kenjit, I have done exactly what you outlined above. I started off by testing multiple cabinet materials by building the same cabinet from MDF, baltic birch, 3D printed carbon fiber reinforced ABS, fiberglass over a nomex honeycomb core and carbon fiber over a nomex honeycomb core. I have since tested and added laminated bamboo to this.
I used the same drivers, crossovers, ports, binding posts, wire and quantity of polyfill. The only deviation was a nominal volume difference driven by the thinkness of the internal brace. All variances in cabinet wall thickness were external. For example, I used 3/4" MDF but only 1/4" thick fiberglass or carbon fiber panels. This means that the external dimension of the MDF cabinet was 1" bigger in all directions while the internal dimensions were the same.
Frequency response using white, pink and brown noise showed no obvious difference in performance. There are small deviations but nothing that seemed to denote a significant difference in sound profile.
So I listened to each of them and there was an obvious difference in performance. The way I did my listening tests was to choose a song that has a ton of detail in it, specifically Don't Talk by 10,000 Maniacs off of the MTV unplugged live. Several of the cabinet materials did require an adjustment in volume to make sure volume was the same. I also added in a competitive benchmark speaker that I have owned for several years that is generally pretty highly regarded.
The net result was that each material behaves differently.
The CF cabinet with the simulation based crossover was perfect. Flawless. I heard details that I was unaware of because I had never heard that song on a speaker that performed that well before. Significantly outperformed benchmark.
In fiberglass, I couldn't hear the strings in the song and Natalie merchants voice sounded recessed. They didn't sound bad, but definitely didn't sound good. Polite or forgiving would be how most audiophiles would describe them. With a warm amp, they would sound horrible, maybe with a brighter amp they would have been okay. These underperformed benchmark without a tweak to the crossover outlined below..
Baltic Birch lost more detail. Portions of the piano were now missing in addition to the string, her voice very recessed in the soundstage and these sounded bad. Underperformed benchmark even after tweaking the crossover.
MDF was different. There was a hair less detail in terms of the instruments. Strings and portions of the piano were missing. However, Natalie Merchants voice sounded less recessed. Underperformed benchmark even after tweaking the crossover.
3D printed carbon fiber reinforced ABS was atrocious despite being 3x as expensive at fiberglass. My wife said it sound like I had taped a sock over the tweeter.
Ultimately, to get the fiberglass cabinet to sound good I need to make a single value adjustment to the resistor in the crossover. This eliminated the loss of detail and the slightly recessive sound of her voice but also made the speaker the tiniest bit bright but it was still less bright than my benchmark product. A great speaker.
For the Baltic Birch and and MDF cabinets I need to adjust two values of resistor and one of the capacitors and the speaker was relatively bright. No worse than a lot of commercial speakers but not what I would define as exceptional. I would only pair them with a warm amp as these were brighter than my benchmark and less appealing to me. The test drivers aren't optimal for that material.
Nothing could save the 3D printed cabinet. They were just awful. If I brought those products to market I would deserve to be beaten with a rubber hose.
I have since tested laminated "soild" bamboo and found that it performs closer to Fiberglass than MDF or Baltic Birch. A small tweak to the resistor and one of the capacitors was enough for them to outperform my benchmark. With a higher mass tweeter, it was a single value adjustment to the resistor and they sound amazing. Ultimately though, these fall short of the composite cabinets in terms of absolute detail and the size and depth of the soundstage.
This testing helped me feel confident that as the cost of cabinet materials increased, the quality of my products increased.
In the end, there are a bunch of variables that will make a speaker sound better or worse. The cabinet material is one, very important variable. Whether it be cast phenolic resin, extruded aluminum, carbon fiber, fiberglass, MDF, baltic birch or solid wood, there are pros and cons to each material and each manufacturer is going to do their best to minimize the downsides of each material.
I personally can't make a wood (MDF/Baltic Birch/Solid Wood) based speaker sound as good as a composite. Others with bigger budgets, better technology, etc... may be able to but I will say based on my own testing and experience I am heavily biased in favor of cabinets made from advanced materials.
