ADV7850 TMDS signal conversion time

Hi, 

   It will depend on the stability of the incoming video, So we don't know how much time it will take to stabilize and for video the pixels maybe delayed from 12-50 pixel clocks depending which video processing modules are turned on.

    Please note that, It depends on the controller that we are using. If customer want the exact result then the customer can use the API's to measure the exact time (i.e In Software customer can add the api like GetCurrentMsCount(); between the appropriate locking registers to get the exact locking time).

Thanks,

Poornima

Hi Poornima,

Thanks for your information. So that mean there is no way to adjust the transmit time, right?
Because our product has two sources: DVI & HDMI, where DVI goes through ADV7850 before converting out, HDMI is directly output. In the same source(one side through HDMI to DVI adaptor), from the time of connecting the source to the screen output, it took more than 10 seconds. So I wanted to ask you if there is a way to improve.

Best regards,

Oden Chang

Hi,

 Kindly note that, whether it is HDMI or DVI source locking time will depend on the stability of the incoming video, So we don't know the exact time how much it will take to stabilize.

 Under normal conditions the PLL should lock in a couple of video frames (i.e. PLL had indeed locked within about 50mS).

 As per expert comment, " 110mS is the ample time for the PLL to lock & 110mS is plenty of time to achieve lock assuming a good video input ".

Note: HDMI and DVI are the same with the exception that HDMI carries info-frames.  The existents of an info-frame marks the stream as HDMI. The main difference between HDMI and DVI.:DVI signal with no info-frames & It can support only RGB input color space.

Thanks,

Poornima

Hi Oden,

You must be an HMDI and HDCP adopter before ADI can approve selling any ADV7850's to you, so these questions might just be academic questions.  This even applies to the evaluation board.  We'd need your company information for validation with the HDMI and HDCP organizations.

Poornima's comments are correct. Normally the ADV7850 will lock on to new video in 2-4 vsync periods.  Delay through will always be less then one Hsync period except when using the 3D comb filter or deinterlacer, then it will change to ~2 frames delay

If you are really seeing 10 second delays then more then likely it's an unstable video or the processor is delaying format changes.

BTW: are you testing this on the eval. board or some other piece of hardware?

Hi Poornima and GuenterL,

Please let me explain our product and the situation we encountered in more detail.
Here is our product's block diagram. The main function of this board is image capture, and ADV7850 is used with Valens' IC.Our source input have HDMI and DVI-I (including digital and analog signals). HDMI does not need to go through the Scaler IC because it is a TMDS signal, but DVI-I includes an analog signal, so we use this Scaler IC ADV7850 to convert the DVI-I (digital + analog) signal into TMDS for VS010TX (because Valens IC only supports TMDS input).

We did the following experiment: The input source is given through the laptop's HDMI port, and the source is the same. If it is connected to the HDMI input, the signal can capture the picture in about 2 seconds; but if it is connected to the DVI-I input (through the HDMI to DVI adapter), it takes about 10 seconds to capture the picture. The difference between them is only that DVI has gone through a Scaler IC ADV7850. Because the source is the same, I don't think it is the unstable video that causes the delay.
I wonder if it is possible that the MCU code is not written well enough to cause the delay problem? Do you have experts who specialize in MCU to help us check it? Thanks for your support.

Best regards,
Oden

I would only suggest monitoring the ADV7850 lock bits, They should go to full lock shortly after the HDMI switch outputs to the ADV7850.  Just ping those bits at 1/2 second intervals and it will tell you how fast the 7850 locks to the incoming signals.

The fact it takes the capture 2 seconds to capture makes me question the real response of the capture hardware.  Your delays might be there.