FDA: Alientek D8

Alientek D8 inside

I've opened my Alientek D8 and discovered (rather unsurprisingly) that it has much less discrete parts than recommended by the STA326 datasheet. In the attached picture those omitted are colored red, those present are green, and in blue I colored some changes in the schematic which I noticed. I did not check SMD capacitors, hence their actual values are unknown to me.

Guys, do you think it's worth trying to restore the scheme to the version intended by ST? Will it improve sound quality and maybe at least lower that horrible hissing noise which is audible a meter away from speakers?

Also, I noticed a little issue with XMOS USB connection. After a dozen seconds of silence the XMOS driver drops USB connection and it takes the driver about half a second to establish it again, which results in a corresponding pause between e.g. some Windows event and its sound. This does not affect any prolonged sounds (like music or games), but it is still pretty annoying. Is there a way around it?
 

Attachments

  • STA326 in Alientek D8.png
    STA326 in Alientek D8.png
    165.2 KB · Views: 1,750
This is a low budget solution chip, for teenagers parties or similar, it is function rich, but in the sense of including a digital limiter for protecting boomboxes and budget 2.1 systems from overload, keeping decent sound when loud. It also allows some trade-offs in speaker driver selection with DSP and some biquads per channel.

The datasheet shows 0.1% THD for non-bridged operation, and not better than 0.03% for BTL. High noise was probably allowed in design process. 100dB SNR is stated, but there is no noise floor plot, there could be a rising noise characteristic towards 20khz. Datasheet says something about 3-level modulation, this surely involves a trade-off for increased noise.

The thing they call DDX is also another trade-off for noise, going to analog with a CODEC, and then amplification, achieves 112dB SNR easily without noise increase at HF, so the thing called DDX is involution rather than evolution, for cost reduction.

There is a third factor, there could be HF peaking due to no post-filter feedback, and this could be solved by looking at output with oscilloscope while connected to speakers, feeding 10~20khz from a signal generator, low amplitude, and tuning a series RC in parallel to speaker to flatten the HF portion of the impedance, where success is observed as no peaking when sweeping from 10khz to 20khz.

There could be a fourth factor: Self interference, but being a chip from ST, the self interference limiting factors are probably inside the chip and not in this PCB implementation.

The parts omitted have to do with optional EMI filtering on outputs, but this is far more related to disturbing other equipment than to self disturbance.
 
Datasheet says something about 3-level modulation, this surely involves a trade-off for increased noise.

The thing they call DDX is also another trade-off for noise, going to analog with a CODEC, and then amplification, achieves 112dB SNR easily without noise increase at HF, so the thing called DDX is involution rather than evolution, for cost reduction.
Actually, here and here is what manufacturers say about this "Direct Digital Amplification" a.k.a. "DDX®" - it is just a peculiar version of PWM involving 3 states of the output signal (fully positive, fully negative and fully grounded or "damped") instead of two.
 
Lests call it "increased THD, increased noise floor, and no damping factor, for saving DAC and some more silicon" for boomboxes. Because consumer electronics is no longer hi-fi, and that kind of equipment usually mounts tiny lo-fi speakers into loose plastic frame/case used as acoustic enclosure.

At least CRT TVs used bigger speakers and bigger acoustic enclosure, resulting in better LF reproduction. Now even expensive flat screen builtin speakers are setting the standard for bad thin sound, and this produces a thin-brain syndrome too, for the ones using that "equipment" to calibrate their senses... Chinese takeover hahaha
 
I´m interessted in one aswell, no real info.. some italian sites discuss it but my italian sucks and and google translate doesn´t make it any more clear :) Seems simlar to Q5 Pro / D802

yes the snubber filter on tp1 it s a very good idea. i tested on v200 and on d310. the sound is very much much better.

snubber.jpg

snubber filter
v200tweaklight.png

v200 tightlight without snubber. with snubber is much better.

v200 tight light tutorial :

<<Amplificateurs FDA, episode II - Page 18>> - 30079872 - sur le forum <<Amplificateurs Integres>> - 1030 - du site Homecinema-fr.com

the best resultat : d310 tweak + snubber :

<<FDA D.Sound d310 - Page 106>> - 30064559 - sur le forum <<Amplificateurs Integres>> - 1030 - du site Homecinema-fr.com
 
Last edited:
Okay, I finally killed that nasty hiss of my D8 and not by restoring the STA326 filtering circuit to its datasheet glory (although I did that too), but by inserting an inductor, as large as I could immediately find which turned out to be 15uH torn out from some dead piece of hardware, between the input power line and the 5V DC-DC converter (it operates at 2 MHz). So, the chief culprit appears to be the usual one - a dirty power source, in this case dirt is produced by D8 internally, hence it can't be eliminated by using even the best possible external power supply.
 
I found a way to eliminate the initial lag when using Alientek D8 with XMOS connected via USB: just add a Windows registry key "DeviceStreamingMode" of the type "DWORD" at "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\USB\VID_20B1&PID_302E\0000\Device Parameters\Settings" and put "2" there. With this key set to 2 a USB connection to the XMOS chip is never dropped hence no more lag.
 
hello folks.
I attempted to link DIYINHK's 8ch multichannel board to the D8 via i2s but it didn't work.
inside the D8 is an XMOS based daughter board taking in the USB signal. I removed this to expose the pins and conveniently found labels for i2s on the underside of the board.
It's been suggested (by DimDim) that the master clocks of the DIYINHK and daughter board
could be different, which may very well be the case because the daughter board is an older generation XMOS.
The DIYINHK clocks are spec'ed at 22.5792Mhz and 24.576Mhz which is 8x the BCLK/SCLK.
I don't have an oscilloscope to check the daughter board MCLK line so I want to
ask for someone here to measure their D8 :eek:)
just as with the DIYINHK board the daughter board's MCLK is likely to be an output (to the rest of the D8).
Looking at the unit from the display side (front), the top of the MCLK pin is in the row to the right hand side of the daughter board, 3rd away.

If the clocks are a fraction of the DIYINHK numbers, is it safe to say the D8 simply cannot be driven at those higher rates?

thanks all...
 
Okay, I finally killed that nasty hiss of my D8 and not by restoring the STA326 filtering circuit to its datasheet glory (although I did that too), but by inserting an inductor, as large as I could immediately find which turned out to be 15uH torn out from some dead piece of hardware, between the input power line and the 5V DC-DC converter (it operates at 2 MHz). So, the chief culprit appears to be the usual one - a dirty power source, in this case dirt is produced by D8 internally, hence it can't be eliminated by using even the best possible external power supply.

Audeo, mind sharing some actual pictures of D8 improve mods you did?
can list partlist as well? Thks

globulegl mentioned snubber filter on tp1 it s a very good idea. I guess you are referring pictures provide by Audeo. Does Tp1 refers to Test Point 1 on PCB? Does D8 mods snubber filter refers the 660p Ceremaic cap & 10R resistor? What wattage is suitable for the resistor

Any other mods beside Tp1 is needed?
 
Audeo, mind sharing some actual pictures of D8 improve mods you did?
can list partlist as well? Thks

Those pictures would be really ugly if I were to take them: everything is so crammed and twisted there due to lack of space. The spot where I put an inductor is marked with an X on the attached picture and all other parts are marked red in the schematic I showed previously. I used some film caps and they were not tiny.
 

Attachments

  • inductor_placement.jpg
    inductor_placement.jpg
    383 KB · Views: 754
Last edited:
Thks Audeo for helping, what does Tp1,Tp2,Tp3,Tp4 refers to ? where is it on the PCB?
also what wattage is the resistors used?

I can imagine the mod messing-up; that wld be my challenge as well, as I have no idea where are exact location to solder on, if without actual mod PCB image. Did you also did the blueline connection?
 
Thks Audeo for helping, what does Tp1,Tp2,Tp3,Tp4 refers to ? where is it on the PCB?
also what wattage is the resistors used?

I can imagine the mod messing-up; that wld be my challenge as well, as I have no idea where are exact location to solder on, if without actual mod PCB image. Did you also did the blueline connection?

That schematic was taken from the STA326 datasheet and has nothing in common with Alientek D8 except the chip itself. "TP" must be "test point", I presume (and those do not matter here), and blue lines, as I said in my message accompanying the schematic, indicate not my doing, but that of our Chinese friends - it is what must be undone by some sheer and brutal cutting. You better find the STA326 datasheet and compare it to your actual board layout. You might well find some more discrepancies which were not there in my case. Resistors' wattage does not matter much, I think, because none of them carry any substantial current. I used some smallish SMD resistors everywhere, big enough to be easy to hold and solder.
 
Okay, I finally killed that nasty hiss of my D8 and not by restoring the STA326 filtering circuit to its datasheet glory (although I did that too), but by inserting an inductor, as large as I could immediately find which turned out to be 15uH torn out from some dead piece of hardware, between the input power line and the 5V DC-DC converter (it operates at 2 MHz). So, the chief culprit appears to be the usual one - a dirty power source, in this case dirt is produced by D8 internally, hence it can't be eliminated by using even the best possible external power supply.

So snubber filter at tp1 & tp3 dint help to remove the hiss?

Oh... I think wld just insert the 15uH inductor:
"Input power line" refers to the +Vin from the DC socket, and "5V DC-DC" refers to the 5V regulator +Vin, am i correct?
 
So snubber filter at tp1 & tp3 dint help to remove the hiss?
Even if it did, I did not notice it.

Oh... I think wld just insert the 15uH inductor:
"Input power line" refers to the +Vin from the DC socket, and "5V DC-DC" refers to the 5V regulator +Vin, am i correct?
Actually, the larger the inductor the better, I put a 15uH one there only because I had one at hand. But make sure your inductor can withstand current of 3A or more, just to be on the safe side. "Input power line" here is the wide lane immediately after the diode at the DC socket and "5V DC-DC" can be seen on the photo as a 6-pin SMD chip marked "IV7GK". See attached some schematic from Aliexpress showing its typical application.
 

Attachments

  • IV7GK step-down IC wide voltage input 6-pin power supply chip IV7GF 36V 2MHZ.jpg
    IV7GK step-down IC wide voltage input 6-pin power supply chip IV7GF 36V 2MHZ.jpg
    15.7 KB · Views: 515