From about 10k hz and higher, I see a spreading of the signal trace coming out of a Philips TDA1543. These images (10k and 20k hz sine waves on a cd disk) are from the analog output of the CD player, but the output from the DAC chip looks similar. The spreading is on lower frequencies, but it gets steadily worse as freq. increases.
The power supply looks okay (+5 volts, 2 mV ripple and noise), and I tried a few different TDA1543's (it is a single chip DAC, in an old Philips-based player), and the signal traces are spread in the same way. Even a brand new chip has this problem.
I used the same disk in another cd player and the traces look good (single trace).
Any ideas what this could be from? Is there a term for this pattern?
The power supply looks okay (+5 volts, 2 mV ripple and noise), and I tried a few different TDA1543's (it is a single chip DAC, in an old Philips-based player), and the signal traces are spread in the same way. Even a brand new chip has this problem.
I used the same disk in another cd player and the traces look good (single trace).
Any ideas what this could be from? Is there a term for this pattern?
It seems to me what you're seeing is imaging, as the TDA1543 has no (or a very shallow) anti-imaging filter. Therefore when outputting 10kHz a frequency of (44,100-10,000) = 31,100Hz is also present. The spreading at lower frequencies is lower because of the sinc form of the zero-order hold function.
Thank you, your answer is very helpful.
I wonder if NOS DACs using this chip have this imaging problem?
The audio does not sound bad, like there is something terribly wrong. But this player has always sounded harsh to me, since it was new, and lately I have been looking into improving it to use it as a spare player.
I wonder if NOS DACs using this chip have this imaging problem?
The audio does not sound bad, like there is something terribly wrong. But this player has always sounded harsh to me, since it was new, and lately I have been looking into improving it to use it as a spare player.
Your anti-imaging filters are interesting. I will try the anti-imaging filter you have labeled as ".ac dec 600 1k 200k", but I have a few questions that may be helpful to others that want to try this:
1) What are reasonable specs for the inductors? (max. dc current, impedance, etc.).
2) Is there a resistor missing at L4?
3) How tight are the specs for the small caps? (e.g., a 3.1nf is hard to find, but 3.0 and 3.2 are standard). Do I need to combine caps to get to these values?
4) Each channel gets a filter, and V1 goes to AOL (or AOR for the right-channel filter) on the DAC, and AC2 from each filter goes to VREF on the DAC?
Thanks!
1) What are reasonable specs for the inductors? (max. dc current, impedance, etc.).
2) Is there a resistor missing at L4?
3) How tight are the specs for the small caps? (e.g., a 3.1nf is hard to find, but 3.0 and 3.2 are standard). Do I need to combine caps to get to these values?
4) Each channel gets a filter, and V1 goes to AOL (or AOR for the right-channel filter) on the DAC, and AC2 from each filter goes to VREF on the DAC?
Thanks!
Thanks for your interest 
I take it you mean the 4 inductor one in the 'Better sounding filter' post.
The 10mH inductors I used were very cheap ones, bobbin wound. I bought a bag of a few hundred and measured their values on my LCR meter to get ones which were within 1%. The losses (series resistance of the wire plus core losses) are the most critical factor in choosing a suitable part. You'll note my inductors are 14ohms at DC and around 25ohms @ 20kHz. Lower values than these are acceptable as series resistors can be added if necessary. Max DC current is very unlikely to be exceeded on any practical 10mH inductor as here we're running at low single digit mA.
The apparent missing resistor is intentional - resistors are being used for damping but for L4 I need the highest Q (lowest damping) and the AC losses of the inductor itself provide it without any additional resistor.
Cap values do need to be within 1% if you want to get the stop band rejection shown in the plot. Its the difficulty of doing this (caps in parallel, selection of inductors) that has caused me to abandon this topology of filter (elliptic) in favour of Chebyshev.
To use this filter with TDA1543 I suggest providing a stable voltage reference with a TL431 - see the post about 'current source elimination' for that DAC.
I take it you mean the 4 inductor one in the 'Better sounding filter' post.The 10mH inductors I used were very cheap ones, bobbin wound. I bought a bag of a few hundred and measured their values on my LCR meter to get ones which were within 1%. The losses (series resistance of the wire plus core losses) are the most critical factor in choosing a suitable part. You'll note my inductors are 14ohms at DC and around 25ohms @ 20kHz. Lower values than these are acceptable as series resistors can be added if necessary. Max DC current is very unlikely to be exceeded on any practical 10mH inductor as here we're running at low single digit mA.
The apparent missing resistor is intentional - resistors are being used for damping but for L4 I need the highest Q (lowest damping) and the AC losses of the inductor itself provide it without any additional resistor.
Cap values do need to be within 1% if you want to get the stop band rejection shown in the plot. Its the difficulty of doing this (caps in parallel, selection of inductors) that has caused me to abandon this topology of filter (elliptic) in favour of Chebyshev.
