[IpsilonSound]

I managed to reduce the THD to as low as 0.017% on my TDA1543-based DAC, but these harmonics are still troubling me. They are not random, as you can see, and for 1kHz tone additional 2, 3, 5, 7, 9, 11kHz tones are visible on the graph. Despite that, the sound is very very good after lots of optimization and meticulous part selection.

The i/v stage is still a passive one, I'm going to start working on an active one very soon.

Any idea what causes these harmonics, and if they will likely go away if I switch to active i/v stage, reduce the load on the output of the chip?

 

[Wodgy]

Those harmonics are inherent in the design of the TDA1543. You're already getting slightly better THD numbers than the datasheet indicates (0.018% typ) so it's not reasonable to expect that you'll be able to eke out any better performance.

 

[IpsilonSound]

Quote:

Originally Posted by Wodgy Those harmonics are inherent in the design of the TDA1543.

I already thought about it... any special reason, why these harmonics are designed for? it's like dithering the sound, maybe, to make it sound more pleasing to the ear?

I saw the graphs for TDA1541-based DAC and there were no harmonic sounds for it. Interesting,eh?

 

[Wodgy]

Those harmonics are not "designed in" -- they're a byproduct of the conversion process the TDA1543 uses. Every DAC has some level of this going on, but on more modern DACs it's usually less in absolute terms. The harmonic distortion profile will definitely contribute to the distinctive sound of the DAC. If you're seeing: 1) that even harmonics are generally greater in absolute terms than odd harmonics, and 2) that there is a natural rolloff, i.e. |2nd harmonic| > |3rd harmonic| > |4th harmonic| > etc. then there's a good chance that the sound will be agreeable to the ear. It won't be as clean or as black as a DAC with better specs, but those reasons are not generally why people are experimenting with these kinds of DACs; people just seem to like the overall sound.

If the graphs of a TDA1541-based converter show no harmonics, then the DAC was not measured properly, or the measurement equipment was flawed. The TDA1541 may very well have better specs than the 1543 (I haven't looked into this), but no DAC has a THD of 0%. (And especially not low-cost DACs developed 15 years ago!)

 

[drewd]

Another thing to consider is that the level of these harmonics is pretty low - ignoring the 2nd harmonic for a moment, look at how low the rest of the harmonics are...somewhere less than -95dB. That's really quite low, especially considering that the DAC you're using isn't state of the art. The second harmonic, even at -80dB shouldn't be too troubling. That's a relatively low level and, to the ear, even harmonics are perceived much more musically than odd harmonics.

To be honest, I think that you're getting pretty good results from your DAC.

-Drew

 

[IpsilonSound]

Quote:

Originally Posted by drewd To be honest, I think that you're getting pretty good results from your DAC. -Drew

Thanks, I wasn't expecting such specs myself, initially thought that it's the software or the soundcard. But the big difference was reducing the i/v load resistors(affects the voltage levels, of course = 1V output now), and choosing better power regulators, as well as proper caps for power and filtering. The spikes(harmonics) were bigger with 2V output, so was the THD. I guess, using the active opamp stage with little load on the chip would dramatically improve the performance and reduce the harmonics. will see

 

[aos]

I belive you can improve THD of those TDA chips by parallelling them (piggyback). And adjusting the load resistor until the THD is minimized. I have a DAC here with those chips that I ordered from www.dddac.de but I haven't built it yet. He does have a fair bit of measurements over there so you can see for yourself.

Those are old designs however and you shouldn't expect miracles as far as THD is concerned. Besides between DACs that I have, there are some that sound much better than the other ones that measure better.

 

[IpsilonSound]

I tried 2 chips in parallel but didn't like the sound, though it was a little more powerful and dynamic. The focus and image wasn't that sharp anymore, plus the 2-chip combination would drain the battery quicker, since the TDA1543 alone consumes about 60mA. I'm working on an active opamp i/v stage at the moment, it will allow reducing the load and decrease the THD at the same time. The dynamics, especially bass will improve with an active stage too.

The opamp choice will be LM7172 or THS4062, I'm more inclined towards the latter. I tried THS4061 and THS4081 chips in my ppa a couple of days ago and was pleasantly suprised, if not to say more. The resolution and clarity was phenomenal, excellent highs with cymbals going all the way to the ultrasound region instead of being mudded with other instruments. The separation of the instruments is a way better than AD8610, the bass is sharp clear and very very deep.

Overall, with THS4081 the music seems to be created effortlessly and it makes you feel like listening to the same music over again. It's like looking thru a foggy window, cleaning the glass and looking thru the window again.

Anyone, give it a try with these chips, let me know.

 

[jefemeister]

The picture is gone. Did the harmonics decay at a nice exponential rate? If so, the DAC will sound good. If one of the higher harmonics sticks out above the previous ones, something will sound off. Harmonics always fall off exponentially in nature.

I would expect a proper I/V stage will help too. Although it's tempting, hanging a resitor off the current output pin is a bad idea. DACs want to see a zero impedance load.

 

[IpsilonSound]

The THS4061/81 don't work with Diamond Buffer from Larry, giving terrible oscillations

. It works great with HAL5002 though.