There are some reliable Chinese DAC seller out there (take hifimediy as example), you can get a good DAC module, power supply including shipping perhaps less than the price of Asus STX. I don't know with India, but in my country where the custom service is not at all reliable, I never had any problem with shipping from China.
And yes, speaker upgrade will be more significant. There's no point in upgrading the source if the speaker cannot reproduce the improvement brought by the upgraded source.
How exactly is it noticeable though?
In a switching circuit one absolutely needs to have local decoupling and the charge carried by the capacitor will provide a low impedance reservoir of energy should it be required, but we're talking about MHz here where rise times are particularly quick.
How exactly does a lower supply impedance or 'flat' supply impedance affect the audio quality? At a very basic level the supply impedance will affect the voltage seen at the power supply pin of the opamp lets say. This will vary depending on the frequency that the opamp is asked to reproduce, but lets say there's a 0.5V difference (a gross exaggeration) between the supply potential for a 1kHz signal vs a 20kHz signal. This wont have any effect on the output wave-form unless we're working with a signal level that's extremely close to the rail limits. Naturally though we've got local decoupling in place so when the high frequency signal comes along it can draw off of the coupling cap for the instant that the PSU and trace internal resistance would act to lower the rail potential.
If the supply and trace impedance were going to have any effect on the signal passed by the device then it should be easily measured as distortion on the output waveform.
Interestingly though, if you look at the measurements performed on super regulators with regards to the reduction in output impedance vs local and remote sensing, the drop is 'considerable'. But we're talking about a drop, at 20kHz, from 0.02 ohms down to 0.0005 ohms. This might sound like a lot, but when you consider that 1cm of PCB trace with a standard thickness of 0.25mm on 1oz copper has a resistance of 0.01ohms...the power supply impedance is going to be completely dominated by the PCBs traces. Even standard 3 pin regs have output impedances of around 0.1 ohm at 20kHz, again the PCB traces are going to be the dominant factor in a multi-chip product. The only real place where the remote sense and therefore very low output impedance feature of a super regulator might prove beneficial is in single chip designs such as some phono preamps etc, where you can remove sense directly to the opamps supply pins.
Because any modulation is imposed on the output. Of course if you're only playing with sine waves, carry on as you are and have fun.
Did you hear about a thing called PSRR? That makes your supply "modulation" irrelevant.
Plus you didn't say how is a PS impedance helping with the inductrances of the traces/wires. Local decoupling is the only real help you can get, PS is "important" only in the 10-100 Hz domain.
Last edited:
Basically yes... for high-frequency digital signals, anything else is overkill. Your super-cool power supply won't do squat for a digital IC situated at the end of 5 inch of wires and traces.
Local decoupling is all you need to do properly:
www.analog.com/static/imported-files/tutorials/MT-101.pdf
Local decoupling is all you need to do properly:
www.analog.com/static/imported-files/tutorials/MT-101.pdf
Last edited:
If digital signal tranmission wasn't "bit perfect" then not only would your ananogue output be distorted and not a faithfull reproduction or the origional material, but computers, the internet and most entertainment devices would also not work, or work as reliably as they do. That is the whole point of laying out digital circuits, so that the signal gets there in a state that will allow the reciever to correctly determine the data stream.
That is why when I'm layoing out a digial board I have at hand about £36K of Cad software, as well as all the tools the engineers use. Its more cost effective to simulate all aspects of the design before commiting to a physical PCB and assembly. Having an unreliable product can ruin a company, look at Amstrad in the early days of PC's and their disk problems...Very few Amstrad PC's these days.
Most new PSU development is in maximising efficiency, shrinking the PSU PCB footprint and reducing heat loss. This is being prompted by the ever decreasing size, yet increasing power of mobile devices, and the requirement for longer battery life. Some of the new devices from Ti, Linear etc, in conjuction with some cool FET packaging
http://www.ti.com/lit/ds/symlink/csd16321q5c.pdf
You can get a lot of power density in a small footprint. Then there are the numerous local supplies, that todays digital boards demand, with some chips requireing upwards of 4 seperate supplies, the increasing practice is to cascade local 3 pin LDO devices for local supplies (there can be upwards of 22 supplies on a board) after a main SMPS. Dependant on the design there can be 12V 5V 3V3 and 1V8 (or 1V2) main supplies with chip cores and IO supplies being local to the device. Then AS pionted out, its all down to decoupling, and here very small chip package (0402, 0201) X7R MLCC caps are your best friend, with the odd COG for the local decoupling of a oscillator. Modding any decouling of a digital board is more likely to cause problems rather than improve things IMO. How many refer to Henry Ott, Electromagnetic Compatability Engineering before embarking on their modifications, or the numerous design notes and app notes available for nearly every chip
That is why when I'm layoing out a digial board I have at hand about £36K of Cad software, as well as all the tools the engineers use. Its more cost effective to simulate all aspects of the design before commiting to a physical PCB and assembly. Having an unreliable product can ruin a company, look at Amstrad in the early days of PC's and their disk problems...Very few Amstrad PC's these days.
