4

I'm trying to repair a Nakamichi CD Player 3 that I got recently. Service manual from ElektroTanya

Relevant section of the schematic:

DAC schematic section

Initially, it would not open and just show an error. Replacing belts and bending the home position spring of the stacker to do the job of a broken off lever on the end switch fixed that.

Then it would also intermittently glitch (fail to read the disc and start spinning it in the wrong direction at full speed it can, or reboot and show garbage on the screen). That was fixed by reflowing the linear regulator for the digital 5V rail, because it ran too hot and over time desoldered itself apparently.

After that I tried to play some music and it sounds like it's reduced down to 1 bit. So I recorded a CD with some 1kHz sinewaves and silence, and this is what I got on the output

  • Sinewave 50%FS 1kHz: 1kHz 50%FS
  • Silence: supposedly-silence (pretty much white noise)

The signal is the same on the output of the DAC, so the output stage analog filter and OpAmp are pretty much ruled out.

When playing silence, I notice when probing the line with a scope, every now and then on the digital filter data input line there is 1-2 high bits, while I assume silence should be all low bits. (Shorting that line to ground indeed makes the DAC shut up)

Given this situation, I'm assuming the DAC is OK (but not completely ruling it out -- maybe the 2 bits sometimes are normal and are part of some sync or inbetween the clock pulses, so they are not being latched in).

So I turned my attention to the laser mechanism and the DSP that processes the EFM signal. I would assume the problem to be of a mechanical or optical nature, but the readout of CDs and even CD-Rs is very robust, even slowing the CD down with my finger doesn't upset the tracking (unlike some other decks I have). All the timecodes and track listings also show correctly for both pressed and recorded discs.

What could it possibly be, or am I wrong in ruling out the DAC this way? (The DAC is "behind" an opamp, and the rest of the digital circuitry is even further "deep" into the machine, so I don't understand why would an IC go wrong in there)

All the voltages and frequencies seem to be spot-on as per the service manual, minus some slight leeway for my entry-level measurement tools.

Update2: Alrighty, now that I've sorted out my scope issues (and gave it a firmware update), time to dump all the waveforms in the chain I've mentioned previously and try to source the replacement chip, as well as point out any obvious weirdness. How they say — It's the Final Rundown!

  1. RF output after the amp, measured on the RF testpoint (CN-105:3) while playing a pressed CD

RF out

  1. EFM output from the amp as seen from the U102, taken at R170, with a pressed CD

EFM out

  1. The PLL clock, CN-109:2 (~PLCK signal), must be 4.2MHz as per the manual

PLL

  1. Bit clock IN U801:17 (2.?68MHz in the manual) and Word clock IN U801:16 (44.1kHz in the manual): 24 bits per word, isn't it too much for 16 bit CD? or am I reading it wrong?

Two clocks from the DSP

  1. Bit clock IN and Data In (U801:18) while playing a silence track on a CD-R, to hopefully highlight the problem. We can clearly see 10 clock cycles are in a logical 1, whereas silence should be 0? Or is it some weird different format?

Bit clock and data of silence

  1. Bit clock IN and Data In while playing a music track on a pressed CD

Bit clock and data of music

  1. Oscillator out (U801:4) to the DSP, 6.9244MHz in the manual

DSP oscillator

  1. Data Left (U802:12) and Data Clock (U802:13) on the DAC while playing a silence track on a CD-R. This is data for the DAC so it clearly should be zeros and yet!

data left data left closeup

  1. Data Right (U802:16) and Data Clock (U802:13) on the DAC while playing a silence track on a CD-R

data right data right closeup

I was about to take some power voltages and the DAC outs as well but while I was at it the -5VD rail went, and the fuse seems intact so I'm puzzled. Either way without it the DAC is completely silent. Duh, will look into it later.

I have a feeling that otherwise these waveforms either makes it clear that either the U102 is toast or I'm totally lost.

akasaka
  • 141
  • 3
  • If the DAC just plays what it gets in correctly, then it gets garbage in. That garbage could come from anywhere, as there are many systems on the digital signal path, such as a digital filter. And there may be some fault as regulators rarely desolder themselves from heat, so unless you measured that there are correct voltages provided and currents are low enough for normal operation, the problem can be that something still pulls too much current and there is not enough voltage for the device to work properly. – Justme Mar 04 '23 at 14:07
  • Your sinewaves look like square waves. – Andy aka Mar 04 '23 at 14:43
  • Can you find some schematics? What DAC is it? What do the data lines look like (show waveforms)? – Tim Williams Mar 04 '23 at 15:19
  • If that's truly meant to be playing 1 kHz sine waves, something's wrong--those are square waves. Something might be saturating? What does the same CD look like played on a known-good player? – Hearth Mar 04 '23 at 16:47
  • @TimWilliams the link to the schematics is right in the first line of the question, and I've added some scope shots of the data lines to the DAC in the update. The signal is anything but correct, but the reason is still unclear. – akasaka Mar 06 '23 at 00:27
  • Ah right, indeed. The waveforms, is that with 1x probe? Beware the bandwidth is especially low in that mode. Also the scope bandwidth itself should be enough to resolve the signals (I don't recognize which scope this is; >20MHz should resolve these signals well enough). – Tim Williams Mar 06 '23 at 00:34
  • @TimWilliams it's a Gwinstek GDS-70122 claiming to be 25MHz (I assume that means 25MS/s in fact? the firmware is fairly jank to be honest, doing EFM tuning with it's persistence mode was a 2 person job). And yes it's a 1x probe... I didn't think that a 1x/10x probe actually means a 0.1x circuit in/direct feed actually :D – akasaka Mar 06 '23 at 00:41
  • Also could you label which components (U801, etc.), pins and/or nodes (when named) are probed? I added a section of the schematic, which I hope is relevant. – Tim Williams Mar 06 '23 at 00:41
  • 1
    @TimWilliams labels added to the graphs – akasaka Mar 06 '23 at 00:43
  • Are you sure? GW Instek doesn't make anything below 50MHz, and that number doesn't turn up any results. – Tim Williams Mar 06 '23 at 00:43
  • @TimWilliams my bad — it's a 71022, 25MHz 250Ms/s – akasaka Mar 06 '23 at 00:47
  • 2
    Stop using the probes in 1x mode. Only use 10x mode unless you know what you are doing and need 1x mode. The non-continuous clock is a surprise, but then again, it matches 18 bit data upsampled by 8x, it likely has MASH or some other upsampling. – Justme Mar 06 '23 at 00:48
  • @Justme but is the clock supposed to flop around like this in the 2V area? – akasaka Mar 06 '23 at 00:51
  • @akasaka 25 MHz is *entirely* different from 25 MS/s. A 25 MS/s scope can't possible have a bandwidth greater than 12.5 MHz by the Nyquist-Shannon sampling theorem. – Hearth Mar 06 '23 at 00:51
  • @Hearth yeah I know, but I've had too much "portable" scopes go through my hands that listed the sample rate instead of the bandwidth on the front plate for marketing reasons – akasaka Mar 06 '23 at 00:53
  • 1
    @akasaka I don't what you mean by flop around, but if you are using 1x probe setting on the probe, it will look like that kind of garbage. – Justme Mar 06 '23 at 00:53
  • @Hearth I have a TDS-460, with 350MHz bandwidth and 150M sample rate; but it does equivalent time sampling (at a maximum 50GSps equivalent), which isn't popular these days (for good reason). (The total bandwidth fact will still be true; ETS just spreads it into many harmonic sub-bands.) – Tim Williams Mar 06 '23 at 01:00
  • Indeed it looks like just discontinuous clock when probing with a 10x probe. I tried overlaying the in clock and in data (U801 pin 17 and 18), the clock is stable and clean but on a silent track there are 2 loose '1' bits for whatever reason and not even always locked in to the clock rate. I assume the DSP (U102) is the next suspect — the SUBQ logic being fine but something in the audio part went nuts? – akasaka Mar 06 '23 at 01:11
  • @TimWilliams Ah, yeah, you can use trickery like that--forgot about it, probably because no one uses it anymore now that multi-GS/s ADCs are available. – Hearth Mar 06 '23 at 02:29
  • Given the new info about how to use a scope I've added a bit more waveforms – akasaka Mar 06 '23 at 13:07

0 Answers0