Author Topic: HDMI image grab with microprocessors. Is it feasible? About HDMI Demystification  (Read 9366 times)

0 Members and 1 Guest are viewing this topic.

Offline hamster_nz

  • Super Contributor
  • ***
  • Posts: 2803
  • Country: nz
Yeah, that's me..... :-)
Gaze not into the abyss, lest you become recognized as an abyss domain expert, and they expect you keep gazing into the damn thing.
 

Offline donkeyTopic starter

  • Regular Contributor
  • *
  • Posts: 51
  • Country: ru
I'm honored to meet you, hamster_nz
 

Offline donkeyTopic starter

  • Regular Contributor
  • *
  • Posts: 51
  • Country: ru
Ultra HD HDMI to MIPI CSI-2 (Camera Serial Interface) converter chipset
https://toshiba.semicon-storage.com/info/docget.jsp?did=28743&prodName=TC358870XBG
 

Offline donkeyTopic starter

  • Regular Contributor
  • *
  • Posts: 51
  • Country: ru
HDMI Oscilloscope requirements

Credit goes to Guenter Lehmann :-+ from Analog Devices, answer on

Question : "Where do
8 GHz minimum bandwidth and a sampling rate of 20 Gsample/s for HDMI 1.4b signal
12.5 GHz minimum bandwidth and 40 Gsample/s for HDMI 2.0 signal
requirements are comming from ?"


Answer :
1.) HDMI data rate = 10 * PixelRate * (ColorDepth /8)
1080p60 (pixelrate=148.5MHz), for 8 bit color depth -> data rate = 1.485 Gbit/s on each TMDS data pair
12 bit color would give us -> data rate =  2.2275Gbit/s on each pair.

The 10x multiplier comes from the 8b10 encoding use to push one color plane down a TMDS pair

For 1080p120, 8 bit color depth -> 10 * (2 * 148.5MHz) * (8/8) = 2.97GBits/s (3G)

For 4Kp30, 8 bit color depth -> 10 * 297MHz * (8/8) = 2.97GBits/s (3G)

For 4Kp60, 8 bit color depth -> 10 * 594MHz * (8/8) = 5.94GBit/s (6G)

Or 4Kp60, 8 bit color depth -> 40 * 148.5MHz * (8/8)  5.94GBits/s (6G)

Note for 4Kp60 the TMDS clock line runs at 148.5MHz but the multiplier is changed to 40x

 
2.) When we look at square waves with scopes, the bandwidth of the scope needs to include the 3rd and 5th (and sometimes 7th) harmonic (Fourier) of the fundamental frequency you are trying to measure.  Ideally we’d like the scope bandwidth to be 10x the fundamental frequency we are trying to measure else we’ll see a distorted image on the scope.  So for 1080p60 where the fundamental is 1.485GHz, the scope bandwidth needs to be greater than 1.485GHz * 5 or 7.245Ghz.  Digital sampling scopes need to have a sampling rate 2 x bandwidth (Nyquist).  The same applies to HDMI 2.0, just faster.


I wrote this guide as a simple set of rules PCB designers can use when laying out HDMI traces.  The key to HDMI routing is to keep the trace flow smooth, no stubs or tight turns and keep the differential impedance to 100 Ohms.  Eye patterns are used to understand why your routing doesn’t work or meet HDMI specifications (theoretical background).  Many layout people don’t have signal integrity tools to simulate the eye pattern and test engineers don’t have test equipment to measure the eye pattern.

 
Hope this helps

Guenter Lehmann
Sr. Application Engineer
Analog Devices
« Last Edit: October 19, 2018, 04:06:02 am by donkey »
 
The following users thanked this post: dreamcat4


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf