Don't suppose you have the parts lying around to make a few more SW/ESB kits, do you Clay?
Sent from my iPhone
On Feb 22, 2011, at 10:50 AM, "Clay Cowgill" <c.cowgill@comcast.net> wrote:
Worth noting that there is a checksum in the data written to the 2212. If the checksum and data do not match, the NVRAM is wiped clean by the code on the next startup. I'm still kinda liking the "more capacitance on +5EAROM" idea for the time being. Of course there might have been a reason why Atari had that value so low (like to prevent the NVRAM from trying to power the rest of the board from +5EAROM back through the address/databus pins or something).
I guess you could always just double the value of C93 (for twice the load) and be relatively safe (as opposed to 'too much').
Another thought-- do you have *another* new 2212 to try using? Maybe the one you have is just particularly finicky? Or also possible-- maybe your C93 (or C94 mod) has significantly degraded and you're just on the bleeding edge of "good enough"...
Mark's idea about the C94 mod is a good one too.
-Clay
From: owner-vectorlist@vectorlist.org [mailto:owner-vectorlist@vectorlist.org] On Behalf Of Mark Shostak
Sent: Tuesday, February 22, 2011 8:42 AM
To: vectorlist@vectorlist.org
Subject: Re: VECTOR: Assistance with Star Wars w/Clay's ESB Kit --- SW novram clobber?
Or... Here's a completely different possibility:
Have you tested or verified the presence of "anti-corruption capacitor" (my words) C94?
This cap "...maintains E2PROM data integrity..." "...during power-up or power-down...", quoting Xircor.
It would be interesting to get a measurement on this cap, as it could potentially explain a lot of "flaky" NVRAM behavior in SW boards.
HTH
-Mark
On Tue, Feb 22, 2011 at 9:38 AM, Jess Askey <jess@askey.org> wrote:
So if switching to ESB during power off does it then the only things I can think of would be that SW is actually getting corrupted while ESB is writing to the NVRAM. So... try this...
a. Do what Joel suggested first
1. put in SW mode... put up a high score.
2. Power Down.... remove the SW NVRAM (on Clay's PCB if I read that right)
3. Switch to ESB... turn on game
4. Put up an ESB high score
5. Power down... replace the SW NVRAM
6. Switch to SW (don't forget this or you will have to start over)
7. Bring up SW and data should be good still.
So, with that, like Joel said on his check, you can see if it is chip specific then you can see if there is something funky in my process. If my process works and SW memory is still OK, then there must be something going on with the chip select between the two NVRAMs which is letting ESB write data to the SW NVRAM... when SW is powered back on and the data isn't checksumming correctly, it will restore it to DIP settings and overwrite it. If the memory in SW is still corrupt after this process, then it has something to do with the powerdown with the select switch being in different positions.
??
jess
On 2/22/2011 8:22 AM, Joel Rosenzweig wrote:
It would be worth swapping the two 2212 chips between SW and ESB to see if the failure moves with the chip or stays with the socket. If it moves with the chip then that's one debug path. If it stays with the socket then that's a different one.
Joel
Sent from my iPhone
On Feb 21, 2011, at 11:56 PM, "Altan (altan@aol.com)"<altan@aol.com> wrote:
Well, I think the reason this is hard to figure out is because the behavior is just plain screwy.
I replaced the SW 2212 a couple days ago and it definitely started storing settings.
For several days, it would power up with free play and my settings.
About 4 hours ago, I got a high score (yeah! ... #3). I turned the game off.
I just turned it back on and the settings were lost (as well as the high score). Boo!
So we can most likely conclude from this that switching from ESB to SW isn't the root cause. Maybe I was just unlucky that the failure occurred at those times --- leading me to assume they were correlated.
What's odd is that the novram test passes. FWIW, the novram test also passed with the original 2212.
What does that tell us when the novram test passes? Anyone know what Atari actually did for the novram test? Did they write a pattern into the ram and then tell the chip to store it, then clear the pattern from ram and then ask the chip to restore it from novram? I'm wondering if knowing why the novram test passes might help explain what I'm seeing.
Thanks again to everyone is hash chimed in --- or will do so :)
... Altan
On Feb 21, 2011, at 11:46 PM, Altan (GAPAS) wrote:
Joel, you have it correct.
SW saves fine.
ESB saves fine.
Switch from ESB to SW (with power off) and SW loses settings
... Altan
On Feb 21, 2011, at 11:40 PM, Joel Rosenzweig wrote:
If the timing was wrong, wouldn't it fail to save the score in that 2212 all the time? I thought he said that SW saved the scores now, but, when switching back from ESB, SW was re-initialized.
Maybe I'm misreading him again.. don't know.
Joel
On Feb 21, 2011, at 11:33 PM, Jess Askey wrote:
I did lots of fanagling with the NVRAM on my I,Robot when it had trouble saving settings and indeed, Atari seemed to balance the /SAVE line with the timing of the +5V NVRAM supply *just right* to make the NVRAM save data right before power disappeared. Technically, if you doubled the load, the power supply would drop in about half the time so putting a big cap in parallel with C93 seems like the best bet... at least to eliminate that possibility.
On 2/21/2011 8:42 PM, Clay Cowgill wrote:
I'm assuming this is the 2212 used for SW novram. Is that right?
I'm 99% sure that's right. (Sorry, it's been a long time since I had to
look at/think about that. ;-)
The idea was that you could take your existing Star Wars settings and high
scores and move them over to the daughterboard without losing anything.
The only thing that comes to (my) mind at the moment is that maybe with two
NOVRAMs sharing their connections (except for the chip select which is muxed
by the game select switch) maybe the 'new' SW 2212 doesn't have enough time
to store the SRAM copy of the memory to the NV portion of the chip before
the EAROM power goes away.
For grins, you might try something like putting a big electrolytic cap
across the "+5EAROM" power and ground pins of the 2212's on the
daughterboard. (Like a 470-1000uF or something-- you could also put it in
parallel with C93 on the SW board-- careful with the polarity.)
-Clay
P.S. My apologies to anyone contacting me in the last few months if I
missed replying to you. We *just* finished getting Ground Kontrol back to
our permanent location after a huge three month remodel, so it'll be several
weeks still before I get my head above water again. We were doing 120+ hour
weeks there towards the end. ;-)
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
Received on Tue Feb 22 18:01:20 2011
This archive was generated by hypermail 2.1.8 : Tue Feb 22 2011 - 19:50:00 EST