{ "numMessagesInTopic": 13, "nextInTime": 1518, "senderId": "cLZyRwBszcuOAYqHv6RbBOimIV0dJaDwGCRGmDxiN75duBmDHQug7OLdLy64dkVFZ5J7LCTztjEFlAv0pazOSiBzAF5PcYSmGDFF", "systemMessage": false, "subject": "Re: cable noise", "from": ""yeangchng" <yeang_chng@...>", "authorName": "yeangchng", "msgSnippet": "Thanks for the primer on shielding & electromagnetism :) I m actually not so much concerned about the hand-wired wire braces on the home-made box as the", "msgId": 1517, "profile": "yeangchng", "topicId": 1503, "spamInfo": { "reason": "12", "isSpam": false }, "replyTo": "LIST", "userId": 39184597, "messageBody": "
> Test: determine maximum single-link bandwidth achievable withoutthe
\n
> Kramer box, with each cable set. Insert the Kramer box, and see if(which
\n> there is any reduction. When you approach the bandwidth limit
\n
> is not a brick wall limit, but a gradual curve, which will changewith
\n
> amplifier/receiver temperature -- colder is better) you will seeclocking
\n> shimmering or a kind of sparkling on the screen due to bit
\n
> instability. This is evident when I run my T210 at QXGA 60Hzsingle
\n
> link.I ran just 1x single link through the Kramer box + associated cable
\n
\n
>distortion of
\n> Both ferrite beads and shielding mitigate external noise
\n
> the original signal. A ferrite beat is a small but powerfulinductor
\n
> that prevents high frequencies from passing. Such high frequencynoise
\n
> in DVI-D would appear as periodic data corruption. It is sort oflike
\n
> when you are watching analog over-the-air television, and somebodyDifferential
\n> uses a cheap electric motor like a hairdryer, blender, or vacuum.
\n>
\n> A ferrite bead prevents very high frequency noise "spikes" from
\n> capacitively propogating through the system, and wreaking havoc in
\n> other circuits. Since DVI is a TMDS (Transition-Minimized
\n
> Signaling), it is a differential signal which rejects noise commonto
\n
> both the + and - lines, and it is fairly immune to cable noise. Aanalog
\n> ferrite bead would have the greatest effect on the single sided
\n
> RGB lines of a DVI-I cable. In short, ferrite beads will not giveyou
\n
> increased bandwidth, particularly since they actually serve tolimit
\n
> bandwidth.Shielding
\n>
\n> Similarly, shielding works to keep out electromagnetic noise.
\n> Shielding of the entire cable does not pertain to bandwidth.
\n
> of individual signals works to establish a constant impedance(1)
\n> wavefront. Impedance [Z] is established by, among other things:
\n
> the gap between the conductor and the shield effecting inductance[L]
\n
> and capacitance [C]; (2) the dielectric insulator between theof
\n> conductor and shield effecting capacitance; (3) the resistance [R]
\n
> the conductor and the shield effecting the LRC filtercharacteristics.
\n
> As discussed above, the extent of shielding from outside noisedoes
\n
> not determine DVI-D bandwidth. The impedance of the cable is veryCat7
\n> important, and 50 ohm cable is no big deal to make. So, it is
\n> primarily consumer hype to tout the shielding of a DVI-D cable.
\n
> ethernet cable is shielded twisted pair, but at 100ohm impedance.DVI-
\n
> D is 50 ohm.lower
\n>
\n> The most likely culprit, of bandwidth limited to less than the
\n
> specification of either transmitter or receiver, is impropercircuit
\n
> design/implementation (i.e. did not follow the factory applicationyou
\n> note for the driver/receiver) or poor PCB layout of the circuit
\n> causing impedance errors. For example in the IBM converter box,
\n
> see the signal pairs always together, so their impedance will beas
\n
> similar as possible. There is probably a ground plane in thecenter
\n
> layers, and the dielectric of the insulation between outer andinner
\n
> layers may be controlled.bandwidth,
\n>
\n> Bear in mind, that longer cables will tend to reduce the
\n
> and the impedance of DVI connectors are not nearly as controlledas
\n
> cable. A connector basically has a short distance where a signalpair
\n
> is unshielded and untwisted. The more connectors, the more likelythe
\n> there are impedance/bandwidth problems. HOWEVER, the bandwidth of
\n
> connectors should be much higher than the DVI spec, and a 2%reduction
\n
> from something 10%+ higher than spec should be of no consequence.I
\n
> suggest a total length not more than 3m, shorter is better.pair
\n>
\n> Bottom line: all you can do is make sure that your hand wiring has
\n> these characteristics: (1) shortest possible length of unshielded
\n
> (since the impedance changes when the shielding is gone); (2) eventhe
\n> without the shielding, keep the pair twisted together as much as
\n> possible; (3) do not coil the unshielded twisted pair.
\n>
\n> Test: determine maximum single-link bandwidth achievable without
\n
> Kramer box, with each cable set. Insert the Kramer box, and see if(which
\n> there is any reduction. When you approach the bandwidth limit
\n
> is not a brick wall limit, but a gradual curve, which will changewith
\n
> amplifier/receiver temperature -- colder is better) you will seeclocking
\n> shimmering or a kind of sparkling on the screen due to bit
\n
> instability. This is evident when I run my T210 at QXGA 60Hzsingle
\n
> link.bandwidth
\n>
\n> Test: try a different graphics card to see what is the max
\n
> for the rest of the system. In the past, ATI has tested betterthan
\n
> third party nVidia. (see, test at Extreme Hardware dot com). Forthe
\n
> most part, the DVI/TMDS driver on the graphics card is the mostchip
\n> critical part of the entire transmission system. Try cooling the
\n
> with freeze spray (from a distance to avoid excessive stress from
\n> differential temperature coefficients), and see how that effects
\n> bandwidth.
\n>
\n> BTW, no rush on the photos.
\n>