P2 decoupling - as seen on Hsync signal
Tubular
Posts: 4,622
I noticed some RGB video signal leaking into the Hsync signal, while testing HD VGA these past couple of days.
At first I thought it might be IR drop related, then I noticed it was worse with text and lines (=high freq transitions), because the graphics tests with solid graphic blocks didn't have the same issue.
So, I'm guessing its decoupling cap related (open to other ideas though). I think one thing to try is to replace one of the two VIO caps in a block of 8 pins (= 1 LDO regulator) with a much smaller value, eg 47nF and see if it helps. So there would be 4u7 || 47nF.
This may be also affecting analog sigma delta operation
--
Here's the scope shot of HSync (yellow trace) when there is text or lots of random lines on the screen
And here's the scope shot of HSync when there is solid blocks of color on the screen. Note there's only a spike where active video starts and stops.
In both these the blue trace is a zoomed in, ac coupled 1v8 rail. It also could do with some decoupling, or improving cro grounding
At first I thought it might be IR drop related, then I noticed it was worse with text and lines (=high freq transitions), because the graphics tests with solid graphic blocks didn't have the same issue.
So, I'm guessing its decoupling cap related (open to other ideas though). I think one thing to try is to replace one of the two VIO caps in a block of 8 pins (= 1 LDO regulator) with a much smaller value, eg 47nF and see if it helps. So there would be 4u7 || 47nF.
This may be also affecting analog sigma delta operation
--
Here's the scope shot of HSync (yellow trace) when there is text or lots of random lines on the screen
And here's the scope shot of HSync when there is solid blocks of color on the screen. Note there's only a spike where active video starts and stops.
In both these the blue trace is a zoomed in, ac coupled 1v8 rail. It also could do with some decoupling, or improving cro grounding
Comments
One thing I was wondering, with my setup there's about a 20~25cm long section of ribbon cable from P2-Eval header to VGA socket. From there it goes through a 1.5m decent quality coax vga cable to the monitor. But it occurs to me there could also be capacitive coupling in the ribbon cable, and I should solder something more shielded to eliminate this possibility
-Phil
Definitely possible that some noise is due to lack of ground plane to VGA header
I was thinking we should construct a test screen which has vertical bars at different spacings/frequencies
There's still a small residual amount of crosstalk when the cable is unplugged altogether, but it looks like it is indeed mostly crosstalk. Thanks guys
You could try disconnecting the vga and making a connector that plugs directly into the board with termination resistors and see if it goes away completely.
I'm keen to see how much an intermediate ground reduces crosstalk. Will give these ideas a try tomorrow