An idiot's blog to design and build speakers

This is the start of my blog documenting my (lack of) progress building a set of speakers from the ground up.

I am writing this here mostly for myself, and maybe other starters can learn from my (probably numerous) mistakes.

First: my background.
I am not completely new to speaker building.
I started a few years back, and I listened to the advice of the smart people in this game :
don't start off wanting to design your own 3-way speakers, start by building existing kits designed by people who *do* know what they're doing.

So for my first project, I built a pair of Needles:
https://www.diyaudio.com/forums/full-range/65061-range-speaker-photo-gallery-369.html#post4933327

Then I got the taste of it, and wanted to go bigger.
So I built a pair of Heissmann Acoustics "Samuel HQ"'s.

And while I was at it, I built my own class D amplifier based on the TPA3116D2, applying all mods mentioned in the diyaudio.com TPA3116D2 topic.

My audio family in one photo:
https://i.imgur.com/nglynTp.jpg

The Samuels were placed in my kitchen, which is 5x6m, and were placed in the corners because well, we have to be able to walk around in the kitchen, right?

This made them sound less than awesome.
Then by pure chance I was gifted a Umik1 by a friend.
I started doing measurements using REW, and found out that using PEQ I could get my Samuel HQ's to sound much better in my less than optimal listening space / kitchen.

(I use a Raspberry Pi with Hifiberry Dac DSP+ to perform the final PEQ)

So then - everything is fine and good...

But now I've been infected... and I want to *design* my own speakers!
So I've been reading on this forum and others, and the more I read, the more I realize I don't know about... well, *stuff*.

At every single point in time where I think "all right, now I know what's going on - I'm ready to start designing my own speaker", I read one more post just to realize "****, back to the drawing board".

This has been going on for a year or so, and I have now reached the point where I say : **** IT.

I'm not going to spend more years reading up on things while always being behind the times.,

I am going to put something together.
I am going to buy parts.
I will be told I've bought the wrong parts, I will be told I'm taking the wrong approach, I will be told it's not going to work..

..and that's fine with me :)

I will learn as I go along, but at least I have a starting point now :)

SO - my starting points are:
- I have decided I like the idea of full range drivers - no crossovers in the vocal range.
- I do like a little bit of bass
- DSP is awesome - I don't know **** about analog crossovers, DSP is my magic bullet for crossovers, room-correction and everything else


Thus I have now decided to just buy some drivers and make **** work. And if it doesn't work out, it's a learning experience.

The drivers I am going to work with are:

- HiVi RT1C-A tweeters
- Tectonic TEBM46C20N-4B (BMR) for full range
- Tang Band W6-1139SIF for (Sub) bass

My current idea is to use the Tectonics for the full vocal range - crossover for (sub)bass between 200 and 300Hz, crossover for the Tweeter between 4Khz and 8Khz, testing as I go along.
I will use class D amplifiers to actively power them (TPA3116 or better).
I will use a Sure / Wondom DSP board (ADAU1701) and SigmaStudio for crossovers and PEQ to match room measurements.

Once I get my first order from Parts Express this blog will continue.
 
Last edited:
Not sure why you’re writing all of this. Looks like you have everything figured out already and are not really interested in comments from anyone else.

But if you are trying to impress us with this effort I’m afraid it may fall short. There are lots of people who have done what you are trying to do. Some have had good results, some have not.

Anyhow, good luck. Hope you achieve something that pleases you.
 
Not sure why you’re writing all of this. Looks like you have everything figured out already and are not really interested in comments from anyone else.

But if you are trying to impress us with this effort I’m afraid it may fall short. There are lots of people who have done what you are trying to do. Some have had good results, some have not.

Anyhow, good luck. Hope you achieve something that pleases you.

I'm sorry if I gave the impression that I wanted to impress anyone.
I started this post to document my travels into speaker design, thinking it might be of use for anyone else going down this road.

I have no idea of what I'm doing or where I'm going with this.
I just decided to go "**** it, I can keep reading forever, let's just start and see where I end up".

I've invested by buying some drivers now - next up buying the dsp backend, and then hobbying around trying to figure out the cabinets to put them in.
Let's see where it leads.
 
Last edited:
Some notes-

The 1139 rolls off just shy of 600Hz on its own. The TE midbass (if used in a 3-way with said 1139) should be happy to take advantage of this rolloff and take some stress away from it in terms of Xmax. I would wager 400-600 is a better transitional point for a 3" to 6" for tonal balance. If you were to use the TE lower, the midrange would not have the weight required to flesh out the range and sound a bit thin. The planar you are going to use is a higher sensitivity tweeter, but takes padding well with a CR filter (passive) required to 'raise' the top-end. It also drops like a rock below about 4.5k.

How do I know? I actually did a 3-way with a 3" mid and the 1139. I also did a different 3-way with the planar in question.

How big do you want to go? My suggestion is about 1.25ft^3, tuned to 25Hz with a pair of 2" ports. Use a 12dB rolloff at ~600Hz. You do not need steep slopes to make a good sounding speaker.

You will likely have to go 2nd order filter on the HP on the mid, and 3rd order on both the LP of the mid and HP of the tweeter for best alignment.

Later,
Wolf
 
It's always so nice to see people step in with constructive comments and encouragement as we are not all pro's and perfect but are trying to learn.
You never know when some one will come up with comment that will end up and make it all come together for and individual...:checked:
 
Pygmy, let's crack on and look at your proposition. TBH, the 6" woofer looks troublesome. And little 2" fullrangers won't dig very deep, certainly not down to 200Hz.

The tweeter is some sort of vertical ribbon or AMT. Sort of thing with little vertical dispersion.

What I am seeing is a speaker not a million kilometers from Troels' 3 way:

SP38/13

If I was using a ribbon type, I'd put it in an MTM:

733281d1549059344-speakers-parallel-serial-wagner-net-au-kit-260-5-inch-mtm-jpg


Keeps the dispersion cylindrical. Which adds a horizontal PA-like projection to the sound. More of the speaker, less of the room.

Short of a WMTMW design, you can get away with a single woofer,

Overall, I think you should double up the little mids. Then it should be doable. Because of the 6dB SPL gain from doubled drivers, distortion is less.

Wharfedale E70:

605562d1489602538-visaton-2-tower-monitor-audio-bronze-5-a-wharfedale-70-jpg


Celestion Ditton 25:

605544d1489600197-visaton-2-tower-monitor-audio-bronze-5-a-celestion_ditton_25-jpg
 
A tip: Try to sim the speakers cabinets individually with winISD. Most use hornrep here, but that is a very complex and difficult to learn package, that only a few really fully master. WinISD works very well, is simple to learn and can tell you a lot about your drivers and how to use them. It does not offfer all the possibilities that hornresp has, but for a starter that is not an disadvantage.

WinISD - Linearteam

I did sim hundreds of speaker enclosures that i didn't build in winISD and i learned that way how a speaker react, wich T/S parameters show what you can and can't do with it, and so on...
 
Great project. I made something similar with 2x and 4x 3.3" Visaton frs 8m (€10@Thomann) and Dayton mini planar's (passively crossed at 5kHz) in pieces pvc pipe. They have a super wide horizontal dispersion and awesome detailed sound; love them. I have something similar in a bigger size in the making; dual 6.5's (alphalite 6) + planar that will have to cross at max 3khz.
Both are sealed, cross around 150Hz and have separate sealed bass/subwoofers.


If you use more then 1 mid, the vertical off axis response will narrow and better match the planar. You also get effortless mids at high levels as they share the load. IMO you get best of big and small drivers with many small ones in a vertical line.
 
All right, so I'm getting to the point where I'm going to use WinISD to calculate an appropriate cabinet for the 1339SIF subwoofer driver.

Since the Tang Band W6 1139's parameters as specified in the TB documentation give the famous "driver data fails integrity check" error, when entered into WinISD I'll have to enter some data myself and let WinISD calculate the rest.

Reading the WinISD manual tells me to start by entering Mms and Cms, by which it can calculate Fs. If I use Tang Bands data (Mms = 39.9g, Cms = 0.599mm/N) the calculated FS is 32.55Hz, 7.2% lower than the Fs specified in the TB documentation.

On the other hand, if I fill in the Fs using TB's data, the Mms WinISD calculates is 44.8g, which is a whopping 36% higher than the value specified (39.9g) in the TB documentation.


Reading WinISD tutorials most tell me to start by filling in Fs, WinISD's help itself tells me to start by filling in Mms and Cms...

Who's right?
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
Thus I have now decided to just buy some drivers and make **** work. And if it doesn't work out, it's a learning experience.

Great attitude!! :) Experimentation in conjunction with your readings is the way to really start to understand what is going on, and yes when things don't work, working out why, will accelerate your understanding.

Tony.
 

Nice!

I've taken his information to try and learn a bit more about WinISD and ports.

I see he's using a 16 liter cabinet, using a 1.5" x 8" port.

Using a 16l box I enter some guesses for the box tuning frequency and find out that with a port radius of 1.5", tuning the box to 30hz will have WinISD suggest a port length of 8.25".
So If I'm correct, he's tuned his box to around 30hz.


When I input that into WinISD (16l, 30hz) I see the often recommended max port velocity of 17m/sec being reached at around 6.2W system input power, so I thought it would suffer from port chuffing.

But then I remembered to check the output level - at 6.2W the SPL is around 91.2dB.

Reading about volume levels at WHAT ARE REALISTIC VOLUME LEVELS? it seems to me 91dB is pretty loud for a living room (they specify 80dB as "Very loud music playback at home").

So if my assumptions are correct his setup would work well for normal usage without port chuffing.
 
Anyway, as I was saying, I first want to try simulating a sealed cabinet.

So this is my current go, can anybody tells me if what I'm doing makes sense:

- because the 139sif isn't in WinISD, and the T/S parameters from the TB specs give data integrity errors, I've created a new 1139SIF driver in WinISD using the measured data posted here: Measured T/S Parameter Thread Sticky, because I figure that even if my unit differs from the measured one, it's probably still closer than the TB "polished" specs..?

- I created a new project with this driver for a sealed cabinet. I've set the cabinet size to a very modest 12 liters, just to see what happens, which results in this graph:
xTVGMiG.png

So to start out, an F3 of around 50hz.

- I slap on a 4th order Butterworth high pass filter at 20hz to remove unwanted lowend and prevent unnecessary cone excursion.

- Since I'll be using DSP, I add a Linkwitz Riley transform to boost low-end. I have completely pulled the figures for this one out of thin air, and end up with this graph:
SsixijY.png

This results in an F3 of around 30hz and a low-end boost of 1.5dB around 75hz,
cone excursion stays just witin limits with an input power of 18W / output SPL of 93dB which is much louder than I'll ever use these speakers for.


Does this make sense or am I saying / doing really stupid stuff here?
 
Last edited:
Us regulars tend to get to know each other... :D

Wolfie is a very safe and experienced pair of hands:

Some notes-

The 1139 rolls off just shy of 600Hz on its own. The TE midbass (if used in a 3-way with said 1139) should be happy to take advantage of this rolloff and take some stress away from it in terms of Xmax. I would wager 400-600 is a better transitional point for a 3" to 6" for tonal balance. If you were to use the TE lower, the midrange would not have the weight required to flesh out the range and sound a bit thin. The planar you are going to use is a higher sensitivity tweeter, but takes padding well with a CR filter (passive) required to 'raise' the top-end. It also drops like a rock below about 4.5k.

How do I know? I actually did a 3-way with a 3" mid and the 1139. I also did a different 3-way with the planar in question.

How big do you want to go? My suggestion is about 1.25ft^3, tuned to 25Hz with a pair of 2" ports. Use a 12dB rolloff at ~600Hz. You do not need steep slopes to make a good sounding speaker.

You will likely have to go 2nd order filter on the HP on the mid, and 3rd order on both the LP of the mid and HP of the tweeter for best alignment.

Later,
Wolf

My intuition of 6" woofers, is they like about 20 Litres, here due to a Qts of 0.4 in reflex.

But that woofer is definitely in the sub-woofer category. Hence Wolfie says cross it low. I think it's the tiny mid that will struggle with this. Why I suggested doubling up the mids.