how to flash ram SPD when it seems to be read-only?

Hi all,
___________________________________________________________________
EDIT: READ THIS POST ( I THINK I FIGURED IT OUT):
http://forum.notebookreview.com/showpost.php?p=5275126&postcount=4

also read these posts:
http://forum.notebookreview.com/showpost.php?p=5277863&postcount=15
http://forum.notebookreview.com/showpost.php?p=5278057&postcount=20
http://forum.notebookreview.com/showpost.php?p=5278138&postcount=21
___________________________________________________________________

So I am able to flash my kingston ram with spdtool. Works great.
Although for some reason spdtool does not recognise my SMbus driver unless I open setfsb and press getfsb. I know this is strange right? But it's not the problem.

The problem is that when I try my samsung 1gb 667mhz ram it seems to have a read only SPD.
When I flash it, during the verification it shows errors. The errors basicaly say that no values were changed. something like "blah blah blah verification failed".

I also have 2 sticks of 2gb 800mhz samsung ram which also seems to have a read-only SPD.

I would really like to be able to flash this ram (I want to see how far I can push my cpu but right now my ram is getting unstable), I am hopeing there is some way to do it. Maybe take off a resistor somewhere on the stick of ram or some other kind of mod I can make to let the SPD be flashed.

I am currenty looking for datasheets of SPD chips and also ram datasheets.
Hopeing to see something obvious that may help.
I haven't found a usefull datasheet yet.

If anyone needs it I can read off the numbers on the SPD chips (I believe I have spotted them on the ram, looks like they have 8 pins). I can also tell you anything else about the ram and I can post screenshot of the errors if needed.

thanks,
MH

 

bump. someone help me out please

 

Well i'll try to help (once I am able to send my memory for replacement/reflashing) by taking the heatsinks off and looking for differences.

But I can only send for replacement after the 7th (next month) and it will take 2-3 weeks to get them back.

 

OK I FIGURED IT OUT!!!!!!!!!!!!

The EEPROM does have write protection (pin7).

Please read this PDF:
http://jungle.at.tut.by/pdf/MemoryModuleSPD_Intro.pdf

look at table 1 on page 3.
It has the EEPROM pin description.

Here is the table:


Now it is time to modify that pin to dissable write protection so I can flash my ram

I believe I have to get pin WP tied to Vdd.

 

more technical documents can be found here:
http://jungle.at.tut.by/download.html

*just scroll down*

Its the official thaiphoon burner website.

 

Now I need to connecy pin 7 to Vdd(GND).
luckely the pin next to it (pin 8) is GND.

So it should be a very easy modification.
I would somehow have to dissconnect pin 7 from Vss. I will do some tests with a digital multimeter, hoping to find a resistor connected to pin 7.

Otherwise I would have to lift pin 7 from the ram.

I have attached a photo of my ram, I have pin7 and pin8 labled on the EEPROM.

 

So how do you plan to install a pull-down resistor when you get the pin disconnected?

Just asking, because connecting something directly to ground is not always a good thing.

 

pin 7 is an input pin. So I though if I lift it from the board it would get no signal.

Then I would connect pin 7 to pin 8 to set logic 0 to pin 7.

I'm a n00b at this sort of stuff, please tell me what part of it is wrong


Also I'm not sure if I have the pin numbering right can someone look at the following and tell me?



Edit: Do I have to use a resistor if I want to connect pin 7 to GND?

 

Ideally, you could put a 10K resistor between any pin and Vdd/Gnd. Tying two pins together should be okay as long as there is some kind of external pull up/down resistor connected between those two and Vdd/Gnd.

If there are weak internal resistors in the chip, and no stronger external resistors on the PCB, do NOT connect the pins together unless you have the spec sheet on the chip. One could default to pull up, another a pull down, and that would be...well...bad.

That circle indicates which pin is pin1, and yes your pin numbering looks correct.

ANSWER: Typically you do use pull down resistors to GND for I/O signals.

UPDATE: That pin 8 is a supply pin itself...it will NOT have a resistor acting as a pull-down. You need to run a 10K between the two pins. You should be able to solder a 0402 10K resistor between the two pins, but you'll need some soldering equipment and a microscope.

QUESTION: Can you see any PCB trace connecting to pin 7? You might be able to just etch that out with a razor blade (yes, I have done that) but you need to be careful with that. Do not try to snap the pin or anything, chances are you'll just break the EEPROM.

 

thanks for helping

one more dumb question, do I have to disconnect pin7 from Vss?

I mean could I just connect pin 7 to pin 8 with a 10K resistor and call it a day?

