eMac 800Mhz -> 1.33GHz conversion

MrFX · Mar 8, 2005

Some more...

lbodnar · Mar 8, 2005

madmaxmedia said:
Do you think you would be able to identify the jumpers for overclocking the G3 eMac, if MrFX were to post a higher resolution photo of the motherboard?

Steve, I think I have found one out of five:

R270

If the bottom pad of R275 is connected to the ground then R275 is the second one. It would help if somebody could measure few resistor values there. Anyone?

lbodnar · Mar 9, 2005

That 2.5v off downconverter

I have spent some time on it now and the only visible consumer of +2.5v supply that comes off downconverter is two Video RAM chips from Hynix.

I want to get rid of DCB altogether so I was going to do it cheap and nasty by getting 2.5v from 3.3v via an inline diode do drop 0.7-0.8v however I am now starting from making a proper voltage regulator on LM317 and a couple of resistors (think 2.5v output 7805). Input voltage should be 5v, as 3.3v indicated on the picture is too low for 2.5v output.

Metal mounting tab on LM317 is connected to its output (2.5v) so the heatsink (if used without mica isolation) should not be in electrical contact with metal case or any other obect connected to system Gnd.

Do you guys think average project builder will be OK with that?

madmaxmedia · Mar 9, 2005

Thanks Leo!

So does it look like the various jumpers are scattered on the board, rather than aligned in a row?

How are you able to figure it out?

Thanks, Steve

lbodnar said:
Steve, I think I have found one out of five:

R270

If the bottom pad of R275 is connected to the ground then R275 is the second one. It would help if somebody could measure few resistor values there. Anyone?

lbodnar · Mar 9, 2005

madmaxmedia said:
So does it look like the various jumpers are scattered on the board, rather than aligned in a row?

How are you able to figure it out?

My thinking was the along the following lines...
Apple uses connector to either monitor or control CPU PLL multiplier. It is J500 on Gen1 and Gen2 eMacs and J11 on Gen3.
You can see that each of J500 pins 1 to 5 is mapped to one of configuration resistors.
If I would be Apple I'd carry over the pinout of the connector to Gen3 intact.
So we assume top 5 pins of J11 should end up on a configuration resistor pads.
Now we can see that pins 2 and 4 have traces leading to blocks of four resistors each. At some point in time Apple obviously was going to implement live switching of PLL multiplier using transistors Q11, Q12 on Gen3 or Q1501, Q1502 on Gen2 but left that idea out together with associated extra two resistors per each PLL_CFG bit.

So we end up with at least two (one pull-up to +3.3v and one 0 ohm "jumper") or four (two extra missing ones from legacy switching circuit) per each bit.

1.25GHz CPU on 167MHz bus should have a pattern on J11 like

+< pin 1 J11
+< pin 2
+< pin 3
-< pin 4
+< pin 5

where "+" means jumper installed and should measure 0 ohms to Gnd with a ohmmeter.

Now compare pins 2 and 4 and their associated resistor blocks - indeed we have R270 present and R275 missing.

That's it so far...

MrFX · Mar 9, 2005

Hi again,

i will measure the values today. Leo, which should i take?

So long.
MrFX

jspivack · Mar 9, 2005

as an "average project builder"...

lbodnar said:
I am now starting from making a proper voltage regulator on LM317 and a couple of resistors (think 7805 for 2.5v output).

Do you guys think average project builder will be OK with that?

as an "average project builder", I have to say yes. although I don't have chips like that floating around my house, after a couple minutes of surfing I see that I can get them both at my local shop and online at Radiospares. Piece of cake. What sort of load do you anticipate? I see that it comes in at least 0.1A and 1.5A versions...0.1A would sound like enough to me for a couple of RAM chips, but I'm a Mech. Eng., not an Elec. Eng. ...

lbodnar · Mar 10, 2005

jspivack said:
Piece of cake. What sort of load do you anticipate? I see that it comes in at least 0.1A and 1.5A versions...0.1A would sound like enough to me for a couple of RAM chips, but I'm a Mech. Eng., not an Elec. Eng. ...

I will use 1.5A version of LM317 or LM317A in TO-220 case (the one with a metal tab and a hole on top) it is easier to work with and might not even need a breadboard to put on.

Datasheet says two chips can draw up to a maximum of 0.9A when in continuous burst mode whatever it is. If we get our power off +5v (3.3v is too low for 2.5v output using LM317) then at 0.9A the dissipated power is about 2W and with 50C/W thermal resistance of TO-220 case it might get quite a bit hot (120C!). So it looks like some simple heatsink will be needed.

lbodnar · Mar 10, 2005

MrFX said:
i will measure the values today. Leo, which should i take?

