When is good too good?

[JonathanPenner]

Hey all,

Disclaimer: I know driver selection isn't everything and it's the design of the speaker as a whole that determines final quality. You can't just throw money at drivers and expect something great at the end.

I recently finished my first speaker build using TCP115s and ND25FAs. They're pretty inexpensive drivers and I know their quality is pretty low even compared to the RS line of Dayton Audio drivers, let alone super high end ones like Scan-Speak. I have a few other build ideas in mind, mostly using inexpensive Dayton Audio drivers, but eventually I'd also like to make a higher end 3-way speaker with drivers in the 30$-80$ price range.

I remember listening to KRK Rokit 5s (nearfield reference monitors) with a reference subwoofer and hearing how they made good recordings sound absolutely incredible. The songs opened up and came alive. But I also remember listening to other music I liked that wasn't recorded very well and hearing just how awful they sounded. Not even a verse into a song I wanted to shut it off, where on other systems I'd have it on repeat.

Since speakers are ultimately to listen to music we like and not just reference tracks, I'm wondering if there's a middle ground. At what point does a speaker become so clear and detailed that it makes good sound great and bad sound awful? Is it possible for a speaker to make every song sound great, even with their flaws? Is it to do with driver material design (metal, paper, poly, etc), distortion characteristics, etc.? Or are there voicing decisions that help mask flaws in tracks?

Or do I have a misconception that all high end speakers are going to have the same issues I heard from the KRK setup?

[johnnyrichards]

I have a pair of the Rokit 5 G3 in my home bar. They used to be my desktop speakers. They are pretty nice in my considered opinion, hard (if not impossible) to beat for the $300-350 you can pick up a pair.

This is the progression we almost all experience - enjoy our music on middling quality systems - upgrade the speakers and then search for better recorded music and so it goes. The only recommendation I have is to experiment with the EQ when listening to music that doesn't sound good on a certain pair of speakers. There is zero shame in doing so. I know people that have remastered songs/albums in software like Audacity so they sound better. Obviously you cannot restore dynamic range, but you can definitely add/remove emphasis on frequencies etc.

Unless you want to listen to Diana Krall the rest of your life, you will eventually realize that your experience is all that matters and if that involves "tampering" with the engineers end result to make it listenable than go for it.

At the extreme (silly) high end of things, companies like Wilson etc will sometimes actually introduce linear distortion in their speakers. This is no accident. They make them bassy or forward or trebly to try and make sure if a lottery winner with a taste for 2000's hyper-compressed pop music walks in to a showroom they will drop the $100,000 on a pair of speakers. So there is a precedent for modifying the output signal on a pair of speakers to target personal tastes.

[maynardg]

Please no peeling of grapes.

[AEIOU]

MAYNARDG Don't try to peel the raisins either!

[DeZZar]

At what point does a speaker become so clear and detailed that it makes good sound great and bad sound awful?

I don't think its really an issue of clarity - I personally don't believe there is anything considered "too" clear. Clear being the absence of harmonic distortion and the ability to reproduce fine detail.

More often than not it will come down to the distortion characteristics of the drivers and how the final system is 'voiced'. A lot of people don't find a perfectly flat response that pleasing to listen to. But I've never come across anyone that wanted more harmonic distortion in a music reproduction environment.

eventually I'd also like to make a higher end 3-way speaker with drivers in the 30$-80$ price range.

As with many things the law of diminishing returns applies here. By the time you've paid the money required to have good basket construction & materials, decent cone materials and a strong motor system with low distortion characteristics - anything costing more is unlikely to materially improve on the results. What's considered cheap vs expensive is actually not a definitive thing though, its relative to each person - but regardless, in my experience, the above rule applies.

Is it possible for a speaker to make every song sound great, even with their flaws?

Define great? It's possible for something to make everything generally listenable (like for example a Sonos system) but will it be the absolute best and most enjoyable reproduction of any one song - probably not. That why you start to get into questions around what sort of music you like listening to when you walk into shops selling high end gear. They know that a system that's going to make you feel like Diana Krall is singing right in front of you in your living room is probably going to make metal by comparison sound like a bag of snakes.

[wogg]

"But I've never come across anyone that wanted more harmonic distortion in a music reproduction environment."
Tube aficionados. They may not explicitly say they want more even order harmonic distortion, but that's exactly why they like their tubes.

[Geoff Millar]

Interesting topic.

Our music collection ranges from pop/rock to classics to jazz, much of which - I would say at least 40% - isn't that well recorded or mixed. Some of it, on LP or CD, dates from the 1920s, such as early Louis Armstrong or old classical recordings. We have a few badly compressed/limited modern rock records which I find un-listenable on anything. I don't have an equaliser so can't follow JR's good suggestion above, but I use Audacity to clean up and adjust the gain on some of our music.

When trying to choose which speakers to build, I made a CD which ran the gamut of genres and recording quality and listened to a few options in a DIY speaker showroom. In particular, I looked at metal v paper woofers and fabric v metal dome tweeters. Of course, their sound depends on the implementation , but paper/fabric sounded more natural with our range of music.

I still couldn't decide what to build, so I asked the designer of one of PE's favourite kits what he would suggest for our requirements and budget - music taste, room size, receiver power etc. His recommendation was spot on: MTMs with RS180P woofers and Morel tweeters. They sound great with 95% of our collection - detailed, great bass, vocals, cymbals and violins sound just right, etc. Not 'high end' cost, but definitely high end sound.

For the other 5%, mainly low-fi bootlegs and 1920s - 1940s recordings , I find the Classix IIs cope better with material such as Dylan's Basement tapes, early Beatles and stuff like Django Reinhardt. There are bum notes, rumble, hiss, '78 crackle' etc and the MTMs reproduce those extras well as they tend to extract everything out of a recording. The Classix use relatively cheap drivers, but as Paul Carmody says on his web page, they tend to 'gloss over' less well recorded music.

That said, the music is what matters: my brain soon tunes out from hiss, tinny drums or rumble and I just enjoy the track.

Geoff

[jhollander]

Also depends how the designer voices or controllers the driver with the crossover. There are some revealing speaker drivers such as aluminum cones that need extra attention to prevent them from sounding harsh and fatiguing. Agree also there are also a lot of bad recordings or "dolby"ised recordings that are unlistenable.

[Geoff Millar]

Most people have their 'top ten' recordings, but a list of 'bottom 10 unlistenable recordings' could be interesting, rather like the 10 worst films ever made.

Or not...

Geoff

[fpitas]

Yeah but often the films are fun. Seldom so with poor recordings.

[fpitas]

I agree that you can't overdo clear and detailed. But don't make the assumption that high buck commercial speakers are "better". Often the frequency response is mediocre to frightening.

[Geoff Millar]

I think that depends on your collection: if all we had was 60s or before, or poorly recorded and /or mixed material, I'd rather have less clear and detailed speakers.

I found retail high priced speakers disappointing, one reason why I decided to go DIY. Re frequency response, I think there's one nameless but famous company which takes legal action against people who publish its products FR graphs....

[jim85iroc]

I've never really been a big subscriber of the "good speakers make bad recordings sound worse" kind of thinking. There are a lot of recordings that sound bad when I listen on a nice set of speakers, but they don't sound any better on crap speakers... it's just that the good recordings get brought down closer to that bad recording's level. The bad recordings still sound bad.

There's nothing about good speakers that will make a bad recording sound worse except more relative output in a range that better exposes the recording's flaws. If you wanted to design a speaker that helped to "smooth over" the flaws of a recording, I'd still use high quality, low distortion drivers, but design the system to have a bit more of a fletcher munson curve as opposed to flat. Take away a little midband, especially around 5khz, and the speaker will likely be more forgiving of bad recordings, but will also have less perceived detail on everything.

[fpitas]

I've never really been a big subscriber of the "good speakers make bad recordings sound worse" kind of thinking. There are a lot of recordings that sound bad when I listen on a nice set of speakers, but they don't sound any better on crap speakers... it's just that the good recordings get brought down closer to that bad recording's level. The bad recordings still sound bad.

There's nothing about good speakers that will make a bad recording sound worse except more relative output in a range that better exposes the recording's flaws. If you wanted to design a speaker that helped to "smooth over" the flaws of a recording, I'd still use high quality, low distortion drivers, but design the system to have a bit more of a fletcher munson curve as opposed to flat. Take away a little midband, especially around 5khz, and the speaker will likely be more forgiving of bad recordings, but will also have less perceived detail on everything.

