Interesting, thanksCould you explain how you measured the relative performance? Was there some test equipment involved or just listening tests?
Also, might be worth mentioning which of those clocks include built in divider circuits so they can natively clock the DAC and servo with just 1 unit.
Thanks,
James
The comparison is basically by my ears and also deploying a scope to compare some of the basic parameters as I don't have the necessary equipment to check the jitters.
Clocks with dividers are FA's C2 clock (now out of production), Audio GD's low jitter JZ-1 clock and the one made in Taiwan (but no more listed in eBay).
thank you for this post... maybe that is the reason why I have also more noise in the right channel
OK gang, I've fixed my channel imbalance problem! All CD player gremlins are now dealt with! It turned out to be just not a great connection to the right channel analogue power supply and/or decoupling cap. I resoldered both and now DC offset in stop is 6.65 left, 6.70 right. Noise floor is now low in both channels and stereo image is centred again. Thanks again, Ray, for your troubleshooting.
Good to hear you fixed it!
The analog DAC ground should be connected to the DOS board input side, because that's where the input signal goes. Signal goes in, but it needs a return as well
If you connect the DOS' PSU ground to the player's board ground somewhere (PSU starpoint or at the output sockets) you should be fine.
Last edited:
Can you tell me something about these clocks in comparison to the standard crystal in the CD player and if such a clock would be an improvement?
The DAC has more ground pins, which are still connected to the player's board. The current can find a way to ground through this other route. Multiple return currents are mixed in this case, and analog ground reference level will get polluted by the digital return current.
I can only say there is a BIG BIG improvement in the SQ. Don't just believe in what I said please go through all the previous threads in this forum and confirm it yourself.
Thanks! I just crammed my DAC PSUs back in, which are the bane of my existence when working on my player due to their spaghetti wiring. They do however seem to make a huge improvement over running off the servo supply, so I guess they're staying. I'm sure 14VA + 88,000uF is overkill for powering the DAC, but I don't know how much by.
Hi guys,
I know I'm not exactly setting the thread on fire at the moment, but I'm feeling energized by recent triumphs in thwarting the CD63 gremlins
I run tests on my player from time to time with a number of test CDs, including the freely available Wave Corrector Test CD.
Until recently, some problems in my player (maybe the ground loop?) caused some audible clipping and distortion of low frequency sine waves. 10Hz - 60Hz tones would produce an audible rattle. However, my speakers now play lovely smooth tones at all frequencies without any kind of break-up, so that's good.
I tried analyzing the output noise of my player last night using my computer sound card and 1khz tones at various different amplitudes. Audacity (also freely available) has a handy tool for spectral analysis. The results can prove quite accurate when working with known test frequencies. For example, analysis of a constant amplitude test sweep yields a nice flat graph from 20Hz to about 19KHz before rolling off.
I had hoped to measure the noise floor of my player, both in terms of amplitude and frequency, but I was unable to calibrate my sound card to accurately measure a 0dB tone without a tiny amount of clipping. I did find though was that while the noise level measured low and even at around -90dB, I consistently saw a pronounced, narrow 9dB 'hump' with a peak at around 4kHz.
Has anyone seen anything similar? I wondered if it could be a narrow band of mains harmonics let through by regulators or insufficient PSU smoothing. I am going to super-regulate my DOS soon and will report back on any improvements there. I did wonder though, if the PSU has a very narrow band of elevated noise, might it be possible to add a capacitor of a particular value to filter it out? My DOS supply has pre-reg caps of 22,000uF, 0.22uF and post-reg of 0.1uF, 220uF.
I know I'm not exactly setting the thread on fire at the moment, but I'm feeling energized by recent triumphs in thwarting the CD63 gremlins
I run tests on my player from time to time with a number of test CDs, including the freely available Wave Corrector Test CD.
Until recently, some problems in my player (maybe the ground loop?) caused some audible clipping and distortion of low frequency sine waves. 10Hz - 60Hz tones would produce an audible rattle. However, my speakers now play lovely smooth tones at all frequencies without any kind of break-up, so that's good.
I tried analyzing the output noise of my player last night using my computer sound card and 1khz tones at various different amplitudes. Audacity (also freely available) has a handy tool for spectral analysis. The results can prove quite accurate when working with known test frequencies. For example, analysis of a constant amplitude test sweep yields a nice flat graph from 20Hz to about 19KHz before rolling off.
I had hoped to measure the noise floor of my player, both in terms of amplitude and frequency, but I was unable to calibrate my sound card to accurately measure a 0dB tone without a tiny amount of clipping. I did find though was that while the noise level measured low and even at around -90dB, I consistently saw a pronounced, narrow 9dB 'hump' with a peak at around 4kHz.
Has anyone seen anything similar? I wondered if it could be a narrow band of mains harmonics let through by regulators or insufficient PSU smoothing. I am going to super-regulate my DOS soon and will report back on any improvements there. I did wonder though, if the PSU has a very narrow band of elevated noise, might it be possible to add a capacitor of a particular value to filter it out? My DOS supply has pre-reg caps of 22,000uF, 0.22uF and post-reg of 0.1uF, 220uF.
Last edited:
Measuring noise with a soundcard is tricky, especially if it's an internal one. There's a lot of internal noise generated by the PC, so the (background-) noisefloor is already quite high. You can do a null-measurement first, with no input signal. Is the 4kHz peak still there?
