Popular Post Soul Analogue Posted November 20, 2020 Popular Post Share Posted November 20, 2020 After the Fanless Full LPS AMD EPYC server build (refer to the Building a DIY Music Server thread), I have some spare time to make some write up to share my experience in this aspect. I will not go deep on schematics, but I will share my views to rectify some misconceptions on LPS in layman terms. Please feel free to ask if you have any question in your mind on LPS. 1) What are the basic components of LPS? Normally a LPS is to convert the alternating current (AC) from the power mains from your electric socket, to direct current (DC) to power your electronic audio / HIFI device. To achieve this task, a LPS should comprise of: a. Power transformer - to convert the mains AC voltage (100/110/220/230v) to lower or higher AC voltages, that suit the use of the power supply circuit; it also acts as a barrier to the mains voltage b. Rectifier diode or bridge - to convert the AC into DC c. Filtering mechanism - to filter the (AC) ripple noise residual from the rectifier, usually comprises of capacitors, resistors and chokes... d. Voltage regulation mechanism - to keep the voltage at an acceptable voltage range of a device / circuit / chip, etc. Here I do not specify a regulator, because a LPS may not really need a regulator! =) e. Connection wirings - simple enough, to connect the the above parts as a circuit and to connect to the target device. 2) Do all LPS sound the same? Hell no! I would regard LPS building as an art! In fact, every components (as in Question 1 ) of a LPS can give different outcomes / favours to the sound. Different kinds of power transformers (EI, C, O, R, torridal cores), rectifier (schottky, hexfred, Silicon carbide, tube, etc....), capacitors (film, electrolytics, etc), voltage regulations (monolithic, discrete components and unregulated) ..... all sound differently! There could be hundred of thousands of combinations. So the sound character from a LPS is the cumulative effect of all components in the circuit. So the building of LPS is like a chef, who uses the ingredients (components) to cook a dish, with their preference seasonings. A good chef is one that possesses proper knowledge/experience on the ingredients and can skillfully combine/manipulate the ingredients into a delicacy. If you agree that different chef cooks differently, LPS from different builders should sound differently. 3) A "low noise" LPS must a good LPS? Yes and No. Generally a well-built LPS should be one that the ripple noise is low enough for the circuit/device to function properly and to sound beautifully. However to achieve the term "sound beautifully", "noise" is not the only parameter to evaluate a power supply for audio. Experienced audiophile should have encountered audio gears that sound very clean and clear, but at the same time... lifeless, sterile, flat and boring. A "noiseless" power supply can give you a clean sounding character, however there are other PSU parameters that contributes to the other sound signatures for enjoyment of our ears, such as bandwidth, speed, imepdance, transient response, frequency coherence and tone. to be continued... elan120, Nenon, RickyV and 6 others 1 7 1 Builder of Linear Power Supplies Link to comment
Popular Post Soul Analogue Posted November 23, 2020 Author Popular Post Share Posted November 23, 2020 4) What is regulated and unregulated Power Supply? which one sounds better? Regulated Power supply means there is an "active" voltage output limiter, ie. a regulator, in the circuit to limit the output voltage, regardless of the input voltage at the AC and the current drawn by the target device (the load), so that the device would not be damaged by over-voltage. The voltage would still be dropped if the input voltage is lower than the regulator's "dropout" overhead. The lowered voltage would cause instability but the device would usually be resumed working when the voltage is back to normal. Unregulated PS means there is no "active" regulation. The voltage is controlled by the resistance in the filtering mechanism, from chokes and resistors. The voltage at the "load" would vary if the AC input voltage or the current drawn by the load change. So unregulated PS does not work if the current drawn of the load keeps changing, such as most digital chips, CPU...etc. It is usually used in Class A amplification circuits, like tube B+, filament, where the current drawn is more or less constant. To use this kind of PS, careful calculation of the Ohms' Law (Voltage = current x resistance) should be done, accounting for the current drawn, resistance (DCR), diode drop, AC input voltage, topology (capacitor input vs choke input). In my opinion, unregulated PS is more difficult to implement. As for my opinion on sound quality, I would prefer unregulated PS to regulated one if I can choose freely. While regulated PS easily sounds more clean and quiet, unregulated PS sounds more natural, expressive, dynamic and harmonic rich under AB comparisons (on condition that both are properly built). The theory "Less is more" applies, as the power goes thru less transistors, which may possibly altering the power integrity, bandwidth, frequency response....etc. Purely passive components, resistors, chokes and capacitors, have less impact than active ones. 5) What is monolithic / discrete component regulator? Which one is better? Monolithic regulator is a single IC (integrated circuit) voltage limiter. 