Hello, I have a costumer that needs to know if PAL M inputs are truthfully tested by ADI on ADV7282.
Since it is a Brazilian only modulation standard, It might have never been tested.
I received reply from specialist.
test team have have fed Pal-M into the ADV7282 without any difficulties in standard autodetect mode.
The ADV7282 may not autodetect PAL-M in fast switch mode. If the customer is using PAL-M in fast switch mode then they must manually force the standard using the vid_sel bits.
according to datasheets and design support files ADV7282 and ADV7282-M Design Support Files
It is supported.
I can check with the specialist to see if any additional information can be provided.
Thanks for the answer, indeed there is PAL M info on the documentations you informed. I had also researched for this before posting, and my costumer is going to buy our kit, but he would like to know from ADI if they somehow where tested. He is asking for that information because he had an older kit that was sold by ADI and in this kit the decoder is from ti. I have seen myself that one, and have pics that I attatched. In the decoder datasheet they state compatibility with PAL M, but in practice we get jittering and noise when the input is PAL M.
Now with the ADV7282 evalboard he is going to do the same decoding - encoding test, and he is affraid to get the same issues. It is a good chance to take out the competitors chip, but I was looking for some addicional information in order to be sure.
Thanks for your attention.
Thank you Dominic! I will deliver that information to my costumer.
This is the report I have from Tecys on the EVAL-ADV7282EBZ:
Standard Video Pattern Generator using Composite input on ADV7282 and output of ADV7390 using composite also, configured as:
REG 80 = 0x02
REG 82 = 0xC3
REG 87 = 0x00
REG 88 = 0x02
All other configurations are factory default.
The input signal from the generator is PAL-M with pedestal. He mentioned the output result in a dynamic image is smudgy on the transitions.
What I need is some directions on this support, if there are registers we can modify the values in order to make it work better. I can provide all details needed and also I will go there and try to make it work by myself but any remarks would be helpfull.
Attached there is an example of the EVM output multiburst circle and the generator pure output. I have other images as well but these are the very best distortion example.
Juliano K. Cioffi - FAE
Latinrep South America - Analog Devices
I am the applications engineer for the ADV7282. I will ask my colleague about the encoder settings that you are using.
All analog devices video decoders (e.g. ADV7180, ADV7182, ADV728x models) all support the PAL-M standard. The PAL-M standard will be automatically detected when the video decoder is programmed with our standard autodetect scripts. Standard scripts are available here: ADV728x Design Support files
So yes using our standard scripts the ADV7282 evaluation board should work fine.
If you use the fast switching script (this is a mode generally used in Automotive applications to allows for faster lock time) then the video decoder will have to be manually programmed in order to lock to the PAL-M standard.
This is done by programming the VID_SEL bits. See the datasheet for more information.
Could you please share some output images from the ADV7282 evaluation board with examples of this 'smudging' the customer is describing.
I suspect that the smudging is caused by 2D comb artifacts in the ADV7282. The ADV7282 is designed for automotive applications where lock time is critical. It therefore uses a 2D comb luma/choma separation filter.
Better picture can be obtained using 3D comb luma/choma separation filter. This however adds cost and delay. Parts such as the ADV7850 have advanced 3D combs to remove artifacts.
There is not alot that can be done to remove 2D comb artifacts i.e. 'smudging'. I will know more after seeing the output images.
Could you also please read User Map registers 0x10-0x13 of the ADV7282 and share these results.
Thanks and Regards,
Senior Applications Engineer,
Analog Devices Inc.
Thanks for the answer. And the attention. Here follows the information required:
Customer is using DVP 10-14-11 --> ADV7282CUST -->Script 02 CVBS Single Ended Autodetect.
0x10 = 0x1A
0x11 = 0x42
0x12 = 0x00
0x13 = 0x5C
First one is the output from generator and the second is the output from the eval. You need to look into the circle in the yellow bar, where the distortion is more apparent.
