Any Diy PC Power Noise filters ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
ok, lets leave the clerics marce, perhaps they have discovered the 'dark bit' ? one that makes it through an EAC ripping undiscovered and the sneaky little blighters then make it through several buffers and reclocking stages to ruin the audio by removing actual music bits, but only the audio, doesnt seem to ruin other data, must hate music? only seems to have an effect if you believe though...

just like you can have caps that are 'too good' for the opamps, because the 'analogue current' (not to be mistaken for digital current) sounds better through the higher impedance decoupling cap, which must have leads (actually this is where they switch to technical language and call it linearity)
 
Last edited:
but getting data from a to b has been worked on for decades now,
Just don't expect a .wav file sent sent to the other side side of the world via a Server such as Filemail to sound quite as clean as the original, with low level ambience unaffected.
Sending a .wav file as an UNCOMPRESSED Zip as Cookie Marenco does, helps to prevent low level degradation due to servers and numerous optical repeaters.The .flac container also appears to offer protection.
There has been extensive testing done in this regard with numerous recipients worldwide, including John Kenny and the designers of the well respected Windows software player "jPlay"
O.K. I am no longer a participant in this thread.
Alex
 
Don't ask for things you are never going to get :)

Please let's return to the useful and contributive focus of the original thread, I found it very interesting.

its only useul to a point, marce is right, the only and best way to deal with the PC noise problem is to disconnect it, add isolation and forget about tweaking it because you will never solve all the problems, we all know the wav will survive intact, just isolate then reclock locally at the dac.

so while its an interesting enough intellectual exercise, IMO its a waste of breath/resources
 
haha that coreaudio place is ridiculous, if you fully load their music server it comes to well over 50K and its most expensive upgrades have absolutely no benefit to playing audio, in fact they raise the noise and ripple by introducing components that are faar too fast, when a micro fanless PC running an ARM will have all the power needed to play audio.

as to the price going up, I guess they made some sales and people gushed, so they figure they can rip people off for more.
 
I will when I get time describe the real world noise we do see in digital designs, ainly experiece gleamed from doing more up market designs where noise can be a big issue, especially if it is so bad it is broadcast from the device(EMC).
The filters mentioned earlier, having had a closer look are I would guess a standard largeih value C l or C L C for lower frequency filtaration, then the magic RF filter being a pi filter with two low value small caps (nF pF range) and the inductive part created using a SMD ferrite bead, probably somthing with an inductance around 60 - 200Ohm at 100MHz. There is nomagic to this design, it is pretty much standard with todays digital designs and multiple voltage supplies. The cap ferrite bead cap filters are used extensively for low noise isolated power islands (for a clock say) and are a cheaper option than an LDO, though high end and critical designs will often have a LDO.The cap ferrite cap filters are available as a three pin device, but more often in the ferrite cap ferrite (t filter) format, which is just as good as a pi for high frequency (RF) filtration.
Depending on the design and the engineer, will depend on the lavel of on board filtration and power islands, ranging from the basic required to get a device through certification (CE)for cheep commercial designs, to well thought out, simulated and tested designs with the right amount of filtration and decoupling, to the belt and braces aproach (a pi filter on every power pin!!!!) that quite often cause more problems than they solve due to multiple resonences.
My worry about the SATA filter is has it been designed to cater for some percieved audiophile problem or has it been designed to cater for the digital signal transmission(as I do believe there is often a difference of opinion).
more thoughts later...:)
P.S. I just hope there is no low level degrading of this reply as it whizzes through numerous servers...
 
I would not put thhem on the back end of a SMPS, wrong technology, size to large and too much parasitic inductance (never mind the capacitance), SMPS's are high frequency so require high frequency filtering techniques, and adding huge banks of capacitors dosn't do them any good either.
The filter shuld be designed to the fundimental switching frequency (usually catered by Cout, with further filtering for the harmonics (which considering the hih dI/dt and square wave switching can be pretty vicious.
 
Filtering the DC power that comes from the PC side of an ATX SMPS psu is completely and utterly pointless, greater men than we have tried (HAM's specifically) and the only thing that anyone can come up with is to design your own SMPS psu from the ground up and make sure its emission is sufficiently snubbed.

SMPS power supplies are the king polluters of the RF world and it would be exceptionally expensive to find one that has zero noise on either side, you would probably have to buy a multi-thousand dollar unit that is rated for no-EMI/RFI radiation military standards.

If I were to make the ultimate PC filtering board I'd get a M4 ATX psu, power that with either 13.5v off a regulated 300VA Linear PSU (SLA Float voltage) or a 300VA unregulated Linear psu /WITHOUT/ the SLA battery that I'd make myself.

