I H8 inductors by Tiido

I H8 inductors

Tiido

14 August 2015 at 08:25:09 MDT

This is the voltage regulator for my digital formats music player project. There's a bit of tidying up to do and I also need to find a different shaped heatsink but for the most part it is complete.

It takes ~14...40V input, in my case 20V from a laptop brick and turns it into 12V, 5V and 3.3V, theoretically up to 5A on each rail.

3.3V and 5V I had no problems with, 3...4A load and no issues whatsoever, the little inductors got a bit hot at that point but rails remained stable... but on 12V bam voltage loss. And eventually I made my own inductor, giant core that is hard to saturate and thick multistranded litz wire with finetuned inductance to get stable output even at higher loads. I'm pretty proud lol.

The chips used are LM2679, pretty neat stuff !

Couple hours later I return and the thing still works, the heatsink is much warmer but overall great results !

Submission Information

Views:
318
Comments:
6
Favorites:
0
Rating:
General
Category:
Visual / Photography

Comments

  • Link

    SMPS noise should be kept out of audio signal lines if possible because it can cause noise and distortion even at a few tens of mV due to intermodulation and rectification from the delicate feedback loops in audio circuits. Galvanically isolated link from server to DAC such as optical is a common way of sidestepping this.

    • Link

      I'm well aware of these effects. The sound card part is going to be well shielded and also pretty much custom made and also will be powered by set of linear regulators to take care of power rail noises. Unfortunately the sound chipset I use won't have digital output present unless I build a bus sniffer + FIFO to capture the DMA cycles and output them myself and the other problem is that there's several other sound sources too such as two synthesizers which would require me to have a digital mixer going on and that's too much effort for no meaningful gain.

      • Link

        Cool. Can I ask what is involved in a custom made soundcard and what is the benefit?

        • Link

          The main thing is that it is not fixed sample rate like pretty much all cards are now days. it plays all sample rates natively, without conversion. There's also two synthesizers for games and certain music formats, you also get full DOS games compatibility which is a sweet bonus. ADC and DAC are really lovely too. Yamaha YMF719 chip.
          And the other main factor is that "I made it" hahaha. That's where all the fun is.

          • Link

            Even at native samplerate you are still relying on the quality of the DAC's oversampling. Oversampling is like most resamplers in that it is lossy, and attempts to guess the correct sample or filter out quirks. FFT-based resmplers on the other hand are more exact, but only available in software. Maybe the best solution would be to find the best samplerate for your DAC and then set a dedicated software resampler to upsample to that rate when the music has a lower samplerate. For instance your DAC may have more distortion at high samplerates. So you might get the best fidelity if you resample to a rate lower than the max, or even the lowest samplerate. A 192KHz file could actually sound better when resampled to 96k or 48k because the DAC performance would improve. That is my theory anyways.

            • Link

              The sound cards of nowdays are pretty much all locked to multiple of 48KHz, derived from a 24576000Hz clock. Stuff like 44.1Khz are not supported at all and the conversion is done by the driver in software. Hardware side only works at one specific rate all times, and the input controller typically has options to repeat the sample sent to get 192, 96 and 48 options, driver controls that but it can also ignore that and convert all rates to the native one itself. Some dedicated cards do have actual hardware based resamplers, typically dirty stuff, but the DAC and ADC still run at fixed rate.
              I haven't profiled my card too thoroughly, but it has way better output at all sample rates I tried than any of the other cards I had around, any arbitrary waveforms come out clean and undistorted when viewed from a scope, while all the other cards had visible ringing at the signal edges (worst were creative cards, and their drivers also do noise reduction in software for input, you can see lots of artifacts on spectrograms, absolutely dreadful). And it was most probably caused by the oversampler in the DAC. Lot of the DAC datasheets I have looked at have fair bit of ripple at the end of pass band, and huge mess at the stop band.
              Good chunk of the Sigma Delta DACs also get pretty nonlinear at high amplitude input, reducing volume in software by 0.1...0.5db will fix that (that is if the music touches 0db at all, and most stuff these days tends to...).
              My music production cards all use R2R DACs, so no kind of oversampling or other digital filtering (unless you use an effect) goes on at the hardware level, and they also allow the clock to be adjusted manually to target any specific input formats cleanly and can be driven from external clock also. But they are kinda too big and expensive to hack up and use in the music machine, the multitude of inputs and hardware effects processing they got will be all wasted also, also they lack support for DOS stuff.