8192 baud 1227808 project.

[Peter Gargano <peter@techedge.com.au>]

Sorry for the length of this post, I got carried away...

Ross Myers wrote:

Subject:   Re: OFF-LIST: Non-efi question
> Peter, how's that 8192 '808' project going?.

Well, as we know, a 1227808 is a 1227165 without 8192 baud capability.
Actually, an '808 also comes in a shorter case than a '165.  Anyway,
U2 is missing on the '808. To get 8192 you could drop in U2, and this
is eminently do-able, as long as you get pin 1 correct :-). There is
a slight problem that some Australian boards don't have 5 SMD capacitors
and 6 SMD resistors, and it makes the job a lot harder (but not 
impossible), but there are another couple of ways out...

1. the 1227808/1227165 etc. has three unused I/O pins B4, D13 and D14.
   (I have recently confirmed this from an "inside" source). It would
   be possible to use one of these pins (say B4) and the existing pin
   A8, as the Rx and Tx signals for 8192 signals. It makes sense to use
   A8 for the Tx signal as this is naturally the 160/8192 OUT signal when
   just the 10k diagnostic resistor is used between A9 and GND. The
   B4 pin then becomes the 8192 Rx (ie. the processor's Rx) signal.

   So, what we have is a non-multiplexed Rx and Tx signal - this would
   be fed into (say) the Rx and Tx pins of an externally mounted MAX232
   (or similar circuit), and EFILive wouldn't know the difference, as
   long as you're using an 8192 aware memcal. The MAX232 circuit has
   been described before, but the lines are not connected at the ALDL
   side of things as they are for "normal" 8192.

   Now kiddies, don't try this at home without adult supervision. I have
   not had a chance to actually try this out yet on a moving vehicle, but
   there should be no problems with this approach as long as the wires to 
   the interface circuit are kept as short as possible - perhaps even 
   permanently wired a couple of inches from the A/B connector. The reason
   for this is obviously to ensure a high level of noise immunity that 
   U2 would normally provide.

2. An alternative 2 transistor circuit, not unlike Sanjaya Vatuk's (see
   http://www.andywhittaker.com/ecu/aldl_hardware.htm circuit can be 
   readily fitted into an '808 - again I have not finalised values for
   my recommended circuit. The '808 would then be functionally identical 
   to a '165 other than lacking capability to be combined with other ALDL 
   aware devices (ie. ABS, etc. - this is because U2 does some nifty 
   sensing that a simple 2 transistor circuit can't do).

   This approach is both cheap and avoids having to solder a 28 pin
   SMD and those SMD R's and C's mentioned above.

> I showed a friend EFIlive running on his VS work ute, he was very impressed,
> but sad to hear that his VP Senator was unable to achieve such things...

The other part of the equation is to either:

a. Update to a newer memcal that has 8192 built in - but as you point out,
   none of the Australian 8192 aware memcals has auto code. Not a problem if
   you have a manual!

b. Retrofit the 8192 aware code into an earlier calibration. This approach
   has a number of advantages such as not having to do any tuning if your
   current calibration is working for you, and not having to worry about newer
   things like anti-theft stuff (but OTOH anti-theft ...).

   I think this is really the way to go, considering that there are so many 
   different memcals that differ in just the data, and retain exactly the same
   code (ie. The Camira 2.0 and VN 3.8 both use the same code AMXY/ASBX, etc.)
   So that once the HC11 code is written, then a little program could be 
   devised to read an "old" calibration, add in the 8192 aware code, calculate 
   a new checksum, and write out a new BIN - 8192 baud ALDL for almost free.

Approach b. also has the possibility to add in a Kalmaker (sorry Bruce for
using that word) style tuning-on-the-fly approach. And this only requires a
modified memcal to work too! 

More on this tuning-on-the-fly (TOTF) approach when I've actually got it all 
working...

regards, Peter
----------------------------------------------------------------------------
To unsubscribe from gmecm, send "unsubscribe gmecm" (without the quotes)
in the body of a message (not the subject) to majordomo@lists.diy-efi.org