​There is so much misunderstanding in this industry when it comes to the HDMI 4K/60 deployment and the type of transmission lines that are required to support it. The average Joe still doesn’t get it and there is a lot to get!
First, HDMI’s mission is to provide a transmission system that can grow and still have backwards compatibility. As every new revision surfaces a smooth and seamless transition should take place over the interface. In a perfect world this is very true, however, when you have literally thousands of developers trying to build HDMI products things can and do happen. This is no different than what we experienced during the “Dark Days” as I call it when the failures were through the roof. But every time a new revision pops up our case studies here in the lab increases due to in field failures. We determined why the failures occur and update our procedures with additional tests to keep ahead of the curve. By discovering these anomalies it allows DPL to keep a handle on it while at the same time report it back to the manufacture to correct. That is how they earn their DPL Mark.
That being said, let’s jump on 4K. Believe it or not, the fundamentals of today’s Rev 2.0a High Speed 4K/60 (greater than 10.2Gbps) was in place way back in 2006 under Revision 1.3a. Now this gets a little deep so stay with me.
When HDMI made the jump with Rev 1.3 from 5Gbps to 10.2Gbps (3.4G per channel), they introduced both Standard and High Speed cables. There was really no magic sauce in doing this. It was primarily wire gauge, metallurgy and topology. Were there differences? Sure there were, but it was due to some firms just making poor products and others that followed the rules and lived high in the black, far away from the commodity types of products. Short cables required larger gauge as would longer cable. This is a no brainer, they just had to make certain limits in bandwidth. This is where sales and marketing would announce claims that were not necessarily confirmed.
Let’s keep this simple, Figure 1 pretty much explains it all. The red line depicts the natural roll off of cable, any cable, they all roll off, some better than others. You can see at 3Gbps the estimated roll off is a whopping 15dB. This is where we lived before Rev 2.0a. At that time HDMI instituted a change in HDMI receivers, primarily displays. Each receiver now includes a built in equalizer to compensate for the losses over this curve. If no equalization was used, cables over three meters would be jeopardized. Fig 2 shows the added EQ correction in Blue.
Now this is where things go a little crazy. Continue to look at Figure 2, Rev 2.0 was introduced being 4K/60 but only at 3Gbps (9G total), the same bandwidth we had under 1.3 and 1.4. It was a transitional period but many hardware people just claimed 4K/60, never really understanding we were not there yet. The bandwidth reduction was accomplished by reducing the color subsampling rate and the removal of deep color. So in essence, we robbed color to double the frame rate, which dropped the necessary bandwidth from 18Gbps to 9Gbps. It wasn’t until early 2015 did we start seeing High Speed 4K/60 under Rev 2.0a for HDR and 6Gbps (18G total). But also notice that the EQ correction only extends out to 5.1GHz, not 6Hz where it needs to go!
So how can it be true that we can use existing 1.4 transmission lines with Rev2.0a? It is all in the math and it is called “The Squeeze Theorem” represented by the equation SinX/X (Sine of X Over X). By calculating SinX/X and our Nyquist operating area we get Figure 3. Better yet, check the harmonics out beyond 6GHz. Now don’t get all warm and fuzzy thinking all is good. There are some pitfalls here. One in particular is jitter. This, in most cases, is an impedance issue. Jitter became one of our biggest enemies when it comes to High Speed 4K/60 at 18Gbps (6Gbps x 3 Channels). So what does this look like in the real world of hardware? Thus far we have experience an extreme high failure rate when attempting to go long with active cables.
Fig 4 shows a response curve for a 1 meter 30AWG cable. We embedded the natural insertion loss with a red line. This is the limit line that cables need to be above. As the picture says, “In the Red You’re Dead.” Obviously, the cables response is pretty good except for the minor dip at 1 and 2Gbps respectfully. However, notice what happen to the 6GHz Eye Test? This cable failed under test where jitter became a big issue.
Figure 5 shows a 2 meter with 30AWG also. Quite a big difference between these. Reason being was the metallurgy and the build.
Figure 6 demonstrates what happens if the product really breaches the Red limits. Here a 3-meter cable is our DUT with 28AWG. Again, material and construction can be blamed here. But in this case it breached so bad it is effecting both 3.4 and 6GHz.
Fig 7 demonstrates what happens when a cable is built right. It is almost parallel to the insertion loss curve. But notice it falls short in getting to 6GHz. The Eye looks great, so why?
Figure 8 is an example of what to expect in the coming months with active products extending cable lengths well beyond 7 meters to estimates as much as 20 meters.
We shall see. Companies are already sending in their new 18Gbps cables for testing with some claiming 15 and 20 meters. Keep an eye on the DPL website, if they pass you will see them posted there.
Jeffery Boccaccio is President of DPL Labs in Ormond Beach, Florida.