180MHz 603ev CPU upgrade for 5300ce?

[Trash80toHP_Mini]

With a little digging, I found this little sucker! It's a UMax C500 CPU 603ev 180MHz Module that I lucked into some quite some time ago. Started looking at grafting it into my 2300c some thirteen years ago, but the bus multipliers I was looking at didn't seem to give it much in the way of legs.



I think I was mistaken at the time, 603ev is the second generation of the CPU family, having a far more capable bus multiplier input:


If the 5300ce system is stable with a 9.29% clock chip, I think the 180MHz CPU might be stable running at 200MHz as well? That would equal the lower levels of the PCI Architecture 'Books. But at 180MHz, my 5300ce would be the champion of NuBus Architure 'Books by a fair margin. The 1400c tops out at 166MHz so an 8.43% advantage in the previous generation ain't too bad at all. With plausibility of a 200MHz overclock limit that would make for a 20.48% advantage over the 1400c/166.

First gen 603e CPU's PLL configs were limited to 4.1:1 and so topped out at 100MHz.


Recent gear shift makes a 5300c from the collection seems a better candidate for this than my beloved 2300c. With its beautiful 800x600 panel, PCMCIA on board. external display supporting VidCard and full gamut of I/0 hookups built right in, no dockage requirements the 5300ce is the winner hands down
_________________________________________________________________________________

OK, 'nuff said, hit me with the slings-n-arrows of rational thought about these assumptions/prospects?

 

[croissantking]

Interesting idea, but the 603ev is lower voltage than the 603e so it wouldn’t be a drop-in replacement.

Even if you could get around that, it would still be cacheless, therefore a 1400/166 would be much faster.

 

[Trash80toHP_Mini]

Dang! That'd be impact from the sling on a trebuchet.

I thought it was the same from the info I've found so far. Third gen is definitely low voltage. Have you got linkage to the 603ev datasheet? I've yet to find it.

Good point about the Cache, my runnig1400s are all G3 equipped, so benchmarking wouldn't be in the cards anyway.

Thanks for the reality check.

 

[croissantking]

Dang, that was about impact from the sling on a trebuchet.

I thought it was the same from the info I've found so far. Third gen is definitely low voltage. Have you got linkage to the 603ev datasheet? Apparently I've yet to find it.

I don’t, this is anecdotal (see the post here)

Good point about the Cache, my runnig1400s are all G3 equipped

Yeah, the Nubus PPC PowerBooks are hobbled by a missing L2 cache (except for later 1400s) and a 32-bit data bus. Clock speed isn’t the whole story.

 

[Franklinstein]

I wanted to transplant a 120MHz 603e into a 100MHz 5300 but i botched it pretty good. Not sure if it's salvageable but maybe one day I'll take another crack at it.
I went with a 120MHz part because a. it was available (from a ruined 52/53xx board) and 2. I figure I could pair it with a 40MHz bus clock up (from the original 33MHz).
Because of the way these machines are designed (it's all based on derivative clocks), I'm not sure it would tolerate a change to the bus clock without screwing something else up, so it may all be for naught anyway.

 

[croissantking]

I wanted to transplant a 120MHz 603e into a 100MHz 5300 but i botched it pretty good. Not sure if it's salvageable but maybe one day I'll take another crack at it.
I went with a 120MHz part because a. it was available (from a ruined 52/53xx board) and 2. I figure I could pair it with a 40MHz bus clock up (from the original 33MHz).
Because of the way these machines are designed (it's all based on derivative clocks), I'm not sure it would tolerate a change to the bus clock without screwing something else up, so it may all be for naught anyway.

If the PB 500 series and 1400 series can handle a 40 MHz bus I’d assume 5300 series could too. But there’s only one way to find out.

Alternatively you could just set the multiplier to run it at 117MHz.

 

[Franklinstein]

Yeah one day when I'm better at smd soldering I'll go back to it. IBM made some of their 603es in the very pretty blue epoxy package (which is what this chip is) so I'd definitely like to save it.

 

[croissantking]

I could potentially help with this, I've saved many of these kinds of fine-pitched chips.

 

[croissantking]

As I do more digging into these machines, I am pretty sure that the 5300 Series would additionally be firmware limited to a 4x multiplier – so 133MHz max on the 33MHz bus.

 

[Trash80toHP_Mini]

Oh well, a 133MHz 5300ce might still be a fun project? Wondering what features/hardware might be impacted by the kick in the @$$ of a 40MHz bus upgrade?

 

[croissantking]

Oh well, a 133MHz 5300ce might still be a fun project? Wondering what features/hardware might be impacted by the kick in the @$$ of a 40MHz bus upgrade?

Well you could clock the bus up to 40MHz for 160MHz operation and it would probably work if you can find a QFP 603e (not ev) faster than 120MHz - there might be some 133MHz chips floating about but I doubt more than that.

People have clocked the PB1400 to 40MHz successfully so I’m very confident the 5300 would take it as well since it’s the same architecture.

 

[Trash80toHP_Mini]

Is there a greater problem with the 603ev aside from the need for a 3.3V hack for the ev processor's lifted power input legs?

 

[croissantking]

Is there a greater problem with the 603ev besides the need for a 3.3V hack for that processor?

No, that’s the key issue. You’d need to incorporate a 2.5V level shifter somehow for VCC. You’d also have to see if 603ev is 3.3V tolerant on the I/O, if it’s not then that’s a total deal breaker.

If it’s just the power input legs then yeah you could do a messy hack!

 

[croissantking]

Here we go. I/O is 3.3V tolerant, just PLL and Core supply needs to be 2.5V. I guess you could do it by bending those 14 legs (13x Vdd, 1x AVdd) back and wiring them up through a level shifter.

 

[Trash80toHP_Mini]

Cool beans! So we're up to a 40MHz overclocking for 160MHz with onboard multiplier setup?

Habitually clueless here:
Is there any possibility of hotwiring resistors to the the CPU leggage kludge to enable the 603ev 5.5x multiplier to yield a very slight 183MHz overclock of this 180MHz CPU? I was looking into that hotwired nastiness to bypass the (unidentified at that time) on board multiplier in Duo dreams about a slight overclocking of onboard CPU (110MHz) and extended 603ev hackage.

 

[croissantking]

Cool beans! So we're up to a 40MHz overclocking for 160MHz with onboard multiplier setup?

Yes

Habitually clueless here:
Is there any possibility of hotwiring resistors to the the CPU leggage kludge to enable the 603ev 5.5x multiplier to yield a very slight 183MHz overclock of this 180MHz CPU? I was looking into that hotwired nastiness to bypass the (unidentified at that time) on board multiplier in Duo dreams about a slight overclocking of onboard CPU (110MHz) and extended 603ev hackage.

It’s not that you can’t set the correct multiplier on the logic board; you can but the real issue is that you’ll get no boot. For whatever reason, support for multipliers above 4x needs to be enabled in firmware.

If you knew what you were doing you could try to disassemble the old and new PB1400 ROMs and analyse what they changed to enable 4.5x-6x support and then modify the PB5300 ROM with those changes. This would not be for the faint of heart though.

Crazy thought, maybe a 1400 Rev B ROM would actually boot a 5300?