ADV7188/ADAU1761 - HUM noise questions

In regards to the ADV7188, as long as you have proper layout and power rail filtering for the video chip, any audio hum should not couple into the video and any video noise should not cross couple into the audio.

I'm going to get this thread moved over to the SigmaDSP threads, They may be able to answer the audio hum questions better then I can.

Hi-

I moved this thread to the SigmaDSP Processors and Development Tools community. Someone here should be able to help with your question about the ADAU1761.

Regards,

AndyR

EngineerZone Community Manager

Thanks for your answer. In my mind, the PAL video signal can have a hum noise from the 50/60Hz. I don't know how to remove this from the CVBS without any disturbance : is there some typical schematic (analog filter) or digital filter ?

If 50/60 Hz noise is getting on your video signal then it's usually related to grounding issues within the system as a whole. That's why I mentioned layout and power supply filtering. Our eval. boards do not show this type noise.

Hi Laurent,

lberitau wrote:

- The input audio level of the board is -4dBm to +16dBm. What is the recommended circuit to handle this power range without quality loss ?

Audio on the PCB level is not typically referenced with dBm. dBm is a reference of power, and therefore requires a certain load to be defined. Audio signals on the PCB are generally referenced in terms of voltage measurements, like dBV, because they are independent of the load. Can you redefine your audio signal in terms of voltage?

I am having a lot of trouble understanding the system. What is the audio source? PAL CVBS signals only contain video data, not audio data. Where is the audio coming from?

If the hum noise is being generated by the video decoder, then the best way to isolate the audio signal path from the video path is to simply keep the traces separated on the board and use a solid ground plane with robust power supply decoupling for each IC. If you still have some residual crosstalk on the audio signal, it might be best to try to remove it within the DSP instead of externally with analog filters.

Hi,

You're right the dBm is a reference of power. In my application the audio input level is -4 dBu to +16dBu (for a 600 ohm load in balanced mode). If you prefer the input voltage is about 0.5V < VIN_rms < 5V.

In my application I use a video decoder (ADV7188) to treat the input video signal and I also use an audio codec (ADAU1761) to treat the audio input. So there is 2 connectors (one for video and the second for the audio). Next, both digital signals (video and audio) are processed by an FPGA to remodulate the stream in UHF.

I understand the need to isolate the input video signal and the audio input signal separately to avoid any crosstalk. But when the input signal has a 50/60 Hz hum noise, I don't know how to remove it because a rejector filter may destroy some useful information at this frequency (the luma for example). Tell me if it's wrong. I don't know if the hum noise is a real problem in audio/video because it's a new domain for me.

50/60 Hz hum in the video and audio is usually caused by ground loops in your system as a whole, not individual components. Before you can eliminate this hum you have to figure out it's source. Rejection filters may not remove the noise if it's a ground loop issue. Here's a couple of links that can possibly help you isolate the cause of the hum.

http://www.epanorama.net/documents/groundloop/

http://www.audioholics.com/tweaks/connecting-your-system/ground-loops-eliminating-system-hum-and-buzz

I'll let Brett respond to the audio questions

Hi,

Actually, video is an area where I have very little experience, so unfortunately I'm not much of an expert. I think the main point is that your audio traces will be susceptible to high frequency noise sources from digital signals on the board, but this can usually be avoided by following some basic PCB design techniques:

Floorplan (initial placement of ICs before routing traces) is very important. Try to keep ICs oriented so that you can route signal traces as directly as possible without using vias.

Keep the trace lengths as short as possible, wherever possible.

Properly terminate digital signals using source and sink termination techniques so that you can eliminate overshoot, reflections, and ringing.

(In almost all cases...) Keep a single solid ground plane so that all signals have short ground return paths.

Place proper decoupling capacitors as close to power pins as possible, creating a ground return path with the lowest area possible... This is your best bet in reducing EMI.

I have a few extra suggestions here: https://ez.analog.com/docs/DOC-1889

That being said, 50/60 Hz hum noise is usually not due to crosstalk issues... it is more related to power supply isolation and grounding problems. In order to remove that noise, you need to make a power supply design that is very robust and capable of converting from AC to DC while removing the AC component as cleanly as possible. In this case, I would refer to power supply design techniques from the company supplying your regulators.

With regards to the ADAU1761 and your analog audio levels, I wanted to mention that the ADAU1761's ADCs and DACs are designed to have an analog full scale reference of 1 Vrms when the power supply is 3.3 V. This means that your audio signal will clip unless you apply some kind of attenuation in the analog domain prior to inputting the signal to the ADAU1761.

Thanks you for your answers!

- I have already seen that the ADAU1761 input level is 1Vrms => Is there some equivalent device with higher input level (codec or audio ADC) ?

- I need to handle the 50/60 Hz hum noise in case of bad input signal coming from another equipment. This noise is not generated by my board but it's a requirement for my design to process a noisy input signal. It is the reason why I focus on this point. I though there is some typical schematic for professional audio/video device but may be not...

The ADAU1701 or ADAU1702 would offer a larger input voltage range: http://www.analog.com/en/audiovideo-products/audio-signal-processors/adau1701/products/product.html

They actually use a current-input scheme that allows you to scale an external resistor to allow for bigger inputs. However, I have never tested the ADCs with signals over 2 Vrms.

If the 50/60 Hz is coming in on a signal from another board, then unfortunately the only thing to do would be to have some kind of tuned filter on the input signal. That would unfortunately have the unintended side-effect of removing the signal information at or around those frequencies. At the moment, I don't know of any typical schematic that would be appropriate. I expect it might be a somewhat complicated 2nd order active bandstop filter.

Here's an example of that kind of filter topology: http://www.cepd.com/calculators/bandstop_ana.htm

Like I mentioned before, it might make more sense from a cost perspective to just implement a bandstop filter in the SigmaDSP.

Thank you very much for your advices ! You gave me some keys to solve my problem.

Hi Guenter,

I have some questions about the ADV7188 with a noisy CVBS input (AC coupled):

1) What is the chip behavior with this kind of signal?

2) Is the "Clamp operation" (page 30 of the datasheet) useful for this type of noise? Is an additional filter required in this case?

Regards,

Laurent

Hi Laurent,

1) Do you have 2 caps in serial in the analog line? The reference schematic already shows that the analog inputs are AC coupled through a 100nF cap. If there's a 60 Hz voltage component across the 75 Ohms load resistors, it will show up in the output image. The chip decodes what is across the load resistor.

2) The clamps limit the amplitude that the ADC sees. In this case the ADC can only handle1.6V range. If the input range is to big then you can change the diver network ratio. Looking at CVBS1 input it has a 19 & 56 Ohm input divider. If the raw CVBS signal is twice what it should be we can change the 19 Ohm to 28 ohm and the 56 Ohm to 47 ohm. This give us 1/2 the input amplitude while keeping the equivalent series line resistance to 75 Ohm.

An analog filter would be hard to built, it's got the same problems Brett outlined above. I would check your power source, is it really isolated, including the third pin. (I think France uses a two prong, unearthed socket)

Hi Guenter,

Thank you for your advices, you give me some important information. I will design a board and if you want I will sent you my schematic for a double check.

In France, we use 3 points connector with the earth pin for power source.

Regards,

Laurent

I'll be glad to look over your schematic. Also it would be good to look over the layout, layout is important for clean video.