Modding Taobao digital amps

Jump to Latest
#122
FYI I just attempted to resolder the STA333BW (before glueing on a heatsink) and with all the tricks I know I can not get the solder from between the pins 🙁

Of course I am out of solder flux today. Well, a new one is on its way 😉
 
Last edited:
#125
jean-paul said:
Not on the contrary! As said, the designer made a nice design but forgot to include a 100 µF bulk cap and a few 1 µF caps....

I already noticed the reg as there is no other semiconductor to regulate 3.3V.

* "662" is Torex XC6206 3.3V 200 mA 0.25V dropout LDO reg.... Not even noise specs in datasheet. Uh oh 🙂
Click to expand...

Thanks. How do search for those things? Agree, no noise spec, not good.
Its so small it quite hard to change. I have a ultra low noise adp151 of similar size but 5 pin, too bad.
However, I will replace the 78m05 to LT3088.

So far, just cap tweak and the sta333 is sounding quite nice.
I would say this board is the current value champ to beat for value vs sound quality, dac and amp all in one.🙂

PS. there is 6 x 0603 smd caps for each channel of speaker output. 3 are extra to the datasheet, I plan to replace those with higher quality, probably solder them under the board to the 0.47uf film cap pins.
 
#129
abraxalito said:
You'll definitely want to do the tweak to the level shifting caps I detail on my blog. That's a design fault which could easily be responsible for the graininess on vocals you're hearing.
Click to expand...

I have add 10nf to one channel and the difference is not that large. I notice Intersil app note for another D2 chip c50 position uses 100nf (note I am just guessing its the same function the 2.2nf cap as in d2p)
I wonder if that cap can go up to 100nf as well.

https://www.intersil.com/content/dam/Intersil/documents/an14/an1435.pdf
 

Attachments

  • Intersil an1435 d2 84.JPG
    Intersil an1435 d2 84.JPG
    63.5 KB · Views: 274
#130
When you say the difference is not that large, were you looking on a 'scope with a load attached to the output? There, the difference is huge.

Good find with the app note. 100nF is huge overkill in that application though its probably an X7R which means at 25V its capacitance is much, much less than advertised (10nF potentially).
 
Last edited:
#131
abraxalito said:
When you say the difference is not that large, were you looking on a 'scope with a load attached to the output? There, the difference is huge.

Good find with the app note. 100nF is huge overkill in that application though its probably an X7R which means at 25V its capacitance is much, much less than advertised (10nF potentially).
Click to expand...

I don't think X7R value drops that much.
I added some 10uF smd to around the LM2596 and it seems to smooth out the things a bit. But during the higher vocal dynamic the sound became hard, so a thing just occur to me the culprit might be the inductor. Those 15uH unknown inductor might not be suitable for class D.
I am going to replace it with a better 22uH inductor.
 
#133
abraxalito said:
I checked the output inductors on my LCR meter, they measure fine.
Click to expand...

It's not the inductor value, it's the quality/distortion at the mhz range and core salutation at high current. I had a green/yellow toroid on one of the 3116 board which has correct value but the sound sucks, hard and grainy, esp during loud passages, replacing that with red -2 type toroid made a huge difference. The sound of the D2P at loud vocal passage has a hint of that bad yellow/green inductor sound, so that is why suggest the existing inductor could have a bad affect to the sound.

My knowledge in inductor is far too basic, but I think most of the inductors out there are not really suitable for class D output.
Those shield type can range from a few cents (taobao price) to a Wurth for 2.46USD at digikey (which also has a self resonance frequency of 26mHz)
7447471220 Wurth Electronics Inc | Inductors, Coils, Chokes | DigiKey

Unfortunately there is not much space in the D2P to use other large type inductor.
 
#134
A red type -2 toroid is most likely a carbonyl core, very low mu hence hard to make it saturate. Yet for an 8ohm load with the stock PSU the output current is always going to be under 3A. Dunno what green/yellow cores might be, never met those.

Plenty of space above the PCB if you stand off the inductors with solid core wires.
 
#135
abraxalito said:
A red type -2 toroid is most likely a carbonyl core, very low mu hence hard to make it saturate. Yet for an 8ohm load with the stock PSU the output current is always going to be under 3A. Dunno what green/yellow cores might be, never met those.

Plenty of space above the PCB if you stand off the inductors with solid core wires.
Click to expand...

D2P. I modified one channel with Coilcraft 1210 22uH (had to solder a wire to made it into a stand up thru hole) and the 2200pf to kp1830 15nf cap. Strangely the modified channel is louder by several db, I wonder if this could be due to the much lower DCR of the Coilcraft inductor?
Maybe there is a touch of improvement, but could can be 100% sure.
I think there si more improvement by changing LM3596 inductor from 15uH to 100uH (note the datasheet spec around 100uH for high input voltage) and I also change the 4 power cap from 470 to Panasonic FM 1000uH . But I think more improvement is from adding some 10uF and 100nF across those 4 power caps.
 
#137
abraxalito said:
Upping the 2n2 to 15nF would produce a significant increase in volume into a normal speaker for sure. That's because the FET when driven via the 2n2 barely gets switched on, it can pass roughly 1A before coming out of saturation.
Click to expand...

Thanks for the explanation.😛
Anyways, I would recommend the 100uH inductor mod for lm2596 and some small caps bypass for the 470uf power caps for the fets. I add a 10uF 100v x7r + 100nf COG for the bypass.
 
#138
Good catch on the LM2596 inductor, I reckon even higher than 100uH might be worth a punt as the total 5V supply current is going to be under 0.6A (the LH edge of the graph). The DSP is taking 0.3A typical but the rest probably doesn't amount to much at all. I'll give 150uH a shot.
 
#139
A perfect DIY DAC?

Hi,

I've been following this post with interest and now have a STA333 board:

Would anyone be kind enough to summarise the steps necessary to convert this board into a 'top notch' DIY DAC please?

I feel that most posters here have a much higher knowledge of electronics than I do!

I can follow instructions/order parts and solder :up: ?

Happy Christmas everyone!
 
You must log in or register to reply here.
Top Bottom