I agree with most of that. A little overall downward slope, like Troels Gravesen often does, also helps "warm up" crappy recordings, without hurting the good ones very much. Once upon a time the listener just tweaked the tone controls a little. Now we have the illusion of audiophilia while the manufacturers save by not installing tone controls.

[JonathanPenner]

Thanks everyone for your input! A lot of great responses.

Some takeaways so far:

• Metal cones can sound harsher without proper care in the crossover
• Paper/fabric cones are generally more forgiving
• The amount of HD is less important than the characteristics of it. (I remember an interview with Dr. Earl Geddes where he said something similar)
• Speakers lend themselves to certain genres of music. A set great for jazz won't necessarily work for rock.
• To sound more appealing for most music (à la Sonos) voicing choices like downward slopes and a dip in the midrange can be used, at the expense of making other music not sound its best.
• EQ can be used similarly, either by modifying the track directly or using external EQ.

About HD, are there specific things to look out for from a HD measurement graph? Right now I only know to broadly look for amplitude across the frequency range (e.g driver X has lots of HD <80Hz).

For my purposes, I suppose a way to approach it is voice a speaker to make 70% of my collection sound good, then use EQ in some capacity (or build another set of speakers) for the remaining 30%. But I'm also in this for the fun of the actual design and build process, so maybe as long as I get something enjoyable I shouldn't worry too much

I might have opened a can of worms by associating price with quality. I absolutely know the cost vs quality scale is no where near linear or even accurate. What I was more implying is other than space characteristics (Vas, diameter), most can agree the RS180P-8 sounds better than a TCP115-8. The RS180 also costs almost 4x as much. The idea by saying "the 30$-80$ price range" was looking at things like an RS180P-8 or W5-1138SMF. That said, given how little experience I have in this hobby I'm still not familiar with how a lot of drivers can sound. I would also be interested in threads about excellent value drivers.

[rpb]

Thanks everyone for your input! A lot of great responses.

Some takeaways so far:

• Metal cones can sound harsher without proper care in the crossover
• Paper/fabric cones are generally more forgiving
• The amount of HD is less important than the characteristics of it. (I remember an interview with Dr. Earl Geddes where he said something similar)
• Speakers lend themselves to certain genres of music. A set great for jazz won't necessarily work for rock.
• To sound more appealing for most music (à la Sonos) voicing choices like downward slopes and a dip in the midrange can be used, at the expense of making other music not sound its best.
• EQ can be used similarly, either by modifying the track directly or using external EQ.

About HD, are there specific things to look out for from a HD measurement graph? Right now I only know to broadly look for amplitude across the frequency range (e.g driver X has lots of HD <80Hz).

For my purposes, I suppose a way to approach it is voice a speaker to make 70% of my collection sound good, then use EQ in some capacity (or build another set of speakers) for the remaining 30%. But I'm also in this for the fun of the actual design and build process, so maybe as long as I get something enjoyable I shouldn't worry too much

I might have opened a can of worms by associating price with quality. I absolutely know the cost vs quality scale is no where near linear or even accurate. What I was more implying is other than space characteristics (Vas, diameter), most can agree the RS180P-8 sounds better than a TCP115-8. The RS180 also costs almost 4x as much. The idea by saying "the 30$-80$ price range" was looking at things like an RS180P-8 or W5-1138SMF. That said, given how little experience I have in this hobby I'm still not familiar with how a lot of drivers can sound. I would also be interested in threads about excellent value drivers.

Everything matters.... at least a little, and sometimes more. If you use a sub, the demands on the mid-bass are greatly reduced. If you don't want a sub, then don't try to get bass from a 4" woofer. You will be dis appointed. I have some 6.5" woofers that work pretty well in my room. The bass is good, but they need a 1 cu-ft box. I have a smaller pair of speakers with a 5.25" woofer. They are nice with a sub, and the size looks like I want. A few months ago, I temporarily assembled a 3-way with a 10" woofer, 6.5" mid, and 1" tweeter. It probably smokes anything else I've built, but it's rather large. My wood working skills are not too good, so it would be big, and ugly. Small and ugly + sub is easier to live with!