It doesn't need to be regulated (still needs large filtering capacitors) if you are hooking it up straight to the M4 as it has a wide voltage input range, but it does if you are charging SLA batteries. (with feedback diodes + reverse polarity diodes)

Oh and to keep sudden dips and rises in mains voltage from affecting your computer I'd put a large enough filtering capacitor in-line with the unregulated PSU.

My PC is a Core i3 2120-T inside of a Norco RPC-430 and a Gigabyte GTX650 OC 2GB on an Asus P8Z68 Deluxe Gen 3

Power consumption (not including monitors) is pretty damn low...

Get underneath that 150 watt limit of the 300VA Linear PSU and you should be on your way.

Oh one last thing, take care with DC, insulate every connection as if its MAINS and fuse both your loads and your batteries and your supply, on both positive and negative wires, and cover bussbars with something thats insulated after you've installed them....

High amperage DC can and will start fires, explode or melt metallic items that you accidentally drop onto the terminals (exploding them in your face), etc...

And make sure you inderstand wire gauges and how many amps they can carry.
 
Last edited:
I lwil hewn I get itme beesdcri hte lera wordl onise we od ese ni idgital sgedsni, ainly eepxrciee agedmle rfmo doing more up mtkear sedsing hwere noies cna eb a big iesus, aleslypice fi it si so adb it si aobrsatdc from the evdeic(CME).
The srtflie mnendeito arerile, ahving dah a lcores okol are I uowld eugss a tasdandr lahrgei value C l ro C L C orf lowre frequecny ailoftntiar, tehn the gmaci FR ifletr iegnb a pi irflte with tow low value lsmal caps (Fn Fp raneg) dan the vdienicut aprt drceeta isung a MDS efritre dbea, lyrpobba osmthign ihwt na icnnaedtuc raodun 60 - 200Omh at 100Hmz. Theer is gioanmc ot isth desing, ti is ryptet hcmu atsadndr with odtyas ildgtia sesdign adn lleumtip ltoagve supilpes. Eht pca ferirte ebad apc etrifls era sued tlvyeiexens rfo low senoi isolated wpore isldnas (orf a oclkc say) and era a echpaer otpnoi htan an DOL, hhugot ihgh den nda alicrtic sgedsin liwl enfot hvae a OLD.The cap trfiree pca ltferis are lavbaliea as a three pin vedcie, but omre tfoen ni eht erftire acp ertfrie (t iflret) tmoraf, hwihc is ujst sa godo as a ip orf high erfucqney (FR) tiloinfrat.
Ingedpnde on eth dsegni adn het ereiengn, lliw edpned on eht valle fo on oabdr flitratoin nad pewor silands, gianrng ofrm the ciabs redqeriu ot etg a iceedv ruhgtoh nieicctairfto (EC)for hcepe lcocmaimre edsgins, ot ewll ghuhtto out, letdmuisa nad etsetd dsegins hwit the ithrg aomnut fo riitotfnal and dglcineopu, to het tlbe nda braecs apoarch (a ip tlfier on yeerv roepw pin!!!!) htat qutie foetn eascu ermo rpoelbms athn htey solve eud to melulpit snreoneces.
Ym worry boaut the TSAA flietr is ash it been gsiendde to catre for esom preiceved heuaildopi mprolbe or ahs ti eben endsgide ot certa ofr hte dlgiiat sginal iarotimsnnss(as I od bleeiev theer si neotf a fncdierefe fo opinion).
omre ohtsugth aerlt...
P.S. I just eohp reeth is no wlo ellev dedgranig fo siht erpyl sa it wshziez rhtugoh suuremon sersevr...
:confused:what?

yes (its OR sit) a problem with todays internet. oh yeah (its OR sit) just a very basic filter, (its OR sit) a complete iroffp and (its OR sit) simple design is bfosucetda for those (how OR who) dont know better and rafmed as magic to the leiuglbl. (there OR three) is (no OR on) doubt the idea has merit in some lpaonptacisi, but those ppaintlcoisa should already have effective figltnier and ictaksng up flseitr designed for a iecnger circuit will not yield good results, it may even cause harm. it should be designed and etdun with a specific placement in mind.
oh ****, wheres my credit card :eek:

jokes

here is the uncompressed version

yes its a problem with todays internet. oh yeah its just a VERY basic filter, its a complete ripoff and its simple design is obfuscated for those who dont know better and framed as magic to the gullible. there is no doubt the idea has merit in some applications, but those applications should hopefully already have filtering and stacking up filters designed for a generic circuit will not yield good results, it may even cause harm. It should be designed and tuned with a specific placement in mind.

