I remember some of this from back in the day.
16 bit technology was absolutely cutting edge for domestic use while 14 bit DAC's were much cheaper and easier to produce and Philips already had these up and running. The 4X oversampling Philips used pushed the resolution close to but not quite 16 bit standards and as a bonus allowed very simple filtering after the DAC while the 16 bit system needed the 'brickwall' filters. The Philips players were reputed to sound better though.
All the early Sony players used to 'time share' the DAC meaning it did both left and right channel decoding while Philips with the cheaper 14 bit DAC used one per channel. I remember the fuss of the time sharing and it could (did) lead to a delay in one channel and possible image shift.
In the end Sony won out though and 16 bit was standardised.
Hi Karl,
Yes, exactly. The interleaving of L and R channels in the DAC caused a 90° phase shift at the highest frequencies. I measured it many times back in the day.
The original Philips using oversampling and two DACS did sound better (which is why I bought a Nakamichi OMS-7 - TDA1540). Nakamichi was one of the cutting edge companies with digital reconstruction a was Revox (I have some Studer machines). Shortly later Nakamichi came out with non-Philips DACs, dual and oversampled. They also notched out the transient glitches before others did.
The first brick-wall filters were L-C tuned units. Later we got into active filters, but they all sounded "not good". Upping the sampling frequency allowed much lower slope (order) filters that were less critical for component value (price), fewer sections (price) and of course had nicer phase characteristics that sounded a lot better. So it was cheaper and performed better. I really do not understand what people see in NOS DACs, and I have seen they do not use the proper filter (smooth move guys!!!). You need those high order filters with NOS DACs.
You'd remember the PCM-56 DAC, the first popular one widely used. You could find that in single (one DAC for both channels) and machines that used dual DACs. Later we got into single bit type DACs, but the good multibit ones always sounded better.
Yes, exactly. The interleaving of L and R channels in the DAC caused a 90° phase shift at the highest frequencies. I measured it many times back in the day.
The original Philips using oversampling and two DACS did sound better (which is why I bought a Nakamichi OMS-7 - TDA1540). Nakamichi was one of the cutting edge companies with digital reconstruction a was Revox (I have some Studer machines). Shortly later Nakamichi came out with non-Philips DACs, dual and oversampled. They also notched out the transient glitches before others did.
The first brick-wall filters were L-C tuned units. Later we got into active filters, but they all sounded "not good". Upping the sampling frequency allowed much lower slope (order) filters that were less critical for component value (price), fewer sections (price) and of course had nicer phase characteristics that sounded a lot better. So it was cheaper and performed better. I really do not understand what people see in NOS DACs, and I have seen they do not use the proper filter (smooth move guys!!!). You need those high order filters with NOS DACs.
You'd remember the PCM-56 DAC, the first popular one widely used. You could find that in single (one DAC for both channels) and machines that used dual DACs. Later we got into single bit type DACs, but the good multibit ones always sounded better.
The Sony CDP101 actually had slightly different component values in the left and right analogue stages after the DAC which I think was an acknowledgement of this problem and image shift some could detect. Its not a misprint in the manual, they really are these values in the real player. The brickwall filter is in the blank rectangle.
A very well liked DAC as I remember
I still have a Bitstream (1 bit DAC) Micromega Stage 2 player that uses two TDA1305T DAC's and that player has always sounded great and it can still hold its own today. Very smooth and detailed... analogy sounding.
Hi Karl,
That component change can't possibly shift the phase enough, and it doesn't have that issue anyway. I would think it affects gain more than phase anyway. That is a dual DAC. The analogue switches appear to be deglitching and maybe there was an issue with each switch (IC512) or PCB payout they are compensating for. The dual DAC receives information, L then R, but it is synchronized and is output in phase. That eliminated the 19 KHz phase shift between channels (but you also had the filter affecting that region too).
Most of the worst offences in D/A conversion were in the analogue stages (or you could hide problems there). By far the best DAC chip I have heard is the PCM 1702 / 1704. My Denon DCD-S10 uses a pair in co-linear, and each is a co-linear DAC. That's four DAC chips in that machine. Low level linearity is no longer an issue when you do what Denon did, expanding on what Burr-Brown did. I don't know if you can beat that DAC today for sound. I did go over the analogue stages in that machine and it sounds even better.
Hi caisson rj,
Okay, do a documentary on the 104.
That component change can't possibly shift the phase enough, and it doesn't have that issue anyway. I would think it affects gain more than phase anyway. That is a dual DAC. The analogue switches appear to be deglitching and maybe there was an issue with each switch (IC512) or PCB payout they are compensating for. The dual DAC receives information, L then R, but it is synchronized and is output in phase. That eliminated the 19 KHz phase shift between channels (but you also had the filter affecting that region too).
Most of the worst offences in D/A conversion were in the analogue stages (or you could hide problems there). By far the best DAC chip I have heard is the PCM 1702 / 1704. My Denon DCD-S10 uses a pair in co-linear, and each is a co-linear DAC. That's four DAC chips in that machine. Low level linearity is no longer an issue when you do what Denon did, expanding on what Burr-Brown did. I don't know if you can beat that DAC today for sound. I did go over the analogue stages in that machine and it sounds even better.
Hi caisson rj,
Okay, do a documentary on the 104.
Hi Chris, the player is the first generation CDP-101 and the DAC was time shared on those. The DAC is the CX20017 and it is multiplexed between the two channels giving an 11 microsecond difference between channels on playback. This was one of the big criticisms of the player.
I think the multiplexing is why it is called 'dual' in the data sheets but it does alternate between converting the left and then right samples before outputting them. I can find very little technical info on that DAC now.
I think the multiplexing is why it is called 'dual' in the data sheets but it does alternate between converting the left and then right samples before outputting them. I can find very little technical info on that DAC now.
It was all very much first generation stuff. I've found an internal block diagram of the DAC but beyond that nothing more. I wish I had kept that first player now... it got traded in against a Denon DCD1500 (the original 1500) which was audibly better. The Sony won out on transport quality though. Typical first gen, built like a tank.
Lol!
I have a Denon DCD-3300 sitting here that's mine. In the original box even.
That diagram .. it should have the R and L outputs timed in sync, not shifted by one cycle. That's what the timing generator and latches do. I'll have to see if there is a phase shift on the next player I look at.
I think the very best transport I have ever seen was the NEC used in the Nakamichi OMS-5 / 7 players, also Alpage and a few other high end machines. It was too expensive to produce, a devil to set up. But it produces the very best eye patterns I have ever seen. Generally the Sony transports are better than the Philips ones. Many will argue. Then on top of that, the Sony is repairable.
I have a Denon DCD-3300 sitting here that's mine. In the original box even.
That diagram .. it should have the R and L outputs timed in sync, not shifted by one cycle. That's what the timing generator and latches do. I'll have to see if there is a phase shift on the next player I look at.
I think the very best transport I have ever seen was the NEC used in the Nakamichi OMS-5 / 7 players, also Alpage and a few other high end machines. It was too expensive to produce, a devil to set up. But it produces the very best eye patterns I have ever seen. Generally the Sony transports are better than the Philips ones. Many will argue. Then on top of that, the Sony is repairable.