Re: Very low cost 20 bit 125 MSa/s 64kB AWG
|
Reginald was proposing to compensate for monoticity errors with calibration though. I guess the danger there is that there's a hole : there may be no combination of bits that provides the wanted value if, for example, a bit's weight is larger than all the lesser bits together. Perhaps they also don't track thermally to 1ppm.
I think some problems are? reduced by using R-2R rather than a full set of individual binary weights but I'm unsure.
If you want to avoid monotonicity errors, resistor tolerance (actually matching) has to be within 2**(-n), which would be 0.1% for a 10-bit converter, and 1 ppm for 20 bits. This is achieved in integrated DACs through the nearly perfect matching obtained with lithography. Non-monotonic DACs are really interesting to watch on a spectrum analyzer as you vary the amplitude of a sine wave, and even more interesting if you add a bit of dc offset to the digital waveform.
Still, it¡¯s a fun and cheap way to build a DAC of modest resolution and you will learn a lot along the way!
Gary NA6O
|