78xx/79xx, 317/337, 1083/1084/1085 are the classic examples, as well as the modern ones LT3042/3045, LT1117, LT1963, etc...they can be implemented simply by provision of input and output capacitors, resistors to set the output voltage. Discrete component regulator limits the voltage output by circuits that consist of separated transistors and mosfets. It is more tedious to build and more technical knowledge is involved in the design. However it is more flexible for adjustments to the designer intend. It is possible to bias the individual transistors, select better quality parts, to achieve the sound desired. From my experience, a good designed and built discrete component regulator can sound a lot better than the monolithic ones. However it highly depends on the quality of discrete circuit and the selected components, as it is open to more variables for sound tuning. It can sound worse than monolithic ones if it is not properly designed. 6) Is more capacitors inside the LPS the better? No! Capacitors inside power supplies are necessary for ripple smoothing and noise filtering. They act like reservoir to temporarily store the energy and drain out as needed. However time and additional energy are required to fill up the reservoir. The larger the reservoir, more time is needed to refill it. As a result, the power supply would be less agile to dynamic responses. Also capacitors have filtering purpose. As the name suggests, they would filter out something from the incoming energy, no matter good or bad. Just imagine you add a UV filter to block the UV lights on a window, if one UV filter already gets the job done, would you add two filters? If two filters were added, some of the details you can see from the window would inevitably be blocked. Hence to preserve every possible nuance and micro-details from the music, less capacitance is better than more capacitance. Sometimes for sound tuning purpose, higher capacitance in power supplies gives more bassy sound, while the highs would be less detailed. kyoya78, NanoSword, mtavares and 4 others 1 4 2 Builder of Linear Power Supplies Link to comment
Popular Post Soul Analogue Posted November 23, 2020 Author Popular Post Share Posted November 23, 2020 7) Bypass or not bypass capacitors? Capacitors of high capacitance, especially electrolytic capacitors, filter low frequency noise more effectively than high frequency noise. The lower the capacitance value, the higher is the frequency noise they are effective at. As such, in theory, to achieve a linear filtering action to the entire frequency spectrum, smaller capacitor can be added in parallel to the bigger capacitor. In my experience, there is not much issue if the bypass capacitors are of the same type and series from the same manufacturer. However, I found issues of the sound if electrolytic capacitors are bypassed by small value film capacitors, especially plastic film capacitors. The speed of their filtering actions are highly different. As a result it would sound like the highs come out quicker and more pronounced than the mids and lows, frequency coherence, uniformity and hence the musicality are destroyed. I would be pretty cautious on adding bypass cap. There is no rule of thumb on bypass caps and must be auditioned to confirm the effect. 8) Does LPS with higher current capability sounds better than lower ones? I cannot speak for the parts and topology selection of the LPS manufacturers in the market. However for the ideal LPS design which aims at the best possible sound quality, the choice of parts is dependent on the current/power drawn of the load. For higher current output, more capacitance is needed to filter the ripple noise to acceptable levels (in the sacrifice of some details, as described in Question 6 above), and higher power transistors in the discrete design. However higher power mosfet / transistors always have higher gate/base capacitance that makes them slower to transient response, than the low power counterparts. Also the transition frequencies, the response to high frequency, of high power transistor is lower. As a result the speed and bandwidth of the power supply would be compromised. So the best way is to have the LPS designed specific to the load requirement, selecting the suitable, capacitance, power transistors/mosfet (with safety margins of course). NanoSword, mtavares, Exocer and 7 others 1 4 5 Builder of Linear Power Supplies Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now