Is there a delay when using BMD SDI Routers?

Hi all,

I wonder if the BMD Router family would introduce some delay (latency) in the HD-SDI signal.

F.e. If I use a a BMD router in front of an ATEM switcher like feeding the camera signals into that router and feeding the ATEM with the outputs of that router. Will that add any delay to the signal?

Usualy you get 1 frame of delay when using unsynced cameras through the ATEM. Will that increase if I use the router in front of the ATEM?

Thanks

pro.

We run all 8/10 cameras that way and don't ever see one enough to notice it.

Thanks jparker for the reply - but I need some reliable answers regarding the amount of delay.

Is there anyone from Blackmagic who can chime in?


Thanks again and have a nice weekend!

[Kaspar Ko]

The short answer is that it will less than a frame.

The Videohub family will switch signals at the top of the next cycle, which means under a frame in any scenario.

Hello Kasper,

thanky for your reply.

But I think you misunderstood me. I don't want to know how fast the videohub will switch.

I want to know if there is any difference in timing regarding the input and the output of the videohub. I mean: will the latency increase when I connect a camera to a videohub and then the videohub to the ATEM switcher?

The ATEM switcher has 1 frame of latency from input to output (if you don't use any DVE). Will that be increased if I use a videohub in front of the ATEM?

Especially for IMAG use you need a low latency setup and any additional frame will mess up your system.

Could you answer that please?

Thank you!


pro

[Kaspar Ko]

Nope, latency will not increase. The only time you would see any latency from the Videohub is when you are switching inputs/outputs as I mentioned earlier. Else any signal passing through the Videohub will have negligible latency in the magnitude of lines. And since HD has 1080 lines per frame and consdiering the amount of frames per second, you can see why it's not an issue.

If there is no/"minimal" latency, why do the Videohub's have a genlock input? Though the videohubs don't have framesyncs so the delay can't be more than a few lines, probably no more than 4 like the miniconverters

[Kaspar Ko]

Reference is only applicable when signal routes are being changed. If for example if I had the follwing scenario: reference is tri-level at 1080i59.94 and I have four different inputs i.e. two 1080i59.94 and two 720p60 sources.

If I am changing the routes between the two 720p sources to a single output, then the route change will happen at the top of the next cycle irregardless of where it is in the frame. This can potentially lead to visible tearing of the image as the routes are changed if the next cycle happens to be in the middle of the frame.

Now say you are changing the routes between the two 1080i sources, since it's the same as the reference applied, then the route change occurs in the non visible raster (I believe line 6) so there will be no visible tearing between the two changes.

Hope that helps clarify.

Hello Kasper,

thanky for your reply.

But I think you misunderstood me. I don't want to know how fast the videohub will switch.

I want to know if there is any difference in timing regarding the input and the output of the videohub. I mean: will the latency increase when I connect a camera to a videohub and then the videohub to the ATEM switcher?

The ATEM switcher has 1 frame of latency from input to output (if you don't use any DVE). Will that be increased if I use a videohub in front of the ATEM?

Especially for IMAG use you need a low latency setup and any additional frame will mess up your system.

Could you answer that please?

Thank you!


pro

You are asking about he propagation delay throughout the router, so in a switcher like the atem, signals are frame sync'ed therefore a delay occurs, from input to output, as the cust. rep explained in a router they are not frame sync'ed they are switched at the top of the reference frame, this is only while switching. I haven't seen BM publish their system propagation delay, but Ross NK routers which are a Codan product, (australian), very similar architecture , you have a 50ns propagation delay. A typical crosspoint switch FPGA will have a 7ns delay. So once the switch is made and the input to output relationship is established, the delay is no more than a distribution amplifier would be.
An easy enough test would be to use the super out of a deck, run one copy to your atem, and one to the videohub then route to the atem, record the multiview output and compare the timecode burn, on the 2 windows, you can even run the output to the videohub to an input to an output to an input looping it in itself several times and you still wouldnt see the frame difference.