My 1st unlocking results - what do you think?

[Miramonti]

This is my first shot at unlocking an athlon (well, unlocking a cpu I should say ) and I don't know what to make of the results.

This is what I get when I boot to different multipliers for a 1900xp with an Epox 8k3a mobo (4/3 bios I think) with multipliers from 7x-16x:

7x = boots at 7x
7.5x = boots at 7.5x
8x = boots at 8x
8.5x = boots at 8.5x
9x = boots at 7x
9.5x = boots at 7.5x
10x = doesn't post, must shut down/unplug, reboot, and then it posts @ default (12x)
10.5x = doesn't post, must shut down/unplug, reboot, and then it posts @ default (12x)
11x = boots at 11x
12x = default, boots at 12x
13x = boots @ 12.5x (I know this is a bios issue...)
14x-16x = doesn't post, must shut down/unplug, reboot, and then it posts @ default (12x)

I used superglue to cover the laser locks and then used the ink from a silver conductive pen, applied with a needle. I didn't do that great of a job at first, so after I got these results, I carefully scraped off the silver with a razor blade and reapplied it (I did an excellent job the 2nd time, using scotch tape to align it etc.)

But after redoing it, I got the exact same results as the 1st time.

Could one of the laser locks exposed and grounding 1 or 2 of the bridges, or what could cause this 'partial unlocking'???

Ugh.

 

[johncar]

Results thru 9.5X suggest the 2X Bit Value circuit has been "hard grounded" therefore set LO "permanently" so that the 2X Bit value can never be reset HI when resetting Multipliers...the 2X Bit Value is not needed until 9X...OR its L1 bridge may still be OPEN, unable for mobo to reset the default LO to a HI to enable the 2X Bit Value when needed...in both cases "permanently" LO.

But therefore we would also expect the 10X and 10.5X results to post and boot at 8X and 8.5X respectively, so we have no explanation for those "no boot" results, esp since they would be "lower" than default 12X.

14X and higher is probably too high for the cpu and/or its voltage and cooling...but the fact that all the no posts do reboot at default 12X suggests the mobo/bios may be involved with this...a form of "auto resetting"???

"Hard grounding" means connectinging "directly" to ground, unlike grounding thru current limiting Pulldown resistors such as the 1K resistors on the cpu, which then does allow LO signals to be reset HI via voltage divider circuits on the mobo.

http://www.beachlink.com/candjac/index.htm link to Final Decoding Palomino article to "understand" the Bit Value concept as it applies to Multiplier decoding and which L1 bridge is related to the 2X Bit Value...(the middle one)...decoding unexpected results becomes a piece of cake then.
John C.

 

[johncar]

dupe...sorry

 

[johncar]

Results thru 9.5X suggest the 2X Bit Value circuit has been "hard grounded" therefore set LO "permanently" so that the 2X Bit value can never be reset HI when resetting Multipliers...the 2X Bit Value is not needed until 9X.

But therefore we would also expect the 10X and 10.5X results to post and boot at 8X and 8.5X respectively, so we have no explanation for those "no boot" results, esp since they would be "lower" than default 12X.

14X and higher is probably too high for the cpu and/or its voltage and cooling...but the fact that all the no posts do reboot at default 12X suggests the mobo/bios may be involved with this...a form of "auto resetting"???

"Hard grounding" means connectinging "directly" to ground, unlike grounding thru current limiting Pulldown resistors such as the 1K resistors on the cpu, which then does allow LO signals to be reset HI via voltage divider circuits on the mobo.

http://www.beachlink.com/candjac/index.htm link to Final Decoding Palomino article to "understand" the Bit Value concept as it applies to Multiplier decoding and which L1 bridge is related to the 2X Bit Value...(the middle one)...decoding unexpected results becomes a piece of cake then.
John C.

 

[Miramonti]

Results thru 9.5X suggest the 2X Bit Value circuit has been "hard grounded" therefore set LO "permanently" so that the 2X Bit value can never be reset HI when resetting Multipliers...the 2X Bit Value is not needed until 9X.

How does the 2x bit value circuit get "hard grounded"?

14X and higher is probably too high for the cpu and/or its voltage and cooling...

understandable

but the fact that all the no posts do reboot at default 12X suggests the mobo/bios may be involved with this...a form of "auto resetting"???

The cpu's default multiplier is 12, running the 1900 at 1600@133 fsb.

"Hard grounding" means connectinging "directly" to ground, unlike grounding thru current limiting Pulldown resistors such as the 1K resistors on the cpu, which then does allow LO signals to be reset HI via voltage divider circuits on the mobo. http://www.beachlink.com/candjac/index.htm link to Final Decoding Palomino article to "understand" the Bit Value concept as it applies to Multiplier decoding and which L1 bridge is related to the 2X Bit Value...(the middle one)...decoding unexpected results becomes a piece of cake then. John C.

Thanks for the info and link - this is totally new to me (and a 'bit' mindboggling )


Anybody else?

 

[johncar]

Really sorry about the dupes above, but more info/another possibility re 14X result...(also 14.5X).

14X requires 1X + 2X +8X Bit Values HI, all others LO. But if 2X cannot be reset HI, (see previous posts), then the setting becomes 1X + 8X HI which is remapped to 4X, a "battery saving" setting for the Mobile Athlon 4 cpus...which your mobo/bios may not be programmed for.

Also...we're not familiar with your mobo...can it do resets above 13X, or is it like some early Athlon mobos limited to 12.5X/13X internal cpu remapping??
John C.

 

[Miramonti]

The mobo is a new via kt333 chipset board from epox.

 

[johncar]

The 2X Bit Value circuit can most likely become "hard grounded" during the L1 bridges closing operation...the laser opening of the the bridges is reported to expose a "ground plane" below the surface...which is the reason an insulating material "filler" in the cavity is recommended before applying the conductive ink/bridge closing material. So we'd think about redoing that L1 bridge closing mod...it's either grounded or still open.

Other sources of "grounds"....all surface labeling and icons are reported to be at electrical ground level...so care is needed in applying closing material on the surface as well...but would expect the leftmost .5X Bit Value bridge to be the most vulnerable in this regard.

Stick with the article's Bit Value concept, it enables intuitive fault analysis, and it's not rocket science...just a 5 bit logical code with some remapping...all detailed in the article.
John C.

 

[Miramonti]

Fixed! I redid the unlocking, this time removing the crazy glue from the laser locks as well. I then found there was a small exposed copper area between a couple of the laser locks that I didn't notice before. I must have scraped too deep when previously removing excess glue. Anyhow, I covered that up well with glue, as well as the laser locks, then cleanly redid the bridges with some silver ink again, and VOILA! I have all multipliers. (except of course 14 and 15, which are 5, 5.5)

I have to be more careful with the razorblade next time when cleaning the surface. Thx. again tho johncar for the bit value info - I'm going to try to understand it better by the time the next unlocking job comes around.