Can't get 133MHz from my Tbird 1.2!

[worms]

Here's the specs:
Blue(ish) Tbird 1.2 (100MHZ) AXHA open L1's!
KT7A-R, Crucial 133/3, Lapped and smeared Taisol, Antec 300W, 40C
fail-safe or optimal BIOS

Here's the problem:
Multipliers from 5 to 10.5 kill the system requiring unplug and CMOS clear (multipliers from 11 to 12.5 work fine and FSB's are good up to about 117x11=1300).

Here's what I've tried:
Scribbling all over the L1's - no change.
Plugging a scribbled Tbird 1.0 into the same system - works fine at all multipliers from 5-12 and will post but not boot at 12.5. The lower multipliers of course allow for any FSB up to 140 or so. 8.5x133=1133 is about the maximum all around setting at which it will boot from the RAID.

Here's what I've noticed:
1.2 1.0
L2 11 11
L3 1001 1001
L4 0101 1010
L5 1 1
L6 0001 1101
L7 11011 11011

L4(BP_FID) and L6(FID) are different and I suspect that maybe they are somehow limiting the multiplier range to 11-12.5. Is this possibly a new attempt by AMD to keep vendors from selling the 100MHz models as 133's while still allowing overclockers to underclock a little.

Anyone else experience this? Anyone know of a cure?

 

[Gametime69]

Well.. I have been hearing that some people have Thunderbirds and Durons that can notbe unlocked because the pins are removed underneath. I seen this at the Over Clockers web site.. another web site had pictures.. side by side of one with all pins and the other with pins removed. strange huh..

Overclockers.com

 

[DarkMajiq]

I think I saw someone post a while ago who was having a somewhat similar problem on the same board, it seemed that a BIOS update fixed things, so maybe try that, although I could be wrong.

 

[eeut]

don't have a kt7araid but kt7a w/tb800, which does same thing happen me. 100-115fsb no probs. tried 7.5*133 on default settings no go, but optimal settings stable which is weird. wanting higher fsb so tried 7*145 go still on optimal. on 7*150 no go. lowered 7*146 default settings no go cleared cmos and tried on optimal settings, went straight to windows, redid again a fresh install no prob. stable for 2 days(7*146). weird and awesome. build it for my wife.

 

[eeut]

forgot this one. heard newer thunderbirds overclocks better when lowering vcore. try playing around there.

 

[johncar]

Worms,
We notice that your 1.2 had the L1 bridges open. If that is the case, then the symptoms point to the 2 L1 bridges in the 2X and 4X circuits as still being open.

Closed L1 bridges allow the mobo to over-ride default L3, L4 settings of a 4 bit binary code to the FSB Multiplier. Each circuit Adds .5X, 1X, 2X, 4X respectively from L to R when set HI. And the setting algorithm is........

TotalX = 3X base all LO + sum of "Adders" set HI.

Except add 8X to TotalX = 3X thru 4.5X to get 11X thru 12.5X.

So a 1.2 Tbird has only the 1X Adder set HI, ie, 3X +1X +8X = 12X, and all the 3X thru 4.5X settings for 11X thru 12.5X need only the .5X and 1X Adders...which you are able to get. This says the .5X and 1X L1s are closed since mobo is able to "change" their LO and HI default settings.

But 5X thru 10.5X alsso need to reset the default LO 2X and 4X Adders, and since you can't do that we suspect those L1 bridges are still open.

Go here http://members.nbci.com/candjac/index.htm Duron OC article for pics, circuit diagrams and full explanation of the above, including tips on page 2 on how to verify/test closed L1 bridges without destroying the material used to close them.
John C.

 

[worms]

Thanks Johncar you were right on!

I had unlocked the 1.0 and the 1.2 at the same time with the same 2B 0.6mm mechanical pencil (sharpened by rubbing on paper) and the 1.0 worked perfectly on the first try. I re-penciled the 1.2 twice and still saw no change from stock. Recall that the unmodified 1.2 had the same 11-12.5 multiplier range as the penciled version. Note also that the stock 1.2 did respond to multiplier settings whereas the stock 1.0 would ignore any BIOS settings and always run at 10X.

I decided to re-do the unlock on both processors via conductive ink. At $15 across town or $10 ordered from Radio Shack I found the pen search to be too time consuming and expensive so I opted instead for the $7 defogger repair paint from the auto parts store. I wiped off the pencil lead with mineral spirits (came right off!) then gently sanded the contacts with 2000 grit paper to be sure and cleaned again with spirits. I found that the defogger paint dried almost instantly and I could not get it applied with a screwdriver but it worked pretty well with a toothpick (blotting once on a non-absorbant surface reduced the drop size to a much better proportion). I did this to both processors and both now support the full range of multiplier settings. 140x9.5=1330 at 42C seems to work well on the 1.2 in preliminary testing.

 

[eeut]

that's great. me just waiting for the prices to go down for the 1.2s/1.3. bought mine when they started selling tb800s bought 2 right away incase i messed it up. did full research on this forums mostly on 3people w/c i considered maturely knowing it all. JC nice explanation.