Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
HiFi exercises

Boo Apple! Leaving a hole anywhere near the speaker's moving cone effectively creates acoustical short-circuit. Low frequencies go down the drain immediately. Speakers should be mounted flush with fascia and not 1/10" away. Mounting them on rubber suspension (as they are) is also a bad idea.

It's not a HiFi but still!
 
eServe

Is there too much of me here? :cool:

I am fighting the temptation to put it all into pizza or 1U rack box. Features:
  • 1.33GHz
  • 1GB RAM
  • 3 internal HDDs (1 system + 2 RAIDed)
  • DVD±RW
  • 2 x standard SVGA ouputs
  • almost silent
  • boots into OS 9 and OS X
I have identified space on the motherboard for network LINK, SPEED, TRANSMIT and HDD/CDdrive activity LEDs. Can also run leads and mount them on the front panel like power-on LED. No drilling needed thanks to translucent body. Anybody interested?
 
eMac bus speed control

Solved! Follow the logic!

Investigate Soonbaek's picture of an old eMac's PLL clock IMIC5003BY.
Assume IMIC5003BY == CY28503 (numbski's datasheet is here) e.g. look at Vdd/Vss pins.
Conclude:
R301 = FSEL0
R302 = FSEL1
R303 = FSEL2

CPU/SDRAM bus frequency is:
Code:
MHZ R301 R302 R303
100  no   no    no 
125 yes   no    no 
133  no   no   yes 
140 yes  yes    no 
150  no  yes    no
Soonbaek, just in time for the weekend:D
 
I'm afraid I'm not following your logic here....

vdd/vss pins don't run anywhere conclusive. According to your logic, shouldn't R303 be bridged by default?

Not to mention that according to the data sheet, 133mHz should be:

no yes yes

not

no no yes
 
Wait a sec....you're presuming 1 == UNBRIDGED on the data sheet, aren't you???

If that's the case, I need to go crack my iMac open, because I might have the 133mHz problem staring me in the face...
 
soonbaek shows old eMac's 100MHz bus photo, in new 133MHz eMac R303 is installed.

Datasheet columns are reversed order - FSEL2 FSEL1 FSEL0 and show logic level. Level 0 means it is pulled to the ground with the resistor/jumper.
 
I've got it now. The table on the data sheet is backwords to what you typed up. Both ways. 0 is bridged, 1 is unbridged, and you typed it up physically backwards. Ugh. Okay, time to go tear apart my iMac. Back soon with news. :p
 
Okay, I'm sitting here with my iMac open. I can't see any sets of 3 open resistors that my can pass continuity with my DMM. I'm getting the sinking feeling that it's beneath my modem, which I can't remove because it seems to take a super-small hex wrench to get at. :(

The only other area I can see that it might be is near my PLL resistors, but there are no open sets of 3 there. This is driving me nuts. That, and my camera is taking all either blurry or totally washed up pictures. a 2Mbit camera no less. ;\

I'll try to put them up anyway, i don't know what use they'll be though.
 
The best I can do is show you my dillema from the existing images:

fsb_pins.png


As you can see, the pins don't go anywhere, and I've been probing all over the place trying to find where they spring up. :(
 
I wonder....Leo, suppose I can't find the resistors. Is there anything preventing me from soldering some kynar wire to a known ground point, to a dip switch, and solder directly to the chip's pins? It'd be tight (I don't savor doing it), but it'd work, right? All we're doing is grounding the pins...
 
Leo,

I have same assumption as well. I have requested pin layout for IMIC5003BY to Cypress for validation.

Unfortunately, I have a wedding to attend to this weekend and I'll be out of town. :(

Hopefully Cypress will reply soon so that you guys can benefit from this discovery. Thanks for all your help guys!

Soon
 
Originally posted by Numbski
<snip>
That, and my camera is taking all either blurry or totally washed up pictures. a 2Mbit camera no less. ;\

Most likely you have to enable macro on your digicam to get close-up pictures. The macro setting is (on most cam) the flower symbol. Also, most digicam requires (minimum) 6-12" to focus even with macro setting. Try holding the camera back. Good luck!
 
I had it in macro mode, but you're right. I was too close. Next time it's open I'll try backing off some.