Someday I hope to be able to experiment with other materials like stone, concrete, aluminum, titanium, 3D printed Ultem, etc... but costs for me to work with them are just too high at this time.
I used the same drivers, crossovers, ports, binding posts, wire and quantity of polyfill. The only deviation was a nominal volume difference driven by the thinkness of the internal brace. All variances in cabinet wall thickness were external. For example, I used 3/4" MDF but only 1/4" thick fiberglass or carbon fiber panels. This means that the external dimension of the MDF cabinet was 1" bigger in all directions while the internal dimensions were the same.
Frequency response using white, pink and brown noise showed no obvious difference in performance. There are small deviations but nothing that seemed to denote a significant difference in sound profile.
So I listened to each of them and there was an obvious difference in performance. The way I did my listening tests was to choose a song that has a ton of detail in it, specifically Don't Talk by 10,000 Maniacs off of the MTV unplugged live. Several of the cabinet materials did require an adjustment in volume to make sure volume was the same. I also added in a competitive benchmark speaker that I have owned for several years that is generally pretty highly regarded.
The net result was that each material behaves differently.
The CF cabinet with the simulation based crossover was perfect. Flawless. I heard details that I was unaware of because I had never heard that song on a speaker that performed that well before. Significantly outperformed benchmark.
In fiberglass, I couldn't hear the strings in the song and Natalie merchants voice sounded recessed. They didn't sound bad, but definitely didn't sound good. Polite or forgiving would be how most audiophiles would describe them. With a warm amp, they would sound horrible, maybe with a brighter amp they would have been okay. These underperformed benchmark without a tweak to the crossover outlined below..
Baltic Birch lost more detail. Portions of the piano were now missing in addition to the string, her voice very recessed in the soundstage and these sounded bad. Underperformed benchmark even after tweaking the crossover.
MDF was different. There was a hair less detail in terms of the instruments. Strings and portions of the piano were missing. However, Natalie Merchants voice sounded less recessed. Underperformed benchmark even after tweaking the crossover.
3D printed carbon fiber reinforced ABS was atrocious despite being 3x as expensive at fiberglass. My wife said it sound like I had taped a sock over the tweeter.
Ultimately, to get the fiberglass cabinet to sound good I need to make a single value adjustment to the resistor in the crossover. This eliminated the loss of detail and the slightly recessive sound of her voice but also made the speaker the tiniest bit bright but it was still less bright than my benchmark product. A great speaker.
For the Baltic Birch and and MDF cabinets I need to adjust two values of resistor and one of the capacitors and the speaker was relatively bright. No worse than a lot of commercial speakers but not what I would define as exceptional. I would only pair them with a warm amp as these were brighter than my benchmark and less appealing to me. The test drivers aren't optimal for that material.
Nothing could save the 3D printed cabinet. They were just awful. If I brought those products to market I would deserve to be beaten with a rubber hose.
I have since tested laminated "soild" bamboo and found that it performs closer to Fiberglass than MDF or Baltic Birch. A small tweak to the resistor and one of the capacitors was enough for them to outperform my benchmark. With a higher mass tweeter, it was a single value adjustment to the resistor and they sound amazing. Ultimately though, these fall short of the composite cabinets in terms of absolute detail and the size and depth of the soundstage.
This testing helped me feel confident that as the cost of cabinet materials increased, the quality of my products increased.
In the end, there are a bunch of variables that will make a speaker sound better or worse. The cabinet material is one, very important variable. Whether it be cast phenolic resin, extruded aluminum, carbon fiber, fiberglass, MDF, baltic birch or solid wood, there are pros and cons to each material and each manufacturer is going to do their best to minimize the downsides of each material.
I personally can't make a wood (MDF/Baltic Birch/Solid Wood) based speaker sound as good as a composite. Others with bigger budgets, better technology, etc... may be able to but I will say based on my own testing and experience I am heavily biased in favor of cabinets made from advanced materials.
Someday I hope to be able to experiment with other materials like stone, concrete, aluminum, titanium, 3D printed Ultem, etc... but costs for me to work with them are just too high at this time.