N scale CPR/BCR Royal Hudson build thread

[craigolio1]

I agree with you Otto.

Scaled down sound is key. Think of how we have to be from the tracks for the prototype to look like our N scale models do. What do they sound like at that distance? They certainly don't pound the ground like they do when you are standing beside the tracks.

We are not dealing with a producing the most bass possible scenario. If we wanted that we could put a stationary sound decoder hooked to an amp and 15" woofer under the layout. Consist that with your train and voila, realistic ground pounding. But that would be cool for about 10 seconds before it became invasive.

We're dealing with a producing the most bass we can from a tiny speaker. Because our speakers really don't produce any bass, the lower we can get them to extend the better as there are lots of things around the 300-400 hz range that would sound better if we could get just a bit more volume in that range. So the displacement becomes important because the more a speaker can move air the louder it is and this goes for the lower end of its frequency response as well. Given the choice between a high excursion aka long throw speaker and one that is not, the long throw will sound better.

I didn't know the Sugar Cube is a long throw. I'm glad there is one in the mix and look forward to seeing (and of course hearing ) the difference.

Craig

[craigolio1]

OK talk is cheap so here we go. 

I got around to measuring today and wound up with sets of measurements.

Here are the constants:

All speakers are mounted in a sealed enclosure of equal volume, with exception of the Zimo as it came with an enclosure already attached.
The speakers are all sealed to their enclosures with silicon.
Any of the speakers not pointing through the floor, were pointing forward through the front of the tender with the exception of the dual speaker test which had one facing forward at the front of the enclosure, and facing backwards at the rear of the enclosure.
The speaker enclosures are all sealed to the frames with silicon.
Enclosed tender, and tender with the front open to the cab refer to the tender body that fits over the speaker enclosure.
The bodies were not sealed to the frames.
The tender frames were suspended above a piece of track at the height they would be at on trucks to provide the same reflection.
The same amplifier and volume settings were used for each test.
All speakers are 8 ohm impedance.
In the dual speaker tests the speakers were wired in parallel resulting in a 4 ohm amplifier load
Each speaker received a maximum of 1 watt of power.
The mic was placed at the same location for each test.
Even if a tender body was used that did not project sound forward into the cab, the cab was placed in front to provide for the most uniform testing.
Testing was done from 200Hz to 20 KHz since these tiny speakers don't produce much usefull sound below 300Hz.


Here's how it was set up during testing. I put the mic in a position where it could collect sound coming from under the tender and also from the front near the cab so that I could use the same mic position for every test.



Types of installations tested:

There were two types of speaker installations, a speaker mounted in a cut out through the floor, and a speaker mounted inside the tender.   

There were two types of tender bodies, one with the sound playing through a tender body with a cut out in the front of the tender, and an enclosed tender body with no purpose made openings.

Combinations of the above mountings and bodies resulted in the combinations I tested.


Cut out in the floor.

Knowles-Fox Mounted facing through the floor, with the enclosed tender
Knowles-Fox mounted facing through the floor, with the tender open to the cab
Loksound oval speaker facing through the floor, enclosed tender

Mounted inside the tender, with an enclosed tender body

Zimo Sugarcube
Knowles-Fox single speaker
Knowles Fox pair of speakers

Mounted inside the tender, body open to the cab

Zimo Sugarcube
Knowles-Fox single speaker
Knowles Fox pair of speakers

Results:
On the left side of the results you will see a range of decibels.  Decibels are used to measure SPL (Sound Pressure Level), or volume.  Decibels are something that have always confused me a little and honestly they are still a bit of mystery to me.  They are not an increment of volume, but more like a ratio of one volume level in comparison to anoter.   To double the volume equates to an increase of 10dB.  Example.  20dB is twice as loud as 10dB.  30dB is twice as loud as 20dB, etc.  In car audio we always said that to double the volume you needed 10x the power so 10x the power would in theory yield and increase of 10dB.  Doubling your power, or adding a second identical speaker using the same power, would increase 3dB.  Good, now that you are as confused as me we can proceed.

Below are the results all plotted together on the same graph.


 It looks like a big plate of spaghetti, and so I will be splitting them up for comparison, but it just goes to show you how much of a difference there is in the resukts of each of the tests. 




Here is the same graph spread out a little and zoomed in so you can see more definition between each of the results.






Next is the individual speaker types on their own.
Here are the results for the two Sugar Cube tests.




Loksound.




Knowles-Fox through the bottom.




Single Knowles-Fox.



Double Knowles-Fox





According to the software, the installation with the loudest average SPL was the Sugar Cube in the fully enclosed tender. Keep in mind though that it's a difference of bareley one dB, and this is an average.  Huge dips in response at 6k, 18k and 20k ruined the curve.  The fact that those frequencies, at such narrow bands, are basically non existent to your ear wouldn't change the volume you hear.  Looking at the graph you can clearly see that the open through the front test produced more perceivable volume.




The installation with the lowest SPL was the Knowles-Fox facing out the bottom with an open tender front.




The installation with the flattest frequency response was the Knowles-Fox facing out the bottom with the open tender front.

http://i1203.photobucket.com/albums/bb383/craigolio1/Speaker%20tests/Flatest_zpsmrwdplyk.jpg

Here are the best three installation for flat frequency response.



And the worst three.




All speakers showing comparison of large dips in response at particular frequencies. Spread over three pictures for better spacing.






All speakers in the enclosed tender.




All speakers in the open front tender.




All speakers open through the floor.




All Speakers bottom end or "bass response"




Above Average bottom end.




Here are some of my observations. 

In the installations that play through the bottom, there is a giant hole at 12KHz.  In same cases a drop of over 35dB.  Consider that a drop of 10dB means half the volume.  That's a huge hole.  Also the three most flat responses, which is considered ideal, came from the installation where the speakers pointed unobstructed out through the floor.  This could have been predicted I think.  It's exactly why speakers are designed with the speakers facing the listener as Peteski and I discussed before.

In the installation with the two speakers there is some funky stuff going on.  My guess is that since the two speakers are mounted 180 degrees out of phase (same as wiring one backwards if they were side by side facing the same way) and the sound is bouncing around inside before finding an exit, that there is some cancellation going on there (as predicted by John).  It may have a different result if I had the room to put the speakers in phase with each other and side by side.  Cancellation is normally more prevalent at lower frequencies but in this case it's happening all over the place wreaking havoc.  Basically, as I have installed it, it sounds like crap and could be much better with alterations.

The graphs all show that the more open the speaker is to the outside, be it through the front, or even better, through the bottom, the better the high frequency response.  Makes sense.

I find it interesting that the lowest "bass" response came not from the largest speaker (as I would have predicted), but from the two double installations (as predicted by John) and from the tiny singular Sugar Cube. That being said, you can see that it rolls off steadily after 800hz and by the time it reaches 200hz its down about 30dB. The extra extension might not even be audible.

Finally because graphs are just a bunch of pretty lines and can't tell you what something actually sounds like, I did some quick  listening with music.  The speakers that played out the bottom sounded the best.  Any of the ones enclosed were missing a lot of highs (seen in the graphs), the ones with sound allowed to escape through the front sounded better than the enclosed but not as good as the open through the floor (seen in the graphs). 

I still have more testing to do but I'm interested in finding the room to get the Sugar Cube pointing out through the floor. As it is it's a little tall.

Craig

[peteski]

I didn't know the Sugar Cube is a long throw. I'm glad there is one in the mix and look forward to seeing (and of course hearing ) the difference.

Craig

Well, this is strictly from observation.

I just gently prodded the speaker cones with my fingers (or some non-magnetic rod) and a also moved the speaker cone electrically (by applying voltage to the terminals). The flat cone rectangular speakers we call sugar cube have very softly suspended cone with quite a long throw.  They also use very strong rare earth magnets which will also aid in deflecting the voice coil.   Most of the very small size conventional speakers used in the N scale sound installs in the past seem to have much stiffer cone suspension, smaller cone deflection and weaker ceramic magnets.

[craigolio1]

I find it interesting though that the Zimo Sugar Cube tested produced much better results than the Knowles-Fox.  They look the same and are of the same design, but obviously they are not the same.  One thing that is different is the KF is in an enclosure I made which is at least twice the size of the one the Zimo is in.  Perhaps I need smaller enclosures for the KF?

Perhaps I'll retest in a similar size enclosure and compare.

Craig

[peteski]

Awesome post Craig!  You sure have some cool tools at your disposal.  Interesting conclusions (but not unexpected, at least to me). I like the scientific approach.  They sugar cube speakers sure changed how we do sound in N scale.

[peteski]

I find it interesting though that the Zimo Sugar Cube tested produced much better results than the Knowles-Fox.  They look the same and are of the same design, but obviously they are not the same.  One thing that is different is the KF is in an enclosure I made which is at least twice the size of the one the Zimo is in.  Perhaps I need smaller enclosures for the KF?

Perhaps I'll retest in a similar size enclosure and compare.

Craig

Yes, enclosure size will affect the sound reporduction.  Ideally it would an infinitely large enclosure fully isolating the front and back of the speaker cone.

[craigolio1]

Thinking back I wasn't clear on why I want to work more with the Zimo. I was testing a couple of the installs, with music, against the Loksound speaker and one of the other through the floor speakers sounded better, as did the Zimo without a tender body on it.

I'm thinking I'd like to cut another hole beside the KF thats through the floor so that I can have a pair of them side by side, through the floor, and physically in phase.

I'd also like to hear the Zimo through the floor.

Craig

[jdcolombo]

Yes, enclosure size will affect the sound reporduction.  Ideally it would an infinitely large enclosure fully isolating the front and back of the speaker cone.

