H4 & H5 series BGACK pullup update fix

[exxos]

MANDATORY FIX

I am calling this fix recommended as it is known to cause issues on original machines, but at the time of writing there are no known issues reported on the H4 or H5 platform relating to this resistor value. It could possibly cause DMA issues with the hard drive or non-critical issues with the blitter.

In light of recent developments in @ijors blitter investigation , It has been realised that some of the pull-up resistors on the H4 & H5 platform may have not been the ideal values.

@ijor Has generously tweaked his program for us to test if any boards (including original machines) which have a inadequate BGACK pullup resistor. Typically this will be 2.2K. In Atari's own documentation on the later STE machines this was updated to a 1.2K resistor. But in order to use simpler /more common value, I would suggest a 1K resistor pack.

The H4 had the resistor missing and there was already a bodge resistor fix in place for this which I suggested would be 2.2K. Which should be swapped out for a 1K value. But it is also possible people fitted other values.

On the H5 series the resistor pack is here. RN12.

RN12.PNG

It will probably be a PITA to change. So placing a regular 0603 2.2K value from BGACK to 5V may be a simpler solution. It will give 1.1K pullup which is also fine.

bgack.PNG

I think because the resistor was missing on the H4 , I probably "forgot" to look into its value. So the typical value of 2.2K was used. Because no issues were reported , I went along with this value on the H5 builds as well. When in actual fact this resistor was modified on later STE machines by one of Atari's own fixes.

IMG_20151024_095629.jpg

Of course we are back to the original DMA investigation again. Where those 33pf capacitors did not resolve the problem. That particular problem was resolved by changing the pull-up resistors on the address and databus etc. Where IIRC D4 always seem to suffer from the most noise on original machines, at least on the STFM series anyway which I saw a lot. Main point is the 1.2K resistor value anyway.

[exxos]

I have changed this to a mandatory fix as I have found a test case involving a H5C1 with blitter and THIS Simple booster, which causes corruption when blitter is turned on in GB6.

The fault shows up in 16MHz mode when the blitter is active. When loading GB6 (blitter off) screen is fine. Turning blitter ON , most times it totally corrupts the GB6 screen or "randomly" places text where it shouldn't be etc. ( and after fitting the resistor and removing it again, it now does not happen, which is typical when I want to take a screenshot!!)

Oddly in a previous tests, if I booted in 8MHz, turned the blitter ON, then ran tests, it was also fine :shrug: In anycase, the system tends to crash with the blitter turned on when running GB6 tests.

With a additional 1K (as I used for tests as I was to lazy to go find a 2.2K) The problems went away.