I'm carrying on coversations in two places. I was just posting on Tycho's boards, and he just verified that I could indeed affect the FSB using the bypass method I just described. Note that I don't like the idea of doing it. Those pins are tiny and too close together for my taste. I'm really hoping he actually has that board, and perhaps can do some educated guessing so far as where to probe around. I'm flying blind in that respect. I've tried near the CPU, near what I think the clock crystal might be.

I guess I know what to do as a last resort now anyway.
 
Re: eServe

Originally posted by lbodnar
Is there too much of me here? :cool:

I am fighting the temptation to put it all into pizza or 1U rack box. Features:
  • 1.33GHz
  • 1GB RAM
  • 3 internal HDDs (1 system + 2 RAIDed)
  • DVD±RW
  • 2 x standard SVGA ouputs
  • almost silent
  • boots into OS 9 and OS X
I have identified space on the motherboard for network LINK, SPEED, TRANSMIT and HDD/CDdrive activity LEDs. Can also run leads and mount them on the front panel like power-on LED. No drilling needed thanks to translucent body. Anybody interested?

O-yea!!! This is one I would love to do... : )
 
Numbski! I am pleased to announce the location of iMac's CPU bus setting resistors! Tadammm!

They are under SODIMM module!

FSEL0 0=R627 1=R625
FSEL1 0=R635 1=R628
FSEL2 0=R637 1=R636

The bad thing is that on my Flat Panel iMac the whole PLL chip is missing so you are lucky!
 

Attachments

  • imac-cpu-bus.jpg
    imac-cpu-bus.jpg
    88.2 KB · Views: 590
reverse engineer extraordinaires

wow guys

What say you/we figure out how to OC every Mac CPU and bus and make a website?
 
WTF??? Under the sodimm? :p

Okay, okay. So Leo, beverage of your choice. Give me an address and it's all yours. :)

Sometime tomorrow I'll take down my machine and run some jumper wires. Then the fun will begin.

I'd have done this already, but I haven't found any good dip switches to use. Looks like I'll be using Radio Shack's.

This'll be fun. :)
 
Just for clarification, you highlighted 6 pairs. In order to change the setting, it's not a matter of on or off, you have to move the resistor from one location to another? Just a little confused as to how this is supposed to work out.
 
Originally posted by Numbski
WTF??? Under the sodimm? :p

Okay, okay. So Leo, beverage of your choice. Give me an address and it's all yours. :)

Sometime tomorrow I'll take down my machine and run some jumper wires. Then the fun will begin.

I'd have done this already, but I haven't found any good dip switches to use. Looks like I'll be using Radio Shack's.

This'll be fun. :)

Check out Digikey. They cary a huge supply of DIP Switches, not to mention EVERYTHING else you could ever want... Digikey DIP Switches :D
 
:) Numbski, I'm glad we are moving forward!

In theory you might need to ground only FSEL0 to get 125MHz and FSEL2 for 133MHz so the first thing to do is to measure resistor values for R625 and R636. If they are zero then you would need to move them to R627 and R637 respectively. But if they are something between 1kOhm and 10kOhm, then they will act as typical pull-up resistors and you will only need to short R627 and R637. Even more since the ground wire is the same you might get away with only three wires and two switches.

Do you get the idea?

Measure the resistance first! Hell, it is all a theory yet! Can you confirm that you have R625 R628 and R636 mounted on yours? If yes, all the rest is OK.

The same thing applies to main CPU speed settings. You can have 1-10kOhm pull-ups and just ground frequency selection pins when needed with a simple switch.

I just had a look at CY28503 datasheet - it already has internal 250KOhm pull-up resistors. Quite high resistance though... Maybe external ones are needed to make sure it will survive nuclear explosion EMI? :)
 
Do I get it right that Apple seems to have two sorts of FP 800MHz iMac - one is just that and the other is "133MHz ready"? Japanese pictures and my iMac have the whole bus clock PLL chip missing and Numbski has it?!

The other thing - if I recall correctly from PC days, the CPU has a better chance to run at specific speed with higher bus speed and lower clock multiplier than at lower bus speed. Is it correct?

If stability is just the heat dissipation issue then it should not matter as Mototola documentation and common sense says that heat dissipation is approximately proportional to the processor speed. Even more the bus subsystem runs at higher speed too so there should be even a little bit more heat at 133MHz bus.

However internal CPU PLL comparator circuit will run at lower ratio so the chance of having a glitch there is lower too. It is actually quite interesting, WHAT exactly causes glitches in CPU at a too high speed?:confused:
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.