Tweaking the TDA1543 SDcard player
Posted 6th September 2013 at 12:05 AM by abraxalito
This board is really a fun one to tweak mainly because its so simple and relatively self-contained, so results of mods can be verified very quickly.
In stock form, the sound is fairly typical 1543, nothing unpleasant but an almost total lack of soundstage depth. On recordings with a high ratio of ambient to direct sounds (Nimbus) the acoustic cues are muddied. The flatness of presentation is one of the reasons I was so slow to accept 1543's other virtues during the course of development of my DACs. Soundstage depth is a biggie for me. There's also a lack of 'finesse' on the HF - lack of 'poise' or 'delicacy'.
I started by fixing up the grounding to enforce a star earth on the analog outputs and power supply. Then went on to filter the DAC's supply regulator (7805) with series ferrite beads and additional lytics and ceramics. It later transpired that the beads were being shorted by a parallel track on the other side of the board. I installed filtering prior to the 7805 in the form of a 220uH inductor (damped with 120R) and a 10uF ceramic between input and gnd pins. This gave a substantial improvement in the delicacy. The wall-wart PSU (9V) supplied is a switcher so emboldened by this success I installed ferrite beads on the input and replaced the 560uF input smoothing cap with paralleled Rubycon ZLH 1000uFs. More delicacy and drive to the sound was forthcoming.
Soundstage was still an issue though so I turned to the DAC's output circuitry. In the past I've had success with removing output caps - typical implementations use 2.2nF Wima polyprops across the I/V resistors. I decided to change these to NP0s and modify the operation of this stage. One of the mods which has worked for me in the past is disabling the internal current sources of the 1543 and doing the job externally. This time though I went for a tangential approach - a TL431 set to 3.75V and resistors to the DAC hung down from this reference. The result was that the music took a step back from the speakers but lost some of the poise. With a passive filter on the output of the 431 (120R damped 68uH inductor, 2 * 220uF to gnd) the poise came back and some. I plan to investigate and optimize this filter, it seems to have a major bearing on the sound. Perhaps try a 4th order filter....
Anyway now the overall sound is quite addictive, it only reveals its shortcomings when A/B'd with my Ozone (based on TDA1387s and with a passive filter) - the soundstage is nowhere near as deep and well-defined and the HF droop becomes apparent. For about 15UKP its bloody amazing
In stock form, the sound is fairly typical 1543, nothing unpleasant but an almost total lack of soundstage depth. On recordings with a high ratio of ambient to direct sounds (Nimbus) the acoustic cues are muddied. The flatness of presentation is one of the reasons I was so slow to accept 1543's other virtues during the course of development of my DACs. Soundstage depth is a biggie for me. There's also a lack of 'finesse' on the HF - lack of 'poise' or 'delicacy'.
I started by fixing up the grounding to enforce a star earth on the analog outputs and power supply. Then went on to filter the DAC's supply regulator (7805) with series ferrite beads and additional lytics and ceramics. It later transpired that the beads were being shorted by a parallel track on the other side of the board. I installed filtering prior to the 7805 in the form of a 220uH inductor (damped with 120R) and a 10uF ceramic between input and gnd pins. This gave a substantial improvement in the delicacy. The wall-wart PSU (9V) supplied is a switcher so emboldened by this success I installed ferrite beads on the input and replaced the 560uF input smoothing cap with paralleled Rubycon ZLH 1000uFs. More delicacy and drive to the sound was forthcoming.
Soundstage was still an issue though so I turned to the DAC's output circuitry. In the past I've had success with removing output caps - typical implementations use 2.2nF Wima polyprops across the I/V resistors. I decided to change these to NP0s and modify the operation of this stage. One of the mods which has worked for me in the past is disabling the internal current sources of the 1543 and doing the job externally. This time though I went for a tangential approach - a TL431 set to 3.75V and resistors to the DAC hung down from this reference. The result was that the music took a step back from the speakers but lost some of the poise. With a passive filter on the output of the 431 (120R damped 68uH inductor, 2 * 220uF to gnd) the poise came back and some. I plan to investigate and optimize this filter, it seems to have a major bearing on the sound. Perhaps try a 4th order filter....
Anyway now the overall sound is quite addictive, it only reveals its shortcomings when A/B'd with my Ozone (based on TDA1387s and with a passive filter) - the soundstage is nowhere near as deep and well-defined and the HF droop becomes apparent. For about 15UKP its bloody amazing
Total Comments 5
Comments
-
Nice work, Richard. The fact that such good sound can be retrieved for so little dosh is remarkable - I've had this experience too many times in various ways to not be astonished, but it's always good to hear other's experiences.
On that note the Aldi TV is providing great amusement, yesterday I got it to a better level again, by cheating - simply shut down more problem sources of interference as a quick fix. And was able to run heavily compressed current recordings, and a weird Metallica thing, at deafening levels relatively cleanly: zero real bass, and bits of plastic rattling away depending on the note struck, but it worked! Microscopic speakers, those nothing oval things you get on monitor screens, but the sound was pumping through the house very nicely.
It's the sort of thing that most would consider totally unbelievable, unless they experienced it directly ...
Edit: Speaking of Nimbus recordings, with deep acoustics, I can recommend Franck/Faure, The Medici String Quartet/Piano, NI 5114. The piano is in a "monstrous cavern", far, far away; also an excellent test for general string tone ...Posted 6th September 2013 at 04:17 AM by fas42
Updated 6th September 2013 at 04:40 AM by fas42 -
I'm left wondering who is getting the most amusement from your Aldi TV - you or SY
Posted 7th September 2013 at 12:57 AM by abraxalito
-
To add a thought, and you've commented on this before - where the cheap stuff, or that which has not been well implemented, has problems, IME, is that optimum performance only occurs when the circuitry and associated hardware is fully stable, in hard, continuous, 'working' mode.
I've experienced this quite a bit, and the Aldi combo is the worst in this regard. From cold it sounds quite terrible, one cannot fathom that it has any chance of sounding good - you just have to wait, :). And on the other side, if you let it 'cool off' it rapidly degrades -- so, after, say, half an hour of zero sound out the sound quality plummets badly, it reverts to a relatively lifeless quality.
The TV sound system, digital amplifier plus speakers, seem to be the main culprits in this regard, from my experiments so far. Your favourite whipping boy, S-D digital, has issues with needing to be constantly exercised to give of its best - and as you point out, it shouldn't be that way, the gear should just work 'right', after say 5 mins of warm-up ...Posted 7th September 2013 at 01:09 AM by fas42
-
From SY's comments, I think it would be a bit scary listening to his system - I suspect that only the very, very cleanest recordings would be listenable to, anything a bit challenging could be quite a disaster ... ;)Posted 7th September 2013 at 01:14 AM by fas42
-
Further on warming up, conditioning cheap gear - I did pretty well on ranking Pano's last round of silly interconnect samples, on ordinary desktop and PC speakers. But, to give away a 'secret', that I've hinted at many times, I wouldn't have had a chance of doing that from the normal startup state of the gear - before seriously listening to the versions I hammered the speakers for about an hour or two, on continuous repeat, with stressful material to make sure all the electronics and drivers bits were in prime shape ...Posted 7th September 2013 at 01:51 AM by fas42
