|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
82.156.128.160
In Reply to: RE: Question for Gstew posted by Jolida on August 23, 2011 at 03:41:39
Hi Jolida,Like you, I’m also in the process of providing the Julia (digital part) with a better power quality. I’ve red that the digital part only needs 3.3 V.
As suggested by other inmates I de-soldered the stock regulator and ordered a Tentlab shunt regulator..
http://www.tentlabs.com/Components/cdupgrade/shunt/index.html
http://www.tentlabs.com/Components/cdupgrade/shunt/assets/ShuntAppNoteAN04V02.pdf
Van on filtering, demping, ect
Van on filtering, demping, ect
Unfortunately I didn’t study the specs of this regulator very good before i bought it. After hooking up the Tentlab shunt regulator too the 12V P4 output of my Pico PSU, the Tentlab shunt only came up with 2.2 Volts.
I first thought I had somehow made a mistake with the grounding arrangement or so. But no matter what I tried, not in any situation it produced more than 2.2 Volts. :-( . After reading the specs more carefully, I realized that I probably bought the wrong regulator. The specs show that the Tentlabs shuts regulator only can deliver a max of 100m Amps at 3,3 Volts.
As I could not find anywhere, how much current the ESI Juli@ digital actually needs, I decided too slide in my Thurlby Thandar: TTi PL330TP triple linear PSU and diretly feed the digital part with 3.3 volts.
This was a very time consuming operation as I have too take my cMP setup out of a very small wall mounted cabinet which is hanging at the wall just beneath the Flat Screen TV.
Van on filtering, demping, ect
On the display I could read that the Juli@ digital part consumes 113 mA at 3.3 Volts.But I am glad I ‘unmounted’ the cMP setup from the insides of the hanging cabinet, because I was struck by the sound quality improvement. The sound stage now is HUGHE ! For the first time I use this cMP setup the music is now completely free from the speakers (3D like). Also the level of micro details exploded. For the first time lots of audible information about the venue or stage where the music is recorded can be clearly heard (echo in the venue, breathing of the artist, coughing audience, plectrums touching guitar strings, screeching music stools, ect, etc.)
Of course, I also tried filtering these power supply lines with ferrites and caps. And yes, again also this time, the sound quality improved. Despite the fact that the power is coming from a linear PSU filtering still gives a worthwhile SQ improvement. Tomorrow I will also try too ad some extra filtering through adding ferrite based coils (40 mH) onto the power lines coming from the TTi.I think I will settle on a simple small 3.3 Volt linear bench PSU or so. There is just enough space left inside the cabinet too fit an extra small linear bench PSU. It will also cost around 50 euro’s. However a shunt regulator is much smaller and will not require any space outside your HTPC case.
About the feeding the P4.
I boot my cMP setup with a bus speed of 110 mHz, core speed of 660 mHz and cpu volted at 0,95 Volt. When my cMP setup boots I can see shorts spikes of around 1,3 amps. When starting up XP these spikes reduce too 0,35 amps. When at rest XP in my cMP setup consumes 0,19 amps on the P4. When playing music (through LAN) this rises too 0,22 – 0,23 amps at the P4. (GA G41-ES2L MoBo (Intel G41/Inch7), Core 2 duo E7300) The new I3 core mobo’s will probably use 20% less energy and also sound better.Because I can hide an extra linear PSU in the same cabinet as my cMP setup is in (so looks, outside design and aesthetics play no part, I think I will settle on an extra small linear bench PSU instead of buying another small regulator
This is my experience so far in powering the digital part of the ESI Juli@. I'm sure Gstew will/can provide you with much more detailed information as he is very skilled in building his own PSU’s
Mark
NB On some forums inmates say: if improving the power on the P24 or the P4 improves sound quality, than there must be something wrong in other parts of the setup.
In their view improving the quality of power supply too a MoBo is futile, because also on the MoBo itself are lots of switching power supplies installed and also the logic on the MoBo is heavily spoiling the power and ground.