The player's output signal could be too strong for the card's input, this would explain the clipping. Have you measured it with a scope? It should be around 2Vrms with 0dB sinewave. Try an 1:1 attenuator made up from two identical resitors. If the clipping remains, then it's in your player. Maybe the amplification of the DOS is set too high? It has plenty of headroom though.
The player's output signal could be too strong for the card's input, this would explain the clipping. Have you measured it with a scope? It should be around 2Vrms with 0dB sinewave. Try an 1:1 attenuator made up from two identical resitors. If the clipping remains, then it's in your player. Maybe the amplification of the DOS is set too high? It has plenty of headroom though.
Hi Ray and higlander,
I dream of owning a scope... One day.
I'm certain that the player is not clipping, or else I'd hear it on my speakers. I'd even be able to see it on the 10Hz tone. It did clip at one point in the past, but not any more.
I did null measurement. The distribution of the noise floor frequencies without the player attached is dead even between 10-19khz. The 4kHz peak only appears in the background 'silence' of my CD63. It does not modulate the output signal, which also measures nice and flat when I analyse the audio frequency test sweeps on the above-mentioned test CD.
The output of the DOS is undoubtedly too high for the card at 0dB, but only by the tiniest fraction. To illustrate, with the waveform zoomed to fill my screen, I'm only missing a couple of pixels. A couple of resistors before the RCA would do the trick as you say. That was something you advised a long time ago, Ray, and I still haven't done it.
Some assorted guesses for the source (or lack of suppression) of the noise peak is one of the following:
- 7812/7912 regs on DOS (upgrading soon)
- 'Off-piste' LM317 regs on DAC analogue supply
- Residual noise of the DAC
- Filter characteristic of the DOS
If it's the regs, I guess it could be mains harmonics somehow creeping through.
If I max the volume and put my ear to the speaker, it sounds like a smooth, evenly distributed hiss. As background noise goes, it's definitely of the more pleasant variety. There is no 60Hz hum any more, which I'm ecstatic about. The noise floor is higher than I remember having with LM4462HA, so I'm going to suspect the 7X12 regs on the DOS first. I'll report back after I change them.
I dream of owning a scope... One day.
I'm certain that the player is not clipping, or else I'd hear it on my speakers. I'd even be able to see it on the 10Hz tone. It did clip at one point in the past, but not any more.
I did null measurement. The distribution of the noise floor frequencies without the player attached is dead even between 10-19khz. The 4kHz peak only appears in the background 'silence' of my CD63. It does not modulate the output signal, which also measures nice and flat when I analyse the audio frequency test sweeps on the above-mentioned test CD.
The output of the DOS is undoubtedly too high for the card at 0dB, but only by the tiniest fraction. To illustrate, with the waveform zoomed to fill my screen, I'm only missing a couple of pixels. A couple of resistors before the RCA would do the trick as you say. That was something you advised a long time ago, Ray, and I still haven't done it.
Some assorted guesses for the source (or lack of suppression) of the noise peak is one of the following:
- 7812/7912 regs on DOS (upgrading soon)
- 'Off-piste' LM317 regs on DAC analogue supply
- Residual noise of the DAC
- Filter characteristic of the DOS
If it's the regs, I guess it could be mains harmonics somehow creeping through.
If I max the volume and put my ear to the speaker, it sounds like a smooth, evenly distributed hiss. As background noise goes, it's definitely of the more pleasant variety. There is no 60Hz hum any more, which I'm ecstatic about. The noise floor is higher than I remember having with LM4462HA, so I'm going to suspect the 7X12 regs on the DOS first. I'll report back after I change them.
Hm, not sure where the 4kHz peak may come from. Not the DAC, because all the aliasing images are far higher in frequency and they will be blocked by the filter. Maybe a voltage regulator? Although the frequency seems a bit low for oscillation.
Or an artifact from the clipping? Clipping creates a flat wavefrom, which is full of harmonics. Remember a square-wave is made up of uneven harmonics of the fundamental frequency? This could be something in-between.
Noise floor is higher when compared to the original opamp-stage, that's normal. The passive filter is more relaxed and attenuation is a bit less. But it should not be audible, only if you crank-up the volume and put your ear to the tweeter.
No, just bog standard 7X12s, and yeah, I know - I expected a higher noise floor. I just wondered if anyone know what could be causing that narrow blip of frequencies 2X louder than the rest of the spectrum. I'm going to super-regulate the DOS and then will measure again. I might also invest in one of those cheapo scopes so I can do a FFT with higher sampling accuracy.
I did find this post from Richard Murdey here: FFT Noise Spectrum of the LM7812 three terminal fixed voltage regulator
His measurements do seem to show a pronounced peak in the output noise of those regulators near 4kHz.
The harmonics that you describe were very evident in my FFT of an 0dB tone as clipped by the input of my sound card, but it was the 'silence' that I was more interested in. At ~-90dB, that signal was not clipped.
Yes, that's more or less what I am doing. It's a bit louder than that, but my amp is very sensitive. None of this is audible at anything remotely approaching safe listening volume. However, the noise is currently quite evenly distributed throughout the audible spectrum, save for this 4kHz blip, assuming it exists and isn't just an artifact of my sampling.
Anyway, if it's just me then don't worry - I just saw it and thought it might be something that affects other people and could be filtered out if the source could be identified.
Cheers,
Ben
Last edited: