|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
76.122.133.81
In Reply to: RE: Effect of ferrite beads on dc power lines too P24, P4 and/or too Pico PSU’s posted by hfavandepas on May 27, 2011 at 15:03:22
I can't get below 145 stably. How do you do it? I know you mentioned that you thought that running ssd os and a NAS allowed you to get there. I was running 143 and I kept going wobbly on stability and I've got the same Kingston ssd as you. What do you run V core at? I run .875 volts but it reads .83 on cpu-z. Maybe you could just list all your settings cause I'm curious why I cant get below 145. Yes I have 120 volt ac and it usually runs 117 or so in the summer. I set my dirty psu at 11.95 volts and that powers my cpu and all my disks. My other clean psu (yes a switcher) is also set at 11.95 volts and runs only my p24 input.
Follow Ups:
Hi Theo,
I made a lot of changes too my setup and gear before I noticed I could go that low.
Here’s a report on what I changed over some period of time before I discovered it.
Too make a long story (below) short, I think it is because the Audyn caps (bipolar elko glatt) I put on the 3.3 and 5 volt headers of the ESI Juli@ digital part.
But hereafter a complete report on things that I all changed before I noticed I could go this low.
* caps on Juli@ digital part
I had nothing else lying around when I wanted too try some caps on these power pins. The only thing I had lying around than was:
- 1 polymer cap (the red one) 1uF
- 1 Oscon 47 uF
- 1 bipolar Audyn cap 47 uF (intended for use in speaker crossovers)
- 1 bipolar Audyn cap 100 uF (intended for use in speaker crossovers)
I bundled the 1 uF + 47 uF Oscon + 47 uF Audyn and put that on the 3.3 V
And the 100 uF Audyn on 5 Volt. See picture.
Van on filtering, demping, ect
Looks very messy because I taped the sides of the Audyn caps and the Audyn connection wires, as I was afraid for a short circuit somehow when dropping them in place. I noticed I nice little pop in sound quality. And that’s it and just left the caps there.
* new HTPC casing
Only a few months ago I switched over from the Zahlman case too a SilverStone SST-ML03B case. Launched in feb 2011. Perfect timing from Silverstone as we where about too switch over too an new AV cabinet. This new AV cabinet would no longer rest on the floor, but would be hanging from the walls. That new AV cabinet is mounted against the wall just beneath the flat screen TV, which is also mounted against the wall.
However these free hanging AV cabinets are not so spacious and also much less deep than the an AV cabinet that rests on the floor. So I had to look for a much less bulky HTPC case. I found the SilverStone SST-ML03B too be ideal for this. It has enough space inside and also in exactly all the right places where you need that space. It has standard audio gear size. And best of all it only 69,- euro !! It’s a steel. Don’t know why it took the PC casing industry so long to come up with this format.
This time I not only put bituminous felt pads on the outside bottom and inside lit of the HTPC case. But I also but a slap of bituminous felt pad inside the casing underneath the mobo. So this case is, also like I had the Zahlman case, heavily dampened with bituminous felt pads.
* shielding all 230 AC cables (complete re-wiring)
I also think it has nothing too do with a complete re-wiring of all (!) 230 AC wiring. When I moved all gear into this new wall mounted AV cabinet, all wires involved where very close to each other. It was a messy, bulky spaghetti of wires running criss cross everywhere inside and on the back of the cabinet.
Nice cable dressing was particularly very hard too achieve in the small 20 mm space between the back of the cabinet and the wall. All wires strictly have too reside in the narrow space behind the cabinet and the wall, because it is a really ugly and messy sight too see them hanging down the wall from behind the AV cabinet. So because of the limited possibilities for good cable dressing, I decided too make a bold move and too replace all regular 230 wires with shielded 230 AC cable for every apparatus that is in need of 230 AC.
(Flat screen TV, HD TV decoder, linear PSU’s, Klein & Hummel, Lavry DAC s, ATX PSU, ect)
It took al lot time too finish this operation. Luckily al most all components had such ‘strain relief through puts’ (see picture).
Van on filtering, demping, ect
So is was not all too difficult too remove the regular 230 AC wires from the inside, and too re-use the ‘strain relief through put connector’. As the shielded 230 AC cables has a bigger diameter, the shield has too end here at the outside of the casing. However the wires inside the shielded cable can pass trough, which allow for the original ‘strain relief through put’ too be re-used.
* strict separation of dirty and clean real earths
All shields on the new 230 AC shielded cables are grounded at the plug-side too a real earth. Also this shielding is strictly divided in two groups of apparatus: a ‘dirty real earth’ for dirty components and a ‘clean real earth’.
The whole operation took almost a month before I had complete finished the transition too this new wall mounted AV cabinet (process time off course was much shorter).
I also looked at the LAN connections of the cMP setup and the powering of the 3COM switch I use. In this article ‘The trouble with wallwarts’ http://www.elmac.co.uk/pdfs/Trouble_with_wallwarts.pdf , I red that even this can be important. So the wallwart (in the meter board down the hall) also got an extra ferrite on the DC output too the 3COM switch. However I didn’t notice a sound quality change.
It’s only after all these changes that I noticed that I could go so low all of a sudden. My guess is, that it’s the caps on the juli@.
However in short time I expect too know. With the ferrites for flat ribbon cables, I also ordered some new oscon caps. Next week (or so) I want to remove the Audyn speaker crossover caps and put these oscon caps on the Juli@ digital part. Then I will know if these Audyn speaker crossover caps caused it.
I don’t think my under volting settings will matter much. I pay no special attention too it.
I have the CPU undervolted at 1 volt V core. Just preventing the CPU too get hot. Sometimes I run it at 0,95 V. But this makes little difference in power consumption (current) or produced heat. I don’t hear any SQ differences.
All DC power from both linears too both pico’s is at 12 volt DC. Same for 12 V DC from Antec ATX.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Thanks Mark. I have external power to at least the Juli@ 3.3 volt supply. I also have more caps on the 5 volt supply but I use .875 volts V core. Maybe I'll try 1 volt and see how low I can go for core speed.
I've got the G31 mobo but I doubt its that. Right now I am trying .95 volts and 140 core speed. I'll run this for afew days then try 135.
Hi Theo,
* caps on the Juli@ digital part
Today I removed the bulky Audyn elco bipolar speaker crossover caps and replaced them with two Oscon 330 uF caps that arrived yesterday (along with the flat ferrites for ribbon cables).
I still can warm boot into 100 and cold boot into 110 – 115 (I don’t know why this isn’t always the same value). So the Audyn caps on the digital part play no role in it.
These Oscon caps again brought a clear noticeable improvement in SQ. Hearing what SQ improvement is achieved only trough changing caps, I think i’m going too closely study how I can power the digital part by battery like you do. I expect some significant SQ improvement.
* About the ferrites on the P24: use them with decoupling caps.
I feel I have a clear picture now on how ferrites on the P24 best can be used. It is just as one can read in several articles: ferrites are best used together with decoupling caps.
So I placed caps on the 3.3 , 5 and 12 volt lines (between the Antec PSU and the ferrites).
This results in a very significant SQ improvement!
Without the caps in place, ferrites on the P24 give a nice little subtle improvement.
But with the caps added it’s a bigger improvement than adding a SSD. So not small!
What is even more astonishing, I do these experiments with the Antec ATX. Because the Pico can’t handle big caps (4700 uF + 47uF + 0,47 uF) on the P24 because pico’s can’t deliver the big rush in currents too the 4700 cap when switched on.
* After 4 weeks of tweaking with ferrites the preliminary conclusion is:
With a lot of ferrites + decoupling caps on the Antec ATX the SQ is very good.
I don’t feel any need too swap the Pico back in place.
The Antex + (a lot of) ferrites + decoupling caps sounds very good too me.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
I added another by pass polystyrene cap across the 5 volt section of Juli@. Yes it added to the sq improvement provided by the ferrites. Thanks Mark!
