|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
82.156.128.190
In Reply to: RE: Fantastic posted by theob on June 11, 2011 at 03:55:22
Hi Theo,Thankx you for reporting these results.
I hope other inmates will start too experiment too with ferrite beads too as I find the results in my setup truly amazing. And best of all: ferrite beads are relatively easy too apply. Especially applying them on the P4 power line is a no-brainer.
Clamping the P24 I find more complicated as I still haven’t made up my mind op on which ‘clamping strategy’ is most simple and/or most effective.
When applying ferrites on individual wire pairs on the P24 special attention is needed is too make sure the ferrite “sees” equal and opposite DC currents so that there will be zero net magnetic flux density inside the ferrite bead. This way avoiding the ferrite too be less effective or even without any effect (saturated) through DC effects.
I think the paper from:
Tim Williams, Elmac Services, ‘Using ferrites for interference suppression’
http://www.elmac.co.uk/pdfs/ferrite.pdf
and/or the paper from
manufacterer Steward on ‘The Use Of Ferrites In EMI Suppression’
http://www.steward.com/pdfs/emi/technical/Use%20of%20Ferrites%20in%20EMI.pdf
are both very clear on these pitfalls.In what way or in what arrangement did you use the ferrite beads on the P24?
I would very much like too hear from other inmates on how they solved ‘clamping the P24’ problem.Roughly I see three methods:
1. Brut force
Clamping 6 to 8 Ferrite beads (hole diameter 12 mm) in a row on the P24.
Downside: some individual wires deep inside the P24 bundle of wires, will never be in close contact with the ferrite. Although twisting the P24 bundle of wires helps some what to bring individual wire too the surface of the wire bundle.2. clamping individual wires pairs or groups of wires on the P24.
Downside: how to prevent mismatching the wires that will result in a non-zero sum of magnetic flux density which will make the ferrite less effective or may be even saturated.3. Using a flat bead ferrite for ribbon cables with use of a P24 extension or so.
See: http://www2.conrad.nl/goto.php?artikel=501962
All individual wires of a P24 (extension) cable can be flattened out side by side and than put through a a flat bead ferrite for ribbon cables.
This way all individual P24 wires will be in close contact with the ferrite. And also the brut force method could be applied by using 2 or more flat bead ferrites for ribbon cable in a row on too the P24.
Downside: May be: bulky. But clamping the P24 is a bulky mess anyhow. No matter what method is going too be used. A lot off ferrite is needed.
All ideas welcome.
Mark(suffering from a bad case of ‘Ferritus Clampingditus’)
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Edits: 06/11/11Follow Ups:
All I can say is that I'm looking for optimum clamping too. Using my random strategy has yielded good results so I'll stay here a while and then try moving them around.
Hi Theo,
I had planned too do some current measurements, So here are the results from my GB G41M ES2L mobo.
Yesterday I cut a P24 extension cable in half, tinned the naked wire ends and screwed all the 24 wires back together with help off little screw-on terminal blocks. This way I can insert my multi meter in each wire individually and measure current in that specific P24 wire. Pff the whole operation toke almost 4 hours. Too put the multi meter on a next wire, the computer has too be shut down. Once the multi meter is moved too the next wire, than the computer can be started up again. Ect. So it toke 24 shut down/start ups.
Especially since the first round of measurements where puzzling to me. So I decided too do a second set of current measurements.
Which again needed 24 shut downs and startups. :-(
But the current measurements stayed the same and somewhat puzzling.
The current is not evenly distributed amongst the wires of one voltage group.
Hmmm I find this somewhat strange.
Or could it be that I created ‘artificial facts‘ due too a faulty measurement setup or faulty measurement proceeding?
Anyhow, since I twice did fry a Gigabite G31 MoBo I now use a Gibagyte G41M-ESL2L
My last P24 measurements originate from the G31M mobo.
On that MoBo the 5 volt rails had the most current running.
