• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Wondom Jab5 4 Channel Double Tda7498E DSP Integrated Amplifier Bi-amp Aptx-HD Fir Crossover

Audiomaniac3

Member
Joined
Nov 8, 2024
Messages
54
Likes
26
Hello everyone,in this video:

I got a four-channel amplifier from Wondom for review on two TDA7498E microcircuits and with a built-in adau1701 dsp module with a maximum sampling frequency of 48000khz 24bit.

The amplifier has many options for connecting and transmitting a signal to it.The board has a line input, Bluetooth 5.0 Aptx-HD, aswell as an i2s input that allows you to transmit a digital signal directly to it.Esp32 wrover module with squeezelite and lms server,can be used for transfering audio over Wi-Fi with i2s protocol.

Thanks to a wide range of capabilities, the board can be used for 4 channels, for 2 channels each amplifier in mono mode, or in 2.1 mode.

There is also an i2s output that allows you to connect another Jab5 amplifier and have an output for 8 channels.
The Sigma studio software allows you to write a project for your needs, specifically in this video, I am writing project for bi-amping with a phase-linear crossover, using room eq wizard and rephase.
The programmer is sold separately, icp(1)(3)(5) will do.

About output power and hissing,according to the datasheet of the TDA7498e microcircuit, the lowest signal distortion is up to 50 watts per channel.

Also, the cooler on the radiator does not work constantly, but turns on periodically as needed for cooling.In Master mode with Bluetooth connected, the hissing is barely audible if you get close to the tweeter, in Slave mode the hissing is a little stronger if you listen closely.

In general, an excellent thing for its money. Link to Tda7498E datasheet: https://www.alldatasheet.com/datasheet-pdf/view/470927/STMICROELECTRONICS/TDA7498E.html

JAB5 amplifier:https://store.sure-electronics.com/product/756
1000021135.jpg
1000021122.jpg
1000020732.jpg
 

Attachments

  • My Project_1.gif
    My Project_1.gif
    3.8 MB · Views: 60
  • Screenshot_20241224_083838_Gallery.jpg
    Screenshot_20241224_083838_Gallery.jpg
    2 MB · Views: 68
So basically I've managed to get most out of adau1701 and adding linear phase crossover for tweeter and bass speaker,thats how step response and frequency response look like.

Also small video of me showing esp32 wrover with squuzelite hooked up to jab5 with i2s audio transfer :
1000021454.jpg
1000021453.jpg
 

Attachments

  • Screenshot_20241225_104216_YouTube.jpg
    Screenshot_20241225_104216_YouTube.jpg
    813.8 KB · Views: 85
  • Screenshot_20241225_104152_YouTube.jpg
    Screenshot_20241225_104152_YouTube.jpg
    1.2 MB · Views: 85
Also little trick how to fix ported speaker phase shift,all you need to know is which frequency your port is tuned to,in my case edifier r1280t port is tuned to 75hertz,now in rephrase all I had to do is add reversed 2nd order allpass filter in rephrase and that's it.
1000021464.png
 
Lifehack with speaker drivers time matching up to 0.01 ms,run 5-10 repeated sweeps 20-20000 in rew for bass speaker only,and then for tweeter only same way,get rid of faulty measurements,now look at step response,now set desired time delay,or in rephase add sufficient amount of taps to achieve right time delay.As you can see my tweeter was delayed from my bass speaker about 0.12ms,now they are perfectly time alligned.

Speaker time alligment with step response:
 

Attachments

  • step1.jpg
    step1.jpg
    521.1 KB · Views: 62
  • step2.jpg
    step2.jpg
    518.4 KB · Views: 62
  • step3.jpg
    step3.jpg
    517 KB · Views: 60
Last edited:
Ofcourse having linearphase crossover with almost perfect time matched speakers is not enough for enjoyable audio experience,final step is to measure speaker frequency response and add linear phase crossover,big advantage of linear phase eq is that it doesn't add ringing and distortion to the signal,only adds small audio delay,in my case it's 125ms,and having option to create as many impulse files for any kind of target curves,without messing speaker phase response is really nice.Even compressed audio over Bluetooth protocol sounds great with adjusted frequency response,so adjusting frequency response is mandatory thing to do, only requirement is equalizer apo with convultion plugin or some media player like Jriver.
 

Attachments

  • z2.jpg
    z2.jpg
    324 KB · Views: 74
  • z4.jpg
    z4.jpg
    521.7 KB · Views: 74
  • z3.png
    z3.png
    102.5 KB · Views: 77
This one is for people trying to hook up ESP32 wrover module to Jab5,basically if you're listening to music from YouTube with 192kbps bitrate audio or Spotify,you don't need this, just stick with Jab5 integrated Bluetooth,aptx HD has 576kbps bandwidth,which is more than enough,but if you're like me with lossless music collection,this guide is for you.

Even latest (Aptx lossless)
codec has max of 1mbps bandwidth and some of my flac files have 2mbps,so Bluetooth is not an option,only Wi-Fi will work for transferring lossless audio.

Basically esp32 allows to send digital audio over i2s protocol straight to DSP,which is good,we are sending digital signal to our unit,in order for it to work over Wi-Fi,we need a wrover esp32 unit and a machine with LMS server,pc or phone with android will work fine.

I am using WROVER esp32 unit,not !!!WROOM!!! And I have no clue how and will other units work.

I made I2S cable from cable that came in cable kit,it works fine,easy to connect and disconnect it, probably I will isolate it later with tape.

Here is also wiring diagram:

JAB5________ESP32
1-LRCLK____25
2-BCLK_____33
3-DATA1___32
4-GND_____GND
5-+5V______5V
6-MCLK____0

So In order to program esp32 with squeezelite firmware just use this URL: https://sle118.github.io/squeezelite-esp32-installer/
You will need a PC and USB cable.Chrome,Opera browser or Microsoft Edge,one of them should work.

Later use your phone or laptop and connect to squeezelite Wi-Fi,password should be squeezelite,while you're connected to esp32 in web browser or through Wi-Fi settings go to default gateway IP address and connect your esp32 to your preferred Wi-Fi network.

Now all you have to do in NVS editor is to copy paste this to dac config.

dac_config: model=I2S,bck=33,ws=25,do=32,mck=0

Choosing between 16 bit and 32 bit versions,to my ear 32 bit version sounds more pleasing.

Now enter max sampling rate: 48000,44100

So last thing is to install LMS server,for PC users it's easy,install LMS,wait few seconds and esp32 will connect automatically to your server and you can add your music library and play music from LMS.

For installing LMS to your android device use this guide:

I like using Neutron player for music playback,since I have all my playlists there,in order to cast music to LMS server all I had to do is install UPNP plugin from lms plugins,and basically I am casting music from Neutron to LMS server running on my Phone.

I like Neutron because,I could set sampling frequency for stream and format,so I am using 44100 flac stream,since my project in Jab5 is made for 44100.

I had to Google this issue,because esp32 didn't boot properly from Jab5.

If you want esp32 to work from Jab5 internal voltage,you must press EN button on esp32 each time after you power on Jab5,or solder 10uf capacitor instead of this button and it should boot properly.

Theoritacally any i2s master device should work with Jab5 if it could provide MCLK clock 12.288mhz.

If you'll try to cast anything more than 48000khz without setting max sample rate in esp32 web settings,you will hear terrible noise,so just remember 48000khz is max sampling frequency.

Here is a video tutorial how to programm esp32 with squeezelite and install LMS server on android device:
 
So one more thing about esp32 squeezelite 16 bit firmware and 32bit firmware difference.
I didn't understand,why I get different frequency response on 32 bit version,but it turned out that FIR crossover I've set in my jab5 project(44100khz),acts as intended on 16 bit version,and acts totally different on 32 bit version,blue frequency response is 32 bit firmware and to my ears 32 bit version sounds better,maybe later I will make project for 48000khz and see how FIR crossover behaves.

For now only few things left to do,I am planning to power up jab5 from Li-Po battery and also I am waiting for FIR equalizer implementation,either squeezedsp developer will implement it properly,either the Neutron media player developer.
1000030451.jpg
 
Stock I2S cable from cable kit suits perfectly,no need to solder anything,I just had to remoove plastic connection on one side and attach cables to esp32.
1000030242.jpg
 
Thank you for this Brosef:cool:

I too own an assortment of these Sure/Wondom boards. Well built, work as advertised, feature rich and bang for the buck are all adjectives that rise to the fore. Good stuff.

Thank you OP. Thanks for the measurements/data points, I've always wondered what they'd look like. Not too shabby at all :)
 
Try to find QC QCC BT SoC/SoM (with or without additional DAC that will in USB debugging bridge (driver) mode work with QACT and you get much more than what Topping exposed so far (speaker EQ 10 PEQ's) including IIR, crossover and so on. The old and only one not discontinued yet from it's generation QCC5125 is amongst more popular one's you will find in such (as it's dirty cheap now).
 
Try to find QC QCC BT SoC/SoM
I guess I already have everything I need,esp32 Wi-Fi and i2s do not compress flac files,and Jab5/ESP32 have Bluetooth for lossy music.Neutron player has like 60 band peq available,so it's more than enough for my needs.
1000030712.jpg
 
I guess I already have everything I need,esp32 Wi-Fi and i2s do not compress flac files,and Jab5/ESP32 have Bluetooth for lossy music.Neutron player has like 60 band peq available,so it's more than enough for my needs.View attachment 429072
As you wish I told you how to improve it on many aspects including integration and DSP capabilities on device and cost. To program and leave as it is. Neutron ain't either free or very stable and can't mitigate standard Android sound output path which brings in a lot of problems and limitations. Hopefully James DSP for Android will become more mature including no root one in time if that's OS platform you wish to persue (don't go with ALSA Linux) for higher level CPU FPU based approach.
 
Neutron ain't either free or very stable
1000030714.jpg

Actually I've tested how neutron player equalizer performs,all I had to do is generate 20hz-20000hz 0dbfs (44100)wav sweep tone and put it on my phone,just run rew measurement on muted output and play this file from neutron at same time.

The Neutron equalizer performs flawlessly,same as equalizer apo for PC,I got same results from both,so I can submit that Neutron player equalizer works very well.

Neutron doesn't cost that much and I think it is best android music player available.
 
View attachment 429085
Actually I've tested how neutron player equalizer performs,all I had to do is generate 20hz-20000hz 0dbfs (44100)wav sweep tone and put it on my phone,just run rew measurement on muted output and play this file from neutron at same time.

The Neutron equalizer performs flawlessly,same as equalizer apo for PC,I got same results from both,so I can submit that Neutron player equalizer works very well.

Neutron doesn't cost that much and I think it is best android music player available.
I am a little bit more into it than you are. 64 bit FP engine is transparent to pretty much anything you do in it however falling back to OS pipeline and it doing the conversion by far isn't. On embedded you will be limit to 32 bit both integer and floating point (as for application no one sees the need for 64 bit).
It's getting interesting how to bypass the limitations and this is great example thinking out of the box to do it.
With JiT enabled in Chromium or Chrome OS build (you compile) you get execution speeds as on Android and full Linux capabilities of course.
 
Last edited:
however falling back to OS pipeline
I will argue here,neutron has direct USB dac access,so everytime you plug any USB dac it will ask permission to mount it,so audio pass-through is not limited to Android capabilities,it has direct audio pass-through with 64 bit processing,with DSD up to 1024 upsample and pass-through,you can see settings screenshots below.
 

Attachments

  • Screenshot_20250216_130504_Neutron Player.jpg
    Screenshot_20250216_130504_Neutron Player.jpg
    152.8 KB · Views: 23
  • Screenshot_20250216_130449_Neutron Player.jpg
    Screenshot_20250216_130449_Neutron Player.jpg
    62.2 KB · Views: 24
  • Screenshot_20250216_130433_Neutron Player.jpg
    Screenshot_20250216_130433_Neutron Player.jpg
    121.4 KB · Views: 21
It's a player, not system wide. Sorry for bothering you, one day you will understand what I am talking about.
Disclaimer: I don't encourage use of any property software solution or DSP architecture (which by all means must be on such to program it). Don't like or favor QC solutions just pointing out you can get them on the QCC SoC and fairly more capable (including one's with 2x 32 bit DSP if that's what you need) on more efficient package. It doesn't have sense to go with old separate SPARC DSP's and equally bad or worse property software. Irony is amplifier board has older QCC BT SoC already and with DSP (which it doesn't use as they didn't put effort in so you could).
 
Subsonic filter for speakers,basically Sigma studio allows you to implement Subsonic filter(highpass filter) into your project,since most of the cheap 2 way speakers have high resonance on low frequencies,playing music on louder than average volume,will add sufficient distortion,for example my Edifier r1280t speakers keep moving back and forward like crazy on 40-50hz.

Simple highpass filter will solve this problem,if implemented correctly, with properly adjusted frequency and q value it won't completely cut off low frequencies,but lower it by few decibel and ofcourse will cutoff infrasound,making speakers sound better with less distortion on high volume.

As of positive gain you can see in my project,adau1701 has +24db headroom and very handy tool called level detector,adding level detector in your project will show whenever signal is clipping,everything under -3db on level detector scale is fine,if it is more than that,just decrease volume in your project and you're good to go.
1000026219.jpg
1000030882.jpg
1000030884.jpg
 
You do it cuple Hz above woofer Fs and not with high pass but self filter and of course when setting sub crossovers. Future more if you can't do it as it would cut too much port output you do it just under the port tuning going with speakers only to cut highest distortion caused by ported design. Soon enough you will realise how bad design choice is to try to extend them too much under the woofer response and bring it in front as it losses the function at certain SPL range and simply isn't what it tries to mimic. Keeping it a little bit under on both axes to extend response and under it to boost the actual one staying in front. None of home speakers is done properly in that regard. You can use Linkwitz transform in case of closed enclosure and carefully to do a same thing not altering the decay order much. In car you can use open bufle for the sub in the trunk and it as the box lowering Fbox and it's not a problem as it's small volume all in all.
 
Good news everyone! I've managed to get FIR squeezeDSP plugin,with LMS server installed on my windows laptop work,I've tested different FIR files and they worked properly,so for now windows users could enjoy this quite useful feature.

The folder for FIR files location:
C:\ProgramData \Squeezebox\prefs\SqueezeDSP\Impulses

 
Back
Top Bottom