Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[FBz]

Prompted by OS923’s recent 1.42 CPU swap to a lesser MDD
… so of course mine, would be different.  The 867 MHz DP
has a 133 MHz bus speed, as opposed to the 167 MHz of the
later MDDs. (Except for DP 1 GHz FW800 MDDs /which also
had / has a 133 MHz bus?)

Mine, from last night:


Now I’ve seen the thread by ProtoolsLE5Guy & MacTron
http://macos9lives.com/smforum/index.php?topic=1652.0
-which is very informative but it still remains fairly unclear.
There’s another old thread over on MacRumors.com that
deals with the same matter and it suggests the removal
of a single solitary resistor (R676) on mobo’s backside
to enable bus speed change from 133 Mhz to 167 MHz.

https://forums.macrumors.com/threads/guide-to-changing-mdd-bus-speed-cpu-upgrade.1708627/

So before I remove the mobo and the (R676) resistor…
anyone have any actual insight into this little escapade?

Sure it would’ve likely been easier to trade a 1.42 CPU
with a 1.25 DP CPU MDD - but first, the 867 MHz MDD
might benefit more. (Especially if only requires removal
of a single resistor.) Changing the bus speed should do
the trick, without changing anything on the 1.42 CPU?

(The 1.42 GHz CPU into 1.25 DP MDD swap will be next.)

Here’s a 1.42 GHz FW800 mobo close-up and you may
note the absence of the R676? Which = 167 MHz FSB?

[Protools5LEGuy]

Now I’ve seen the thread by ProtoolsLE5Guy & MacTron
http://macos9lives.com/smforum/index.php?topic=1652.0
-which is very informative but it still remains fairly unclear.
There’s another old thread over on MacRumors.com that
deals with the same matter and it suggests the removal
of a single solitary resistor (R676) on mobo’s backside
to enable bus speed change from 133 Mhz to 167 MHz.

Consider yourself lucky. You didnt had to deal with the MacBidoule article in a time when Google translate just didnt  do a thing right(excuse french folks if I mispelled)

Certainly the post is long (2014 started) but I think I almost killed it in 2020 with the HD pictures of the green and the blue boards.

Long read... yes I have to admit. Fairly unclear? Strongly disagree.

Read it twice and excuse the offtopic around.

Dont feel bad for re-reading. I have read 15 times opencore legacy patcher wiki and still havent a clue. 

[Protools5LEGuy]

Changing the bus speed should do
the trick, without changing anything on the 1.42 CPU?

Yes. But when you read the macbidoulle article translates as the opposite.

Also it didnt helped that the representation of the resistors in some schemes are "X" marked.

Here’s a 1.42 GHz FW800 mobo close-up and you may
note the absence of the R676? Which = 167 MHz FSB?

Yes

[Protools5LEGuy]

Also is a bit unclear on the "overclock 866 dual daughtercard" vs "unlock 166 bus on 133 motherboards". Those are 2 subject.

2 resistors vs 1 resistor.

One on the backside of the motherboard + One on the CPU daughter card Vs Only the one on the backside 

[FBz]

Well, it worked! Simple.
Removed only the R676 resistor from the DP 867 Mhz mobo
and effectively changed the FSB from 133 MHz to 167 Mhz.
(Apple 820-1445-A mobo)



But, not at first.
After removal and reassembly it wouldn’t boot.
So then re-soldered the resistor back. Still no boot.
Placed PSU & CPU back into 1.42 GHz FW800 MDD.

Still no boot.
(Samsung PSU had failed!)

All over again this morning. Removed R676.
Then re-installed mobo along with CPU & new PSU.
BOOT!!!



I’ve more FW800 machines than FW400 - and prefer the latter.
This 867 MHz DP machine is a welcome addition…
with a 1.42. GHz DP and a copper heatsink.

Results from an Inland 128 GB SSD w/ Bribge adapter:



Thanks Protools5LEGuy!

[FBz]

Similar story here after resistor removal… except boot-capable.
And this 867 / 1.42 recognizes 1.75 GB under Tiger 10.4.11.
Maybe just a consequence of the process?

Although full 2.0 GB of RAM is installed. (512 MB x 4.)



However OS 9 boots provide the following:



Any “service technician(s)” here?
AHT “red flags” J20 too.

Both OS 9 and X “see” the J20, 512 MB RAM as 256 MB.
ALSO, using PC2700U RAM… instead of the noted PC2600U.

And have done the pipe cleaner + alcohol cleaning of slot J20.
Will eventually pull mobo and check all J20’s solder points.
For now, will just leave the J20 slot empty.

Even with the J20 socket empty…
machine boots and runs MUCH faster than before!

[refinery]

It's probably a bad stick. OS9 machines will report there is a full 2GB ram installed, it just won't use it. Instead it will always say that finder is using 500+ MB of RAM.

[FBz]

It's probably a bad stick.

Thanks.
Exactly what I initially suspected.
But further testing with 12 other
(known & tested good) sticks also
provided the same results.

It's either that J20 slot itself - or a
result of that R676 resistor removal?



It recognized the full 2 GB prior
to the removal of that resisitor
and the change from 133 MHz
to 167 MHz FSB.

[peeperpc]

It's either that J20 slot itself - or a
result of that R676 resistor removal?

My 820-1445-A board also has the resistor removed but still uses the stock 867DP. All 2GB RAM recognized.

[FBz]

My 820-1445-A board also has the resistor removed but still uses the stock 867DP. All 2GB RAM recognized.

Thanks!

Just discovered two more candidates (from storage) for the
1.42 CPU & copper heatsink swap. One’s an 867 DP MDD
and the other… a 1.25 SP MDD.

In that graphic above (funny) but before the resistor removal,
the 1.42 CPU recognized the 2 GB of RAM before the switch
to 167 MHz FSB. I’ll do this other 867 DP next and check it
all again before adding the 1.42 CPU to compare results.

Might be that Mr. Big Fingers damaged that J20 slot?

[peeperpc]

To make it clearer (in case you wonder if all the slots work), all four memory slots on the board work in both OS X and OS 9. The RAM are 1GB, 512MB, and two 256MB sticks.

Thanks!

I also thank you for giving the specific info that the 820-1445-A board is compatible with the 1.42DP CPU card (820-1470-A).

Years ago, I tried to use a 1.25DP card (820-1472-A) with it and failed (the power light went away when I released the power button). The card came with some damaged capacitors which I'd replaced but it still didn't work. Yes, I might do it wrong. But the Apple service manual implies that the 820-1445-A is one of the boards that have a compatibility issue with some CPU cards.

Without some clarity (is it the faulty card or the board is inherently incompatible with any other CPU card that is not 820-1310-A?), I decided to stop and put back the stock 867DP (820-1310-A) card.

So your info is very helpful.

[FBz]

"The RAM are 1GB, 512MB, and two 256MB sticks.”

This one does recognize the 512MB RAM in the J20 slot - but reports it as 256MB... and Apple Hardware Test also reports problem. [OS 9 hung sometimes @ boot, until removal.] You using PC2700 RAM or still with the PC2100? Was wondering about maybe using two PC2700, 1 GB sticks of RAM and leaving two slots open / if latest cleaning does not solve the J20 problem? (Maybe later.) For now, have just deep-cleaned ALL those RAM slots on that board and should reassemble and re-test it later today. Hopefully it will be “cured”.

Also cleaned the just discovered, other 867 DP and the 1.25 SP MDD boards. If the “new” 867 board tests okay, may try a 1.25 DP MDD CPU on it and let you know results. Goal here is to rob all the 1.42 GHz dual CPUs from the FW800’s and put them in FW400 cases / boards. (May have to re-cap some more PSUs along the way.)

“Without some clarity (is it the faulty card or the board is inherently incompatible with any other CPU card that is not 820-1310-A?), I decided to stop and put back the stock 867DP (820-1310-A) card.”

“Years ago, I tried to use a 1.25DP card (820-1472-A) with it and failed (the power light went away when I released the power button).”

Similar first result here with the 1.42 DP in the 867… until PMU reset.

