimum space to meet the system requirements.

 

Another cross section of this would be the actual system efficiency regardless of system size, but that will need to be looked at separately. More important for systems with space but limited budget.


  • dgage likes this

#35 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 03 March 2016 - 04:00 PM

Added graphs for the normalized burst output per liter. This will only be used for finished amplifier/driver/cab systems with appropriate limiting and amplifier reserves built in. Won't be used for K20 tests with passive cabs for obvious reasons after looking at the data. Currently the graphs are not comparable. Need to make some changes to the site to get them setup as a directly comparable graph.

 

Will be adding...

 Calculated 1 watt efficiency for all passive systems.

Calculated power applied curve during the long term measurements for all passive systems

Possibly calculated current applied as well.



#36 Funk Audio

Funk Audio

    Advanced Member

  • Members
  • PipPipPip
  • 49 posts

Posted 03 March 2016 - 05:22 PM

Added graphs for the normalized burst output per liter. This will only be used for finished amplifier/driver/cab systems with appropriate limiting and amplifier reserves built in. Won't be used for K20 tests with passive cabs for obvious reasons after looking at the data. Currently the graphs are not comparable. Need to make some changes to the site to get them setup as a directly comparable graph.

 

Will be adding...

 Calculated 1 watt efficiency for all passive systems.

Calculated power applied curve during the long term measurements for all passive systems

Possibly calculated current applied as well.

 

How hard would it be to put the output per L in the comparable charts section?



#37 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 03 March 2016 - 06:50 PM

How hard would it be to put the output per L in the comparable charts section?

 

 

That's what I meant in my post above. Eventually it will be and also in the comparable graphs tab but it requires some changes to the backend of the site, which I do not handle. It'll get done at some point.



#38 Funk Audio

Funk Audio

    Advanced Member

  • Members
  • PipPipPip
  • 49 posts

Posted 03 March 2016 - 09:26 PM

That's what I meant in my post above. Eventually it will be and also in the comparable graphs tab but it requires some changes to the backend of the site, which I do not handle. It'll get done at some point.

Sorry I misread that part, and thought it meant something else.



#39 Ricci

Ricci

    Super Bass Overlord

  • Moderators
  • 1,104 posts
  • LocationLouisville, KY

Posted 03 March 2016 - 10:02 PM

Welcome to the forum.



#40 Funk Audio

Funk Audio

    Advanced Member

  • Members
  • PipPipPip
  • 49 posts

Posted 03 March 2016 - 10:25 PM

Welcome to the forum.

Been registered for a couple years now, I was sure I had posted before, I guess not. Well I am posting, now you'll never get ride of me B)

 

It is nice to see new graphs coming, mainly I like to see things to help people who aren't as technically minded compare products in aspects they care about and understand. Like the one you just added. And something like what has also been discussed output/$ at given frequencies, and distortion compare would be nice but that one does get complicated.






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users