[sdiy] Suitable DACs for Demultiplexing

[Simon Brouwer]

Tom Wiltshire schreef:
> Hi All,
>
> I've just been messing about with the dsPIC 33FJ128GP802. I've used 
> the dual channel DAC on this chip for modulation sources before with 
> no problems. It occurred to me that since the DAC can run up to 
> 100KHz, it wouldn't be too difficult to demultiplex the output and get 
> 8 signals instead of two.
>
> I set this up using a DG409 dual 4-to-1 switch with one DAC channel to 
> each switch. The uP controls the DG409 address lines and Enable pin. 
> The Switch outputs go to the usual cap-and-op-amp-buffer that you see 
> on numerous synth schematics from the 80s and after.
>
> However, I ran into a problem. The DAC is a Sigma-Delta type and uses 
> an upsampling digital filter to produce a pulse train at 256 times the 
> sample frequency. What this seems to mean is that the output is 
> filtered. Instead of stepping abruptly from one sample to another, it 
> glides smoothly between them. Obviously for audio (the DAC's intended 
> purpose) this is ideal behaviour, but for demultiplexing it's a 
> nightmare. The output sample has only just reached the level you 
> wanted it to have when it is time for the next one to be sent. The 
> effect on the demultiplexed outputs is for an 'echo' of the previous 
> channel's signal to break through on each output channel.
>
> Can anyone give me any guidance about which types of DACs are good for 
> this application?
>
> I've got a Microchip MPC4922 dual channel resistor string DAC that I'm 
> currently experimenting with instead, but it is only 12-bit.
Note that although the resolution of this DAC is 12 bit, its worst case 
linearity error is 12 LSB, which makes it less accurate than some 8-bit 
DACs (the old ZN426E8 for example which specified 1/2 LSB accuracy).

With cheap DACs you get much less accuracy than the number of bits suggest.

-- 
Vriendelijke groet, 
Simon Brouwer. 

| http://nl.openoffice.org | http://www.opentaal.org |