Great project 
My wish list:
Id like a 0.01% version please.
24bit depth is more then enough for not limiting the SNR
can you design a variable max voltage/currant through the R2R to achieve the best SNR zone and/or control the Vout/volume in a novel and loss free manner.
Can the opamp output be changed to allow flexible options by using an opamp socket Then users can plug in their chip of choice eg some will want to use the AD844 feedback free route while others will want to "roll" opamps.
My wish list:
Id like a 0.01% version please.
24bit depth is more then enough for not limiting the SNR
can you design a variable max voltage/currant through the R2R to achieve the best SNR zone and/or control the Vout/volume in a novel and loss free manner.
Can the opamp output be changed to allow flexible options by using an opamp socket Then users can plug in their chip of choice eg some will want to use the AD844 feedback free route while others will want to "roll" opamps.
Great project
My wish list:
Id like a 0.01% version please.
24bit depth is more then enough for not limiting the SNR
can you design a variable max voltage/currant through the R2R to achieve the best SNR zone and/or control the Vout/volume in a novel and loss free manner.
Can the opamp output be changed to allow flexible options by using an opamp socket Then users can plug in their chip of choice eg some will want to use the AD844 feedback free route while others will want to "roll" opamps.
I decided on 28 bit, to have headroom for a perfect digital volume control. At -72 db volume you still have 16 bit resolution with perfect linearity thanks to the sign magnitude architecture.
Can't change voltage/current though the R-2R network, is already optimized.
The LME49724 is only available in SMT, and I also don't like sockets.... Normally I like discrete amplifier designs, but the opamps choosen are the best. And the "raw" R-2R resistor string outputs are available for anybody who have different opinions on buffers....
LME49724 looks nice!
0.02% resistor alternative with Silver design and zero EMI? - Take this DAC to the heavens -
"Firstly, nickel limits the working temperature to 160°C and tin has a peak temperature of 230°C. Therefore, by using an inner terminal made from silver/palladium and an outer terminal of platinum/silver, Riedon is able to offer devices that operate up to 300°C. Nickel is also magnetic, so resistors with nickel/tin terminals cannot be used in applications such as magnetic resonance imaging (MRI) where there are strong magnetic fields."
Link --- digikey.com/en/articles/techzone/2013/sep/power-rating-is-just-one-resistor-parameter-to-consider
0.02% resistor alternative with Silver design and zero EMI? - Take this DAC to the heavens -
"Firstly, nickel limits the working temperature to 160°C and tin has a peak temperature of 230°C. Therefore, by using an inner terminal made from silver/palladium and an outer terminal of platinum/silver, Riedon is able to offer devices that operate up to 300°C. Nickel is also magnetic, so resistors with nickel/tin terminals cannot be used in applications such as magnetic resonance imaging (MRI) where there are strong magnetic fields."
Link --- digikey.com/en/articles/techzone/2013/sep/power-rating-is-just-one-resistor-parameter-to-consider
Last edited:
I mean compensation of capacitive resistance caused by stray capacitance. When I.e. switching at 3Mhz and 1pf stray capacitance this will render to 50kOhm capacitive resistance. This must be compensated somehow, otherwise the voltages at the netpoints aren't right.
No problem here, thanks to the very clean and compact SMT design and 4 layer PCB, any stray capacitance are too low to matter much. I even choose to use only 7 bit from each shift register, as the 8th bit are located on the other side of the chip...
And the 470 pF capacitor at the output of the R-2R resistor chain take care of anything coming though.
The results can be seen at the FFT plots, so clean....
How low is your stray capacitance between your traces? In my old I/V DAC design some x pF have mattered.
Also take care after time, where some dirt comes over/between the soldering SMD's..
on your second given FFT plot, there are some spurious power harmonics...
Hp
Kastor, can you please explain to me the relevance, for this confuses me? In my opinion the DAC won't be operated in ambient temperatures higher than 50 Celsius... So I guess the temperatures of any component will not exceed 75 degrees.. So what's the need of components with much higher operating temperatures?
Haven't calculated trace stray capacitance, just did my best to keep things separated at layout and have the analog signals on the top side, with digital signals to the shift registers on the bottom side.
Again, the advantage of the sign magnitude R-2R dac is that MSB bits are NOT switching when not used, so you don't have large signals straying into low levels analog signals. That's also why you don't need de-glitching on a sign magnitude DAC, there simply aren't any glitches....
Those "spurious power harmonics" are more than -140 db down and mostly likely from the EMU-0404, as my initial testing are powered from my lab power supply....
I also really need better test equipment, the choice seems to be between real audio analyzers starting at $10K, or inadequate low cost AD/DA boxes like the EMU-0404. Something in between are needed, that might be my next project
No relevance for temperature, only for EMI or Silver if you believe in Silver.
A transformer has EMI, R-core transformers have the least EMI. Silver has the least electrical resistance, the electron transmission is akin to a direct pathway.
If Silver and EMI are overkill, yes, very likely, but if the price is not an issue then it's an alternative.
I think soekris can use any resistors he desires, this is his project and we are his disciples, any resistor will still be perfect.
Last edited: