MINI USB DAC power supply modification

After I bought the MINI USB DAC , I felt that the sound quality had indeed improved. However, I had an opinion about the power supply when I saw this machine. It is difficult for a hifi product that uses feverish op amps and resistors and capacitors to tolerate such a simple power supply. Although I saw a ready-made power supply kit in the product information, it is not in line with the DIY spirit to buy a ready-made power supply for such a low-tech thing . Secondly, it seems that the power supply is no longer available, so I just went for it myself.

First, let's sort out the ideas. The original power supply part should be said to be sufficient in terms of transformer power, but there are some unsatisfactory aspects, such as: 1. Instead of using a symmetrical power supply, it uses 1875 simulation, which is inevitably unsatisfactory in terms of transients and noise; 2. The voltage is low, and the output voltage after 1875 simulation is barely ±6-7V, which is only one step away from the minimum operating voltage of opa2604 ±4.5V, which is definitely not "powerful" enough; 3. Digital-analog separation is a basic requirement for hifi circuit design , and this is also fully considered from the PCB. However, in terms of power supply design, it is estimated that it is due to cost and cost-effectiveness considerations, and the positive power supply is directly used to power the digital part. Of course, the designer We have also fully considered the problems that may arise from the imbalance of positive and negative power supplies. For example, the capacity of C15 and C16 on the positive power supply is 1000uf, while the negative power supply uses 220uf, and the relay used to switch analog and digital signals is powered by the negative power supply. However, if the analog and digital power supplies can be separated, it will undoubtedly be a complete solution.

According to the above requirements, our power supply should be like this. The transformer needs to have three output windings: dual 9-12v and single 9v. The power line needs to transmit these two sets of voltages, so at least 4 wires are required. The rest of the circuit is simple in design and will be shown when it is made .

After sorting out the ideas, I started to make it. First, the external power supply was used, because the original power supply had enough power. In addition, this power supply has an advantage that its shell is not directly sealed with glue, but fixed with screws, which is convenient for disassembly and assembly, so I planned to modify the original transformer and shell. The circuit diagram of the power supply is as follows:

Disassemble the transformer and the secondary, then find another section of enameled wire with the same wire diameter as the secondary, wind the two wires together, and wind back the entire length of the original secondary wire. In this way, you can get an actual measured AC voltage close to double 8V (because one wire is added to make the winding diameter thicker, so theoretically the number of turns of the original length of the wire segment will be reduced, but because it is already a dual power supply, it is enough to use), which is used to power the analog part. Note: in order to obtain positive and negative power supplies, one end of the two windings needs to be connected end to end as the middle terminal, which is different from the wire package connection method of full-wave rectification.

Next, wind the digital power winding. If materials and space allow, you can use the same wire diameter and the same number of turns (but I don’t have enough enameled wire on hand, and it’s not tight enough to wind it manually, and there’s not enough space, so I use thinner wire). It’s also a two-wire parallel winding method, with the same number of turns as the previous winding, so you get a single group of voltage around 7-9V, and use full-wave rectification to get a single group of voltage around 12V for use in digital circuits.

After the silicon steel sheets are installed (and finally painted and dried if conditions permit), the transformer transformation is completed.

Next, we install the rectifier filter circuit . As can be seen from the left part of the picture below, the original circuit board is relatively simple, and the positive and negative power supply parts are barely soldered on. Since the digital power supply only needs two rectifier tubes, they are directly soldered on. Overall, this part does not have much technical content, so I will not say more.

The power cord requires 4 wires, one more than the power cord provided by hifidiy.net . I used a shielded USB printer cable, which has 2 thick and 2 thin cores, plus a shielding net, as shown below. So I used 2 thick red and black wires as positive and negative power supplies respectively, and 2 thin wires combined as the digital power cord, and the shielding net is naturally the ground wire.

As for the plug, I should have bought a 10mm 4-core plug, but due to carelessness, I not only bought a 3-core, but also a 12mm diameter one. Alas, I will have to work hard to expand the hole when installing it. Since it is a 3-core, I just made a mistake and used the plug shell as the fourth contact point. Fortunately, the host shell is also grounded, but the grounding point is not sophisticated enough. I will change it next time I buy a 4-core plug. Solder the 3-core, pass the shielding net from the back of the plug, and tighten the shielding net with fixing screws.

In this way, the power supply part is completed. The finished power supply is shown in the figure below. The 0.1u capacitor is temporarily missing in the figure, and three 1k resistors are connected in parallel at the output voltage end to avoid the impact of excessive no-load voltage on the host caused by connecting the power first and then plugging the aviation plug.

Modify the MINI DAC host part! Fortunately, as mentioned above, the designer has separated the analog and digital parts as much as possible, so it doesn’t take too much trouble. [page]

From the figure below, we can see the original circuit The power supply is input through 1875 analog ground terminal, and two 1000u capacitors are used as positive power filter (the enlarged part in the figure), and then divided into digital and analog circuits, while the negative power is only sent to the analog part through two 220u filters. Of course, the negative power supply has another use, which is to power the audio source switching relay, which will be discussed later.

After analysis, it can be clearly seen that the purpose of analog-to-digital separation can be basically achieved by cutting off the power supply to the digital part (the red "X" in the figure).

PCB back side

The specific modification method varies from person to person. Generally speaking, it is to remove the redundant components first (refer to the figure below), then cut and jump the wires. Since there is no filter capacitor after the digital part is cut, I added another 470u capacitor near the power line. In addition, since the power supply of the digital part is cut, the positive and negative power supplies no longer need tilted filter capacity. I swapped the positive and negative capacitors. The actual picture is as follows. (Careful friends will find that the circuit board at the original power socket position is shorter. That is the sequelae of the "careless plug" mentioned above. After the 12mm socket is installed, it presses against the circuit board and cannot be installed no matter what, so I have to cut off a part)

With the above changes, it feels like the work is basically done, right? No, the real trouble is still to come.

As mentioned earlier, the audio source switching (D/A switching) relay is powered by a negative power supply. This negative power supply is not taken from the end of the circuit , but after the 10 ohm resistor in the solo headphone amplifier circuit . Similarly, there is a power indicator LED that is powered by the positive power supply through another 10 ohm resistor. To be honest, I am a little confused about the designer 's design, so I decided to change them to the digital power supply as well. At first glance, it should be completed by just reversely connecting the 5.1v voltage regulator tube and jumping the 100 ohm resistor connected to it to the 3rd pin of the 3.3v LM1117, right? See the figure below.

Don't be too happy, the relay doesn't work at all when the voltage is reversed (it seems that there is no built-in diode to filter the reverse electromotive force, I'm too lazy to check the information to find out), and because the grounding end is hidden in the multi-layer PCB , it is impossible to disconnect it. Later, I thought of a way: turn the relay "180 degrees" and connect it with a jumper (the left half of the picture below). Now the relay works, but there is a "sequelae", the contact becomes normally closed instead of normally open, so the D/A switch state is reversed. I can't think of any solution for the time being, so I have to make do with it.

The LED modification is not a big deal. Similarly, lift one end of the 5.1k resistor and jumper it from pin 3 of the LM1117, as shown in the picture above.

At this point, the power supply transformation of the MINI USB DAC has come to an end. As for the improvement of sound quality, since I have neither testing equipment nor a pair of "golden ears", I dare not comment on it. Deep down, I don’t quite agree with overly subjective listening evaluations. I can only wait for knowledgeable people to make up for it.

Finally, a reminder: If you use a three-pin plug like me, remember one thing: the ground wire is connected only after the shell is installed! So if you want to test it with power first, you must solder a ground wire to the plug shell, otherwise it will not work.