Hi Theo,
Whow, same thoughts were coming up on the other side of ‘the pound’.
This afternoon I did exactly the same.
And…. with exactly the same results.
I still had some glare. This bothered and irritated me. I thought: may be some extra HF frequency filtering is needed. So I place two extra caps (Oscon 47 uF + 1uF polymer) on the 3.3 V of the Juli@ digital part. See Picture. And … with exactly the same results !
Van on filtering, demping, ect
I like this.
To me it shows these improvements are not random.
If we both independently achieve the same results.
I also did some further experiments with the flat ferrites for ribbon cables.
See picture.
Van on filtering, demping, ect
A ‘pile of charcoal’ works. But also these flat ferrites for ribbon cable work.
However crucial in both situations is: adding decoupling caps too the 3.3, 5 and 12 volt lines before the ferrites (this way a 2nd order HF filter is created (I think))
The effect of ferrites is 4 times bigger, when ferrites are combined with caps.
If one does that, than ferrites really bring a major SQ improvement
More to come.
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
I got glare or what I call excess highs so I took the cap out. All this stuff(i.e., tweaking etc) IMO is simply finding a combo that sounds real then stop and listen to music. Then after a few days when you get bored with your sound try something else. Isn't this exactly what a diy audiophile does? Its definitely what I do.
I know that sounds a bit anti science. Let me restate my tweaking philosophy. I believe in the theory of constraints. I learned some of this while working in manufacturing for a while but I think it applies to audio as well. As an example (applied to pc audio tweaking) when one adds a cap to the 5 volt supply on Juli@ like I did) and while it should sound better it doesn't, that means something else is constraining or limiting the output (in this case the output is excellent sound). Anyway when I reach a constraint like this I look elsewhere for what may be holding back a breakthrough to better output. Is my foundation for Juli@ ps as good as it can be in terms of acoustic isolation, is my Buf32s in need of tweaking, etc etc I think you get the idea. But in the end its all about getting good music. Oh by the way while accompanying my wife to Bed Bath and Beyond (a US store for stuff for bedroom, bath, kitchen) I picked up a half inch butcher block which I used to replace my quarter inch plywood base on which I place roller bearings for my Juli@ LiFePo4 battery based ps. And so it goes.
Hi Theo
It doesn’t sound ‘anti science’ too me at all. Sometimes also preconditions first must be met too make something work or too make the effect noticeable.
If you like searching for ways too better sound quality than I have some new not so controversial but also some controversial tweaks too try out if you like.
From reports in the past, I know you have a revealing system that lets you hear minute little changes in sound. Like playing a music file from an USB drive or reducing USB polling frequency. All these are very little improvements in sound quality but your system is able too let them hear too you.
I have 3 new to try.
* First tweak: caps in front and after the ferrites.
This one is not controversial at all. It can be red in almost al articles on using ferrites for suppressing conducted EMI. It is about the using caps together with ferrites
Do some googling on ferrites and capacitor and you will find articles about ferrite based filter networks from Texas Instruments, analog devices, altera, ect.
In there articles can be red that there should be a capacitor in front and a cap after the ferrites.
In front of the ferrites the cap functions as a bypass capacitor to provide a very low-impedance path too ground and thus allow HF noise to pass through to ground. The second function can be to help compensate for voltage droop when the mobo demands large currents peaks.
Also a smal(ler) capacitor is needed after the ferrites too suppress possible resonance.
I don’t have enough knowledge too exactly calculate the values of the the caps needed in front and after the ferrites. So I just tried some caps I had at hand.
I do know that the caps used must have good HF qualities. So I used a standard triplet of caps which I also use on the P4 as smoothing caps. The tripplet is a combination of a Panasonic FC 4700 uF + Oscon 47 uF+ 0,47 uF. I hope the large Panasonic will provide a nice current reservoir and I hope the 37 uF Oscon + 0.47 uF polymer combo will provide good HF shunting.
I placed one such triplets on a 3.3 volt line and an other triplets on a 5 volt line of the P24 (before the ferrites) Too my ears it greatly enhanced the effect of the ferrites (doubled / tripled it).
With these 2 capacitor triplets in place on the 3.3 and 5 volt line in front of the ferrites, I also placed such a triplet after the ferrites. This also brought a noticeable change in sound. But I don’t know if I like what I hear.
So today I ordered some new smal 47 uF Oscons to put behind the ferrites. In litterature I read the caps after the ferrites can be 100 to 200 times smaller than the caps before. See if some smaller Oscons after the ferrites will do the trick.
So if you like tweaking, try and listen how caps enhance the effect of ferrits in your setup.
* 2nd tweak.
This is a controversial one. While just playing with my android phone too move the cursor around through LAN with use of the RemoteDroid server program, I noticed that moving around the cursor added less (EXTRA !) latency than moving the cursor around trough use of the USB mouse. Note: the java server app also ads a bunch of latency. But moving the cursor only ads little extra latency on top of this. Somewhat baffled by this I searched for my old PS2 mouse.
I enabled the PS2 mouse settings in hardware restarted and noticed that moving the cursor around through use of the old fashion PS2 mouse added hardly any extra latency!
-> Now here comes the tweak.
From Reylands I learned that reducing the USB polling frequency removed some glare and ‘relaxed’ the overall sound quality. Reducing USB polling also brought slightly more micro details.
Since moving the cursor around trough use of the USB mouse causes highest (EXTRA) latency off all methods, I thought: why not completely remove the USB mouse? No USB mouse means: no USB polling frequency at all, since all USB ports can be completely disabled in bios and software. No extra separate power needed too power the USB bracket. No matter how clean this extra power too the USB bracket might be, the cleanest power is using no power at all. There is also no possibility that HF noise from the dirty Pico will somehow travel through the USB bracket onto the MoBo via the USB socket.
So I decided too disable all USB root ports in the hardware manager and enabled ‘PS2 mouse’ in the hardware manager. I unplugged the USB mouse connection from the MoBo socket and dirty Pico and disabled all USB in the BIOS.
From memory I can remember that in a standard cMP setup the PS2 mouse sounded a little less.
But now after 1,5 years of power supply improvements and with all USB activity disabled, using a good old fashion PS2 mouse sounds better too me. This can easily be heard.
I know, this almost sounds like blasphemy: telling a cMP user too disable all USB activity unplug it from the mobo socket and unplug it from the dirty PSU and instead reconnect the old fashion ‘dirty’ PS2 mouse and old fashion keyboard.
But never the less: would you like too try and let me know what the effect on SQ is in your cMP setup?
* 3thrd tweak
With the USB bracket removed from duty, the dirty Pico in my setup now only powers the little Kingstone 8 Gb SSD. Since the Pico is also a switcher it also throws HF noise around. May this HF noise can travel through the SSD via the sata–cable onto the MoBo.
So I decided too also place ferrites on the 12 and 5 volt lines that feed the SSD. Because the power lines of the molex connector too the SSD are short I only managed too place 2 ferrites on these lines. I ‘coiled’ the lines around a little paper pellet too make sure that the power wires are tightly pressed against the inside walls of the ferrite when closing the ferrite clamp.
May be ‘the placebo effect’ could fool me, but I think a notice a very slight SQ improvement. Same amount as reducing ‘USB polling frequency’ or playing from USB stick. So very small but noticable.
Let me know if you are going too try some of these tweaks.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Hey Mark. I dont have any room on my p24 line for caps or more ferrites so I cant try that. But I do use an old ps/2 mouse and I have put ferrites around the ssd ps. So I already did 2 out of 3.
Excellent follow up, reporting and feedback as always Mark. I'm still good at 140 core speed and .95 v core. I like the sound better than .875 v core and 145 speed...a lot. After a day or so I'll lower speed but feel no need to right now.
Hi Theo,
I forgot. I use a G41M-ES2L.
May be it's the mobo.
But I don't think that's the reason
Also with this mobo type I never could go lover than 135
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
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: