Compact, low cost, active 3-way speaker

This is an idea which has been noodling about in my head for some time. A small portable active speaker, of modest cost. Something I can easily build multiple copies of, something I can give as a gift if I choose. I have friends who are decent woodworkers who have inquired about making a copy of one of my existing speakers, but they have balked at the expensive drivers and electronics I use… so this is something I can point to and say “you can copy this…”

I also like the idea of an expandable speaker where a second woofer cabinet can be added to extend the bass response and give higher SPL capability. I got this idea from the Barefoot Sound, where several of their very fine monitor speakers are expandable in this way.

========== =============

Primary Goals

The finished system shall be a compact, active, 3-way with 8” woofer, with a total internal volume of 16-24 liters, with sealed box bass alignment.

Performance Goals

  • Bass response F3 = 40 Hz.
  • On axis within target response +/- 1.5 dB from 100 hz – 16 kHz.
  • Directivity Index within target response +/- 1.5 dB from 500 Hz – 10 kHz
  • No specific max SPL goal or harmonic distortion goal, other than these performance attributes shall be optimized to the extent possible given the drivers selected and the other requirements.
Secondary Goals

  • Low driver cost and low electronics cost. At this point, “low cost” is not really defined, but low driver costs would be in comparison to Scanspeak Revelator, Satori, Bliesma, Purifi, etc. I was thinking on the order of $250 of drivers per side, but this is flexible. In terms of electronics, “low cost” means significantly less than a Hypex FA-123 ($550 per side).
  • Easily replicable. This means easily built with readily available parts. I may want to make multiple copies to give as gifts.
  • Simple cabinet finishing that does not require weeks of shaping, sanding, veneering, sanding, varnishing, polishing, etc.
  • Easily portable.
  • A modular design which would allow me to design/fabricate a second bass driver/cabinet later. This would extend the bass response and expand the SPL capability.
  • An opportunity to try new techniques and learn new skills which can be applied to my future high end projects.
  • An opportunity to discover new information which would be of use to other DIY speaker builders and developers.
What are the electronics options? One option I have been curious about is the line of equipment from MiniDSP. The Flex and Flex-8 are both serious high performance gear. I might use the Flex-8 on a future project. The cost of the Flex-8 (or a pair of Flex’s) pushes the cost into the same range as Hypex Fusion, so this is not appropriate for this project. However, a good stepping stone into the world of MiniDSP software/hardware is the 2x4HD. It has lower performance than the Flex, which is also a 2-in/4-out device at double the cost. But the 2x4HD is good enough for this project. Another compelling reason for me to work with the MiniDSP 2x4HD is that it would be a good development tool for any future project, allowing me to have a DSP in the loop when I develop prototypes.

A single 2x4HD shared between two speakers will not provide the necessary 6 channels of signal required by a pair of 3-way speakers. But a single 2x4HD could be used in an active/passive speaker. This is something that has interested me for some time. A passive crossover from tweeter to mid, then active crossover from mid to woofer. There would be 2 amplifier channels required. I find the challenge of this architecture intriguing.

The interesting thing about this approach is that the passive crossover does not need to perform any EQ of the drivers within their individual pass band, nor does it need to manage the overall EQ or voicing of the speaker. This can all be done by DSP. The DSP can also provide BSC for the woofer channel and the mid/tweeter channel. The DSP can tailor the bass response of the woofer. The passive network will need to provide a high-pass and low-pass function for each driver, and relative level matching between tweeter and mid. The passive network will manage the relative magnitude and phase of the drivers through the crossover, and as a result of this, it will manage the directivity index.

The passive network will also manage any out-of-passband EQ. The most likely need for this would be a notch filter on the woofer or midrange at a frequency above the low-pass filter. For example, if the midrange passively crosses to the tweeter at 3k, but the mid has a + 9 dB breakup peak at 8k, the DSP will not be useful in managing that peak. Any EQ at 8k would affect the tweeter response, which would be quite unfortunate. Thus the passive network will need to handle that.

So at this point, I believe this is how I will proceed. There are a lot of amplifier options which offer low cost and acceptable performance, so this remains an open decision.

  • Like
Reactions: 1 users
Have you seen the "Class D" topic about TWS or True Wireless Stereo? Some discussion in there about boards with EQ - NVARCHER on AliExpress.
Cheapish woofer choice might be MCM 55-2421. Has excursion for sealed and/or EQ'd bass response.
Could go active 2-way but then passively manage a midbass and tweeter combo.
  • Like
Reactions: 1 user
I like your approach. But why not sure Sure/Wondom DSP/amplifier?