Unfortunately this would be the expected performance of the ADV7282. You can try and modifying the luma and chroma filter responses in order to get a better result. See luma filter and chroma filter sections of the ADV728x hardware manual http://www.analog.com/media/en/technical-documentation/evaluation-documentation/ADV7280_7281_7282_UG-637.pdf
However I doubt you will get any meaningful improvement in performance.
You can use a part with 3D combing such as the ADV7842 or ADV7850 in order to get better YC separation. This involves storing frames of video and analyzing them. This usually involves the use of an external DDR2 memory chip to stores the frames of video for the video decoder to analyze them. This adds cost, delay and complexity to the system. However the YC separation is considerably better.
Note I am surprised that you have any video output. Register 0x10 equal to 0x1A means that the ADV7282 has locked to the color subcarrier but not to the horizontal sync. Register 0x13 also shows that the ADV7282 has gone into free-run mode i.e. ignores input and outputs a blue screen output.
I did not check these register results,I should have done before. That explains a proper functioning of the board because he did not conect input source. The generator is being used for testing procedures in production and will be busy until Thursday.
Since the decoder and the eval is working in the expected performance, and by the recomendations you made, I will just set up a meeting for the monday aug-24 and there I will try to set up some better performance. I will write the registers down and will post it here again just for make sure.
They are very satisfied with all the support, customer mentioned they are re-designing that product I mentioned in the e-mail and placed our DAC. He also mentioned that our ADV7282 works much better than our competitor, and still they are thinking about using ours.
I have learned with this some good lessons about video and performance. It was great to have your support and Dominic's help in the beggining of all. I will let you know more after my visit.
Thanks a lot.
I just came from a visit in our customer and was a very good visiting. We talked about other important projects.
I tested the evaluation and saw personally the performance of our decoder plus encoder evaluation. Was pretty good.
Here follows the registry settings:
ADV7391 - 0x56 I2C Addr
ADV7282 - 0x42 ||
Subcarrier - 1458562593 or 3824150049
We loaded the default script from the DVP 10-14-11: Autodetect_CVBS_Single_Ended_In_Ain_1_YPrPb_Out
I am curious about one thing: When we tested with an NTSC input we had perfect response. With PAL-M input and a color bar together with a multiburst there were two lines of a little distortion in the intersection of those images that was not happening with NTSC. I tried to chance the registers of Address 0x17[4:0] by asserting Address 0x18 as well 0x17[7:5] and I found some better performance using the filters but they were just only visual. With the analyser all the filters were much better with factory default options and the changes listed above.
I left the board there in case there is a possibility for making a better adjustment please let me know. I really got curious about the performance difference on NTSC.
Would you explain this difference for me please, I got curious.
I would have to see an image of the artifact. PAL has a slightly higher resolution than NTSC so usually it has fewer artifacts than NTSC.
Sure Robert, I will follow up the visit and soon I will post the two images from PAL-M and NTSC so you can take a look. As soon as I get it I will post them here, and let you know. The pics I took unfortunately were saved in low resolution so I have to wait for my customer to take some new ones.
Thanks a lot and have a great week!
Has been a little time since I last posted here, but our customer decided to produce their boards with our competitor IC.
Great news: They left it behind. We beat the competition there. He just sent me the images on NTSC for making a comparison with PAL-M.
Making a comparison with previous image on the circle and yellow bar, it is possible to see that there is a little less distortion in the intersection.
This is my first victory against competition in an already designed project and we started this process in my very first month on ADI. I am glad we got better results from the ADI line. Thanks for your dedication.
The point now is to see if we can find a better adjustment for working with PAL-M that can be a little similar to NTSC performance.
Look forward hearing from you!
Juliano K. Cioffi - FAE
Other than the comments I made previously, I do not think that there is much that can be done to improve the performance of the ADV7282 further.
I sent you an email discussing other options.
I think we should close off this EZ post and discuss more over email if necessary.
Thanks a lot for all the support!
Juliano K. Cioffi - FAE
Retrieving data ...