[Diy_efi] Continuing the chip switcher "problem"

[mbelloli at speedymotorsports.com]

     I took a look at a datasheet for the 74xx374 and think I understand
what is going on.  You put your new request of change of data in, and
on the next positive slope of the clock, the old data will be
switched to the new data.  So until the clock rises, the old data
will be there.  And if I'm getting this right, the data is only read
at the negative transition of the clock?  So you will have done the
change during the time you aren't performing a read.  And the system
will be no wiser.  Even if an instruction was in being read, and not
data from the chip it will still be transparent.  Hence the
Transparent latch name.  Could you look at the Not PSEN signal I've
taken from the MCU at pin 23, and the Not OE pin as I've recorded
them on my webpage.
           http://home.comcast.net/~hexibot43/Waveforms.htm
     Is it possible that although the signal never reaches Vss or ground
the computer could still be reading it as Vss or logic state low?  Is
it possible that when it reaches the 2.5 v median point the computer
accepts that as ground?
     I pulled another like computer apart that I hadn't modified in any
way, and the logic of the pins are the same.
     I have checked the ground repeatedly on my scope.  I've tried using
the chassis ground.  I will tomorrow try backing off the time scale
on the scope, becuase 5us/div is the absolute setting on the scope. 
In fact it is a two trace scope, but when taking reading at this
level turns the second channel off.
     I'm thinking of putting one of those 74xx374 in place, as a test and
give it a state change at the inputs, and see if the outputs ever
change.  If they do, perhaps the signal is there and I'm just missing
it.  So if I've got this right I'd run the signal at Not OE into the
clock in of the chip, Hold the output control low, and apply my input
to the D line.

Thanks,
Marcello

You know I can't talk to anyone I work with about this, they just don't
understand. It is so nice to talk to people that know more than me.



> 1. Check ground connections between scope and ecm carefully
> 2. More importantly, if the time scale is 5us/div on your plot and this
> is the best your DSO can do, don't use it, it is not good enough for
> anything more than audio.
> 3. In any case, if there is a changing signal on /OE use the rising edge
> of this to clock your upper address lines into a register like a 374,
> not a 373. This will cause the address lines to affected memory to only
> change when the memory is not being used. If all program code in the
> multiple images is the same and only tabular data is different this
> should make the switching transparent to the cpu.
>
>> More info -  The Not OE line of the eeprom is connected to the not psen
>> line at pin 23 of the MCU.  Now further looking at this line, there is a
>> 10kohm resistor attached between this line and pin 63 of the mcu which
>> is
>> the Vref line.  Pulling it to Vref.
>>
>> Not(Psen) - Program Strobe Enable : The strobe pulse to fetch to
>> external
>> program memory.
>>
>> I'm guessing this is the reason the signal looks so strange.  I'm going
>> to
>> take a look at the signal tomorrow when I can plug it back in the car.
>> I'll take some readings right at the MCU.
>
>
> --
> Robert W. Hughes (Bob)
> BackYard Engineering
> 29:40.237N, 95:28.726W or perhaps 30:55.265N, 95:20.590W
> Houston, Texas "The city with too much Oxygen"
> rwhughe at oplink.net
>
> _______________________________________________
> diy_efi mailing list
> diy_efi at diy-efi.org
> http://lists.diy-efi.org/mailman/listinfo/diy_efi
>

_______________________________________________
diy_efi mailing list
diy_efi at diy-efi.org
http://lists.diy-efi.org/mailman/listinfo/diy_efi