Hi MrFX, I can definitely say R270 and R275 are two of the "jumper" resistors.

To find the rest we need to find everything that is connected directly (0 ohm) to J11 pins 1,3 and 5 and to pin 2. This is a lot of probing but without this I am stuck... I also suspect that pins 1,2,3 and 5 are all connected together

MrFX, is there a chance to get the same excellent quality picture of the are on the reverse of J11. I just seem to be totally missing 12 resistors! By the way rar file was nearly the same size as tiff so if you can, try to save it as jpeg. We'll get it in the end.

Leo

MrFX · Mar 10, 2005

Hello Leo,

i will try this, but tomorrow...

I've packed the picture, because some browsers display tif-images instead of download... but i can try the reverse side in jpeg... it was a tif, because i've take it in raw-mode... so there was no convert necessary.

May the force be with you...
MrFX

lbodnar · Mar 10, 2005

Yes, yes, Yes!

We can throw downconverter board away. eMac will run without it!

I have temporarily tapped in power by soldering directly to the reverse side of DCB connector on the logic board and it powers up and works! My impromptu 5v->2.5v converter (see post number 379 above) gets pretty hot so I might use a better heatsink in the final version but schematic will stay the same (I have used LM317T - can't remember how it is different from LM317A or LM317, was it Vout precision?)

I supply +12v, 5v, 3.3v off a standard 120W ATX power supply.

That's basically it! eMac headless is born. Now where one can purchase bare eMac logic boards for nothing?

netjosh_granada · Mar 11, 2005

eMac Bus overclocking???

hi all again.

I'm already planning a wonderfull case for this eMac project, but I'm planning to make it fit on a standart ATX case too.

I have a question that only 1st gen eMac owners can answer (hopefully lbodnar has mine):

¿Could it be possible to raise the bus speed from 100 Mhz to 133 Mhz or even more?

I think that if it's possible to change CPU multiplier through PLL resistors, there has to be another set of resistors around the logic board to increase bus frequency without risking the CPU (i.e. 700 Mhz = 7 x 100 Mhz would be 933 Mhz = 7 x 133 Mhz --> at first should try 666 Mhz = 5 x 133 Mhz, after that... 933 Mhz or 1066 Mhz would be perfect ;-) )

I know I should need brand new memory on a 133 Mhz bus, but as I only have 128 Mb PC100 this is not a problem.

If anybody has any ideas on how it could be done, let us know...

jm2005 · Mar 11, 2005

Hey Ibodnar. Great Job. Do you think I can upgrade my Powermac g4 400mhz-1.5 ghz. Am I pushing my luck here?

Jm 2005

guylan · Mar 14, 2005

lbodnar said:
Yes, yes, Yes!

We can throw downconverter board away. eMac will run without it!

I have temporarily tapped in power by soldering directly to the reverse side of DCB connector on the logic board and it powers up and works! My impromptu 5v->2.5v converter (see post number 379 above) gets pretty hot so I might use a better heatsink in the final version but schematic will stay the same (I have used LM317T - can't remember how it is different from LM317A or LM317, was it Vout precision?)

I supply +12v, 5v, 3.3v off a standard 120W ATX power supply.

That's basically it! eMac headless is born. Now where one can purchase bare eMac logic boards for nothing?

IMPRESSIVE!!! That is the way to do it baby!!!!
Nice job on the 2.5v converter.

Ok one question about the 2.5v(sense) is this also required to be connected?

lbodnar · Mar 15, 2005

guylan said:
Ok one question about the 2.5v(sense) is this also required to be connected?

It is connected to 2.5v line on the ligic board and is used to feed 2.5v back to the DCB for voltage regulation. Just use it as a second 2.5v wire. It can be fed back into LM317 regulator to get better quality 2.5v if this becomes necessary.

lbodnar · Mar 16, 2005

Vcore adjustment on Gen2 eMacs

Altering CPU Vcore voltage

I never found time to play with it but if someone wishes to try to increase CPU Vcore voltage to achieve higher clock speeds or decrease it to reduce generated heat, here is the information required:

Voltage controller datasheet - http://www.onsemi.com/pub/Collateral/NCP5322A-D.PDF
Controller chip is located on the front-facing edge of the board to the left of Airport connector on the bottom (CD drive) side.
Refer to page 5 for possible settings in the range of 1.1v - 1.85v to the accuracy of 0.025v.
Configuration resistors on the logic board as pictured below are in the order...
Vid0 Vid1 Vid2 Vid3 Vid4 (datasheet table is in the reverse order - Vid4 .. Vid0)
Resistor present = 0, missing = 1
Attached picture for Gen2 shows default setting of Vcore = 1.55v (hirez)

lbodnar · Mar 16, 2005

Altering CPU Vcore voltage

Generation 1 (GeForce graphics) eMacs use different voltage chip:
http://www.semtech.com/pdf/sc2422b.pdf (1.1v - 1.85v range)
Defult voltage is Vcore = 1.75v

rmanger · Mar 16, 2005

Hi everyone!