Most new PSU development is in maximising efficiency, shrinking the PSU PCB footprint and reducing heat loss. This is being prompted by the ever decreasing size, yet increasing power of mobile devices, and the requirement for longer battery life. Some of the new devices from Ti, Linear etc, in conjuction with some cool FET packaging
http://www.ti.com/lit/ds/symlink/csd16321q5c.pdf
You can get a lot of power density in a small footprint. Then there are the numerous local supplies, that todays digital boards demand, with some chips requireing upwards of 4 seperate supplies, the increasing practice is to cascade local 3 pin LDO devices for local supplies (there can be upwards of 22 supplies on a board) after a main SMPS. Dependant on the design there can be 12V 5V 3V3 and 1V8 (or 1V2) main supplies with chip cores and IO supplies being local to the device. Then AS pionted out, its all down to decoupling, and here very small chip package (0402, 0201) X7R MLCC caps are your best friend, with the odd COG for the local decoupling of a oscillator. Modding any decouling of a digital board is more likely to cause problems rather than improve things IMO. How many refer to Henry Ott, Electromagnetic Compatability Engineering before embarking on their modifications, or the numerous design notes and app notes available for nearly every chip
am i missing something, or does that new mosfet linked above have the hugest gfs for a mosfet youve ever seen? is there something about the conditions or the way its specified i'm missing? because 150s is more than 150x the highest gm out of a field of mosfets we chose for the NTD1 reworked Pass D1 IV stage or is gfs a completely different measure to gM? is this just a biproduct of 'designing for minimized losses'?
Supply modulation? This is simply an extension of inductive distortion as described by Self and is visible with sine waves. People seem to like throwing the old 'we listen to music, not sine waves' as some kind of way of proving something wrong, but this has little to do with it. Music is, at the end of the day, simply lots of sine waves added together. One can simplify things for the measurement process and analyse the system using only one frequency at a time, or they can go to the extent of measuring with multi-tone stimuli, but the end results will end up being the same.
The point is that if the supplies and the surrounding decoupling and layout were insufficient then you would see this in the measurements.
Good luck modding the Xonar then.
When you've performed those measurements in 1Hz increments between 20Hz and 15KHz, and then with multiple sine waves at the same time drop me a line.
Qusp, it was just to show the one of the new packaging styles for mosfets, IR have some very interesting packages as well, it was nothing to do with the electrical specifications. Over the last decade component packaging has evolved, pushed by the requirements for more in less space, and the requirement to get the heat away from the actual chip. The new packaging is designed to reduce the junction to case thermal resistance allowing for more efficient removal of heat from the die. One of my favorite MosFet packages is this:
http://www.irf.com/technical-info/appnotes/an-1035.pdf
With the right design with a metal case you can arrange the layout so that the case becomes the heatsink, much easier for automated production than TO-220s.
With a digital to analogue system and the right equipement you can feed a controlled set of step frequencies and automaticly monitor the waveform, and also a controlled mix of frequencies, often done when CS101 tests have to be carried out.
http://www.irf.com/technical-info/appnotes/an-1035.pdf
With the right design with a metal case you can arrange the layout so that the case becomes the heatsink, much easier for automated production than TO-220s.
With a digital to analogue system and the right equipement you can feed a controlled set of step frequencies and automaticly monitor the waveform, and also a controlled mix of frequencies, often done when CS101 tests have to be carried out.
When you've performed those measurements in 1Hz increments between 20Hz and 15KHz, and then with multiple sine waves at the same time drop me a line.
Fourier says that we don't need to... But maybe you know more math than the rest of the world.
When you've performed those measurements in 1Hz increments between 20Hz and 15KHz, and then with multiple sine waves at the same time drop me a line.
I don't think you understand what's going on here. Electronic devices aren't like loudspeakers, they don't show peculiar resonances at unusual frequencies. One does not need to perform sine wave distortion measurements at every conceivable frequency to prove that a design or device has no intrinsic problems. Sure, you need to perform enough measurements to be able to reveal certain trends, like rising distortion with rising frequency, due to the diminishing effect of feedback, but beyond that it's largely pointless to increase the step sizes. The same goes for doing it with multiple sine waves.
Now electronic devices do show resonance issues with high speed circuits, but this occurs at very high frequencies, much higher then you're ever going to encounter within audio work. Of course you could purposefully design in resonances within the audio band using certain filters, but why would you want to do that? And on top of that you'd be aware of it as you put it there in the first place.
Nope, it's just that I think the dynamics of complex passages are not easily represented by the kind of measurements we tend to rely on. But hey, that's an argument that cannot be reasoned out as far as I am concerned.
If the complex passages are band-limited to fs/2, then fourier holds no matter how complex they are.
I'm sure this thread will spiral into tangential inanity regardless, but the important mod to be done on the ST/X cards is replacing the clock - Vanguards are available on ebay. It is a substantial increase in sound quality. This has been covered elsewhere on the web with a lot of positive feedback.
From what I have read elsewhere, power supply mods are not as beneficial.
From what I have read elsewhere, power supply mods are not as beneficial.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.