HDMi lead Version 1.4 vs. 2.0
 

In article ,
Jim Lesurf wrote:

That's a bit of a stranage way to put it. The HDMI cables that don't
work are ones that don't conform to the standard, and you can hardly
expect the standard to classify cables that don't conform to it.

See the problem I outlined earlier.

Two cables behaving differently. One 'cured' the problem between a TV and a
recorder, the other 'cured' the problem between a monitor and a computer.
Neither worked correctly in the other application.

I quite agree that there's a problem with many HDMI cables not working
properly. Either the cables or the devices are not meeting the
standard.

-- Richard

HDMi lead Version 1.4 vs. 2.0
 

On 22/01/2015 10:30, Jim Lesurf wrote:
Two cables behaving differently. One 'cured' the problem between a TV and a
recorder, the other 'cured' the problem between a monitor and a computer.
Neither worked correctly in the other application.

Your symptoms are just weird. Dropped bits on a digital comms link
causing a green cast to the picture?

I wonder if it had fallen back to some kind of analogue mode.

Andy

HDMi lead Version 1.4 vs. 2.0
 

In article ,
Vir Campestris wrote:

Two cables behaving differently. One 'cured' the problem between a TV and a
recorder, the other 'cured' the problem between a monitor and a computer.
Neither worked correctly in the other application.

Your symptoms are just weird. Dropped bits on a digital comms link
causing a green cast to the picture?

I wonder if it had fallen back to some kind of analogue mode.

HDMI cables don't have any analogue connection.

HDMI can use various colour spaces; perhaps a faulty connection caused
the two ends to disagree about which one to use.

-- Richard

HDMi lead Version 1.4 vs. 2.0
 

In article , Vir
Campestris wrote:
On 22/01/2015 10:30, Jim Lesurf wrote:
Two cables behaving differently. One 'cured' the problem between a TV
and a recorder, the other 'cured' the problem between a monitor and a
computer. Neither worked correctly in the other application.

Your symptoms are just weird. Dropped bits on a digital comms link
causing a green cast to the picture?

Yes, it did make me think at the time of the old H2G2 business of being
beyond 'surprise' and having to resort to 'astonishment'. :-) Until it
happened I'd assumed that tales of HDMI leads affecting the result in such
ways must be fancyful. However there it was.

On the same system some screen modes give other kinds of image defects. I
can only guess, but the ideas that came to me we

1) Some kind of clock synch skew problem with parallel bit transmissions
over the three data pairs arriving 'out of synch' and changing some colour
values.

2) ditto for edge effect ringing (mismatch to the cables) confusing the
receiver.

3) crosstalk between the data pairs causing some values to be altered when
received depending on the patterns of other pairs.

From the document I have, I can see that HDMI transfer *isn't* impedance
matched at both ends of the cable. They define the load at the destination,
but the input is essentially a 'current sink or float' - i.e. 'high
impedance'. Hence any mismatch between cable impedance and destination
could play havoc with the voltage and current levels sloshing back and
forth. And in turn aggrivating crosstalk.

So whilst 'bits is bits' there remains the problem of correctly receiving
the levels to determine correctly the bits sent. The bits aren't 'dropped'
so much as the RX simple interprets some of them incorrectly.

And the behaviour will depend on how well a particular cable impedance
matches a given monitor/TV input impedance. Also may depend on how much
the signal pairs are bent or squashed together, altering their impedances
and any crosstalk levels. Laid out straight the cables may be OK. squashed
and bent, maybe not.

In theory, theory and practice agree. But in practice... :-)

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

HDMi lead Version 1.4 vs. 2.0
 

On Sun, 25 Jan 2015 11:09:18 +0000 (GMT), Jim Lesurf
wrote:

In article , Vir
Campestris wrote:
On 22/01/2015 10:30, Jim Lesurf wrote:
Two cables behaving differently. One 'cured' the problem between a TV
and a recorder, the other 'cured' the problem between a monitor and a
computer. Neither worked correctly in the other application.

Your symptoms are just weird. Dropped bits on a digital comms link
causing a green cast to the picture?

Yes, it did make me think at the time of the old H2G2 business of being
beyond 'surprise' and having to resort to 'astonishment'. :-) Until it
happened I'd assumed that tales of HDMI leads affecting the result in such
ways must be fancyful. However there it was.

On the same system some screen modes give other kinds of image defects. I
can only guess, but the ideas that came to me we

1) Some kind of clock synch skew problem with parallel bit transmissions
over the three data pairs arriving 'out of synch' and changing some colour
values.

2) ditto for edge effect ringing (mismatch to the cables) confusing the
receiver.

3) crosstalk between the data pairs causing some values to be altered when
received depending on the patterns of other pairs.

From the document I have, I can see that HDMI transfer *isn't* impedance
matched at both ends of the cable. They define the load at the destination,
but the input is essentially a 'current sink or float' - i.e. 'high
impedance'. Hence any mismatch between cable impedance and destination
could play havoc with the voltage and current levels sloshing back and
forth. And in turn aggrivating crosstalk.

When it comes to _unidirectional_ transmission line systems, it's
only the recieving end that needs to be matched. Provided this is the
case, there will _not_ be any reflection of energy back from the
matched load to trouble the mismatched sending end to cause additional
reflections.

The more predictable mismatched sending impedance is usually chosen
to be low to make it look like a voltage source. The only time it
might make sense to use high impedance constant current sources is
when you wish to combine the outputs of several sources in parallel in
a simple additive summing of signals into the feeder cable.

When the signals being sent are _all_ serialised for transmission
over a single transmission line, the crosstalk problem simply doesn't
exist.

I'm not fully acquainted with the HDMI spec but I can't imagine it
making any sense to use more than a single balanced pair for each
direction of transmission of the digital video signals.

Assuming HDMI supports a 'back channel' for full digital video
return, as long as the rule, "Match the recieving end only" is
applied, the two balanced pairs could be laid up as a quad pair,
saving on unnecessary shielding between the pairs (reserving any such
cable shielding as a common shield for the whole cable).

I'm no expert on HDMI so my 'assumptions' that the designers made
commonsense use of the lessons of "Communications 101" could be
totally wrong (it really aught to be a 'no brainer' to get this sort
of thing absolutely spot on by now).

As for the weird effects sometimes experienced with HDMI connections,
they won't be due to cable issues if good engineering practice has
been followed in formulating the cabling standard. Since the spec
concentrates on matching the receiving end impedance to the cable
impedance, there's a strong suggestion that, for once, the standards
committee do have a fuller understanding of transmission cabling
requirements suggesting that the issues may be more to do with
mishandling of the protocols rather than cable defects.
--
J B Good

HDMi lead Version 1.4 vs. 2.0
 

In article , Johny B Good
wrote:
On Sun, 25 Jan 2015 11:09:18 +0000 (GMT), Jim Lesurf
wrote:


When it comes to _unidirectional_ transmission line systems, it's only
the recieving end that needs to be matched. Provided this is the case,
there will _not_ be any reflection of energy back from the matched load
to trouble the mismatched sending end to cause additional reflections.

Yes, that's all wonderful as theory. Alas, in reality the receiver is
unlikely to precisely match the cable. We're talking domestic consumer
goods here and five-ten quid cables bent and squashed behind the equipment.

So in the real world, as soon as the rx end reflects we will find that the
gross mismatch at the tx end will bounce back most of that again. Whereas
if the tx end was reasonably matched it would eat a fair bit of it
following the first reflection.

The result is a degrading of the waveform shapes at the rx end which is
somewhat bigger than if the tx end was matched to some reasonable extent.

And it means a tendency for some of the energy from one bit in the serial
sequence to be delayed and presented to the rx later on. Perhaps affecting
its ability to reliably understand the next bit correctly.


The more predictable mismatched sending impedance is usually chosen to
be low to make it look like a voltage source. The only time it might
make sense to use high impedance constant current sources is when you
wish to combine the outputs of several sources in parallel in a simple
additive summing of signals into the feeder cable.

And open (high) and closed (low) source will both reflect almost all that
comes back to them. So in this respect a near-zero source (voltage assert)
is as bad as a near-open (current source/sink). They both have a reflection
coefficient whose magnitude approaches unity.


When the signals being sent are _all_ serialised for transmission over
a single transmission line, the crosstalk problem simply doesn't exist.

The HDMI cable has *three* paired-line-in-shield connections running in
parallel to shovel the data across. This is to cope with the sheer rate of
data transfer.


Assuming HDMI supports a 'back channel' for full digital video return,
as long as the rule, "Match the recieving end only" is applied, the two
balanced pairs could be laid up as a quad pair, saving on unnecessary
shielding between the pairs (reserving any such cable shielding as a
common shield for the whole cable).

Alas, people tend to assume that we have things like balanced parallel
pairs and shields and treat them as if in a textbook.

Easy to forget that, say, putting something right up against such a system
can affect its behaviour or that a shield only works when matched and thick
enough. And that any physical distortions also alter behaviour.

Mismatch at the tx end can also mean some currents bounced back up the
*outside* of the 'shield'.

Real World != Undergrad Textbook Theory returns TRUE here, alas.



As for the weird effects sometimes experienced with HDMI connections,
they won't be due to cable issues if good engineering practice has been
followed in formulating the cabling standard.

IF statement duly noted. :-)

Alas reality may not match theory, any more than these sources match their
cables. The real world seems happier with "fling the problem over the
fence". i.e. here "assume the rx is matched regardless of whatever cable
you got and how much you've bent or squashed it."


Since the spec
concentrates on matching the receiving end impedance to the cable
impedance, there's a strong suggestion that, for once, the standards
committee do have a fuller understanding of transmission cabling
requirements suggesting that the issues may be more to do with
mishandling of the protocols rather than cable defects.

I beg to differ. :-)

I assume you do realise, BTW that real wideband cables tend to have an
impedance that is frequency dependent. Somehow I doubt all HDMI cables are
adjusted to meet the Heavyside requirement, even ignoring the presence of
other cables right beside them, etc.

I fear that in reality the people making these cables and systems are more
reliant on "Its digital, innit!" - i.e. taking advantage of the ability of
digital systems to ignore small alterations in waveform shapes, etc. Alas,
this has its limits.

I can see that the HDMI designers have tried to play some useful tricks.
e.g. the balanced pair clearly is aimed to have a constant internal total
current within each shield. That - if achieved - will help reduce problems
like crosstalk, for example. The problem is that this may not help with the
above problems. And relies on accurate synchrony with a tx that is
mismatched by choice.

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html