(Often, just “one more little thing” will solve the problem.)

[peeperpc]

"The RAM are 1GB, 512MB, and two 256MB sticks.”

You using PC2700 RAM or still with the PC2100? Was wondering about maybe using two PC2700, 1 GB sticks of RAM and leaving two slots open / if latest cleaning does not solve the J20 problem? (Maybe later.) For now, have just deep-cleaned ALL those RAM slots on that board and should reassemble and re-test it later today. Hopefully it will be “cured”.

3 of them are 2700, only one (the 512MB stick) is 2100.

By the way, when I tested the 1.25GHz DP CPU, I only used the 2700 sticks.

Similar first result here with the 1.42 DP in the 867… until PMU reset.

(Often, just “one more little thing” will solve the problem.)

I don't recall if I had already tried that but I will try it again, thanks.

[FBz]

After deep-cleaning all the RAM slots and reassembly, it booted twice with no problem. Full 2 GB of RAM present and working!

The third time was not “charmed”.

J20 slot then reported no good and MDD was back again to hanging under OS 9 and X. Started thinking (often dangerous here) that perhaps as the machine warmed up and things began expanding a bit, that maybe all those little contacts in J20 might expand just little more… enough for solid contact with the stick’s contacts to be lessened - might be a factor?

And as RAM is often changed in and out of various machines here, wondered if maybe a very slight amount of wear might have had something to do with this. So after trying all of the previous sticks of RAM yet AGAIN, I opened another MDD and scavenged that RAM too.

First stick in, VIOLA! and it’s been booting and running stable since. (Two hours now.) And it’s highly likely that J20’s contacts have become worn / sloppy over time. The “new” stick of RAM shows none of those slight vertical grooves on its’ contacts like nearly every other stick of RAM ‘round here. And have had no other problems with the previously tested RAM, in any other MDD here.



@peeperpc Nearly everything that I’ve read states that you must change to the PC2600 RAM (just try to find some of that). So I went with all PC2700. Might just be an “up-sell” to sell more RAM / because I did boot with some 2100 RAM in the beginning. Hope that a PMU reset works for you.

Now, it’s on to the other 867 and the 1.25 SP. (Don’t need to remove the R676 in the latter.) Maybe tomorrow?

And it might be a little sad to see the (3) FW800 MDDs cannibalized here but much of their trim pieces, side panels etc. will be going to (hopefully) three FW400 1.42 GHz DP machines. And who knows, might eventually attempt to overclock one of their CPUs - like M.A.R.L. labs / MacTron.

There is a very noticeable difference in boot times as it is… with the 1.42.
(Especially with OS 9!)

[FBz]

It’s been suggested that the J20 slot on this might have had failed or cracked solder points which would only allow the intermittent and / or occasional full contact with all RAM stick connectors(?).

An extremely close inspection of all the backside solder points for the J20 slot was done and no such problems were found.

However, a closer inspection and comparison between the RAM that worked (#13) and those that did not… seems to show a slight difference in overall thickness between the RAM sticks. (I’ll check this with a micrometer later.)

But for now (Day 3) it’s still booting without fail and even avoided hanging yesterday when I forgot to power up the external fan that I use on open cases. It did get pretty warm until I realized this. Slightly thicker RAM = better overall connections / performance?

What appears here to be “thicker” is “TECHWORKS” RAM. (#13)

 

So, there is a slight difference in RAM thickness... see last three images attached below.
This did make a difference.

[FBz]

Yes, there will be pics when all are done.

Rain day here, so out in the garage cleaning another 867MHz and a 1.25 GHz case for two more 1.42 GHz MDD CPU transplants. Rather than rob side panels from better looking MDDs… used quadruple-aught (that’s four zeroes - 0000) steel wool, lightly sprayed with GOO GONE (goo and adhesive remover) to diminish the look of the now typical gouges, scratches and other time-worn abrasions on the MDD side cases… without reducing the normal sheen / shine of the polycarbonate too much. If you just use the steel wool alone, you end up with a dull matte finish where you’ve attacked the targeted areas.

