STE video chip

[exxos]

(Continued from previous page)

@fenarinarsa could you tweak your program to output all 1s to the DMA audio ? It may help me see what the timings are better to work out how much time there is to latch data. Or actually alternate 1 and 0 might also he useful.

Also any ideas on demos or music players or something which will let me do a sample audio listening test ? Then again even a sine wave would probably work to test.

[exxos]

Was just looking into other latch chips and the S374 says

Schmitt-trigger buffered inputs at the enable/clock lines of the ’S373 and ’S374 devices simplify system design
as ac and dc noise rejection is improved by typically 400 mV due to the input hyste

I don't have any type of 374 chips (loads of 373 though :roll: ) Will see if I can order some from somewhere and see if just swapping the chip for a slightly "slower" one can solve the problems.

EDIT:
Now I re-read that, it doesn't seem like the data inputs are schmitt :roll: Will order a S and LS type and just see if either help with the noise problems.

[Steve]

It's exciting to see you solving this so methodically :)

[JezC]

It's exciting to see you solving this so methodically :)

Waiting for a chair smack in response to this! ;)
:hide: :fire: :fire: :fire:

[Steve]

Oh no have I jinxed it?? :)

[exxos]

It's exciting to see you solving this so methodically :)

:lol: At least someone is keeping up with all this :thumbup:

The problem with the capacitors, while it solves the noise problem, it does not necessarily mean it will solve the problem on "every" machine. I guess capacitors could be added on every input but then it may start aggravating things elsewhere. I guess I could try some lower value pF caps if I can find them. Something like 10pF on every input may work well. But while this may work on my machine, it may not be enough for other machines. But I think I would rather solve it with a different chip which just has better noise immunity anyway. Only one way to find out I guess..

[exxos]

Waiting for a chair smack in response to this! ;)
:hide: :fire: :fire: :fire:

I actually don't know how I have got roped into solving another one of these stupid problems on these machines again. I mean the entire motherboard is like playing Russian roulette. Did I mention these machines were barely working out of the factory ? :lol: Did I mention this is why I designed a new motherboard in the end ? :lol: :stars: :chairsmack:

[exxos]

Finally did a before and after shot..

WITH 33pF

IMG_0214.JPG

STOCK (no caps)

IMG_0211.JPG

So just 1.00V to 0.720V was enough to solve the problem.

It got me thinking about if the ground would be enough to solve it also.

IMG_0217.JPG

IMG_0216.JPG

Did not solve the audio problem fully but was an dramatic improvement. There is a lot more " low-level noise" still on the latches than with the caps, but I guess as its only about 200mV anyway its no big deal.

I guess if better wire was used such as very thin multi strand wire shorter more direct wires used, it may *just* be enough to get around the problem. Am still going to try swapping the chips. Though I guess also it is how easy it is to fix this problem because of people's "limited" skill levels on removing chip successfully. I guess the bodge wire route may be the easiest thing to try first. If it does not solve the problem then adding capacitors will undoubtedly will. Choices choices...

[exxos]

Found some stranded wire and rearranged how it was connected as well.

IMG_0218.JPG

IMG_0217.JPG

IMG_0223.JPG

0.74V so not really any different than before. But probably worth trying for a "easy fix" at least. But as said previously, this probably won't solve it for everybody as the whole thing is very borderline. Hopefully when my new chips, it will present as a better solution, albeit, slightly harder to do.

The video is still going batsh*t crazy.. Need to chase up my buffer boards... Also oddly a few pF on the 32Mhz clock seems to now help. I think as the machine warms up the 32Mhz line is to inherently become more unstable. I placed 33pF on it ( it is all I had to hand) and the bulk of the corruption is gone again. The " slight random corruption" still seems to be originating from the 16Mhz clock. Why do I feel like I am going down the rabbit hole again with all this stuff ? :lol: :roll:

IMG_0221.JPG

EDIT:

For PCBs with below options will start to produce on February 8

:cry:

[exxos]

@sporniket How accurate is the flood fill on your board vs the original ?

Capture.JPG

EDIT:

Tried another GND route based on the floodfill in the above image.. Same results are previous. about 0.7x Volts, audio seems mostly calm again now, but the video is messed up again. Though not really surprised.

IMG_0226.JPG

IMG_0227.JPG