Actually, "infinitely large" is not necessarily the best.  The infinite baffle is perfect for isolating the front wave from the rear wave.  But the folks who started Acoustic Research in the 1950's discovered that if you use speakers that are capable of large diaphragm excursion (like our "sugar cubes"), a better design is to make the box small enough that the air inside the box acts as a sort of "spring" to help both extend the diaphragm excursion and to dampen it.  The AR-1 speaker was a revelation for its time, providing smooth, extended frequency response in what was then considered an impossibly small box of 1.7 cu. ft. 

Building a truly accurate acoustic suspension design, of course, requires matching the free air space in the enclosure to the resonant frequency of the drivers and their "Q" - the elasticity of the diaphragm and its excursion capability.  There are formulas for doing this.  But with the tiny speakers we use, none of which are capable of any acoustic output below 300hz, and not much between 300-500hz, a "close enough" approach is workable.  Knowles settled on 1 cubic centimeter of free air space for its speaker line, but I've found by experimental testing that you can get by with about 10% less than that, and that for an 8x12mm, an enclosure with about 700 cubic mm is fine.  And bigger is NOT better in this case, because at some point you lose the "spring" effect of the trapped air.  This may be why the Zimo seems to be the star performer here: it's enclosure size is more accurately matched to the driver, using the "spring" effect to full advantage. 

And remember, the SHAPE of the enclosure doesn't matter.  It can be a box wrapped around the speaker; a long-thin box with the speaker off to one side; a cone; a cylinder; a pyramid; etc. - it doesn't matter.  Only the internal free air volume is critical.

John C.

[CNR5529]

Craig, awesome results. Looking forward to translating your findings into a frame design we can manufacture! As mentioned offline, it will be interesting to compare the results from the final design printed in metal vs these tests on plastic frames.

I still have more testing to do but I'm interested in finding the room to get the Sugar Cube pointing out through the floor. As it is it's a little tall.

I cant remember which Zimo sugar cube speaker I sent to you for testing, but there are a few different sizes http://www.zimo.at/web2010/products/decoderlautneu_EN.htm. This still gives us some extra room to play with.

[craigolio1]

Thanks.

And John I wasn't aware that these tiny speakers have specified enclosure sizes. That's very helpfull.

Based on what John pointed out I think it's a good idea to retest the Knowles Fox in more suitably sized enclosure, something in the 1000-700 cu mm range. It would be a fair comparison the the Sugar Cube I have.

Craig

[Lemosteam]

@CNR5529 , I got my metal tender frame yesterday.  I will share my shrinkage findings later.

What material are you planning on using?  I used stainless steel.

[CNR5529]

What material are you planning on using?  I used stainless steel.

I will be using the stainless steel materials as well, either raw or matte black.

Just to add to the sample size, I have been printing out another part for audio turntables without adjustments and they have been coming back at 97.14% of the target size.

[craigolio1]

I finished the final round of testing using new enclosures for the Knowles Fox speaker, and with the Zimo Sugar Cube positioned to face through the floor of the the tender.

Here is a comparison of the new Zimo test, and the previous Zimo tests:



The difference in frequency response around the top end is obvious and thats reflected in the listening tests.  Pointing the speakers directly through the bottom of the tender produces a much more favourable sound.


Next is a comparison of the Knowles Fox single speaker, in a new sealed 1 cu cm enclosure:



Thanks to John for the information on enclosure size as it's obvious it made a difference.  You can see improvements in low end response, flatness and SPL.


Here is a comparison of the same speaker, but using two of them side by side, both in their own sealed 1 cu cm enclosures.



In this test we can see improvements that most likely stem from changing the physical phase of the speakers from 180 degrees out of phase to in phase with each other.  That took care of a lot of the frequency dip issues in the top end.  However there was a reduction in the response in the bottom end.


This next graph is a comparison of all of the speakers that were tested through the bottom.  This includes the Lok Sound large oval speaker.



While there were improvements in the sound from all of the installations I tested this time, it's clear that the Zimo is the winner.  There's a 5-10dB increase in SPL through the bottom end, higher SPL, and flatter response then most. 

This was obvious during listening tests with both music and YouTube steam train videos (Royal Hudson of course!).

My Royal Hudson model will be using a Zimo Sugar Cube for it's sound installation.

Craig

[Scottl]

Craig,  this is very cool!  I have a pair of ear bud headphones from Sony that are "bass-enhanced".  They really are, yet the drivers are tiny.  How do they do it, compared to the problem you are working with?  Is there a way around the physics here? 

[jdcolombo]

Very interesting, Craig.

I wonder what OEM speaker Zimo uses . . . is it an 11x15 (looks like it from the photo)?  I've generally found that the 11x15 speakers are better performers than the 9x16 (e.g., Fox).  I think the reason for that might be that the diaphragm is more square, and less stretched out in the 11x15, so there is more uniform movement.  The old Knowles Dumbo 13x18 was my favorite, but alas it is long gone.

John C.