This statement could be true. But I can’t see how in my situation the power supply situation too my MoBo can influence my DAC, pre-amp and active speakers. Both setups are completely separated
There is a glas toslink between cMP and DAC. And on the 230 AC Volt side things are also separated as much as possible in any way. Not only both groups are on a separated spur, both groups are also on a different 230 AC Fase. In a home environment one can do no more too isolate two 230 AC groups of equipment from one and another.
See this drawing for information how my setup is layedout ((inter-)connected, powered, wired, etc)
Van on filtering, demping, ect
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 08/23/11Follow Ups:
From the picture it seems you are trying to power the P4 via tentlabs regulator. But from the Spec sheet you posted, they do not make the 12v version. How did you attempt to power the P4 then? And also their maximum current rating is just about 100ma. Seems way insufficient i suppose... I am using the Dexa with 630ma, & i myself am debating if that rating would be sufficient. Also please provide a link from where you procured the Ferrites. I wish to install them as well..
My measurements on the current requirements of the various voltage rails on the Juli@...
On the digital section (I only have a measurement for the +5v which is also feeding the 3.3v regulator and all of the 3.3v-consuming chips... sorry!):
+5v PCI 147.8 Ma idle, 189.2 Ma playing
+3.3v ?? Ma ide;, ?? Ma playing
On the analog section (+-9v measured after the regulators):
+5v DAC 11.0 Ma idle, 84.6 Ma playing
-9v analog in/out 55.8 Ma idle, 69.8 Ma playing
+9v analog in/out 89.7 Ma idle, 51.6 Ma playing
Greg in Mississippi
Everything matters!
Hi Jolida,
I think don’t quite understand your question. This thread subject is about: how too power the digital part of the ESI Juli@. I was not able too power the Juli@ digital part with the 3.3 V Tentlabs shunt regulator. Probably because of it’s max current is limited too 100 mA. When I tried too power the digital part with a linear PSU al works fine. On the meters of the linear PSU I could read that the digital part needs 113 mAmp at the 3.3 V entrée point.
I’m puzzled what makes you think I want too power the P4 with the Tentlabs 3.3 V shunt regulator.
In an optimized cMP-setup the power at the P4 varies from 1,3 Amp (spikes) during boot up and is still 0,19 amps when XP is at rest. So it’s impossible too power the P4 with it. The capacity to deliver current is (much) too low and also the voltage is wrong (3.3 Volt DC). The P4 needs 12 V DC . So what makes you think I want to power the P4 with it ???
The ferrites I use you can buy anywhere. I buy them at www.conrad.nl.
But these ferrites are nothing special. Any online shop arround the globe for DIY electronics sells them.
Please keep in mind that in this situation only using ferrites (without caps), has very little effect. You need too combine the ferrites with caps to create a filter. See graph from a Murata white paper on using ferrites for noise suppression and decoupling.
Van on filtering, demping, ect
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
I'm sorry I must have been mistaken. From the picture u posted, I thought the tentlabs was connected to a p4 plug. I dint realise it was for the Julia.
As I have mentioned, I have made a linear psu for my p4 as well using the 12v Dexa regulator which is rated max at 630ma. Then how is it powering the p4 without any issues. It's been more than 10 days since I'm running my cMp using this newly made linear, & it's doing a fabulous job. If the current consumption of the P4 is more than an amp, then how am I able to get it to work??? I have had no issues with it, my only doubt being if it will not cause any dynamic restriction as opposed to using a higher rated regulator...
Edit :- And as for the ferrites with caps, are u pointing out at a Pi-filter???
Edits: 08/25/11
Hi Jolida,Ow I now see the mis-understanding. :-)
I make use of P4 extension cables for their connectors. I buy a whole lot of P4 extensioncables and than I cut them in half and solder their connectors too the filters I’m experimention with on the P4, P24 and on the power line too the 3.3 V input at the Julia.
