How to Generate a Tri-level signal.

MisterBill2

Joined Jan 23, 2018
18,176
Another concept that would not be limited by analog switch delays is to simply feed the outputs of digital gates through resistors into the inputs of a summing junction of a really fast OP Amp, or even a fast video amplifier. The math to calculate the correct values of resistance may be tedious but not hard, and there are a wide selection of really fast devices available. So the different approach may be an easier solution.
And another thing may be to see what the camera makers suggested circuit is. Circuits are seldom designed for no particular application.
 

ronsimpson

Joined Oct 7, 2019
2,989
I very much understand sync and video. I have read all the comments several times and do not know what your want. Sorry my thick head.
I understand:
You have a camera.
You have a FPGA.
It appears the camera has 2-level sync, but you want 3-level sync.

I also understand you will never give us the name and number of the camera.
You are not going to say what receives the signal from the camera.

The timing from sync (of any type) to video is very critical. Just adding a new level is not going to help. You must keep the timing from sync to a pixel accurate with in 1/4 pixel. If you video rate of 75mhz you have data every 13nS. You need to position the new sync to within 3nS or you will see noise. (maybe you can get away with 5nS noise)

The camera has a "dot clock" of maybe 75mhz. (I can not know what frequency) Your FPGA has a clock of some frequency. The two clocks need to be the same. Not just the same frequency but the same phase to 3nS. (depending on that is looking at the camera) If you have 75,000,000hz in the camera and 75,000,010hz with the FPGA you will see noise role through the picture 10 times a second.

Maybe the receiving device has a three pixel averaging circuit which will hide much of the noise. It will hide some of the resolution. I see "high resolution" several time in the posts so maybe you do not want to average 3 pixels.

I think you do not like "low resolution sync" and want to add "3-level high resolution sync". By looking at the low resolution sync and using it as a standard you are not going to get better. Then adding on top of that the random 0 to 13nS delay in the FPGA you will have much worse sync.
 

danadak

Joined Mar 10, 2018
4,057
A high speed 8 bit bipolar DAC would do this fairly easily. With onboard Vref,
single supply (you can offset its output with a fast amp or use a bipolar version
of DAC), digital interface simple.


Regards, Dana.
 

Thread Starter

lindula

Joined Sep 23, 2016
94
I very much understand sync and video. I have read all the comments several times and do not know what your want. Sorry my thick head.
I understand:
You have a camera.
You have a FPGA.
It appears the camera has 2-level sync, but you want 3-level sync.

I also understand you will never give us the name and number of the camera.
You are not going to say what receives the signal from the camera.

The timing from sync (of any type) to video is very critical. Just adding a new level is not going to help. You must keep the timing from sync to a pixel accurate with in 1/4 pixel. If you video rate of 75mhz you have data every 13nS. You need to position the new sync to within 3nS or you will see noise. (maybe you can get away with 5nS noise)

The camera has a "dot clock" of maybe 75mhz. (I can not know what frequency) Your FPGA has a clock of some frequency. The two clocks need to be the same. Not just the same frequency but the same phase to 3nS. (depending on that is looking at the camera) If you have 75,000,000hz in the camera and 75,000,010hz with the FPGA you will see noise role through the picture 10 times a second.

Maybe the receiving device has a three pixel averaging circuit which will hide much of the noise. It will hide some of the resolution. I see "high resolution" several time in the posts so maybe you do not want to average 3 pixels.

I think you do not like "low resolution sync" and want to add "3-level high resolution sync". By looking at the low resolution sync and using it as a standard you are not going to get better. Then adding on top of that the random 0 to 13nS delay in the FPGA you will have much worse sync.
Hello, thanks for responding to my message. HD Video and 3G,6G,12G SDI video require a Tri-level Gen lock signal: 0.3V, -0.3V and ground.
 

AnalogKid

Joined Aug 1, 2013
10,987
It appears the camera has 2-level sync, but you want 3-level sync.
Don't think so. It sounds like he is designing a genlock signal generator for a hi-def production environment. Back in NTSC times we used a signal called blackburst (also known as "color-black"). These days it is a much less complex signal that relies on the slave equipment developing all internal sync signals and phase-locking its internal master clock to the external tri-sync frequency.

https://en.wikipedia.org/wiki/Genlock
https://en.wikipedia.org/wiki/Tri-level_sync

ak
 

AnalogKid

Joined Aug 1, 2013
10,987
Hello, I landed on a solution that will support 12G SDI Genlock Tri-Level signal:
https://www.analog.com/en/products/lt1399.html

I will connected the VinA to 0.3V, VinB to -0.3V and VinC to ground and control the enables with my FPGA.
An alternate approach is to use switched current sources rather than voltage sources. An advantage is that you need only two sources, since the source-terminating resistor to GND provides the GND output when both sources are off.

ak
 

Thread Starter

lindula

Joined Sep 23, 2016
94
Hello, wanted to share my solution:
Tri_level.jpg
The AD8003 has great Bandwidth and Slew Rate numbers. I use a D/A to generate the TRI4_HI voltage and TRI4_LO voltage. The D/A allows me to adjust the 300mV levels.
 

AnalogKid

Joined Aug 1, 2013
10,987
Hello, wanted to share my solution:
Your three buildout resistors form a combined source impedance of 32.5K. What kind of cable are your driving, and what is its characteristic impedance? And, what is the termination impedance at the far end? If the cable impedance and/or termination are in the 100 ohm range, then the buildout redsistors will attenuate the amplifier output voltages by over 300-to-1.

ak
 
Last edited:

ebeowulf17

Joined Aug 12, 2014
3,307
Your three buildout resistors for a combined source impedance of 32.5K. What kind of cable are your driving, and what is its characteristic impedance? And, what is the termination impedance at the far end? If the cable impedance and/or termination are in the 100 ohm range, then the buildout redsistors will attenuate the amplifier output voltages by over 300-to-1.

ak
Looks like they got those numbers from the reference info on the LT1399 mentioned earlier... except that they're off by x1000. That pesky little "k" at the end changes everything!
LT1398-3920.png
 
Top