KVM switch for DVI

Benjamin Gawert wrote:
> * _DD:
>
> > I'm looking for a 2-channel or 4-channel KVM switch for use with DVI
> > video. 1600x1200 resolution would be good. 1920x1200 would be even
> > better.
> >
> > Any recommendations for models or brands?
>
> I'm using an ATEN CS-1764 4-port USB/DVI KVM which atm is connected to
> two PCs, one Mac and one workstation (HP i2000 Itanium workstation) on
> the computer side and a Dell 2005FPW 20" TFT at the monitor side. I have
> this KVM for 2.5yrs now and I'm very satisfied with it. Changing
> between input channels is just ~0.5s. The ATEN also switches sound
> (which I don't use). It also has an integrated USB hub with two USB
> ports (additional to the 2x USB for keyb/mouse) which can be switched to
> any computer independently of keyb/mouse/video.
>
> I had an analog (VGA and PS/2 keyb/mouse) 8-port Avocent KVM before
> which was really good and didn't compromise image quality but also was
> extremely expensive. Avocent makes great KVMs but for home use it's just
> overkill and too expensive. I also tried some cheaper KVM switches but
> they all suffered from one or more problems.
>
> Benjamin

Benjamin, how is the image quality via your ATEN vs. directly
connecting the DVI cable to the monitor? Can you tell a difference?

Also, how does the independent USB switching work? Are there buttons
for that? Does it support USB 2.0?
 >> Stay informed about: KVM switch for DVI 

* _DD:

> I'm looking for a 2-channel or 4-channel KVM switch for use with DVI
> video. 1600x1200 resolution would be good. 1920x1200 would be even
> better.
>
> Any recommendations for models or brands?

I'm using an ATEN CS-1764 4-port USB/DVI KVM which atm is connected to
two PCs, one Mac and one workstation (HP i2000 Itanium workstation) on
the computer side and a Dell 2005FPW 20" TFT at the monitor side. I have
this KVM for 2.5yrs now and I'm very satisfied with it. Changing
between input channels is just ~0.5s. The ATEN also switches sound
(which I don't use). It also has an integrated USB hub with two USB
ports (additional to the 2x USB for keyb/mouse) which can be switched to
any computer independently of keyb/mouse/video.

I had an analog (VGA and PS/2 keyb/mouse) 8-port Avocent KVM before
which was really good and didn't compromise image quality but also was
extremely expensive. Avocent makes great KVMs but for home use it's just
overkill and too expensive. I also tried some cheaper KVM switches but
they all suffered from one or more problems.

Benjamin
 >> Stay informed about: KVM switch for DVI 

* HockeyTownUSA:

> 1600x1200 - 1920x1200 would be very expensive, if you can even find one.

Nope. Every better DVI KVM does 1920x1200 without any problems. Going
higher which also means Dual-Link DVI is a different story, though.

Benjamin
 >> Stay informed about: KVM switch for DVI 

* Benjamin Gawert:

> I'm using an ATEN CS-1764 4-port USB/DVI KVM which atm is connected to
> two PCs, one Mac and one workstation (HP i2000 Itanium workstation) on
> the computer side and a Dell 2005FPW 20" TFT at the monitor side.

A little remark: the Dell 2005FPW has a resolution of 1680x1050. I also
had a Dell 2405FPW which runs at 1920x1200 for a few days connected to
this KVM switch and it ran fine.

Benjamin
 >> Stay informed about: KVM switch for DVI 

* toronado455:

> Benjamin, how is the image quality via your ATEN vs. directly
> connecting the DVI cable to the monitor? Can you tell a difference?

There is no difference, and there never can be a difference. Unlike
analog signals like VGA DVI is a fully digital transmission and thus
there is no image degradation when using a KVM switch or extension cables...

> Also, how does the independent USB switching work? Are there buttons
> for that?

Yes, there are separate buttons for USB routing...

> Does it support USB 2.0?

I'm not sure because I rarely use this feature and if I do I only use it
with an old USB 1.1 card reader.

Benjamin
 >> Stay informed about: KVM switch for DVI 

On Tue, 02 May 2006 21:28:02 +0200, Benjamin Gawert
wrote:

>* toronado455:
>
>> Benjamin, how is the image quality via your ATEN vs. directly
>> connecting the DVI cable to the monitor? Can you tell a difference?
>
>There is no difference, and there never can be a difference. Unlike
>analog signals like VGA DVI is a fully digital transmission and thus
>there is no image degradation when using a KVM switch or extension cables...

Untrue.
It may be digital but the signal strength still has to be
strong enough to differentiate the digit(s). Flip a few
bits and it can certainly degrade the image, though it's not
so likely it would work at all if it did it on the scale
that analog did.
 >> Stay informed about: KVM switch for DVI 

* kony:

>>> Benjamin, how is the image quality via your ATEN vs. directly
>>> connecting the DVI cable to the monitor? Can you tell a difference?
>> There is no difference, and there never can be a difference. Unlike
>> analog signals like VGA DVI is a fully digital transmission and thus
>> there is no image degradation when using a KVM switch or extension cables...
>
> Untrue.

Nope, it's not.

> It may be digital but the signal strength still has to be
> strong enough to differentiate the digit(s). Flip a few
> bits and it can certainly degrade the image, though it's not
> so likely it would work at all if it did it on the scale
> that analog did.

If a KVM switch does influence the signal so that as you said "some bits
flip" then it's defective. Period.

With DVI the image quality remains constant (there is no degradation).
If signal quality decreases over to a point that violates the
specifications (i.e. because of too long extension cables) the you get
display errors or no display at all.

Benjamin
 >> Stay informed about: KVM switch for DVI 

On Wed, 03 May 2006 07:17:52 +0200, Benjamin Gawert
wrote:

>* kony:
>
>>>> Benjamin, how is the image quality via your ATEN vs. directly
>>>> connecting the DVI cable to the monitor? Can you tell a difference?
>>> There is no difference, and there never can be a difference. Unlike
>>> analog signals like VGA DVI is a fully digital transmission and thus
>>> there is no image degradation when using a KVM switch or extension cables...
>>
>> Untrue.
>
>Nope, it's not.
>
>> It may be digital but the signal strength still has to be
>> strong enough to differentiate the digit(s). Flip a few
>> bits and it can certainly degrade the image, though it's not
>> so likely it would work at all if it did it on the scale
>> that analog did.
>
>If a KVM switch does influence the signal so that as you said "some bits
>flip" then it's defective. Period.

Nonsense, same vague thing could be said about analog cable
and would be equally untrue. The signal level can in fact
degrade and this is WHY there are always (ALWAYS) cable
length limitations for digital signal cables. You must use
a repeater of some sort to overcome the problem at some
length (which may well be longer than the spec calls for, as
it should be, but nevertheless IS necessary at *some*
length). It has nothing to do with the KVM switch (opposed
to not having one), though.

In fact, with ANY digital cable, the signal itself does
ALWAYS degrade. That's just what wire does. The question
is whether this always-present degradation of signal is too
much for the receiver to differentiate the data.

Is there a checksum on the data? That may preserve it (if
enough bandwidth remained for resends), but it does nothing
to prevent the bits flipping in the first place.
 >> Stay informed about: KVM switch for DVI 

Benjamin Gawert wrote:
>> Your very basic thinking of degradation because of "flipping bits"
> doesn't really fit to DVI because the TMDS signalling used in PanelLink
> communications (the technology that is used in DVI) is more than the
> plain transmission of a few bits. PanelLink uses quite complex data
> words with ecc schemes which makes it even more robust.


Then to a certain extent I was wrong, but it still does NOT prevent
bit-flipping... even words are still subject to signal levels. It
would only cause resend but even then, it will require the bandwidth to
do it. If it cannot display the image in realtime terms from resends,
it is still a degradation.
 >> Stay informed about: KVM switch for DVI 

* kony:

>> If a KVM switch does influence the signal so that as you said "some bits
>> flip" then it's defective. Period.
>
> Nonsense, same vague thing could be said about analog cable
> and would be equally untrue.

