Too bad the cs8412 is holding it back. It restricts it to 48kHz where the TDA1543 can do proper 96 kHz. The sweet spot for high res audio.
Also those two blue caps in upright position before the output hold it back. Those are the all important output caps. If you short them out your soundstage and detail will increase tremendously. Those El cheapo caps make it sound veiled and lifeless.
Any normal amp has input caps that do the exact same thing; bring the bias down from 3V to 0. It's a high-pass filter that takes out the low frequencies (including that 0Hz bias). You'll only notice crackling when you turn the volume. That's it.
If that scares you, replace them with high quality MKP or PiO caps 2uF minimal, 10uF nominal, voltage unimportant.
Yes, it's absolutely a very well designed chip, the stupid thing is that is was always absolutely ruined by the digital upsampling and output filters. If I interpret what I see on your board correctly it does not have a fixed low-pass filter that ruins the upper treble. One hole is empty and another one is shorted that contrasts with the original layout of the board.
All the TDA1543 needs is a resistor between signal and ground of the output. That should be that black part.
That 1543 kit will get your feet wet in diy.
About shorting out those output caps…
You can do so with confidence ONLY if after do so you:
— read only a few mV DC on output using DMM
— read only a few mV DC on output during pwr on/off cycles (transients)
Looks exactly the same, only he would use boutique parts in the signal path. The only place where that would really make a difference is the output caps. And if you short them it will sound like no caps, which is better than the most expensive caps you can think of.
Here's another schematic:
The only part that might be changed is the 2200 ohm resistor for a value just a bit lower to keep it from overmodulating on peaks. I use 470 on a quad so just under 2k (1k8 or 2k). Or bridge it with a 20k, same result. That's a thing I found out with trial and error listening to my reference track. With piano strokes peaking I get distortion with 510R, and that's over with just a bit lower 470R. For 1 tda1543 multiply by 4. Just simple carbon 0.25w resistors.
If you remove the caps you will get 3V DC on the output. I never had any trouble with that. I know it's not by the book, but it just works better.
Also: there's a voltage regulator on board. That's good. I had a real hard time getting any difference in sound, let alone better, with a linear power supply. Just so you know. You might want to measure with a DMM if you can on pin 4 and 5 (furthest from the U dent). It should be below 8V and preferably around 7v. Not 5V, then I'd up it a notch or two.
⚠ WARNING : the schematic you linked says 9V. That is a bad idea. Even with a heatsink, like mine has, standard when on 8V, makes it burn out in 6 months. There were many complaints of failures. I've had mine for over 10 years.
I've done around 20 of these dacs and never had any complaints. I am not an electronics engineer so I never learned to limit my thinking inside the box. I am an engineer though and I did learn the basics. I'm not stupid and I will not give dangerous advice. Just in this, age of snowflakes where every plastic bag needs to be printed with 'don't pull this over your head' makes me feel like I'm in the movie idiocracy. I just saw it on TV last week and I had a hard time discerning between the ads, and the movie itself. A lot more than, say 10 years ago.
The only part that might be changed is the 2200 ohm resistor for a value just a bit lower to keep it from overmodulating on peaks. I use 470 on a quad so just under 2k (1k8 or 2k). Or bridge it with a 20k, same result. That's a thing I found out with trial and error listening to my reference track. With piano strokes peaking I get distortion with 510R, and that's over with just a bit lower 470R. For 1 tda1543 multiply by 4. Just simple carbon 0.25w resistors.
Good question. No, it's not the digital part that is clipping of course. That would make it a bad recording. An engineer always keeps it just under 0dB.
It's the output impedance. (ok, I might say something stupid now) I think it pulls just a bit to much current from the chip.
And another admission of stupidity (insert même with people slapping their face). I bought a 2:1 transformer to get a little more oompf and get rid of the DC. Good idea at first. Of course I was eager to test it. It sounded broken. Like a badly tuned radio.
What I forgot is that a transformer has next to no resistance, so this reduced my output impedance to close to 0. Duh! Of course it won't work. Not like this. The resistor should be in front of the transformer, not to ground.
OH well. The best way to learn is to make mistakes.
Good question. No, it's not the digital part that is clipping of course. That would make it a bad recording. An engineer always keeps it just under 0dB.
It's the output impedance. (ok, I might say something stupid now) I think it pulls just a bit to much current from the chip.
And another admission of stupidity (insert même with people slapping their face). I bought a 2:1 transformer to get a little more oompf and get rid of the DC. Good idea at first. Of course I was eager to test it. It sounded broken. Like a badly tuned radio.
What I forgot is that a transformer has next to no resistance, so this reduced my output impedance to close to 0. Duh! Of course it won't work. Not like this. The resistor should be in front of the transformer, not to ground.
OH well. The best way to learn is to make mistakes.
Auto suggestion I guess, as @viggen initially presented a board with oversampling filter chip.
Regarding transformer, I would not allow any DC current to go through. It would need to have DC separation by a capacitor (but we don't like capacitors in audio path) or a balanced active or level shifting buffer with a precision DC servo loopback. But then we don't need a transformer. Transformers are good when a source impedance is high, it is driven by a current source.
Had a look at your dual 1794 Ali unit. It'll do for a convenient all-in-one headphone-amp device. But if fidelity is a important criteria -- over the cost/convenience of that headphone unit -- then there are better dual-1794 options on Ali. E.g., A stuffed dual 1794 kit, which also has avail separate PS modules....
I wish I knew what to do with all them boards and modules... but am deciding to get the 1794 all in one device in lieu of a Merason atm. both using the same dac chip but Mersason is not using a pair.
That all-in-one box of dual 1794 for a headphone out is not up to what is capable with 1794s in my post.
But you may have other priorities wrt time and patience. I’m not going to mention $ , because those one box units just get quickly thrown into drawers. Initially cheaper and easier, yes. But ultimately: Quick to bore; just buy another cheap Ali box .
Whereas with PCB kit, the diy potential is there and that can be BOTH more fulfilling and economical.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.