Scott
+-32,767 in 16bit (if I am not terribly wrong) is +/- one bit less, or 99.99695% DFS or –0.00027db below DFS.
Is this that serious?
And you mean that this limitation happens in the digital domain?
I have read many loose remarks here and there that DACs distort close to DFS
George
Incidentally, EDIROL that Scott mentioned, house asahi-kasei chips inside.
Again and again ! When will we ever learn?
George
Which affordable DAC now hasn't, just as Boulder's filthy expensive music streamer does 1955A's.
(merely illustrating that it's often economics that forces a product to a halt, not the end of the technology line. => Roots supercharger/Turbo compressor)
CD's for verifying performance of players usually have full scale test tones on them, only recently have I seen players that don't play them without clipping, but I also see lots of spurious tones and THD that should not be there. Maybe I'll take some pictures, it's possible the rest of the circuit is at fault.
I know it's only a $100 product but the DAC in question is an Ambery that uses TI/BB parts. It was optically connected to a Toshiba DVD player playing the Denon Super Audio Check CD.
Last edited:
Insufficient information, he shows a boxcar windowed sine wave, the spectrum is continuous and not at discrete frequencies in this case the sidebar makes no sense. I hate to make assumptions but figure 5 seems show a continuous sine wave at one frequency resulting from phase re-arrangement of figure 4. Figure 5 is simply physically impossible from the information given. I fear there might be a total mis-understanding of Fourier theory which would not be a new thing.
Don't worry Joshua, I understand that you know what I am talking about. Many here, apparently do not. We have talked on the phone, as I remember, that is more than enough for me. Your continuing input on this thread is refreshing, reminding me of my 'committed' clients, rather than the critics of hi end audio.
My best guess at this time is that the effective input capacitance of the JC-3 is 125pf or slightly less. I could drop that in half, and I might, in future. The potential advantage may not be worth much. I will think about it.
Excellent idea- the user can always add capacitance or shunt resistance, but it's awfully difficult to reduce capacitance. And we're certainly agreed that loading is VERY critical for extracting top performance, regardless of whether you personally care about MM or not.
If feedback is not anathema, it is fairly easy to reduce capacitance over a reasonable bandwidth. In the case of JFETs for example, a stage with feedback to the source can remove most of the gate-source C. Cascoding can reduce the Miller drain-gate C, and driving that cascode part so the drain sees a replica of the gate voltage can reduce it further.
Negative input C can be generated by positive gain on one side of a capacitor connected to the input, of the right amount, again over a restricted but still useful bandwidth. This means that even excess cable capacitance can be reduced if you like. Of course this is not for the faint of heart, as regeneration is just around the corner.
I prefer the first approach for the most part, as this reduction effect translates into less capacitance variation with voltage, which is a distortion mechanism whether a linear variation or not. Of course most of the time it is also a nonlinear function of voltage, which just makes things worse.
For low Z sources like typical MC cartridges, these are usually minor concerns at midfrequencies. The behavior at well-out-of-band frequencies, the ultrasonic spuriae, and what these may do downstream, is another matter.
- Status
- Not open for further replies.