I've been lurking on these forums for some time itching to overclock my Gen3 eMac. I hope that MrFX submits those pics soon, because I'm starting to reach my limit...

lbodnar, will increasing the CPU core voltage increase CPU performance? Or does this give only more potential performance?

lbodnar · Mar 17, 2005

rmanger said:
I've been lurking on these forums for some time itching to overclock my Gen3 eMac. I hope that MrFX submits those pics soon, because I'm starting to reach my limit...

Apparently not everybody knows that there is an "education" Generation 3 eMac with 1GHz processor, basically underclocked Gen3 1.25GHz eMac. Tim and I have managed to overclock it to 1.33GHz by shorting pads on J11 connector. Unfortunately it is not going to work with stock 1.25GHz model because to overclock it one needs to remove the configuration resistors which needs finding them first...

will increasing the CPU core voltage increase CPU performance? Or does this give only more potential performance?

Yes, increasing the supply voltage makes semiconductor's switching time shorter potentially giving the chance to operate at higher frequency. One still needs to increase clock frequency though! I can't remember exact formula but Freescale hints in 7447 datasheet that 10% Vcore voltage drop causes 20% maximum frequency drop ("derating"). It must be then that MaxFreq ~ Vcore^2. It obviously works both ways so upping Vcore by 10% you can potentially squeeze 20% more speed. Remember though that the CPU will generate extra 20% heat due to voltage increase and 20% more due to clock increase.
Temperature ~ Power ~ Vcore^2 * ClockFreq

MrFX · Mar 17, 2005

Sorry for the delay... i was busy the last days... hope to take the picture today...

So long.
MrFX

lbodnar · Mar 17, 2005

Here is Gen1 eMac (700MHz) running without downconverter board. I have combined 2.5v LM317 regulator and a DCB connector together on a small breadboard. They are just regularly spaced jumper pins. Blindmate connector still needs 12v and 5v supplied to it.

It is not easy to do the same on Gen2/Gen3 eMacs as they have male DCB connector on the logic board.

MrFX · Mar 17, 2005

Hello Leo,

you can download the backside-picture from J11 of Gen3 eMac board here.

So long,
MrFX

lbodnar · Mar 17, 2005

Gen 3 (1.25GHz) Overclock details

MrFX said:
you can download the backside-picture from J11 of Gen3 eMac board here

Great! Here they are then:

Code:

                         1.25  1.33  1.42  1.50  1.58  1.67  1.75GHz
R658 (reverse side)        +     -     +     +     +     -     -
R270 (J11 side)            +     -     -     -     -     +     +
R673 (reverse side)        +     +     -     -     -     -     +
R275 (J11 side)            -     +     +     -     -     +     +
R689 (reverse side)        +     +     +     -     +     -     -

As usual, "+" means either resistor or solder bridge is present and "-" is an open contact or removed resistor.

Just to let you know, Open Firmware might not correctly show 1.58GHz speed (displayed as 750MHz) because no Mac yet runs at that speed. This issue is the same on Mac mini - it has the same CPU and very similar architecture. It is purely indication issue and the solution can be found here

lbodnar · Mar 17, 2005

jm2005 said:
Hey Ibodnar. Great Job. Do you think I can upgrade my Powermac g4 400mhz-1.5 ghz. Am I pushing my luck here?

Hi! I think you can replace the CPU with the one from Gigadesign, Sonnet or Powerlogix. They offer up to 1.8GHz drop-in replacement CPUs but you have to ask folks who had experience with them. I replaced one in Gigabit Ethernet for Dual 1.47GHz. It just worked. Try to open a new thread or look for existing one. There is a CPU upgrades database on xlr8yourmac

rmanger · Mar 17, 2005

Hooray!

Great job, lbodnar and MrFX! I will try to overclock my eMac soon. Wish me luck on good results!

eMac 800Mhz -> 1.33GHz conversion

macrumors newbie

Attachments

macrumors regular

macrumors regular

Attachments

macrumors 68030

macrumors regular

macrumors newbie

macrumors newbie

macrumors regular

macrumors regular

macrumors newbie

macrumors regular

Attachments

macrumors newbie

macrumors newbie

macrumors newbie

macrumors regular

macrumors regular

Attachments

macrumors regular

macrumors newbie

macrumors regular

macrumors newbie

macrumors regular

Attachments

macrumors newbie

macrumors regular

Attachments

macrumors regular

macrumors newbie

Our Staff