Or is this the right way to do it:
1. disconnect pin 7 from the board.
2. connect pin 7 to pin 8 with a 10K resistor

Also If this goes wrong is there a chance to kill my notebook?
I don't mind killing the ram ( already killed one stick by flashing wrong timmings onto it).

 

I would highly recommend disconnecting pin 7 from the rest of the PCB. If there is another pull up resistor somewhere in there, you will have effectively caused the pin to float in the middle. Once you get the new SPD data in, I'd also say it would be a good idea to tie the pin back to where it was to write protect it again.

And with any hardware mod, yes damage can occur. I'm not responsible for that. Make sure you only cut the right trace(s), test connectivity of any nearby pins, and when you turn on the computer for the first time do NOT have the battery plugged in. Be ready to pull the plug out just in case something goes wrong. You should have a second or two to pull the cord before any permanent damage happens (if something goes wrong that is).

 

I would never blame anyone for my own modifications going wrong
I take full responsibility for my actions.

So now I will do some tests with a DMM and maybe try the mod today or hopefully this weekend.

BTW, thanks for the suggestion to undo the mod after I flash the ram. It is a very good idea

Can't wait to see if this works

 

Good luck, and take time to study the layout of the PCB. Try to get a little familiar with the area you are working in. Conductivity/continuity tests with DMMs are definitely the way to go.

Interesting thing you are trying to do here though, not sure what you are trying to do with this though.

BTW, if you need resistors DigiKey is a really good place to hunt those down. Any old 0402 10Kohm 10% or 5% part will do. Those usually cost a few pennies (each) at most.

 

I'v done a simmilar mod to my PLL (where I pulled the TME pin low) to allow overclocking through setfsb (thread in my sig about setfsb and clockgen explains it all in detail).

Now when I get my FSB to 270mhz my ram is very unstable and so I cant overclock any higher.

If I could flash a lower frequency onto this ram, after the overclock it should still be stable.

I'm also going to do this to my 2gb sticks of 800mhz ram if the mod works.
Then I'll flash them to 533mhz. That should allow me to overclock my FSB to 300mhz (hopefully).

 

Ok, I'm half way there.

I first found a lot of points (and lots of resitors) connected to pin 7.

I looked on the other side of the ram, directly under the EEPROM , there was a resistor there which was directly connected to pin 7.

I removed the resitor shown in this photo:


Now I can't find any points connected to pin 7, even the ones that used to be connected arn't anymore.

So I believe I have isolated pin 7.

Now I need to get a 10K resistor and connect pin 7 and pin 8 with it.

 

Might be worthwhile seeing if you can flash the SPD eeprom as it is now. Since pin 7 is isolated, that may be read as a logic (0) write-enabled by the protection circuitry, allowing it to be written to.

 

It would not do that. Without knowing that the input is tied high or low via a resistor, it could easily be a floating input (and who knows what value it will be interpreted as).

If you want to do this right, get a resistor. NEVER leave something not connected when it is supposed to be. Undesirable behavior may result.

 

Well I will take your advice.

But let me just say that when I was doing a simmilar mod to my PLL I accidentaly made a bad connection so that the TME pin was getting no signal.
My notebook just gave me a black screen when I tried to start it up.

So it didn't cause any damage, but I will do it your way since it's not too hard.

 

quick question, do I have to us a 10K resistor?

Because the resistor I removed was not even close to 10K.

can I use a 5K? will it make any difference?

EDIT: I do have a 10K resistor BTW.

 

DONE

here is a photo of the finished mod:

 

It doesn't work!!

I put the ram into my notebook and all I got was a black screen.
Not even the toshiba logo came up.

I checked my work with a DMM, I checked everything many times. It was perfect.
I don't know why it doesn't work.

Anyway I put my good ram back into the notebook and it's fine.

I think the reason it doesn't work is that pin 7 is also used as "temperature event".

From what I understand, the external pin that I modified is the EVENT# pin.
The WP is done internally.

So this mod may work with some ram but with my ram it doesn't work.

Now I have to go out and buy new ram that can actually be flashed.

 

Right now I have undone the modification.
I put the ram into my notebook and it works fine.

But obviously it's still write protected so is pretty much useless for overclocking.

I'm really disapointed in how this turned out. I thought that it would work
I still think there must be some way to modify the ram to remove write protection, but it's too hard for me.

Anyway I'm giving up on this (for now ).

 

Yo! Have ya checked the latest Thaiphoon Burner 6.6.0.0 SuperB! Edition? This proggy can do cool things, bro!

moral hazard,
let's create a list of memory modules with a write-protected SPD chip!

 

Thanks, I'll check it out.

That is a good idea, can you start a new thread and I will post what I know.

 

I had experience with HYNIX.
They are protected.
I was able to successfully flash Samsung and Kingston (Using Elipida chips) modules.

 

Actually one easy way to check is to find the datasheet for the ram you want to buy.

In the datasheet it sometimes will say "SPD programmable".

Also you can't really go by brand name, it's a case by case kind of thing.

 

Nope they don't put that data out in some of them.

 

As you can see, I said "sometimes"

 

moral hazard
There is also a good thread at XS about WP SPD. Here is the link to Spd Update Failed ... Protected EEPROM or what ???. Thanks man for your efforts! You are doing a great job!

 

The most promising part of that thread was the second last post where there was going to be an effort to remove the WP SPD eeprom and solder another on that had been pre-flashed with the desired SPD table. No reply was given whether it was successful or not though the user was somewhat confident it would work. User had previously unsuccessfully tried to un write-protect the eeprom by playing with the WP pin on the chip.

 

That does sound like it should work.

Though I think it's a bit hard to remove the chip from the stick of ram without causing a lot of damage.

 

Thaiphoon Burner 6.7.0.0 can detect the permanent software write protection. To determine whether the SPD write protect feature is enabled select PSW Protection from the EEPROM main menu. For instance, if you have two non-protected memory sticks visible on SMBus as a device 50h and device 51h then both the EEPROM at 50h and EEPROM at 51h checkboxes should be enabled. If your memory modules are write protected then these checkboxes are disabled.

I have two memory sticks, one of them which is 50h is protected permanently, another one which is 51h is not protected. So, the picture should display information as follows.



Unfortunately, PSWP is not reversible.

 

I have been able to flash samsung chips as well. I have a crucial stick with micron chips and its locked. Also have a kingston stick with "kingston" chips its locked as well. There is a kingston stick with Elipida chips at work I can try. Just wish i would have known before i bought 2 other sodimms -.-

We shouldn't give up on this yet Im using the full version of thaiphoon burner and would recommend it for you guys to detect if sticks are locked.

 

I have two Crucial 16HTF25664HY-667E1 SODIMMs in my laptop that I was about to burn, but I'm finding them to be locked too. They have eight pin SPD EEPROMs with E02FR // K813 on them. Thaiphoon Burner 6.7.0.0 shows PSWP not to be grayed out, so that's not the issue; should be some other form of protection... can I circumvent it somehow?

I had just burned a 1 gig Samsung M470T2864QZ3-CE6 stick (also 8-pin EEPROM, this one labeled 302RP // 9717) with no issue, so its clearly these particular sticks, not the system.

I'm willing to bet I have the same issue that moral_hazard was experiencing with his modules in the posts on pages 1-3.

 

It's to my understanding that if Thaiphoon doesn't have the checkboxes grayed out then your fine. What software are you trying to use to write the new spd's?

 

I don't think that's the case. I believe theres a hardware write-protect as was explored earlier in this thread. I've used both SPDTool 0.63 and Thaiphoon Burner 6.7.0.0 to attempt to burn. Both are successful with the Samsung SODIMM; neither is successful with the Crucial (Micron) SODIMMs listed above.

Also, I just tried a Hynix 2GB HYMP125S64CP8-Y5 (which also has an E02FR SPD EEPROM like the Crucial SODIMMs) and it will not write either. They get the same errors that moral_hazard got, which is, they fail on verify, and if you re-read the SPD, you find it doesn't change at all... hence write-protected.

Apparently, some modules have a hardware WP pin set by connection, either internal to the chip or externally by a resistor.

I also have a set of G.Skill 4GBx2 SODIMMs that I think I will try tonight. If those work, I'll just swap them and my dv6500t will have 4GB and the dv9700t will get an 8GB upgrade in the name of overclocking

 

Hmm interesting. Goodluck and let us know how it goes

/drool at 4gb ddr2 sodimms

 

Yup, it did work on the G.Skill F2-5300-CL5-4GBSQ 4GB SODIMM modules. If anyone wants SPD unlocked 4GB modules, I bought them on the 'egg. So now I'm running with 8GB DDR2-833 RAM at 3.13GHz

Couldn't be happier!

 

Guys, please check this article about PSWP. If you have any comments on this, please PM me.

 

That's awesome. So basically, I know that the Crucial modules have pin 7 set high on the ST M34E02 EEPROMs. I guess now if I want to overclock those, I have to disconnect 7. If I ever decide to swap the 4GB G.Skill modules back out, or overclock my other laptop, I'll keep that in mind. Thanks!