From part 2, I mentioned about the first stage of circuit, and here’s the D-latch stage. As mentioned earlier, I choose 74AC574, because of its low internal switch resistance. I measured all 8 channel’s switch resistance and it measured around 9~11 ohm each channel. 74HC574 measured around 22~27 ohm. I paralleled 2 channel together to get half of its resistance , around 4.5~5.5 ohm. The big drawback of this 74AC574 D-Latch are, switching noise will be higher, I need to separate the power supply from the first stage, and also carefully layout the PCB to reduce noise. Power supply need to be either battery or shunt regulator.
The supply voltage into 74AC574 also need to be slightly higher than the first stage. The first stage used 5V as power , and this stage I used 6.2V zener. there are 2 reason I increase the power supply voltage :
- The output voltage are 1/3 of the power supply. by using 6.2V , I can get around 2V of audio output voltage so that I can drive my vacuum tube amplifier directly.
- The IC’s internal clamping diode will effect the sound quality of this stage. When logic signal feed into 74AC574, nearly 5V ( when logic = 1) of this switched signal will be leaked into the power pin of 74AC574 via internal clamping diode, I removed the power supply of 74AC574 and replaced by a 470uF capacitor, the leaked voltage actually charged the capacitor and power up the 74AC574 stage by it self and my DAC can sing without any external power supply feed into 74AC574, but it measured only 4.7V max, and not regulated. Thats why I used higher voltage to switched off the internal clamping diodes to prevent the ” leak ” from first stage.
After the 74AC574 D-latch, the output pin from 74AC574 directly connected to R-2R ladder. I matched all resistors to be 5 ohm lower than normal, to compensate the internal resistance of paralleled 74AC574 D-Latch. I experiments with several combinations of resistors, and I found that 7.5K-15K resistors suit this design perfectly. lower resistors will increase switching current and induced self-noise into this DAC, and any higher resistors will be very bad in rejecting external noise interference too. I used a small capacitor at its output pin as simple filter. The output pin has 2V of DC potential so a coupling capacitor is a must, I didn’t include this coupling capacitor in the PCB because I can always play with different kind of capacitors in between the PCB and RCA jack. The R24 are actually not needed, and replaced by R50, R51 and also multi turn trim pot. R50 and R51 are 6.8K , 620 ohm, and the multi turn trim pot are 200 ohm, I used this to replace 7.5K ( R24 ) because this MSB bit will switch between +Ve and -Ve very fast , to represent +Ve half wave and -Ve half wave. This switching noise can be heard and by turning this resistor, I can null out the switching noise.
If I build this DAC again, I will choose better power regulator because this DAC are very sensitive to power supply quality. My first test play for a month sounds good using 1000uF capacitor between the power supply for 74AC574 but I add more caps up to 6800uF, the sound quality improved a lot. I also try to powered the 74AC574 stage using 4 pcs of NiMh battery, at around 4.8 volts. I have to lowered the first stage power supply to 4.5V because of the internal clamping diode problem mentioned earlier.
Both large capacitor and battery powered versions sounds really good, but battery powered version has lower audio output voltage at only 1.6V, still enough to drive most amplifiers out there.
The output impedance of this DAC are higher than normal DACs, because of 7.5K-15K resistor had been used in R-2R ladder. I have no problem driving vacuum tube amps directly, and I also built First Watt’s Bi 2SK170 buffer stage if I want to drive solid state amplifier.
The minimalist R2R NOS DAC are complete and I’m designing another version which is only 20bit, and with better low noise shunt power regulator design. In fact , I already built several versions and this is only one of all versions I built.