But with this G41M MoBo the most current runs through the 3,3 V group!
So one just can not assume that every MoBo will have most current on the 5 volt group.
With this MoBo both the 3,3 V group and the 5 V group drawn much current.
So which voltage is ‘the power house’ in a cMP setup, depends on the MoBo used..
P24
pin
01 3,3V 0,23A
02 3,3V 0,19A
12 3,3V 0,18A
13 3,3V 0,17A
0,77A total
04 5 V 0,12A
06 5 V 0,12A
21 5 V 0,15A
22 5 V 0,11A
23 5 V 0,15A
0,65A total
10 12 V 0,02A
11 12 V 0,02A
The black GND wires have current almost equally distributed amongst them.
Information about PSU setup for this P24 current measurement session:
Linear PSU -> P4
Linear PSU -> pico PW-200-M -> SSD + USB bracket
Aerthwatts ATX -> extension cable -> P24
The measurements where done with a fully Optimized cMP setup according too cicsmemoryplayer website. So with LAN disabled in BIOS and Device manager. Additionally I also disable serial ATA channel 1 (in bios and device manager). And I also disable PCIe root ports in the device manager.
Although these current measurements give information on voltage and currents, thus making possible better decisions on which wires too group when clamping ferrites, I think all P24 wires need clamping because EMI noise is likely too travel around on all P24 wires.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
Well I am surprised that there is such a big difference between the 41 and 31 mobo's. I guess if you move many of your ferrites onto the 3.3v lines and get better sonics that would somewhat corroborate. I saw you initial measurement on the 31 mobo (which I have) but I did not see that current measurement distributed per each pin. Do you have that? I put one more ferrite on (current count now 7) and am not sure I can fit any more so unless I re-distribute I'm done.
Edits: 06/14/11
Hi Theo,
Back than for the G31 mobo I did it like this. With wires combined per voltage group.
Van P24 current measurements
Now for the G41M MoBo I did it like this:
Van P24 current measurements
So only for this G41M MoBo I have the currents per individual wire.
As I don’t dismiss Bibo01 remarks about ferrite that may loss their effective after a few weeks, how are you sonics after a few day’s now ? Do you think the ferrites in your setup are still effective?
I did a lot of surching and reading on the net, too find out if the effect of ferrites will wear out after some time. But the only thing I came across is, that it is important too prevent saturation as much as possible when (too much) ‘one way’ DC is applied on them. I also came across that ferrites are widely used on PCB boards in all kind of digital electronics too suppress HF noise. The web is literally littered with articles on this subsect. But I didn’t come across one article where there was a warning about wear out of ferrites in applications after a period of time. So lets cross our fingers.
Coming week I will start moving around the ferrite clamps. Since ferrites are very cheap is fun too mess around with them and I again ordered some 8 ferrite clamps with a hole diameter of 12 mm. And also bunch of round little ferrite bars with a diameter of 8 mm. I want to experiment with 3 or 4 winds around these little ferrite bars and than clamp this little coil into the ferrite clamp with 1,2 diameter.
Right now the individual P24 wires still have a lot space/air around them while being inside the ferrite clamp. There should be ferrite all around them. Not air all around them. So may be wrapping them around these little ferrite bars and than clamping the whole inside a ferrite might give better results. This way their will be hardly any air left inside the ferrite clamp and also the wire will be in very close contact with ferrite inside the ferrite clamp. But this I just me messing around with ferrite’s, I don’t dare to predict if it will be more effective. But I let you know in a few days.
Mark
fully optimized cMP2 PC -> ESI Juli@ -> Van den Hul Optocoupler MkII-> Lavry Black DA10 -> XLR Mogami Gold -> Klein & Hummel O300
I have not lost the sq improvement...its still there. In fact I notice my system appears to be less sensitive to daily variations in noise on the powerline (unless its been very good last couple of days which is unlikely in my ac feed). So I will continue to monitor also.
Thank you for your response.
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: