Unlocking Failure Diagnostic Clues ??

My first effort at unlocking on one of newegg's XP1600's, led to a brief but glorius success. I was able to get the chip going stably at 1860mhz on air and at standard default vcore voltages. After a few hours though, it reverted to its XP1600 rating, no matter where I put the multiplier, and it wouldn't restart unless I set the cpu to Default settings in the bios. I pulled the chip, and one of the L1 bridges looked a bit sloppy, so I poked at it to remove the conductive paint, and when it finally came off, one of the bridges shown thru clean and gold-leading me to think that maybe the paint hadn't been is such good contact at that point. And his question is :...Do certain kinds of mistakes in the unlocking process offer up clues as to what's wrong ? Does a chip's reversion to it's default speed indicate an open bridge, whereas one bridge shorted to another might cause different problems-such as incorrectly detected cpu speeds, fluctuating cpu speeds, etc. ? I'll repair the bridge tomorrow, when the lights better, but in the meantime I thought I'd ask if anybody out there has had experience with this ? Is there anyway to tell if a particular bridge is open, perhaps by measuring resistances theu the cpu's pins ?

Welp best bet is to just clean it all off and redue all bridges fro start.

Use a few coats of some strong nail polish across the pits taping off both sides of the pins. Let the nail polish dry for about 5 mins in between coats. Make sure to take the tap off after about 30-60 secs after aplying the coat.

2 coats of some good Nail polish should be enough.. When ya put the conductive paint on, use a single wire thread from an old un used wire.. Like from a old fan. Just one single thread so its skinny enough and not wide.. paint on the bridge in prefect straight lines that touch only the 2 pins your closing. Nothing else..

Let it dry for about 15-20 mins to let the conductive paint to become completely conductive. Then try it out. I also like to cover up the unlocking with a piece of scotch tape covering the bridges so ya dont mess them up later on with applying or taking off thremo compound later on.

jboy said:

My first effort at unlocking on one of newegg's XP1600's, led to a brief but glorius success. I was able to get the chip going stably at 1860mhz on air and at standard default vcore voltages. After a few hours though, it reverted to its XP1600 rating, no matter where I put the multiplier, and it wouldn't restart unless I set the cpu to Default settings in the bios. I pulled the chip, and one of the L1 bridges looked a bit sloppy, so I poked at it to remove the conductive paint, and when it finally came off, one of the bridges shown thru clean and gold-leading me to think that maybe the paint hadn't been is such good contact at that point. And his question is :...Do certain kinds of mistakes in the unlocking process offer up clues as to what's wrong ? Does a chip's reversion to it's default speed indicate an open bridge, whereas one bridge shorted to another might cause different problems-such as incorrectly detected cpu speeds, fluctuating cpu speeds, etc. ? I'll repair the bridge tomorrow, when the lights better, but in the meantime I thought I'd ask if anybody out there has had experience with this ? Is there anyway to tell if a particular bridge is open, perhaps by measuring resistances theu the cpu's pins ?

Click to expand...

We wrote this just for you.....
http://www.beachlink.com/candjac/index.htm link to Understanding Multiplier Code article.....others as you see fit for more detail re signal circuits' diagrams etc.
John C.

That's some really great info John C, thanks for sharing. You seem to know a lot about the electronics side what is your occupation?

Tiger said:

That's some really great info John C, thanks for sharing. You seem to know a lot about the electronics side what is your occupation?

Click to expand...

Thanks for the kind comments about the articles,and very welcome for the info, just glad to see it will be useful., esp since bridge mod failures are so very common.

We are an old retired guy who spent a career testing large steam turbines in power plants, with an electronics hobby. But if you look very carefully you won't find a single transistor in any of these signalling circuits...ie...no electronics at all. They are all what one might call Direct Current Resistor Logic circuits....just one step up from a switched flashlight circuit. Can analyze the circuits with nothing more than the 1st week of DC Electricity, (probably taught in middle school these days), Ohms Law plus the principle of setting signal voltages using Voltage Dividers, ie, voltage between resistances in series is proportional to the ratio of the resistance of the grounded resistance to the total resistance.

Thanks again, John C.

Hey, less of the "old". I'm probably not a great deal younger than you! All it requires is the basics and some good scientific common sense. I'm a semi-retired pharmacist and always had a fascination with electronics. The first computer I ever used was at uni and was a Fortran IV that probably had 1/10^6 of the capability of the machine I'm using and occupied a 2 story building and was totally surrounded by a airconditioned air. We wrote the programs and then used punch cards to execute them.

Thanks for the link, John. There's more there than I wanted to know. Unfortunately for me, I'm both old and digitally challenged, so it's going to take me a while to digest everything on those pages. Can you tell me, without my having to read all the info in the various categories, if there's a way to tell if I've screwed up the pit filling step on the XP's ? I filled mine with epoxy, but it seemed kind of tacky after it set up so maybe the proportions on the mix weren't correct. Anyway, I've cleaned off the L1 bridges, but there's still a little conductive lacquer residue left in the pits, and I'm trying to find a way to tell if maybe I have screwed up, and the residue is causing a teeny tiny short in one of those pits. I used something called Nickel Print to close the bridges-it's some kind of powdered nickel/lacquer mix I got in an electronics supply store, if this helps. Tried to repair a couple of suspect bridges, but things just kept getting weirder after each fix, and I figured I'd better start fresh. I've got to make sure my pits are properly filled, and not shorting. Any helpful advice will be greatly appreciated...

jboy:-
Re diagnosing pit filling failures...best way is the diagnostic procedure explained in the article...list a bunch of "I Set This but Got This" and we'll try to do it for you. But not until you redo them again as you say you've begun to clean them up.

Hard to come up with a visual inspection procedure of pits themselves. Problem is directly connecting the L1 bridges' "dots" to the sub-surface ground plane which then permanently sets a LO which cannot be reset HI.
So mobo/bios loses ability to set what you ask it to.

You could try Multimeter resistance checks from each dot to a "known ground". A very low resistance like a few ohms would mean that that dot "is" probably connected to the sub-surface ground plane. A resistance in the kilo-ohms range or higher would suggest only a normal indirect connection thru the internal electronics...might be OK to refill and connect.
Lettering and icons on the surface are reported to be at ground level...might use those as your ground reference to connect one Multimeter probe to.

Look into the pit with a high power eye loupe?? Then you might try carefully scratching the sides of the pits with a needle to try and create a short insulated length between the dots and any conductive material still in the pits. Conductive material in pit is problem only if it also connects to either dot of bridge.

Try solvents carefully with q-tips to prevent running the stuff all over?? Acetone??
We and many others use conductive solution found in rear window defogger wires repair kit...found at auto parts stores.
Hope this is of some help.
John C.

Tiger:- Small world, we started "using" computers in 1953. IBM 701 Mainframe FIV cards. But never got into the hardware till just a few years ago as a hobby.

Thanks for the response, John. I'll check those resistances tomorrow with a multimeter. Am I to understand, that, if I get a very low reading from anyone of those dots to ground, that my cpu is forever locked to a Lo reading ? Or is just the multiplier associated with those particular L1 contacts locked. More simply put, if any of these contact points are grounded, am I forever more denied the pleasure of overclocking via my motherboards bios ? If this is the case, are there any hardware workarounds,e.g.-fine jumper wires in the socket, bridging other L connectors on the cpu, etc ? I had this chip running stablely at 12X, and would dearly love to do it again. Be back tomorrow..

Stop wastin your time and just recover the pits with a strong nail polish.. It dont matter if ya got some of that conductive stuff in the pits or not... Sd long as the pit is covered it wont touch the stuff you use to connect the bridge.

If what you have used dont work dont use it again.

Either get some of the defogger stuff from an auto store or go back to a differenet electronics store and find a conductive pen...Radio shack may even sell it.

ya know it would take me till next year to understand what is said in that link above.. You dont need all of that... Your not cuttin and connecting bridges other then the ones ya know which is the L1's.

So just clean them up good with some 99% Isopropyl Alcohol and Make sure your dots are clean and not covered up with a film left over.. Check them with the mult set on Ohms and hold one end to a dot and the other end to a ground dot on the CPU.. If ya get a reading them its good... If not then the dot is covered up with something and ya should try scrapin it off... Might want to try taking an erraser and rub it and try again... Do that for each of the 10 dots.

Also yes you should use a good 10x loop or hell use a microscope like i have heh.

Remember dot to dot nothing else.. Dont touch a ground of the L1 next the the L1 bridge.

Acetone is probably the best solvent to use. I really battled to unlock my XP and had done many of the earlier Durons and Athlons with no problem at all. Wrote an article on this which you can find under beginner guides I think (painting L1's with defogger paint). I finally isolated the problem to the dots. Seems they may be covered in some kind of coating. I carefully scraped each with the point of a needle and then unlocked and it has worked perfectly. AMD maybe using some new kind of technique because I note that over the last few weeks this problem seems to have increased dramatically. If you are not sure whether there is conductive paint in the pits then try carefully removing the filler from them with a needle point. I filled mine with epoxy and a 1mm screwdriver blade and then skimmed the excess off with a hobby knife and then closed the bridges with defogger paint also using the 1mm blade.

KnownKiller said:

So just clean them up good with some 99% Isopropyl Alcohol and Make sure your dots are clean and not covered up with a film left over.. Check them with the mult set on Ohms and hold one end to a dot and the other end to a ground dot on the CPU.. If ya get a reading them its good... If not then the dot is covered up with something and ya should try scrapin it off... Might want to try taking an erraser and rub it and try again... Do that for each of the 10 dots.

Click to expand...

If ya get a reading then it "might be good"...depending on which dot you're testing and the level of the reading. The following assumes you've removed any insulating film from the surfaces of all the dots, "as quoted above". Then...

A very low reading like just a few ohms from either upper or lower dot means there is a "direct connection" to ground...most probably 'cause the old conductive material is still connected to sub-surface ground plane and the dot being tested. This sets a permanent LO which the bios will not be able to reset to HI if requested Multiplier needs it to be HI. So this "reading" is definitely not good, must disconnect the ground.

A reading of a few thousand ohms could result from the internal resistance to ground of the Multiplier's signal circuit, should be no problem. But this level reading should only occur when testing the lower dot of an "open" bridge, 'cause the upper dot of an open and not shorted to ground bridge is only connected to its socket pin then, it's not connected to internal circuit electronics when open.

So ungrounded and unconnected upper dots should generate an open circuit/infinite resistance...what the quote above would call "no reading", yet the upper dot is perfectly OK.

So the test in the quote only tests for insulating film on the dots, "assuming" that the bridges are open and all grounds have been cleared first...but how does anyone know that for sure after the mess exists??
Life is not always as simple as we'd like it to be.
John C.

Whew ! Think I'm OK. Got a measurement of about 150 ohms between ground and all the individual lower L1 dots-does this sound like the dots are ungrounded ? All the upper dots were open, no reading) I'll refill the pits, (might try the nail polish thing, but the solvent medium in my trace painting stuff is lacquer thinner, which might dissolve the nail polish).Does anybody know what the solvent base, and the conductive medium, is in the defogger paint ? Before proceeding furthur however, I'd really like to find out if my resistance measurements are OK, as it doesn't make any sense to join the bridges if the dots are shorted. JohnC., you mentioned values of a few ohms for shorted dots, to a few thousand for unshorted ones, Mine fall in between. So if you or anyone knows, or has a free unmodded XP around that they can check with a multimeter, I'd dearly love to know.Thanks for everyone's help.

jboy said:

Whew ! Think I'm OK. Got a measurement of about 150 ohms between ground and all the individual lower L1 dots-does this sound like the dots are ungrounded ? All the upper dots were open, no reading) I'll refill the pits, (might try the nail polish thing, but the solvent medium in my trace painting stuff is lacquer thinner, which might dissolve the nail polish).Does anybody know what the solvent base, and the conductive medium, is in the defogger paint ? Before proceeding furthur however, I'd really like to find out if my resistance measurements are OK, as it doesn't make any sense to join the bridges if the dots are shorted. JohnC., you mentioned values of a few ohms for shorted dots, to a few thousand for unshorted ones, Mine fall in between. So if you or anyone knows, or has a free unmodded XP around that they can check with a multimeter, I'd dearly love to know.Thanks for everyone's help.

Click to expand...

The few thousand ohms was a recall from a Duron we tested when they first came out. 150 ohms is considerably more than a dead short to ground and since you get it for all 5 lower dots, it sounds like that "might be" the inactive/no power internal resistance for your chip, and very possibly "no grounds".

But there's another relatively easy way to test for "no grounds" on a 1600 "after" an L1 bridges "cleaning" operation. Just install the 1600 and boot. IF you get the default 10.5XFSB, THEN the lower dots of the 1st 4 L1 bridges are NOT grounded...because...default 10.5X Multiplier has the 1st 4 Bit Values = .5X +1X + 2X + 4X ALL set HI. So you can't get 10.5X on a 1600 IF ANY of those 4 signal busses have been reset LO by accidental grounds like the bottom dots to the sub-surface ground plane...you'd get something different than 10.5X. So we suggest you pop it in and see what you get. If 10.5X then take it out and carefully insulate/fill the pits and close the L1s...doing the necessary research to avoid any possible dissolving of the insulating fill...we don't feel capable in that chemistry arena.

Let us know what the 1600 boots at, the boot test is ideally suited for the 1600 because default 10.5X has ALL 4 of the Bit Values you'll need to oc with set HI by default...as the boot test with other chips would not test ALL 4 circuits for grounds.
John C.

PS Edit:- If you get 11X, then ALL 5 are LO, the 1st 4 probably from pit grounds, the 5th by default.

Great ! I'll try the test tomorrow, and if it passes, I'll redo the bridges, after cleaning each and every one throughly with some acetone on a toothpick. Which brings up an interesting question; Does anybody know of any source of tools/implements to work on things at this scale ? From the finest needle, to single strands of stainless steel wire from stranded military communications line, everything seems just a shade too large. What do scientists and technicians use to manipulate things at this level ?

Progress Report: Arrrgh ! Damn those bridges are small. Tried twice to redo things, and screwed up both times. Seem to have trouble making this trace painting compound adhere properly. Because it's in a lacquer thinner base, it starts drying immediately out of the bottle-sets up too fast, hard to work. What is that defogger stuff like ? Does it flow easily, and stay wet long enough to get it on right -(like maybe, 2 minutes) ? JohnC.-noticed in your reply to jszents post about his xp2100, you said he could reverse the mod by ungrounding the 8x pin. I thought once one of those dots had been grounded, it remained that way forever-it in effect throws a switch in the chip that cannot be reset. True ? also, just out of curiosity, how hard is it to cut the bridges if they're joined ?

jboy said:

Progress Report: Arrrgh ! Damn those bridges are small. Tried twice to redo things, and screwed up both times. Seem to have trouble making this trace painting compound adhere properly. Because it's in a lacquer thinner base, it starts drying immediately out of the bottle-sets up too fast, hard to work. What is that defogger stuff like ? Does it flow easily, and stay wet long enough to get it on right -(like maybe, 2 minutes) ? JohnC.-noticed in your reply to jszents post about his xp2100, you said he could reverse the mod by ungrounding the 8x pin. I thought once one of those dots had been grounded, it remained that way forever-it in effect throws a switch in the chip that cannot be reset. True ? also, just out of curiosity, how hard is it to cut the bridges if they're joined ?

Click to expand...

Defogger stuff is also fast drying, no way 2 minutes ...must be fairly quick...we held bottle and cpu in fingers of one hand to enable quick move from bottle to cpu...and used at least 5X eye loupe to keep all in focus, save transfer time. Maybe use masking tape to be able to work faster??

No irreversible switches "in" the chip...change the external switching/grounding changes/restores the signal.

Bridge cutting is most difficult operation unless you have proper tool...which is "diamond coated" rotary burr...but with finest "pointed" tip. Then epoxy into suitable tube, use like a pencil, heel of hand held steady, fingers slowly/carefully stroking number "ones" across the bridge, inspecting often to avoid going too deep...does not take long with the diamond tip. Get a used one from your dentist, (we did), or buy at hardware/hobby shops. Finest point is critical as bridges are not very wide...as you well know.
John C.

Jboy ya better stop now while your ahead and wiat to find the proper conductive pen to use.. You keep going like ya are and you'll ware out something or scratch up the chip around the dots and your done.

Call around for a conductive pen.

JohnC:-Does this mean that basically, as long as you confine your activities to the L1 bridge, that you can reverse anything you do, without fear of damage ? Kinda sounds like it, if shorting one bridge to another only results in an indeterminate multiplier, and an accidental grounding from the pit to the dots can be reversed ? No ? Thanks for the poop on the defogger stuff.

Killer, I hear you guy. Just dont feel like spending another $15 on something that might be as big a pain in the butt as the way I'm doing things now. You're the first guy I've heard say anything good about those pens. Doesn't the paint dry in the tip after you use it ? How long do you think those things will be good for-some of these things have a shelf of about one week after you first use them. Think I can get what I'm using now to work, after making my big discovery below.

BIG DISCOVERY: The resistance measurement to ground between properly connected dots on the L1 bridge drops to about 1/3 of what it was before the dots are joined. You can check for a good connection just by clipping a pair of alligator clip leads to you multimer, clipping one of the other end to the gold grounded trianle at one corner of the cpu, the other to a fine sewing needle, and touching each individual bridge. If you get zero resistance, it's shorted to ground. If you get the resistance you measured between the lower dots and ground before you joined the dots, the connection between the two dots has not been made. If you get a reading of about 1/3 tour original pre-joined measurement, the bridge is properly formed. May do a separate post on this, sure makes it easy to diagnose bridge problems. I can smell that Nobel prize already

jboy said:

JohnC:-Does this mean that basically, as long as you confine your activities to the L1 bridge, that you can reverse anything you do, without fear of damage ? Kinda sounds like it, if shorting one bridge to another only results in an indeterminate multiplier, and an accidental grounding from the pit to the dots can be reversed ? No ?

Click to expand...

Yes...But...it may not be easy to remove the grounding material in the pits....sort of like theory vs practice.

It's time for you to just clean up the surface, with acetone as Tiger suggests, being careful to do it with a saturated q-tip to avoid slopping large amounts around and possibly damaging something from reaction with acetone....then plug in the chip and see if and at what Multiplier it boots at. Don't know if your L1s are currently open or closed, but doesn't matter...let's get some "real data" and if it boots we can hopefully decode any "unexpected Multipliers".
John C.