THEN with Meguiar’s QuikWax sprayed on afterwards, buff that out with a clean terry cloth towel. Might not be perfect - but ’tis much more visually acceptable. (Short of spraying them entirely with clear coat.) Not today.

Once I get these two done, I’ll await simple instructions from MacTron, Knez or others (that maybe a six year-old could follow - because it’s me after all)… about just how to overclock one or more of the 1.42’s.

Then maybe I can rest a bit… after attempting to install six SSDs in a Quicksilver.
(Why IIO, why?)

[FBz]

This second MDD 867 MHz DP transition to 1.42 GHz went very well last night.
The R676 resisitor removal was much easier & quicker this time.
Did test a 1.25 GHz DP MDD CPU first and that worked fine… before finally changing to the 1.42.
[So peeperpc, you may have a faulty 1.25 CPU.]

However, the machine would not initially power up with the first PSU installed, even after it had tested functional in it’s previous home. So I swapped out that 360 Watt AcBel PSU for a 400W AcBel  and it booted the 1.25. THEN I swapped in the 1.42.

And then… thinking that maybe the 1.42 required a 400W PSU… I checked the first 867 that I’d successfully transitioned and it had a 360W Samsung PSU that booted and continues to run the first 1.42. Seems that I now have some more PSU work to do.

Or maybe Andy Cuffe down in Austin Texas will get some of my PSU work? (I still have another 1.25 GHz MDD to transition to a 1.42 and I’ll need a functioning PSU.)

*No problem with the J20 ram slot in this machine… like the first-transitioned 867 mobo. AND this one has four, 512MB sticks of PC2700 RAM installed



*Above image, after cleaning and polish.

[peeperpc]

This second MDD 867 MHz DP transition to 1.42 GHz went very well last night.
The R676 resisitor removal was much easier & quicker this time.
Did test a 1.25 GHz DP MDD CPU first and that worked fine… before finally changing to the 1.42.
[So peeperpc, you may have a faulty 1.25 CPU.]

Does this second MDD 867MHz DP also have 820-1445-A board and is the 1.25MHz DP CPU card 820-1472-A? I just want to be sure.

[FBz]

Yes, the 867 MHz DP MDD mobo is an 820-1445-A  ©2002

BUT NO… the 1.25 GHz DP MDD CPU is an 820-1452-A  ©2002
                             Not the 820-1472-A CPU that you mentioned.

Getting conflicting results from a search for your 820-1472-A.
Some note it as a 1 GHz DP while others say 1.25 GHz DP. (?)

I’ve another stock 1.25 GHz DP MDD CPU installed and will check
that CPU number next time I have the heatsink off for re-paste.
Other MDD CPUs here:

1 GHz               SP                        820-1497-A  ©2002
1.25 GHz          DP            also… 820-1497-A  ©2002
1.4 GHz            DP                       820-1470-A  ©2002

[peeperpc]

Yes, the 867 MHz DP MDD mobo is an 820-1445-A  ©2002

BUT NO… the 1.25 GHz DP MDD CPU is an 820-1452-A  ©2002
                             Not the 820-1472-A CPU that you mentioned.

Thanks for the info. 820-1452-A is not often seen. I don't know if it can be called a rare item.

Getting conflicting results from a search for your 820-1472-A.
Some note it as a 1 GHz DP while others say 1.25 GHz DP. (?)

An 820-1472-A card can be 867MHz DP, 1.0GHz DP, or 1.25GHz DP.

I think the code refers to some kind of board layout version. Since it is not always specific to the speed of the card. For more example, an 820-1470-A can be 1.25GHz DP (FW800 MDD), 1.33GHz DP (Xserve G4), or 1.42GHz DP.

I’ve another stock 1.25 GHz DP MDD CPU installed and will check
that CPU number next time I have the heatsink off for re-paste.
Other MDD CPUs here:

1 GHz               SP                        820-1497-A  ©2002
1.25 GHz          DP            also… 820-1497-A  ©2002
1.4 GHz            DP                       820-1470-A  ©2002

All 820-1497-A cards I've seen are SP. Are you sure this 1.25GHz is DP?