Compact, low cost, active 3-way speaker

Huge peak at 1.5khz then a steep drop off. I can't massage that.
Nuts. I just bought a pair too... guess I should throw 'em on a baffle and see how they measure.

I did notice upon initial inspection that the diaphragms are pretty loose. Had one slide partway out and nearly fall to the floor. There seems to be little or nothing securing them in place, and some slop in the front-rear fit which I was thinking could cause some response issues if they can rattle around front to back.
Oh, that's disappointing.

I have a miniDSP digi-8, and am not super impressed with the programming GUI. It's fine for what it does, but a bit limiting. Sigma Studio looks a lot more flexible to me (albeit a steeper learning curve).

Having said that, I think your hybrid approach with passive mid-tweeter crossover and DSP for the woofer-mid is a good one. You'll have all the flexibility you could want for the critical mid-tweet transition, and the woofer-mid transition should be much more amenable to a textbook crossover function upstream of the amplifier.
Yes. The only quasi 3 channel DSP+amp I’ve seen is the Arylic mentioned in post #16
It goes for around $99 for 2.1channels of amplification and internal DSP on a plate amp.

Thinking that this might be an affordable wireless option for potential builders, I purchased a pair of plate amps, and the software to tune the DSP.

Try as I might, the ACP Workbench program would not load past the splash screen, at least on my main PC. :mad:
Last edited:
I always take the risk of building a pair of cabinets based on simulations with ideal piston drivers. When the cabinets are done, I make measurements of the drivers installed. This is a risk, and there is always a possibility I will find that a different CTC spacing or a different roundover/bevel is needed. In that case, I would have to build a new pair of cabinets. I try to do the baseline in-cabinet measurements before I have applied expensive veneer or spent time in finishing... I have never needed to scrap a pair of cabinets, but that is always a risk for me.
Back in post 226 I wrote that, but now I am planning to build a prototype... what happened? In a completely active speaker, I am able to control and manage the timing delays among all the drivers independently of any filtering or EQ. This flexibility allows me to correct or alter path length differences in ways that can make a positive difference in directivity and power response.

This project will use a passive crossover network from mid to tweeter, and at this point I can't simulate with enough accuracy to be certain of the final directivity response. I will need to do simulations with real driver data to design the baffle properly, and that means making a prototype.

  • Like
Reactions: 1 users
Yes. The only quasi 3 channel DSP+amp I’ve seen is the Arylic mentioned in post #16
Tinysine has a 2.1 (2 X 50 + 100 W) board that performs much better than the Sure product. It also has Bluetooth with apt-X and supports TWS. Here's a LINK. I have this board and the Sure JAB5. The Tinysine boards are my speakers (one per speaker) and the Sure board is in a desk drawer. They also have 2.1 versions that support S/PDIF (coax or TOSLINK) instead of Bluetooth.
  • Like
Reactions: 1 user
ASR Did a review of the 1701 DSP chip that powers this board, and yes, it seems the A/D is a very weak link. Best to use one of the digital options. Unfortunately no comparisons between the SPDIF vs RCA have been made, as far as I could find.

Also, one thing to note, the 2x50W+100W version may have crossover restrictions. A German reseller posted that the 2x50 section crosses at a minimum of 150hz, and the 100W section crosses at a max of 150hz. I couldn't find this in the manual, but it may be something to double check on. A 4x50W and 4x100W are also made by this company.
For any more serious systems, another weak link is that a 1701/1401 just has very limited instructions.

It's fine for IIR filters, but you can't fit any FIR with sufficient amount of taps in there.
At higher a higher samplerate (96kHz max), this is even worse.

ADC and DAC are already so-so just straight from the datasheet, don't need reviews/measurements for that.
They work great with just another ADC/DAC/CODEC or for low-end (woofers and subwoofers) it's also fine.
In all cases, PCB design needs to be adequate to get best performance.
(aka, you can't blame a chip to be bad, if the board design is a mess)
  • Like
Reactions: 2 users
To actually put some numbers to this.
The 1701/1401 can do 1024 instruction @ 48kHz and 512 instructions @ 96kHz.

In practice this is a little less because there is a bit of overhead etc.
So roughly around 1000 and 490 instructions (I forgot the exact amount, has been a while).

A 100 tap FIR filter costs about 103 instruction, each extra tap costs about 1 additional instruction.
The frequency resolution = sample rate / number of taps

So at 96kHz, you can only have about 490 taps.

Which is just not gonna work for any high pass filters below 150Hz or so (again roughly).
Let alone for more channels.

It can be used in a bit of hybrid filter system, when not to much is going on.
I have done that in the past.
But that is squeezing every single bit out of it.

It most certainly is not enough for a 3-way system.

Anyway, assuming that FIR filtering is what you're looking for.

A double precision IIR filter block (biquad) is only 10 instructions.
So even at 96kHz, you can have 35-45 of those (plus some additional things).

Most costly are dynamic filters (like compressors etc)
  • Like
Reactions: 1 user
Just a thought. If your topology requires 4 channels of amplification, why not go fully passive, but embed a better quality 2 channel BT amplifier in one of the speakers and connect them? or do you want a "speaker wire free" setup? The cost may be the same, but simpler and potentially open up better Class D amp options? Not revolutionary I know.
Were any measurements made for these boards?
Here are my Sure JAB5 and TinySine 7800B THD test results. These were done 2 weeks apart. And as always, YMMV.

Test Setup:
  • Power supplies: Meanwell LRS-350-36 for Sure JAB5, LRS-350-24 for 7800B

  • USB Interface: Focusrite Solo, 3rd Gen. An adjustable (6 dB steps) balanced attenuator was used to keep the Solo’s input gain constant.

  • Loads: 4 ohm ARCOL 50W power resistors mounted on heatsink

  • 1701 DSP: The SigmaStudio program connected the ADC inputs directly to DACs. No processing.
  • The 4 ohm loads may be considered low for the JAB5 which is specified by Sure at 6 ohms (uses 7498 chips). However, my speakers are 4 ohms. The 7800B uses TPA3116 chips.

  • These were short (~3 sec) sweeps.

  • The loopback tests (Solo output to Solo input) show the system distortion levels.

7800B THD Measurements.jpg

JAB5 THD Measurements.jpg
  • Like
Reactions: 5 users