How do you finalize/tweak your design when it's 'half built' (noob help please)

[nannoo]

This is my first speaker build, is based on drivers from a B&W XT4 speaker, one bass, mid, and tweeter per side.

I used:
Crossover Design Calculators
to calculate 4th order and 2nd order crossovers at 3.5Khz and 350Hz

Now they are half built and testing, they sound terrible (and quiet). Mostly due to a 'ringing' sound in the high end of the mid driver, it sounds peaky at low levels and HURTS at high levels. Is this modal break up of the cone I am hearing?

I've started tweeking the xo’s by manually inserting/removing components in various configns and observing the response. This has increased level output and overall quality, but there’s obviously room for improvement.

here's the quesions:

- What is your advice for tweaking?

- in my position, would you save money and tweak crossovers by tweaking and listening, or would you give in to buying a measurement MIC and build the things properly?

 

[SY]

Measurement software- impedance vs frequency is critical to crossover design. CAD software- textbook formulas only work for resistors, not actual drivers; don't even think about going there. Measurement mike- gotta see if you're doing what you think you're doing! And finally, a book like Joe d'Appolito's to understand how to perform and interpret your measurements.

This all adds up to a couple hundred bucks, well worth it, since once you do a successful project, you'll be itching to do more, and the cost gets amortized.

 

[AuroraB]

the cost gets amortized.

Says who ? -- 'cause that's when it starts to get really expensive !

 

[nannoo]

"the costs get amortized"

This project has already cost WAY too much!... I already have revelator drivers sitting in the cupboard, so costing is just 'oops'.

Tweaked the xo more since my last post, finally the speaker is sounding nice and like a speaker, more importantly the performance is getting closer to what I hoped for - I can now hear details I simply couldn't hear on my previous speakers.

But I'm glad you said 'yes' to the measurement mic, I was very close to pulling the trigger on amazon, so now it's certain.

 

[Jay]

- in my position, would you save money and tweak crossovers by tweaking and listening, or would you give in to buying a measurement MIC and build the things properly?

You seem to be obsessed with speaker design. It's very fun as a hobby (and expensive too), but be aware that you should not expect to be able to create your own good speaker in a year or two. I'm trying to tell you the truth here. Not to dis-motivate you but to put things in the right perspective.

First, build a proven design from a well known designer. Understand the design, then build your own alternative. By doing this you will know if you can do better or not. You will not pretend that you can do as well as the well known designer because you just will know.

Doing it by ears without prior background of doing it by measurement or simulation tools is a waste of time. There is just no way one can make a proper speaker that way, period.

By using the calculator you don't learn anything, and the outcome is very very poor. There is an easier way, that will teach you a lot of thing, and will give better result than that calculator, and will bring you to the right track of speaker design practice.

By using calculator, you usually use only one parameter, the assumed impedance of the drivers. You can do the same thing with Speaker Workshop by using ANY driver that has similar parameter. Just pick any arbitrary driver that physically similar with your driver.

Then you will learn to use more data to get more accurate result. You can start with using data published by manufacturer, and even later measure by yourself the T/S parameters of your drivers (you need the mic).

With experience you should build several skills, including voicing skill (you will understand the relationship between any component change with measurement and how they actually sound). The more you build speakers, the more you may rely on your ears. Measurement tools can only give you average result. You need your experienced ears to go beyond that.

 

[nannoo]

You seem to be obsessed with speaker design.

Hey Jay,

I was always convinced that DIY was the only way to achieve high-end without the high price, and that probably stemmed from reading about and understanding the realities of manufacturer mark-up about 14 years ago, around the time I first discovered the madisound webpage!

But you're right, I am obsessed with design, I have been my entire life, it's what one lecturer at my university kindly terms 'the engineering brain'. It's in our nature to constantly seek a better solution.

I was initially scared away from DIY on the basis of not understanding the electronics side of things, however as I returned to education (i'm currently finishing my 4th year of Engineering) I have learnt much of the theory and fundamentals, in addition to reading many DIY forum sites over the years, build logs, douglas self's excellent works on amplifier design and his loudspeaker design cookbook (which I must admit this is more reference at the moment as I learn to gain an intuitive understanding of things)

Douglas Self convinced me of my initial beliefs as an engineer - I am without doubt an objectivist, and my idea of audio nirvana is an amplification system that take what you put in and amplifies it within your listening environment with as little variation as possible.