See picture.
Van on filtering, de-coupling, demping, ect This way I can easily swap the filters in and out. It’s very easy this way too listen too the effects on sound quality with or without the filter in place in the power line.
I think the DExa can handle a very short spike of 1.3 mAmp. That’s why the short 1,3 spike causes no problem.
I red your doubts on the possible effect on dynamics, bass, ect. But these thoughts and effects are from the analogue audio domian. I think these effect and laws do not apply too the digital domain. Here it is about the effect of noise (high frequency ripple) on the power and on the Zero Volt (GND) and what effect this has on the proper function of: clocks, timers, logic, ect, ect. Which in the end all will result in some sort of jitter. Modern computers with proper drivers and proper software, don’t lose bits anymore. With the right drivers and right software audio can also be ‘bit perfect’. Just like your photo’s, documents and other files also are ‘bit perfect'. But in a modern PC the timing of the bits is mostly not optimized in the way it is important for high quality audio reproduction . Of course timing in a PC is very important for proper technical function. But unfortunately audio reproduction not always necessarily benefits from this in a PC. But sometimes it does. The new I core processors and chipsets do sound much better.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 08/25/11 08/25/11 08/25/11
I am using the i3 core 540 processor for my cMp. Linear psu for Julia & P4. Last evening I bought a few ferrite beads to start with & used just one where the P4 plugs into the motherboard. Same with a p4 extension cable so that I can test with & without it. Sounds slightly better with the ferrite though. You were saying something about adding caps along with the ferrite ???
When I made both the linears, I had kept the one powering the Julia inside the cMp & the one for the P4 outside. Two days ago, I wanted to manage both of them inside the cMp for aesthetic reasons. So I kept both inside the cMp side by side. To my surprise, the sound changed quite a bit for the worse. The liquidity in music deteriorated & the bass became less tight & uncontrolled. I'm still wondering what caused this to happen. Any ideas??
Hi Jolida,
I’m not an expert in this matter.
I would guess you have your home made linears not boxed.
So they are not magnetically shielded
So the (powerfull) magnetic fields of these transformer coils will induce currents (distortions) in there immediate surroundings.
Other than that I have no idea.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
The linears are kept side by side next to each other inside the cMp cabinet which houses the motherboard as well. The cabinet is grounded though. Does each linear psu need to be individually boxed & grounded ??
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Does each linear psu need to be individually boxed & grounded ??It's a bit of a myth that linear PSUs are necessarily a noise-free panacea for the ills of the switcher. Quite the opposite. Way back in 1989, an article in Hi-Fi News on power supply design for pre-amplifiers put it well and, as I happened to build part of the design (I've still got it), I kept a copy. The author has a weakness for purple prose but this extract might be useful:
. . . the standard PSU circuit is a devil. It pollutes the supply by drawing current in swift packets as the voltage peaks. A sideways look reveals a multi-resonant network, snapped on and off by switches (disguised as rectifier diodes) and damped by a load which happens to be an audio circuit in series with a pair of regulators. Each diode commutates (switches) current at least twice in every cycle, producing a pattern of back-EMF spikes with a PRF of (say) 200Hz. Sags and surges in the mains voltage only add to the poly-rhythm. If the spike’s own (oscillatory) frequency is high enough, there’s no guarantee it’ll be spotted with anything short of a VHF lab’s sampling oscilloscope. Still, it won’t have any trouble making itself manifest once it’s inter-modulated with another signal containing components harmonically related to 200Hz - such as music. If this is so, the most likely symptom is modulation noise, noise which follows legitimate audio signals.For a local but equally helpful discussion, see link. It doesn’t give you a recipe but it does explain things really rather well.The bi-phase capacitor ‘smoothed’ supply’s output ripple has high harmonic content. Thanks to brute-force filtering, AC filtration is typically zero at 100Hz, becomes slightly effective (eg -5 to - lOdB) in the audio midband, only to return to nil above the reservoir capacitor’s resonant frequency, between 1 and 10kHz.
