DAC resolution / availability

[Gene Zumchak]

Paul,

    You posed an interesting question.  If you made an 8-bit  DAC using weighted
input resistors, R, 2R, ... 128 R,  the value R which represents the MSB would have
to be accurate to 1 part in 256, or .39% just to give you a 1-bit (LSB) error.  But
with an R/2R network, you only have to buy two values of 1 percenters ( or even one
value if you parallel two 2Rs to give you R).  I have not worked out the math, but
I don't think you need anywhere near that kind of accuracy.  You have to account
for the switch resistance (CMOS latch) driving the R's so R=100K is a good value.
Also, most SIP resistors are 2% nominal value, but I believe extemely well matched
to each other.  Anyway, it is my gut feeling that making an 8-bit, 10-bit or even
12-bit DAC from an R/2R ladder might give very satisfactory results.

Gino



Paul Perry wrote:

> At 12:47 PM 17/03/99 +0000, jbv.silences wrote:
>
> >The cheapest solution is to design your own DAC with a simple R network and an
> >op-amp.
> >This way, you can have as many bits as you want in input.
> >Of course, it should take slightly more space on your PCB than a dedicated DAC
> >chip...
> >
> I dont think this is the cheapest way to make a DAC, if you need 12 bit
> accuracy,
> or even 8.... think about the precision resistors you would need.... unless
> you have
> laser trimming facilities at home...
>
> paul perry melbourne australia