Despite this I do believe certain 'distortion mechanisms' can still be perceived as 'better' (Vinyl source & valve amplification seems like an obvious example) And this is something which I am reading about.

On the subject of intuition, there's only so far you can go with theory and copying others, eventually you need to come up with your own ideas founded on solid theory and intuitive understanding. This has always been my practice and would of course be my aim one day with audio, be it for personal or commercial gains.

These speakers I am working on represent my deliberate 'throwing myself in at the deep end' - becuase no driver parameters are available it forces me to learn to measure them myself, and even just spending time listening to the crossovers whilst making adjustmens has given me abundantly more experience and intuition than reading books would have done.

I fully accept that I am noob, but I am doing my best to learn, and this is a learning exercise!

On the subject of adjustment by ear, I should point out that I've spent a reasonable amount of time working in recording studios (professional studios) producing music (at home) engineering live sound (for some reasonable well known bands) and various other audio-related crafts, so I do have a good understanding of what 'good sound' is and what it should be like. Admittedly only for my own ears, but I've never had any complaints

Sorry for the rant, just thought it would all make a bit more sense if you knew where I was coming from!

I've ordered the measurement mic, so I look forward to getting some real-world measurements of these speakers within a few different rooms, then you guys can tell me how bad I did!

Oh, and the speakers are currently sounding comparable to other £1200 ish speakers I've auditioned with Hi-End Naim gear, and that's from an old Denon AVC-A10SE amp with my computer as source (incidently a stupidly cheap way to get a half decent dac, stereo amp and surround sound decoder in one, mine was £80 off ebay, I also have an AVC A1D which cost me £28 'broken' and I have since repaired, the torroid in the thing would have cost twice as much alone!) ... though this is all highly subjective.

Double Oh, I should point out I started this project cos I was offered the drivers for £120 for the set, which I figured was good value. If I'd have known then what I know now about how long it takes to design, make, test a speaker properly, I'd have just bought a pair of PMC's and been done with it, but hey, I've learnt a lot along the way and had great fun, just a pitty this hobby seems to eat time and money, two resources I'm guessing we're all short of!

 

[Lynn Olson]

All professional designers measure, using the best gear they can get their hands on. Advanced measurement systems used to be very expensive, as much as $150,000 back in the Seventies, when it took an anechoic chamber, an on-staff Fortran programmer, and DEC PDP minicomputer to accurately FFT measure individual drivers and the complete loudspeaker system.

The advent of the PC-based MLSSA system dropped the cost down to about $3500 in the early Nineties, and became the standard of the industry. You can now use most any soundcard, half-decent measurement microphone, and a software program like ARTA to measure loudspeakers, so the cost excuse is gone. However, that still leaves a pretty steep learning curve in how to design loudspeakers, crossovers, and how to measure individual drivers and the overall system.

I don't want to pop any bubbles, but those fancy Scan-Speak drivers are completely worthless without a measurement system and the know-how to use it.

Alternatively, if you don't want to go down the rabbit hole of learning a minor engineering discipline, you can seek out the many projects on the Web, and ask around what builders think of the results. In terms of cost-effectiveness and best use of your time, the Web-project approach is the most direct approach.

Think of building your own race car from scratch. The Scan-Speaks are equivalent to a brand-new turbocharger kit ... not much good without the rest of the car. The measurement system is similar to a dyno ... there's not much point modifying the engine unless you have access to a dyno.

The alternative is a kit car; somebody else has already done all the hard work of making sure the thing works and doesn't catch on fire the first time you drive it. (Fortunately, defective loudspeakers don't catch on fire, they just destroy drivers and sound bad.)

 

[nannoo]

Wow, that's two very negative responses now.
Is there some taboo about getting into DIY? Are you wishing me not to bother?
I made it clear that I am learning, studying, and that it is in my nature to design and develop new solutions. It is not in my nature to take anothers work and call it my own, however I am happy to learn from it.

Please, if you have nothing positive to say, get afk.

 

[SY]

I tried to be positive...

This is a real project, and you're going to have to accept that it will take some additional resource, both financial (though relatively minor) and intellectual (major). But the rewards will extend well beyond a great sounding pair of speakers- you'll have learned a LOT of new things, and gained confidence and fluency in measurements and how to interpret and use them. The next set of speakers you build will be even better and even easier for you.

edit: I should add that a three way speaker is six times more difficult than a two way.

 

[nannoo]

SY

Thank you so much for your input, I have always had a lot of respect for the comments you leave on the forum.

One of those was talking about the 90-10 balance, and how all the work to get a speaker sounding its best comes when you're actually working with it.

Beyond tweaking the enclosure volume/port tuning, and all the obvious things to do with trying to get the overall response as flat as possible via crossover filtering, are there any specific steps/advice you would give that may not seem obvious at first?

(and point taken about 3-way vs 2-way complexity, the mid driver in this speaker is obviously designed to work across the B&W XT range, it extends pretty low, and for a little while I was getting better performance running the design as a 2-way without the bass drivers)

 

[Pano]

Please, if you have nothing positive to say, get afk.

Really? I see nothing but, good, solid, friendly advice in the replies.

If you wanted to build an airplane and not have it crash on the first flight, would you accept words of wisdom from those who have build dozens and crashed quite a few?

It can be fun to throw together a few cheap drivers and parts and make something that plays music, but you've already gone further than that. Take the advice of those who have done this before, they are all encouraging and helpful.

As for the simple stuff, your first posts sound like you don't have the attenuation and levels right between your drivers. And/or you've wired something wrong.

 

[sreten]

http://sites.google.com/site/undefinition/diy
(see if nothing else, the excellent FAQs)
The Speaker Building Bible
Zaph|Audio
Zaph|Audio - ZA5 Speaker Designs with ZA14W08 woofer and Vifa DQ25SC16-04 tweeter
FRD Consortium tools guide
http://web.archive.org/web/200909021...esigningXO.htm
RJB Audio Projects
http://web.archive.org/web/200909022...ve99/Spkrbldg/
Speaker Design Works
http://www.htguide.com/forum/showthread.php4?t=28655
A Speaker project
DIY Loudspeakers
Humble Homemade Hifi
Quarter Wavelength Loudspeaker Design
Pi Speakers - unmatched quality and state-of-the-art performance
The Frugal-Horns Site -- High Performance, Low Cost DIY Horn Designs
6.283 Audio Pages
Linkwitz Lab - Loudspeaker Design
Music and Design

Great free SPICE Emulator : SPICE-Based Analog Simulation Program - TINA-TI - TI Software Folder

Hi, work it out for yourself, rgds, sreten.

 

[ODougbo]

Yes, just thinking the same....all good advice, from folks that have been down the same road.

Post some sketches of your xo, links to your drivers (less work for the folks than can help you).

And yes time consuming but a interesting hobby, that's the point for some of us.

 

[nannoo]

Really? I see nothing but, good, solid, friendly advice in the replies.

re-reading the replies it does look like I interpreted them wrong

I was just hoping for more 'advice' and less 'it wont work' if you know what I mean.

I'm aware it's a steep learning curve and in honesty I'm starting to hit that pretty hard, so it's easy to get discouraged. On the other hand I'm very excited about getting a measurement mic in the post and finding out what these drivers are actually doing!

 

[Lynn Olson]

I was serious when I was comparing speaker design as comparable to performance-modifying a car. You can take your car to a shop and have them do all the work, or you can really dig in and learn engineering. The second path is far more rewarding, but it does mean a serious time commitment and a willingness to learn a lot of weird stuff that's not all that interesting (at first). I'd be the first to admit the arcana of FFT and MLS measurements, and the requirement of correctly setting the time window, aren't all that much fun and don't seem to have much to do with the thrill of hearing your new loudspeaker for the first time.

