DAC resolution / availability

[Gene Zumchak]

Jorgen,

    Actually if the values of 1% tend to the low side, that is good.  One needs to compensate for the on resistance of the CMOS switch (latch) that is driving the R/2R ladder in a DAC.  It seems, however, that 20K is almost an order of magnitude too small to minimize switch resistance error.

Gene Z.



jorgen.bergfors at idg.se wrote:

> > > With 1% metal film resistors, you can normally find 10 resistors out of 200, that are within 1 count on a 4 1/2 digit DVM. That is my experience.
> > If you by them as "amunition belts", you'll most probably find out, that
> in some local region of the belt you get 0.1% without any trouble.
>
> Well, I suppose this is different for different manufacturers. With the resistors that my supplier sells, they seem to vary randomly within the belt. But different belts seems to have different amount of variation. It seems that larger values (above 100 k) have more variation. But I suppose luck plays a part too.
> This weekend I selected five 20k resistors from a 100-belt. I found five with the exact same reading on the 4 1/2 digit DVM, but compared to other resistors that I have measured, this belt was rather bad. It seems that the actual value is almost always slightly lower than the nominal. Of the 20k resistors, only 2 out of 100 was more than 20.000 k. So I used the 20 kohm range on the DVM to get maximum resolution.
>
> /Jorgen