AW: Re: pwm ...

[Haible_Juergen#Tel2743]

> Well, the linearity and even monotony of a 8-bit D/A-converter is usually
> very poor.
[...]
> If uP clock is tens of megahertz and it has a chrystal clock oscillator,
> jitter is no problem. An PWM is linear, that why CD-players use 1-bit
> D/A-converters. Of cource, the bit stream should not depend on the timing
> of the uP-bus.

I am not sure if I want digital signals in the 10+ MHz range in my
audio systems. A 100kHz cycle with 8 Bit resolution would mean
40ns pulses for the LSB. That's a region where crosstalk and
reflections can ruin a lot of theoretically good performance.
In a Bitstream Converter, everything is concentrated to a very
small space, isn't it? But we're talking about a big, multiplexed
system here, so I still have my doubts ...

> Of cource 8-bit D/A-converters are cheap, but are the cheap
> compared to a hex CMOS-buffer and a few resistors and capacitors ? And
> the availability of a spesific DAC, old D/A and A/D chips are hard to
> find.

True. But why not use a few resistors and some CMOS buffers to build a
discrete DAC? It works great in my home-built polyphonic keyboard.

JH.