the problem is OBVIOUSLY the buffer ic's
becouse the ack line doesnt go far enough down to close the gate on the buffer
it reaches from your pic about half as far to ground stretch the display to show low end transitions
and measure its time incrum
the slope isnt right for the ! ack levels
so its the dam buffers
it is a ! ack signal needed between 0.72 and 4.72 volts
on stfm i just measured the ack pin on the dma chip!!! it reaches on my scope .53 low and 5v high perfect
on ste with no corrected buffers i get at the port 0.74 low and 4.2 high {low end too high in value} not even right
but at the chip 0.68 and 4.75 {correct} but using the exact same ic i simply unplugged it from the ste
and moved it too the stfm why is there a .1v descrepancy in low and a .5v in high levels
becouse the ttl is getting in the way nothing more ffs....!
but using an 8 bit block transfer using 16 bit i get .72 low average and 4.13v average for a 16X16 block
they arent ack data at the right phase to then initiate a read/write cycle and this is the buffer ic's
as they introduce latency in needed read write phase and this makes a time displacment between !ack received at the dma
and the next read write phase allignment
early ste dma was in a socket
the stfm and early ste both use EXACTLY the same ic for DMA
made in the same place from the same ic's look on both machines good and bad at the die map
youll see its exactly the same
it saves face for atari just to increase the current to the ack line on the so called good dma chip
to then event a cycle of transfer than to ask users to change 4 ic's or two at a pinch
i think m8 you should change the buffers and report the results then
as well as before
i have three ste here
two didnt work with satan before i changed the buffer
one is the older dma and two the later one
so even the later chip will have issues
and this also shows bad in some hdd config's
if i place a hdd before the satan {i mean ultra satan} i cant see it on a good dma but the bad one see's it
replacing all logic as i said
all three machines work perfect as does satandisk and very well done !!! its a great tool
so... maybe someone else needs to check or jookie needs some act logic...??????
dont argue !! with 30 people change them then tell the other i think about 30 people so far have posted
we are all wrong please...?
ive been bleating about it for i think 7 or more years so far
to replace a 'bad' dma you can just use one from an stfm maybe its bad also ???
its the same die just the buffers added and there timings not taken into account on some lines
this topic is a phew bar for sure !!!!!
adding a different psu may also help as a dc regulated and non regulated supply
will affect the link between the power ground and atari ground images....
and linkage.... in terms of millivolts this matters
with respect to the buffers needs .... and there own intensions
i know for a fact some ttl was bad very bad even reported on the news
especialy amerian makers becouse of there international policy change
this happened {workers werent happy}
this grew in the late 80's early 90's
and affects some ttl batches .... {some makers maybe bribed people also maybe intel who knows !!!}
).
, if the so called "bad" DMA chip (C025913-38) in the early STE's is the very same as the DMA chip in the earlier STFM machines, then why do these not suffer any HDD /UltraSatan problems? 
No wonder they produced a replacement chip... 