Small wonder dedicated audiophiles have taken to the hills, powering their pre-amplifiers and CD players from batteries. . . .
The mundane approach to building audio supplies crumbles further once we acknowledge that all conventional voltage regulators (whether discrete or IC) rely on NFB. It makes them rugged, predictable workers at low frequencies but their ability to ‘filter out’ incoming garbage is progressively impaired with ascending frequency and decreasing periodicity. Conventional regulators (whether IC or custom) are especially hapless in the face of VHF voltage spikes. They need protection as much as the op-amps they serve. Summing up, it seems likely that, if universal AC/DC power conversion techniques influence sonic quality, then they do so by offering cross-rhythms of their own making, as well as introducing others from down the line. Up to now, most ‘good sounding’ PSU’s have relied on empirical turning. That small changes to values, to component makes or mere layout has led to dramatic sonic differences is neatly explained by the exaggerated sensitivity of multi-resonant networks. It’s small wonder high resolution sound systems often sound radically different from day to day.
In short, I wonder if you’re not asking more of your kit than it is able at this point to give you. That’s not a very helpful answer of course.
Why not start by putting the units back where they were and making sure that that restores your earlier ‘inner peace’? Then take it from there. I have little experience in this area but I'm not convinced that just playing about with the grounds is going to be a fruitful approach. On the bright side, there's plenty folk round here who should be able to help.
Edits: 08/26/11
Awesome. I shall try to implement one soon. If the psu for the P4 & the Julia need to be kept outside the cMp cabinet, is it ok to run a twisted wire with a length of about 12 to 16 inches from the psu to the P4 and/or Julia ?? Does this cause any harm in any way? Or is it mandatory to keep the psu close to the load using short lengths of wire. The problem with the latter is that the transformers (AC power) will radiate inside the cabinet. Which of the two would be a wise choice??
Jolida,
Provide or point us to more details on and post some pictures of your linear supplies & how they fit into your cMP case... That will help a lot in understanding what happened when you located your linears inside of the case.
But before I go into a few ideas, let me amplify something Dave suggested... If you want to learn a lot about practical power supply design, read up on what John Swenson has written. In addition to the thread Dave referenced, I suggest the following one on DIYHIFI.ORG: http://www.diyhifi.org/forums/viewtopic.php?f=15&t=1957 and also doing a search for posts by John on low-Z power supplies, mostly in the Tube DIY forum a few years back.
To reduce noise induced in your linear supplies, I can think of several things to try:
1. Use 'transformer-ringing snubbers' as John mentioned in the computer PS article Dave cited or the one I listed above from DIYHIFI.ORG.
2. If you aren't using good soft-recovery diodes (like MSR860's), change to them.
And of course, John Swenson's supply is a great solution. BTW, I believe it is his raw DC supply section (transformer -> snubber -> small cap -> choke -> larger cap) that does the supply noise reduction trick. I suspect that any good regulator will work well after this portion.
Then, I can think of a number of other causes for the sound differnces you hear with your linears inside of your cMP case:
1. How your wires are configured and dressed. For the least amount of radiated field, all wires should be twisted or braided with their opposite... Positive against related ground wire, the three wires coming out of a center-tapped transformer braided, etc. Also, route the wires so they are suspended in free-air as far away from other wires, surfaces, and components as possible. If they must come near other wires, try to cross them at 90-degrees. If they have to touch a surface, use a small block of wood to hold them away from the metal. And take your AC inputs to the transformers out of the case as quickly as possible and especially minimize what they touch and come near.
2. If you haven't sexed your transformers (or don't even know what that is), let me know & I'll post references. But if you haven't and if your transformer's cores are electrically connected to the case of the cMP, circulation currents from your transformer's cores may be polluting your grounds.
3. If your cMP case is steel, then it is a good idea to leave the linear supplies outside... And eventually move all of your cMP out of the case. Magnetic steel cases and hardware (non-magnetic is a little better) tends to add a thinness & harshness to the sound.
