WM8740 output stages in Cambridge Audio 640c mkI vs mkII

[zgtc]

Hey,

I want to convert my Cambridge Audio 640c mkII into a CD Transport, and hopefully build an external DAC reusing the two WM8740 salvaged from the player, and compare it to the other DAC I am building based on TDA1387.

Basic difference between mkI and mkII players is that the newer version includes two DACs instead of just one. That surprised me because the WM8740 is able of differential output (no need of logic glue to invert signals). A bigger surprise is that mkII uses 2 DAC chips and still the output is Single Ended, not balanced.

I'm trying to understand the why's of the output stages, and maybe try some mods. Service manuals are freely available on hifiengine dot com, I just prepared a super simplified summary of the output schematics:

mkI output, left and right channels:


mkII output, left channel:


Any help explaining a bit the output stages or ideas for mods would be very welcome.
Thank you

 

[zgtc]

Sorry I think this should go in the Analog Line Level subforum

 

[zgtc]

any help would be appreciated 🙄

 

[sgrossklass]

The funny thing is... I looked at the service docs and noticed various areas with mod potential... just none relating to the output stage. Well, I did notice that impedances would generally seeem to favor low distortion over low noise and could quite arguably be reduced by a factor of 2 (at which point one may also consider some LM4562s or something). It seems they are using what's basically two "superbal" input stages (Self, Fig. 12) to combine N and P output for each DAC channel, and then another to combine both channels out of phase. Seems reasonable enough. We may not be getting anywhere near the DAC's native noise level, but then how many CDs are properly dithered to begin with, let alone with enough noise shaping to require >100 dB, if the material even permits it...

They seem to be drowning everything in 78/79 series regulators... possibly not even M series parts. I would check what's actually installed and how their noise performance stacks up according to the datasheet. I mean, most of it shouldn't matter that much, output stage CMRR should take care of DAC analog supply noise, and a bunch of 5532s at very moderate gains shouldn't exactly be pressed for PSRR either. I would, however, take a very close look at the DAC CLOCK supply... possibly the DAC digital supplies as well.

I was slightly disappointed to see that the transport has its own master clock independent from DAC sample clock... I never noticed that the WM8740 can apparently take up the slack in such a scenario, replicating or leaving out single samples once accumulated difference gets too big. Interesting... probably doesn't make much of a difference in sound assuming we're talking ppm level deltas, though I wouldn't call it ideal.
I would attempt to injection lock the transport clock... maybe tap off at MCLK_LOCK (CN2) via <TBD> pF in series and run some coax over to the XTALI connector CN17 (I don't think this is being used)... hmm, maybe the coupling cap should be at the receiving end instead, any cable capacitance would be in parallel to C17 (15p).
That is one thing you struggle to do on a transport + DAC setup, unless you get a master / word clock connection back. Otherwise you get all this mess with clock regeneration using PLLs (with VCXOs for best results).

 

[zgtc]

The funny thing is... I looked at the service docs and noticed various areas with mod potential... just none relating to the output stage. (...) It seems they are using what's basically two "superbal" input stages (Self, Fig. 12) to combine N and P output for each DAC channel, and then another to combine both channels out of phase. Seems reasonable enough.

Good! I didn't know that term nor design. Reading about it, thank you

We may not be getting anywhere near the DAC's native noise level, but then how many CDs are properly dithered to begin with, let alone with enough noise shaping to require >100 dB, if the material even permits it...

Well, the point is I want to use the DAC board in an external case, where I'll also have a RPi streaming FLAC from a NAS, so yes, the noise will be important.

I would, however, take a very close look at the DAC CLOCK supply... possibly the DAC digital supplies as well.

I was slightly disappointed to see that the transport has its own master clock independent from DAC sample clock... I never noticed that the WM8740 can apparently take up the slack in such a scenario, replicating or leaving out single samples once accumulated difference gets too big. Interesting... probably doesn't make much of a difference in sound assuming we're talking ppm level deltas, though I wouldn't call it ideal.
I would attempt to injection lock the transport clock... maybe tap off at MCLK_LOCK (CN2) via <TBD> pF in series and run some coax over to the XTALI connector CN17 (I don't think this is being used)... hmm, maybe the coupling cap should be at the receiving end instead, any cable capacitance would be in parallel to C17 (15p).
That is one thing you struggle to do on a transport + DAC setup, unless you get a master / word clock connection back. Otherwise you get all this mess with clock regeneration using PLLs (with VCXOs for best results).

I'm willing to recap the basic suspects, I don't think I will dare to mess with the clock just yet, but all these tips are noted.

Thank you again

 

[zgtc]

They seem to be drowning everything in 78/79 series regulators... possibly not even M series parts.

I'm willing to recap the basic suspects, I don't think I will dare to mess with the clock just yet, but all these tips are noted.

But I do want to update the power supply, so those regulators will go and will probably use Salas BiB v1.3 to feed a cleaner DC.

Thank you again