To use this filter with TDA1543 I suggest providing a stable voltage reference with a TL431 - see the post about 'current source elimination' for that DAC.
You would see the same trace with any DAC chip which does not include a built-in reconstruction filter. Many NOS designs omit this, so the same trace would be seen at the output terminals. The other player which does not show this obviously includes a reconstruction filter.mmerig said:
As abraxalito says, you are seeing images.
Many thanks to abraxalito and DF96 -- I understand what is going on now. I am amazed that imaging (or aliasing) this extreme is prevalent in many NOS DACs, and listeners accept it or even like it. If a preamp had this kind of distortion, I wonder if it would be so embraced. The imaging probably explains why I never liked the way this cd player sounds -- very harsh and fatiguing.
Abraxalito's passive filters were attractive because of the low cost, but the tedious matching was a deal killer for me. I will probably put in a current-output dac to get around the TDA1543's problems. Russ White's COD , for example, has potential.
Abraxalito's passive filters were attractive because of the low cost, but the tedious matching was a deal killer for me. I will probably put in a current-output dac to get around the TDA1543's problems. Russ White's COD , for example, has potential.
Listeners accept the imaging because despite how bad it looks on the trace it still sounds preferable to the noise modulation generated by S-D type DACs. I very much doubt that it explains harshness and fatiguing sound as I and others haven't found harshness when listening to TDA1543 minus an anti-imaging filter.
True that the TDA1543's sound doesn't appeal to everyone, some miss the 'detail' which other DACs bring to the party. That was my initial response on first hearing a NOS TDA1543 - too mellow. The harshness you experience might be some implementation detail of your setup. Plenty of other listeners use TDA1541s with no output filter and don't report harshness.
I grew up with vinyl records, mellow is what I like, and I got it with an Arcam 172, which has a B-B 1716 DAC, until the wiring leading to my house was struck by lightening and ruined the dac (not much else, fortunately).
I have had the same preamp and amp system the whole time -- the only change is the cd player (the old Philips based one and then B-B based). The external dac I have now has a B-B 1794 in it with an SRC4192. Some years later, my house was struck by lightening again, (!) and this ruined the cd player and I have been struggling with used, unreliable transports since, but that is another story.
So that is why I point to the TDA1543 based cd player as the source for harshness. I have read about the mellow TDA154x DACs, but I have only heard the one that I have and would never characterize it as mellow -- even with a headphone in the cd player's jack.
I don't doubt that others have better experiences with TDA154x's but for some unexplained reason, I have not. Maybe it is the clock, after all.
Thanks for your insights!
I have had the same preamp and amp system the whole time -- the only change is the cd player (the old Philips based one and then B-B based). The external dac I have now has a B-B 1794 in it with an SRC4192. Some years later, my house was struck by lightening again, (!) and this ruined the cd player and I have been struggling with used, unreliable transports since, but that is another story.
So that is why I point to the TDA1543 based cd player as the source for harshness. I have read about the mellow TDA154x DACs, but I have only heard the one that I have and would never characterize it as mellow -- even with a headphone in the cd player's jack.
I don't doubt that others have better experiences with TDA154x's but for some unexplained reason, I have not. Maybe it is the clock, after all.
Thanks for your insights!
To me it also seems highly unlikely that clock jitter would make the sound fatiguing. I've paid almost zero attention to jitter in all my DAC developments with multibit chips and never found harshness to be as a result of that.
The fact that you get harshness on headphones is indicating that there must be something amiss with your player. Otherwise I'd have suggested it was a system noise problem between the player and the amp, but this data point rules out such an explanation.
TDA154x are definitely the right DAC choice for you it seems, just rather puzzling to me that yours isn't doing anything for you (or your other half)...
The fact that you get harshness on headphones is indicating that there must be something amiss with your player. Otherwise I'd have suggested it was a system noise problem between the player and the amp, but this data point rules out such an explanation.
TDA154x are definitely the right DAC choice for you it seems, just rather puzzling to me that yours isn't doing anything for you (or your other half)...
Could be that something downstream of the DAC does not like ultrasonic energy present at its inputs?
I have seen this where the amp behaved badly from an IMD perspective when fed with ultrasonic tones and this could well result in in band mixing products that are not harmonically related to the program audio.
Personally I don't do the NOS thing, but designing circuits with large out of band spurs in the outputs just seems to be asking for trouble with the next stage to me.
Far better to be conservative in what you output and to design to be liberal in what you will accept from the previous stage, NOS with no (or poor) output filtering violates this design philosophy.
73 Dan.
I have seen this where the amp behaved badly from an IMD perspective when fed with ultrasonic tones and this could well result in in band mixing products that are not harmonically related to the program audio.
Personally I don't do the NOS thing, but designing circuits with large out of band spurs in the outputs just seems to be asking for trouble with the next stage to me.
Far better to be conservative in what you output and to design to be liberal in what you will accept from the previous stage, NOS with no (or poor) output filtering violates this design philosophy.
73 Dan.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.