4. How you have your linear's parts, especially the transformers mounted, will impact the sound. See the pictures of a couple of my supplies... Note that I've removed the steel cap around the laminations and mount the transformers to a wooden (in this case, bamboo) base using brass hardware & nylon standoffs. The brass screws/bolts go all the way through the wood base & are fastened with brass acorn nuts which act as small 'tiptoes' to better control vibration from the transformers. When I have some noticable vibration on a transformer I use compliant footers (typically from Herbie's Audio Products) to isolate them and their bases from the case.
5. Vibration damping of the case, bases, transformers, and supply components like the diodes, capacitors, and heatsinks will make a difference. I use a combination of Dynamat Xtreme (very good for resonant surfaces, but tends to emphasize & muddy bass a bit) and EAR SD40AL (more neutral than Dynamat).
While I do agree that the radiated field from your transformers can cause a problem too, IMHO these factors will generally make a larger difference in the SQ.
All of this does not detract from Dave's post or the information imparted there. But these are all pretty simple things you can easily do and try.
As for how to run wires from your supply into your P4 or Juli@, of course shortest is the best. You are better to have the regulator as close to the power consumer as you can manage. And if you do need to run wires, use a good conductor and insulation and twist/braid them as suggested above. 12-16 inches is not too bad.
Greg in Mississippi
P.S. Also read John Swenson's comments in these threads to better understand some of the complexities of the mechanisms that impact the sound quality of our cMPs and also help explain why things that seem like they shouldn't make a difference in SQ do:
http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=90268
http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=87759
http://www.audioasylum.com/cgi/t.mpl?f=pcaudio&m=89544
Everything matters!
Great Work. I have made my linears following your instructions. The parts i used are the MUR 860 diodes ( only two diodes doing a full-wave rectification, as the transformer i used has a center-tap),ELNA Silmic II Caps,& a Dexa Regulator. It looks quite different from the pictures u posted above though. I will post some pictures once i go home this evening. I have no complaints with the linear supplies. They have improved the cMP a lot. The only issue i faced was when i kept them side by side inside the cMP cabinet ( Alumimium), which is grounded. Before i put them inside, i had just the linear for the Julia inside the cabinet & the one for the P4 outside. This was very fluid & clean. The moment i brought the one for the P4 inside, it messed up the whole spectrum to a very great degree. Last evening i wanted to do another trial so i moved both the linears about 3 inches apart (inside the cMP itself)& surprisingly it performed a lot better. Im not able to figure out what is going on technically. But by now its quite evident that both the fellows dont like to share the same place. I intend to revert them back to where they were originally, one in & the other out of the cabinet to see if it solves the problem...
Hi,
I have already made a 3.3v Linear psu to feed the digital part of my Julia by de-soldering the onboard regulator & going straight to the 3.3v input. Sounds a big step ahead. Though the caps need a lot of burn-in before they show their true potential. Thereafter i made a 12v linear psu to power the P4. Another big step ahead. My only concern was with the regulator i have used in them. I have used the 3.3v Dexa UWB series for the psu feeding the Julia & the Dexa 12v UWB series for the one feeding the P4. Both these have a maximum current rating of 630ma.
GStew uses the LT1083CP for his P4 which is rated at 7 amps, & LT1085 for the Julia which is rated as 5 amps. That is way ahead in current handling compared to the Dexa. Though the Dexa have a very good Ripple rejection & Noise levels, i am still dicey if their current rating would suffice for those purposes. As per Gstew, the P4 draws less than 0.5 amps, but he has used a 7 amp regulator. Julia draws around 170ma in Play mode but he has used a 5 amp regulator. This was the only part i need clarification about. Will the Dexa have any downside in terms of Dynamics or other aspects vs the LT regulators is what i want to know...
I intend to move on to the ferrite filtering after i snap out of this...
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: