DDDAC 1543 MKII distortion on analog output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I just completed DDDAC 1543 MKII and I have a problem at the output - something is riding base harmonics, and sound is pretty shi**y. While waiting for response from Doede, thought someone might seen this before or have an idea where to look. Boards are supplied with 11V LM317 stabilized, 120VA toroid.

I am feeding it with SPDIF from SB live and CD4000. FS looks quite good on the scope. I tend to suspect TDA1543 tower.
 

Attachments

  • output.gif
    output.gif
    31.7 KB · Views: 543
Member
Joined 2003
Paid Member
hi,

i saw your mal. but i am traveling and for some reason cannot send mail from the hotel network :confused:

any way, we can take this off line when I am back next week. Two questions already, have you put the sound card on 44.1 kHz and 16 bit? any thing else will not work fine

have you tried a "normal" cd player to the spdif input?

regards
doede
 
Hi Doede,

Did some more testing. Sound from SPDIF out of Marantz CD4000 is much better. However, when I put test CD with 1kHz sine wave, I hear low freq "clicks", around 1Hz, even from CD player. I did not catch that on the scope. Same signal/CD plays clean from CD4000's analog output.

I am using the same sound card settings which I used for DiyEden SVDAC without problems. I will doublecheck them.

The signal riding sine wave is too high, around 50KHz so that is not what I heard :) Not present in CD4000 DAC though, but that is probably not my problem.
 
Output format: 44100 Hz, 16 bits per sample, 2 channels
Active buffer size: 2000 ms (352800 bytes)
Device: "02: SB Live! Audio [9400]"
Mixing: hardware, primary buffer: inactive

Buffer playback cursor: 16444 bytes
[===#============================================================]
Buffer write cursor: 334800 bytes
[=============================================================#==]

Data buffered:
Total: 1995 ms (351908 bytes)
Async buffer: 190 ms (33552 bytes)

Buffer locks done: 536
Underruns: 0
Time played: 1:06.093 (11658844 bytes)
Time written: 1:08.088 (12010752 bytes)
Total time played: 2d 4:38:17.673
Volume: 0.000000 dB / 0.000000 dB
 
Member
Joined 2003
Paid Member
Things seems to fall in place. This theme has been around here a few more times before already. If you "listen" to pure sinewaves with NOS, A-Synchrone reclocked dacs, you will have clicks on a frequency which is the difference between the source clock and the reclocking clock. No way around it.

Listening to music (what most of us do) this effect is not noticable and interestingly enough (that is why it is done) produces more detailed soundstage and more spatial sound than with no reclocking (cs8414 is synchrone with source clock with build in pll)
 
Now that I know what to listen to, clicks are audible for sure. I find it difficult to accept it as normal.

If clicks are audible with sinewaves, they are also audible with music fragments containing sinewaves.


I can verify that DDDACs clocking method indeed improves transparency by effectively blocking source jitter, so the concept is valid.


The clicks are caused by the CS8414 receiver buffer under-run / overflow. This problem is caused by slave clock mode (both BCK and WS are inputs), in combination with source and master clock running (slightly) out of sync.

If the source clock runs slightly slower than the master clock, there won't be enough samples to fill the receiver buffer, so samples need to be repeated, if this condition lasts too long, too many samples are repeated sequentially, and audible clicks occur.

If the source clock runs slightly faster than the master clock, there are too many samples, and the receiver buffer overflows, now samples have to be dropped, if this condition lasts too long, too many samples are dropped sequentially, and audible clicks occur.

So the problem with clicks can simply be solved by avoiding receiver buffer over / under runs, in practice this means matching / syncing the master clock with the source clock.

One possible option is to use an adjustable master clock (+/- 100ppm), and match the master clock frequency as closely to the source clock as possible. This should reduce / eliminate audible clicks. Tentlabs offers 2.8224 MHz low-jitter VCXOs (Voltage Controlled Crystal Oscillators). The oscillator frequency of these VCXOs can be varied slightly (approx. +/- 100ppm) by varying the DC control voltage between 0 and 5V. The problem with this method is that the adjustment procedure must be repeated for each source, and oscillator drift (master / source) may require re-adjustment.

Best way to solve this problem is using the VCXO with an appropriate PLL configuration, in order to lock the master clock to the source clock.
 
It makes perfect sense. I did not get too much into the subject so far as this is my first stab at DAC part of DIY, but now makes me wonder how other reclocked designs deal with this phenomenon? I don't think it's by having a screwdriver next to DAC chassis.

For me, sacrifying clean and distortion-free sound for the sake of slight improvement in imaging and transparency is not acceptable. I find myself now more trying to catch artifacts, rather than listening to music.

So far, a very unhappy owner of DDDAC.
 
Which problem exactly did you have, parasite harmonics or clicks? I have already resoldered and reseated the whole tower couple days ago, so unlikely to be the cause of the problem. I even replaced 2 TDA1543 just in case.

See also post above from designer claiming that this is normal behaviour due to reclocking, due to difference in incoming SPDIF Fs and reclocked Fs, which makes perfect sense to me.

Can anyone with working SPDIF DDDAC MKII confirm that there are no clicks with 15KHz or 10KHz sine wave? I really hope it's me who did something wrong here.
 
And yes, I almost forgot - I got the same result with one TDA1543 only, so tower is out of the question for sure.

To summarize, there are two issues here, with CD4000's SPDIF as a source:

1) Low frequency clicks, with any freq sine wave. If I run 15KHz for example, which will be barely audible for me, I hear clicks only. Theory that this is audible in sine waves only and not in music, is kind of ridiculous, and I am not amongst those who can hear difference between cables. If I can hear it, anybody can.

2) Parasite harmonics in audible spectrum as depicted above. They are all over the place. Simple comparison with default oversampled output of Marantz CD4000, puts DDDAC down hard in this respect.

I would very much like to hear from other owners of DDDAC MKII with SPDIF receiver.
 
in my case...

I had same issue ('clicks' on SPDIF input). Was caused by low quality chassis female cinch plug, wired to the SPDIF input. The plug was of really low quality (I choose it because of plastic body for isolation from chassis). Once it was changed to a better one --> no more clicks !
Hope it helps.
Benoit
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.