Nope. With analog transmission image quality is directly proportional to
the various influences over the transmission line. With digital, it's not.

> The signal level can in fact
> degrade and this is WHY there are always (ALWAYS) cable
> length limitations for digital signal cables. You must use
> a repeater of some sort to overcome the problem at some
> length (which may well be longer than the spec calls for, as
> it should be, but nevertheless IS necessary at *some*
> length).

Right. So what? Of course you can't use as much extension cords as you
want with DVI. But that doesn't change a thing on the fact that digital
transmission unlike analog transmission has no image quality degradation
due to longer cables or the fact that there is a KVM switch in the line.

> In fact, with ANY digital cable, the signal itself does
> ALWAYS degrade. That's just what wire does.

Right, the _signal_quality_ degrades. But that doesn't mean the
_image_quality_ degrades, too.

> The question
> is whether this always-present degradation of signal is too
> much for the receiver to differentiate the data.
>
> Is there a checksum on the data?

Yes.

> That may preserve it (if
> enough bandwidth remained for resends), but it does nothing
> to prevent the bits flipping in the first place.

"Bit flipping" as you call it doesn't happen over cables. What happens,
though is that there are various types of influences (crosstalk, wave
effects, reflections, irradiation etc) that are effective in cables (and
any other type of signal transmission). One of the advantages of digital
transmission over analog transmission however is that digital
transmission is way more robust to these influences than analog
transmissions. With analog, the data integrity is directly proportional
to the amount of disturbance. With digital transmission the data
integrity remains constant up to a certain extend where no valid data
can be recovered from the received signal. Since unlike with analog
transmission the data integrity remains constant over a certain area of
noise level the image quality also remains constant, no matter if the
cable is say 1m or 5m or if there is a KVM in the line or not.

Your very basic thinking of degradation because of "flipping bits"
doesn't really fit to DVI because the TMDS signalling used in PanelLink
communications (the technology that is used in DVI) is more than the
plain transmission of a few bits. PanelLink uses quite complex data
words with ecc schemes which makes it even more robust.

Benjamin
 >> Stay informed about: KVM switch for DVI 

* kony:
> Benjamin Gawert wrote:

> Then to a certain extent I was wrong, but it still does NOT prevent
> bit-flipping... even words are still subject to signal levels. It
> would only cause resend but even then, it will require the bandwidth to
> do it. If it cannot display the image in realtime terms from resends,
> it is still a degradation.

Again, you are thinking too basic. There simply is no "Bit flipping" in
cables. There only is a certain amount of noise and signal deformation
because of the cable properties (R-L-C combination). Even if there would
be "bit flipping" the whole data word wouldn't have to be resent. There
is no lag, there is no image quality degradation, there is nothing
visible, period. If you increase DVI cable length the image remains
totally perfekt until the TMDS controller is not able to extract data
any more and the image gets extremely distorted (massive artifacts) or
there isn't any image at all. With analog you can see the image getting
worse and worse when increasing the cable length.

Benjamin
 >> Stay informed about: KVM switch for DVI 

"Benjamin Gawert" wrote in message

> > Is there a checksum on the data?
>
> Yes.

That would come as a surprise to the people (including myself)
who worked on the DVI spec. DVI does NOT include a parity
bit, checksum, or other error-detection/correction functionality
within the data stream, nor is there any provision for error handling
on the link. In short, should the signal degrade to the point where
errors are being introduced into the data (but the receivers can still
maintain synchronization and data decoding), there is no way for
the receiver to detect this.

However, it should be noted that display applications are generally
pretty forgiving when it comes to the occasional error, whether that
comes in via analog OR digital encoding of the video.


> "Bit flipping" as you call it doesn't happen over cables. What happens,
> though is that there are various types of influences (crosstalk, wave
> effects, reflections, irradiation etc) that are effective in cables (and
> any other type of signal transmission). One of the advantages of digital
> transmission over analog transmission however is that digital
> transmission is way more robust to these influences than analog
> transmissions.

Yes and no. As I've said here before, there is really NO difference
between "analog" and "digital" when it comes to robustness in the presence
of noise IF THESE ARE COMPARED AT COMPARABLE DATA
RATES. There's simply no way to get around the Gospel According To
St. Shannon, which sets an absolute and unavoidable limit on the amount
of useful information you can get through any physical channel in the
presence
of a given amount of noise. What most people mean by a "digital"
transmission - wherein a given physical line carries a serial stream of
amplitude-encoded binary - is simply an example of trading off data
rate for noise margin, and that's all it is.


> Since unlike with analog
> transmission the data integrity remains constant over a certain area of
> noise level the image quality also remains constant, no matter if the
> cable is say 1m or 5m or if there is a KVM in the line or not.

However, additional "noise" (meaning anything in the signal which
is not "signal," including all forms of noise, distortion, etc.) is
unavoidable for longer cable runs. You WILL get a higher error
rate with longer cables, all else being equal, and that's unavoidable.

> Your very basic thinking of degradation because of "flipping bits"
> doesn't really fit to DVI because the TMDS signalling used in PanelLink
> communications (the technology that is used in DVI) is more than the
> plain transmission of a few bits. PanelLink uses quite complex data
> words with ecc schemes which makes it even more robust.

Nope. PanelLink(TM) uses a proprietary 8-to-10-bit encoding
scheme whose primary functions are to DC balance the line and
minimize the number of transitions (and therefore signal-induced
noise) in the resulting data stream. It most definitely does NOT
add anything in the way of robustness in terms of the error rate.

Bob M.
 >> Stay informed about: KVM switch for DVI 

"Benjamin Gawert" wrote in message

> Again, you are thinking too basic. There simply is no "Bit flipping" in
> cables. There only is a certain amount of noise and signal deformation
> because of the cable properties (R-L-C combination). Even if there would
> be "bit flipping" the whole data word wouldn't have to be resent. There
> is no lag, there is no image quality degradation, there is nothing
> visible, period. If you increase DVI cable length the image remains
> totally perfekt until the TMDS controller is not able to extract data
> any more and the image gets extremely distorted (massive artifacts) or
> there isn't any image at all. With analog you can see the image getting
> worse and worse when increasing the cable length.

Which is really just another way of saying that "analog" encoding
of information (here's where I generally throw in the observation
that there's really no such thing as an analog or digital SIGNAL -
it's always just voltage variations or whatever on a wire, and what
matters is how we're supposed to interpret those) degrades
"more gracefully" in the presence of increasing noise levels than
does most "digital" encoding. This happens because with
"analog" encoding, the most important parts of the information
(the MSBs) are also those which are most resistant to noise, and
the LSBs are lost first. In the typical "digital" scheme (although not
ALL "digital" representations suffer from this), all bits are equally
vulnerable to noise, and the image continues to look pretty darn
good (although NOT perfect - the data stream does NOT have to
be error-free to look good) until you hit a brick wall in terms of
noise level, at which point everything goes to hell at once. An
analog-video system will often deliver a very usable image well
beyond the point at which noise has completely knocked out a
simple binary-encoded digital interface.

Of course, we also need to note that noise-induced errors in the
video signal are rarely a problem for either system, INCLUDING
analog; what is a much greater problem with current "analog video"
interfaces, when used with LCDs and similar fixed-pixel displays,
is not signal integrity but rather the ability of the receiving display
to generate a correct sampling clock. There's no pixel-level
synchronization information conveyed in analog video systems, so
synchronization/sampling (which is not really a "digital vs. analog"
issue) is a much bigger issue. Take the pixel clock away from any
"digital" interrface, and see how well IT does....

Bob M.
 >> Stay informed about: KVM switch for DVI