The hardest part of his exercise is finding a configurable 2 channel DSP with onboard amplification that provides about ~80W (4 ohm) and 40W (4 ohm for LF+MF and HF respectively) that costs about $100 a side.

I've deconstructed the Kali Audio IN-8 V2 and IMHO this is one to beat at $349 per speaker.
  • Like
Reactions: 3 users
I like your approach. But why not sure Sure/Wondom DSP/amplifier?
Yesterday I reviewed my Wondom JAB5 equipment. I have two Wondom JAB5 4-ch amp boards with on-board DSP. I have two 36V power supplies. I have two wondom programming boards. I have a variety of connectors and wiring harnesses. All said I have about $210 worth of stuff. Originally I was strongly leaning towards using this as the electronics package for this project. Several factors give me pause.

First is the relatively low performance of the JAB5. Although the distortion is quoted as 0.07% at 1k, it rises rapidly above 1k, and is above 1% at 5k. The noise plot that Sure publishes shows a noise floor of ~ -80 dB at 1k, but rising to -60 dB at 20k. The frequency response is not flat, showing a rise of several dB from 5k to 20k. It is hard to estimate the frequency response because the scale is so compressed.

This low performance means that whatever I learn from working with this Wondom equipment will have little or no value to my future projects. It will have little value to other DIY’ers, who I suspect will prefer electronics with higher performance.

The second concern I have with the Wondom is that I will have to learn a variety of electronics technician skills to construct a box to safely house this stuff. I would need to select and modify a “Bud box”, fit it with the necessary mounting hardware, ensure it meets UL safety requirements. I have never built my own PC or other electronics device, never repaired or modified a printed circuit board, I own virtually no electronics test equipment. Constructing electronics is not really my interest. I feel I might be over my head in correctly assembling these parts into a functional, safe box. I also realize that constructing this box is very far from “easily replicable”.

So using the Wondom JAB5 seems like a big investment in time, new tools, new skills, for something which I may never use again.

I took the liberty of tracing the FR graph and converting to FRD to analyse it:

Instead of that misleading 190dB vertical scale, I converted it to a 30dB scale, which, as you know, is the difference between 0.1W and 100W:


There's is a 3dB rise at 14KHz. Furthermore, there a 2dB drop at 50Hz.

The 1% THD @5Khz is far higher than what is achievable by current generation commodity Class D amplifier chipsets like the TI TPA3255, which are better by orders of magnitude (<0.03% THD for 5W @5Khz)

Hmph. Yes somewhat underwhelming. It must be using an older generation or lower spec Class D amp. In retrospect the min load impedance of 6 ohm was probably the first clue...
  • Like
Reactions: 1 user
it really sucks that our options for dsp plate amps is either so cheap it sucks or so expensive it's out of a lot folks reach. shame to see the wondom stuff not performing very well, I was really interested in them.

having done a recent Kali lp6 amp swap I was soooo jealous of the lovely clean pcbs and great layout, smal factor and low heat. makes me look at my complicated dsp 3 way and go "damn...."
  • Like
Reactions: 3 users
If separate DSP and amps is OK, the ADAU1401 "DSPMini" boards are a bargain. You could get one for each speaker and perhaps add a better ADC through I2S as the integrated ADC seems to drag the specs down. The savings could be spent on separate amps for LF and MF+HF.
The MiniDSP/passive hybrid does seem more elegant and "robust" though 🙂
I think there was FreeDSP project, or someone else, who put the inputs of ADAU1701 in parallel to reduce noise, perhaps it was this one:

There is also a recent thread for new 8channel FreeDSP board. These are great I think, but not sure about availability and price, performance ought to be great and there is nice selection of boards. But, it is PCB boards and need psu and tinkering and all that.
This looks very interesting. I had a broadly similar idea but I had some specific constraints around “compact” where I had a width limit of around 15 cm (6”). I did not define the technical constraints as above. I just wanted to get as good a bass response as I could in the space available - hopefully as close to my old speakers that had 10” bass units.

Having read up on Rod Elliot’s ESP audio site, I rather liked the approach of 3 way active as matching driver sensitivities becomes a non issue.

My first effort used Visaton W100S drivers, 4 bass and 2 mid and a Monacor CAT298. I used Rod’s LR 4th order 3 way and Linkwitz Tranform with 4 x TDA7293 amps. (2 bass, 1 mid, 1 treble).

I crossed over at 300Hz and 3.5KHz since the bass and mids use the same drive units there is great scope on where to cross over.

Rod’s articles flag how much power you may need when you use a Linkwitz Transform (LT) hence my decision to use 4 bass units and 2 amps since they only have 30W RMS power handling. I also have baffle step compensation.

I had some help with the cabinet as I didn’t have any woodworking equipment at the time. After building all the electronics and wiring it up on the bench it sounded very promising so I proceeded to install everything in the base of the speakers.

What I hadn’t appreciated at that time is just how important proximity effects are. Once completely built they sounded absolutely dreadful. There was inductive coupling into the small signal areas from the power supply and speaker outputs. Distortion was chronically bad as a result.

I rebuilt with the small signal stuff in small Hammond aluminum cases attached to the rear of the speaker. This did the trick and the distortion went away.

I experimented with different LT eq. With eq to 30Hz the excursion was huge on bass heavy material and ultimately overall volume was limited by bass excursion. 30Hz was just too ambitious! 50Hz was tight and punchy and 40Hz was mostly ok, but some tracks just had too much bass.

That was my first prototype that I did as a proof of concept as I didn’t have any idea of the concept was any good.

It was too tall for WAF, so I needed to rethink it a bit.

Prototype 2 uses only 4 drive units but uses exactly the same electronics. I used 2 Monacor SPH135C 5 1/4” Bass units , 1 MSH115HQ midrange and DT254 tweeter. The bass units have 80W power handling so can easily handle the power requirements from using the LT eq.

This is my final solution and is elegantly compact at 15 cm wide and 1m tall with tweeters at ear height when we’re sitting on the sofa. I am finishing them in real oak veneer and with light grey grilles to match the living room decor.

True bass fans wouldn’t like them I think as you don’t get the visceral chest thump that you can get from 10” or larger, However the quality of the bass is a delight, it is very even and articulate and it goes plenty loud enough and produces enough low bass to shake the sofa. The quality of the presentation at all frequencies is vastly superior to my old 2 way passives.

There are many things I didn’t know I needed to know when I started this!

I will recycle my old drive units into a large active 3 way tower.

I wish you every success with this project and will follow with interest and see what I can learn for my next build.
  • Like
Reactions: 1 user
A couple of weeks ago I bought a new amp for testing drivers and speaker prototypes. My old one was rated at 10 W/ch, and when running a large woofer to Xmax it would get hot and sometimes clip.

I bought this new one from Parts Express
It uses the TDA7498E chipset, and is advertised as 160 W/ch. This chipset has good performance:



A more realistic rating would be 120 W/ch into 4 Ohm at 1% THD. Still, $83 for a 120 W/ch amp is very reasonable. I was using it to listen to pair of passive speaker I recently completed, and I thought it sounded really good... by which I mean I could not perceive any difference between it and my reference amp.

So this amp, or one similar, could provide the 2 channels of power I need for each speaker. The amp and its power supply could be mounted on the back of the speaker or located in the speaker stand.

That would put the electronics cost of this project at ~ $520 for the pair.

QtyCost $Net $
MiniDSP 2x4HD1230230
2 ch TDA7498E amp283166
Passive Xover estimate260120
Total (2 speakers)516

Which is half the cost of the high end option of a pair of Hypex FA123 amps, or a MiniDSP FLEX-8 + 6 channels of IcePower amp.

So this is definitely an option.
The MiniDSP/passive hybrid does seem more elegant and "robust" though
As I design this, I try to keep in mind a vision of the end user... I imagine my Dad using this speaker. I just don't see him, at 78 years old, needing to turn on or manage multiple layers of equipment just so he can watch a blue ray disc... or God help us, boot up a PC :( he would just give up in frustration.
  • Like
Reactions: 1 user
I agree. My 12 year kid wants to talk to his speaker and my partner doesn’t even want to do source selection, let alone switch on 3 amps. (“Why does your speaker need 3 amplifiers?!”)

What’s ideal would be something like the Arylic Plate amp with 2 in 4 out, instead of , the 2.1 channel model.

It’s a plate amplifier with built in DSP, but also wireless receivable, Bluetooth , streaming services capable, and app driven. The DSP is configurable for only an extra $20.

Plain board version measured here:

Perhaps unbridge the TPA3116 for a real 4 channel output and you’re done.

Perhaps they are interested in upselling a model with 3dB more power. I shall enquire with them whether this is in the pipeline and report back.
Last edited:
  • Like
Reactions: 1 user
Joined 2009
Paid Member
This seems one of the big frustrations for a lot of people?

It's like all or nothing.
Also the implementation is very often not great.

Nobody got a Raspberry based solution developped for this purpose: a couple of dac's shield easily coupled to some amp boards in small format factor?
It would allow for evolution both hardware and software. Idk if it would be easy to implement?
  • Like
Reactions: 1 users