OT: UM6502A can it be different in any way?

Hi all,
I'm sorry for the off-topic, if anybody has a better idea on where to post
such a question, please suggest it!
A long story short: I'm repairing a few Atari Asteroids arcade PCBs. This
game uses a 6502A clocked at slightly more than 1.5 MHz.
Two of the boards I have, were populated with a SY6502A and worked fine.
One board was populated with a R6502 (non-A) and anyway worked fine too.
However
I thought I'd use the correct speed-grade CPU, so I put in it an UM6502A.
While the game never
crashes, with this CPU the vectors are randomly plotted in the bad
position. For example, some of the rocks have a side "glitching" between
the right and a slightly offset position as it moves.
Some of the letters drawn on the screen are also "shaking" in the same way.
I've tried a "chinese remarked" 65C02 that I'm not sure what it was as the
old markings have been sanded away, but it's some sort of 65C02 because the
NMOS unconnected pins are in this chip
actually connected to something and the chip has protection diodes to both
supply rails (unlike NMOS 6502s). However the unknown 65C02 also works fine.
A genuine MOS6502 also works fine.
Another "chinese remarked" chip shows the same glitching/shaking behaviour
on random vectors, this one is some kind of NMOS 6502 but also this one has
had every original marking sanded away. Maybe another UM6502?
Since I have two working Asteroids board (one is -04 revision and the other
is -02), I tried also the same CPUs on the other board and they produce
similar
glitches but with different amount of shacky vectors. Board -04 show much
more
glitching vectors than the -02, however, all the other 6502s work fine on
both.
Is there any known difference of the UM6502? Any different drive capability
on data/address maybe?
I'm more inclined to think that these boards do load some output slightly
too much, I don't think it could be a software issue.

Best regards
Frank