Vectrex Generations differences

[Malban]

Hi there,
I am still trying to enhance emulation by examining my vectrex.

I think I have a 2. generation vectrex (SN: 3036448)
and a 3. generation vectrex (SN: 3117403)

but I have not opened them to see what actual hardware is inside.

After reading some older messages (thomas' mainly) I wrote a testprogram for shiftregister testing.

Vectrex 2. Generation displays:

Vectrex 3. Generation displays:



I would STRONGLY be interested in other vectrex output, especially with serial numbers (and if you per chance have opened the vectrex, what version of VIA is used)

Here a link to the testprogram: ShiftTest_0.bin

I really would appreciate some feedback!

Thanks

Malban


PS
Basically the program draws a vector from the middle to right using a scale of $7f, the first shiftreg is "complete" with 18 cycles.
The 2nd shiftreg is cleared after different numbers of cycles from bottom to top after 18,17,16,15... 10 cycles.
What you see is for 2 Gen, that VIA basically ignores clearing of the shiftregister if you clear it after exactly 15 cycles.
and for 3 Gen, that VIA basically ignores clearing of the shiftregister if you clear it after exactly 14 or 15 cycles.

Actually  ignoring is a bit "weak" the shift reg just refuses to shift anymore and outputs the last known value ...


PPS
The "thing" is that in Thrust the title logo "shiftreg" is cleared after exactly 14 cycles. On emulation in 3. gen vectrex it looks really bad. But as of now I have no real hardware cart of thrust (Packrat - hurry up! :-)). And somehow Thrust refuses to work with VecFlash and with Mateos device...
Anybody noticed something "odd" with Thrust who owns a 3. Gen vectrex?

[christophertumber]

I had listed some of my machines in this thread:

vectorgaming.proboards.com/thread/972/hardware-tests-wip

Let me know which machines you'd like me to run this test on.

[jfmateos]

Serial number: 3059228
BIOS checksum: 7931



Serial number: 3013113
BIOS checksum: B746

[thomas]

Well, my no-buzz also shows one long line, #3111385.
I've come across two different 6522 varieties so far, the majority actually indeed does not lock up when using 16/17 (?) cycles instead of 18
(would have to check which one it was, there was a bug because of this on one tester's machine when I accidentally reprogrammed too fast. but on noone
else's), but some do. My initial concern was more the delay in cycles once it's shifted out for a fast textroutine forward/backwards, I've come across 3 different
cycle varieties so far, the slowest is three cycles slower than the fastest chip, and a few have a small but noticeable sub-cycle delay.
There were quite a few manufacturers of 6522, I saw a documentation from one clearly stating that they put 'fixes' in place for the shift reg bug that
'the other manufacturers do not have'. So I expect there are more behaviors out there I haven't seen yet..

Not sure how these carts do bankswitching of 32k roms, did you try doubling the 32k and use it as a 64k rom ?
The eeprom code inside Thrust will kill it otherwise on a 32k w/o automatic doubling but run fine otherwise.

[thomas]

Thrust crashing part 2: or you put an 'rts' at the start of the eeprom load/store, they are at:

eeprom_load LDX #eeprom_buffer *7EAC: 8E C8 80 '...'
eeprom_format LDU #eeprom_buffer *7EC5: CE C8 80 '...'
eeprom_save LDX #eeprom_buffer *7ED8: 8E C8 80 '...'

for the 'old' version and at

eeprom_load LDX #eeprom_buffer *7E99: 8E C8 80 '...'
eeprom_format LDU #eeprom_buffer *7EB2: CE C8 80 '...'
eeprom_save LDX #eeprom_buffer *7EC5: 8E C8 80 '...'

for the recently fixed version, where the bonus game does not crash
(three guesses who found a fix disassembling/poking it)

[hcmffm]

My no-buzz Vectrex (SN: 3108554) shows two long lines (5th and 6th line from top).

[Malban]

Thx for your tests, they seem to be more or less consistent.

@christopher:
Would you mind testing with your oldest Vectrex? The one that gives the strange results with Spectre?

Here a screenshort of Thrust on my No Buzz:




Here you see, that the Thrust logo is also "extended", this seems to happen since ShiftReg is cut off at cycle 14...

Regards

Malban

[thomas]

yes, the failure 'lock' mode is in far less cycle spots than i thought, interesting. But it's not Vectrex 'generation' dependent, after all my no-buzz doesn't show
the 2 lines and I bought mine from the original owner; And he never changed a chip inside, never opened it up.