I just dont see it as completely solvable and it doesnt need to be solved, its just another money pit. any time you added new hardware you would have a new harmonic that needs filtering. there are far better solutions for far less money, for just the cost of that silly filter you could buy the i2s fifo reclocker and isolator board, job done
 
Last edited:
Thanks for all that info.
I have two small questions.

I run mpdpup now flawlessly on an ASUS 1001PX Laptop where obviously I cannot change anything inside. It then feeds waveIO board; I ordered a Sjostrom Regulator (Best ranked on Linear Audio 4) to feed it but up to now it is fed from the USB, and then goes to AD1865 NOS SRPP from Raindrop through Isolated SPDIF from the waveIO.

1- When the ASUS runs from battery, it's perfect. When I plug the charger, there is a high level of distorsion. I guess the noise is coming from the charger ? How to fix it efficiently : rebuild a new Linear PSU, buy a well engineered SMPS which will stay outside, or insert a Felix filter as jtwrace suggests

2- Should I set the WaveIO inside the DAC box (it has its own PSU) or better to have a separate case, grounding to avoid aerial noise propagation ? It the difference huge or second order ?

Jean-Louis
 
I have done multiphase spread spectrum SMPS's for ******* that were very quiet, they had to be, any noise as stated could be deadly, cost was probably a bit more than your average PC supply. But done plenty othersthat were cheaper, for communication devices.
Disagreeto an extent on SMPS, yes they are a major contributer to noise, but when you looki into it, the super fast rise times used by even the simplest devices these days and lack of understanding and simulation of decoupling and power delivery system also means that switching noise is also a big big contributer to the noise spectrum.
LOL Qusp, hats what it looks like before the spell checking, but an excellent example of data that is travelling around half the world:D
JLOP, charging using the supplied plug in cheep chargers designed down to a price, my preference is like medicine dont trat the symtoms cure the problem, better supply with less noise.
 
nothing wrong with SMPS if designed with low noise and high performance in mind, for poweramps it makes high current regulated power viable and the increases in performance are quite significant. the stuff from Audiopower is a good example, the amps using these supplies measure like this... the last pic is dynamic behavior, full power pulse ie 600W @ +/-65vdc the red and blue traces are the rails. at this point they are actually better IMO than linear unless you went nuts and built a 600W class A amp to supply regulated linear power to your amp

lol yeah I know what you mean re spellcheck, I ran your reply through a randomizer, just the words so the right letters are in each 'word' just shuffled.
 

Attachments

  • Distortion Product Ratio 3W 600mA copy.jpg
    Distortion Product Ratio 3W 600mA copy.jpg
    113.4 KB · Views: 366
  • FFT Spectrum Monitor - 3W 8R 400mA copy.jpg
    FFT Spectrum Monitor - 3W 8R 400mA copy.jpg
    148.6 KB · Views: 355
  • Full Power Pulse.jpg
    Full Power Pulse.jpg
    140.6 KB · Views: 353
Last edited:
A few notes on spread spectrum SMPS's, done some funky systems usings these also involing multiphase. Instead of say one 50A SMPS use 5x10A multipahse SMPS with spread spectrum. Instead of noise at one set of frequenies at a high level, you get a broad spectrum of low level noise. The other advantage is power management, SMPS supplies work best in a sweet spot, near there full power output, when you have low current requirements they go into burst mode, noise increases and also drops in frequency content often to levels that will be audible. With multiphase and power management you can switch of banks or seperate supplies to match the capacity to the required output. Again filtering careful layout and good design, SMPS's can be amazingly quiet.
On linear supplies, for main power they are not as good as some may think, I have seen problems with functional and EMC testing where a linear supply let through audible noise during conducted susceptability test, where a SMPS would have filtered it out. Now I know you are going to say filtering should have stopped it, but these wrer high level tests and the thing had some crazy filtering. The problem was actually the linear supply in conjuction with trying a single point 0V to shielding topography! a more intelligent shielding and 0V scheme solved most of the problems, adding an SMPS put all the noise way above both the audio range and the bandwith of the audio circuitry, plus ground and power planes for low level signals and a proper decoupling scheme and power distribution and seperation limited the rest of the noise.

http://www.onsemi.com/pub_link/Collateral/AND8428-D.PDF

http://www.onsemi.com/pub_link/Collateral/AND8045-D.PDF

Qusp, I notice your into photography, now the last kids at home are 13, I've recently started playing about with photography myself, its great fun playing with light. Envy your job as well, PCB design these days is more physics and constraints than the art it use to be, when all jobs were done by red and blue tape.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.