TVP5147 Image Artifacts

Hello Andrew,

I'm really not seeing any significant differences between the two images.

The TVP5147 is not recommended for new designs.  You really should be using the TVP5147M1.

The TVP5147M1 now supports square-pixel mode using patch.  See patch download web site below.

     http://software-dl.ti.com/dsps/dsps_public_sw/dsps_swops_houston/ANALOG_VIDEO/Analog_Video_Decoder_Versions.htm

Also, strongly recommend that you do not attempt to use manual coarse and fine gain controls.  If you want to adjust the overall gain, please use the contrast control.

Regards,

Jim

Hello Jim,

Did you view the image at full size? In particular if you look at the innermost black circle, the reflections are visible as lighter arcs. These appear on an otherwise identical image just by disconnecting and reconnecting the video feed. Then while the feed is connected, if I reset the chip and apply the same exact settings, the image is again identical without the reflections.

As for the TVP5147M1, I had already designed with and tested my design before the P rev had gone nrnd, and we had done a large proportion of our visual image qualification testing before the M1 had square pixel support, unless I can be given any clear reason to move to the the m1 I can't afford to restart all our testing. We're not worried about supply of the P rev, we're a low volume manufacturer, and even if you stop supplying it rochester electronics has a huge stock of these parts.

Why do you say not to use manual gain controls? If they shouldn't be used, why are they provided? As mentioned, we need strict computer control over the image settings, without having to adjust for different agc settings. We are using both gain controls and contrast, contrast alone does not give complete control. I've verified that the video signal is well within the max p-p voltage range of the input. Also, we're using this circuit to replace an existing video capture card which did not have auto gain. It is a requirement for this replacement to match the image produced by the existing card as consistently as possible.

Thanks for the reply,

Andrew 

Hi Jim, 

Thanks for the suggestions. I've been swamped by some electrical compliance issues the last few days so haven't had a chance to get a unit set up for this test yet.

I must admit I don't think I ever tried enabling the autogain to see if it removed the shadowing, although I'm not sure if it would be a suitable long term fix, as the process of setting the manual gain can be enough to trigger the problem, so as soon as I disable the autogain and set the manual gain I forsee that may trigger the issue. I'll certainly try it though as it could be very useful to know.

Now that I think of it though, I guess as part of my startup procedure I disable autogain and set manual gain, and that appears to be working, so perhaps your suggestion will be exactly what is required to fix it for my application.

Thanks,

Andrew

Hi Jim,

I've had a chance to test the autogain, unfortunately there's not change seen. I set up a physical switch to switch between autogain, then back to manual gain. With the shadows present, switching to autogain or back again made no difference whatsoever. 

I did find however that switching the input channel to an unused channel, then back to the correct one would sometimes fix it. It seems somewhat dependant on how long I leave it on the other channel, ie switching there and back quicky (less than half second) only fixes it about 1 in 5 times, leaving it switched longer (2-3 seconds) fixes it almost every time. It's not a reliable fix however, even waiting for 5 seconds didn't fix it every time, and the full chip reset is much quicker than that (about 2 second process to video being active on screen).

I also had some tvp5147m1 on hand so replaced the chip on one of my test boards. Once the patch code loading was implemented I tested it, and I actually can't seem to reproduce the shadowing effect at all, so it's either fixed or at least  vastly reduced in the M1 rev parts. The captured image however is somewhat different to the P rev part, I"d say it's slightly better, with less noise. While this would normally be a very good thing, unfortunately due to the software testing being finished with the current hardware, there's no chance to upgrade the part without invalidating a good part of the testing, deadlines mean retesting isn't going to happen for this release. I'm going to have to stick with the full chip reset for now I think, and just be ready to move to the newer chip in future if/when I get a chance.

Thanks,
Andrew 

Did you ever sort this out, we have a similar problem...

Hi Peter,

It's quite interesting to hear someone else noticing this problem. Are you using the tvp5147p or something different?

I never had any real resolution other than what I posted here, ie after gain changes I quickly reset all the other chip settings. This does cause a dropout of the video briefly, such that there's a video pause every time the setting is changed.

But to be honest I was never really happy with this resolution, it was just the best I came up with before having to move onto other tasks. Looking at my code again now though, I think I stopped doing even that, must have decided the ghosting wasn't bad enough to warrant worrying about.

Strangely coincidental you ask about this now, as we've been having an issue during calibration of the unit this is based on, and now I'm reminded of this problem... maybe it's got something to do with our current troubles.

Andrew