Most of the "sound" of the speaker, assuming no gross errors, comes down to the refinement of the crossover. If the crossover is far from optimum, you won't hear the quality of the drivers at all. When the crossover is close to optimum, say 90% of perfect, then the drivers start to be audible as good versus mediocre drivers ... but it takes experience to distinguish the sound of the driver versus many other things that go wrong in a loudspeaker.

The great thing about learning how the measurement system works is you can start to identify the sound of design errors ... followed by the unsettling realization that many highly-reviewed high-end speakers have design defects that made it into production.

 

[sreten]

Hi,

I like the analogy of a good bicycle. Different approaches for
different purposes but for each purpose there are tried and
tested ways of doing things, and adding anything original
to the ways of doing things, not impossible, but difficult,
bicycle design does keep developing, relatively slowly.

Thing is : there is loads of drivel on the internet about how
to build speakers, some of it hopelessly naive, some of it
hopelessly technically inept, but all very plausible sounding.

Building a better bicycle is difficult, but at least the goalposts
are fairly obvious. Building a better loudspeaker equally so, but
moving the goalposts can make any design seem to be good.

Good loudspeaker design moves along at a glacial pace in
principle, and jerks in terms of driver technology available.

rgds, sreten.

"Crossover Calculators" are the spawn of the devil ....

 

[ODougbo]

Some good thoughts there, not like you can go out an build a Trek Madone, btw, the avid cyclist will spend $5,000 - $7,000 on a good road bike, then there's all clothes, tires, etc. I'd say a speaker hobby is a lot cheaper and "speakers" can last for decades.
A question to ponder is road bike improvement about double every 10 years, what about speakers, are the speaker/drivers that much better than 10 years ago?

Nanno, a couple of ideas for your project; don't was a lot of time building a small, one compact xo board, make three smaller ones, one for each driver.
Also, if the mid and tweeter levels are way off, put in some mechanical (twist type) L pads, then you can listen and adjust quickly.

 

[SY]

Beyond tweaking the enclosure volume/port tuning, and all the obvious things to do with trying to get the overall response as flat as possible via crossover filtering, are there any specific steps/advice you would give that may not seem obvious at first?

Three things:

1. Polar pattern.
2. Polar pattern.
3. Polar pattern.

Don't get overly focused on flat on-axis response at the expense of off-axis response- you're going to put those speakers in a room!

You may also end up spending a lot of time playing around with various trap circuit values (yes, that's also part of crossover design!) to get rid of various frequency response anomalies- including the baffle effects, which you must not neglect. You'll be playing around with anti-diffraction materials as well.

 

[nannoo]

Some really interesting advice there, I can't say I completely get the race-car analogy, probably because of my lack of car-skills!

On the other hand, like a lot of us I did go through the cycling hobby, (though mine was downhill, for anyone who cares mine was a trusty GT STS which was sadly stolen) and the analogy has certainly made me re-consider.

I don't know about road bikes, but with MTBs there isn't a great deal of room for improvement, and certainly most 'new' ideas are actually pretty old - basic geometries and suspension systems are well established, and variations come down to personal preference.

So, how much improvement is there left to be made to loudspeaker design, or is it personal preference? Would I be greatly mistaken to try to 'push' reference type (low distortion, flat response) loudspeakers onto the consumer market?

I guess this kind of discussion should move to a more appropriate thread, but I would love to hear peoples opinions of it...

Thank you to all for your advice, particularly sreten for my reading list!
SY - with regards to polar patterns, will i see more on what I should aim to achieve as I read through sretens list? are there well established goals?

My laymans guess would be to go for flat response at about 5 deg...

 

[SY]

There are a lot of well-established goals, none of which people agree on! But the worst case is a design that has "horns" in the frequency response at off axis angles, coming from the different polar patterns of the lower and upper driver. It's a subject that can fill books and it already has! The important thing is to make an informed decision on what your polar pattern should look like, then design and measure toward that goal. A "layman's guess" will not turn out happy. I haven't gone through sreten's list, but I'd certainly start by reading Floyd Toole and Sean Olive's various papers and presentations on the subjective effects of varying polar patterns.