More on the topic. I have a theory. Much of the older music has limited bass. In a system with a sub, if the blend is not really good, the bass in an old recording may never quite make the sub audible, and the bass that is rolled off at 80hz just fades away. Perhaps a speaker with prominent bass at 60hz is better for such music. A quick look at the RS180p shows very limited bass. (f3 of 74hz.)

I design my speakers to sound good with the music I like the most. If other music sounds so so, then I just don't play it much, if ever. I'm sure there is some compromise involved in my voicing, but generally my response is close to flat. (As measured with my non calibrated mic.)

[DeZZar]

I guess the challenge here is that a more expensive driver "might" be better.

I really like the RS180P and am working with it at the moment. In my country I can get it for around $80 wholesale (and unfortunately this is still a 'budget' driver - 30 bucks won't get you much in Aus)

In my opinion though the SB Acoustics MR16P kicks it to the curb. It goes for $180. I personally still find that reasonable for a top quality driver.

It can come down to the cost of manufacture which is a factor that does not correlate with the quality of the outcome. SB is my go to at the moment. You're getting the classic Danish engineering for the cost of Indonesian manufacture. The build and sound quality are outstanding and I think that represents excellent value.

You really won't do better than the SB by paying double+ for a scan-speak equivalent which suffers from a much higher cost of manufacture!

[rpb]

I guess the challenge here is that a more expensive driver "might" be better.

I really like the RS180P and am working with it at the moment. In my country I can get it for around $80 wholesale (and unfortunately this is still a 'budget' driver - 30 bucks won't get you much in Aus)

In my opinion though the SB Acoustics MR16P kicks it to the curb. It goes for $180. I personally still find that reasonable for a top quality driver.

It can come down to the cost of manufacture which is a factor that does not correlate with the quality of the outcome. SB is my go to at the moment. You're getting the classic Danish engineering for the cost of Indonesian manufacture. The build and sound quality are outstanding and I think that represents excellent value.

You really won't do better than the SB by paying double+ for a scan-speak equivalent which suffers from a much higher cost of manufacture!

The RS180p looks like it would be a good option if used with a sub. I really like the RS270p driver I have. If the 180 is similar, it's a winner..

[rpb]

The PE data sheet shows 74hz in a small ported box. It's good to know that better alignments work for it.

[Geoff Millar]

There's another MLTL design for a RS180P MTM somewhere on PETT, but it's a relatively large cabinet; if I can find the link I'll post it.

Geoff

[wogg]

I get down to 40Hz with the RS180P-8 in my build Supernova Minimus. It's an EBS alignment, so they benefit from a little tone control or wall placement. I've got them set with +6 on the AR bass tone control, measurements at the listening spot without gating show decent bass to 40Hz as well with a nice slow downward slope all the way up. Based on the model with the PE specs, the F3 should have been 60, but they ended up tuned a little lower and perform better than that.

http://woggmusic.com/supernova-minimus-speaker-build/

[fpitas]

Thanks everyone for your input! A lot of great responses.

Some takeaways so far:

• Metal cones can sound harsher without proper care in the crossover
• Paper/fabric cones are generally more forgiving
• The amount of HD is less important than the characteristics of it. (I remember an interview with Dr. Earl Geddes where he said something similar)
• Speakers lend themselves to certain genres of music. A set great for jazz won't necessarily work for rock.
• To sound more appealing for most music (à la Sonos) voicing choices like downward slopes and a dip in the midrange can be used, at the expense of making other music not sound its best.
• EQ can be used similarly, either by modifying the track directly or using external EQ.

About HD, are there specific things to look out for from a HD measurement graph? Right now I only know to broadly look for amplitude across the frequency range (e.g driver X has lots of HD <80Hz).

I've certainly heard speakers that do all types of music well. The Revel Salon2 and the JBL K2 S9900 come to mind, but I'm sure there are others. Very different designs and drivers, but the engineering is solid. Any speaker is the sum of its parts. Drivers get all the attention and press, but good designers also obsess over things like the crossover, the polar pattern, having a sturdy non-resonant cabinet etc.

[Geoff Millar]

Agreed, good speakers should work well with all types of music; some, like Paul Carmody's "Tarkus", could be 'voiced' for rock music but will still play anything well.

Geoff