5x0(c) CPU card differences

[aladds]

Looking at this: http://hackaday.com/2016/08/22/ask-hackaday-calling-all-68k-experts/

I'm wondering if there's just a set of jumpers telling the Mac to look for an FPU or not...

But I can't find pictures of the back of a 550c card to check if what look like jumpers between the LSI chip and the connector are indeed jumpers, and whether they should be set differently if they are. The 540c card seems to have all three unpopulated. (R700/701/702, see 00:30: https://youtu.be/82OkgLW_lnY?t=30s )

Can anyone take a scan/photo of a 550c CPU card to confirm/deny this?

Thanks!

Adam

 

[Paralel]

I have a 550c CPU daughterboard... that I actually put into a 540c. I never bothered to check if the FPU was indeed functional. It might be a limitation of the CPU daughterboard in the 540c itself. I'll have to check, and if it is registering, take that old bitch apart again and take pics of the CPU daughterboard for them.

 

[Elfen]

You need a NOS 68040, not a cheap Chinese clone that they claim is something it is not. There many horror stories, including a few here where the CPU/FPU was replaced and "BOOM!!!!" In fact one of them was in the FPU for the Classic II a couple of daring young members' here created and sold to the those who wanted one! It works beautifully (I need to save my pennies and get one for my Classic II)!

There is no switch (that I know off) to show that there is a FPU in the '040 or not in the 5x0 CPU Card. there should be none as you can swap out the 68LC040 out of a LC/Performa 475, Centris or other Mac using the LC processor and put in a fill in a full '040 in its place and the system will run better; there are no switches or jumpers to set on any of these machines for such an upgrade to be done. Neither should the 550's CPU Card vs. the 540's CPU Card. If anything, such jumpers are for clock/bus speeds, nothing more.

 

[Paralel]

Yep, I was the one who got a bad Chinese knockoff when testing the Classic II FPU/ROM card proto. Sucker nearly melted right in the socket. I'm lucky it didn't damage my system. Fortunately, the card was ok and everything turned out alright in the end when I replaced it with a NOS Freescale 68882 in my Classic II.

After just testing my 540c with the 550c CPU daughter card in it, it has something to do with the daughtercard itself. My 540c recognizes the FPU, and it runs great. In the FPU tests I ran it through the only 68k reference system that handily beat it was the 840AV. So, we know its not the CPU, nor anything else in the system now, it has something to do with the daughterboard itself, either something on it, or the ROM somehow.

Now I have to consider taking my system apart to help this guy. I'm not sold on it. Everytime I take it apart, something goes wrong. Its in a good place now, so you can see my hesitation. However, I would like to solve this mystery. Its a tug of war in my head now. I think the first step should be me dumping the 550c ROM, and him trying that.

I left a comment to the original article, but I don't know if it went through.

 

[Paralel]

Its looking like the guy got another remarked chip. The mask on the chip does not match the correct mask for the FE33A 68040 processor from that time period, but the FE33V version, which does not have an FPU.

Really makes me glad I just went with a 550c processor card right from Japan.

 

[rsolberg]

If this is in fact another FE33V, that's maddening. Would installing a PGA socket be feasible? I don't imagine there's much clearance to accommodate a socket and PGA packaged CPU in the PowerBook.

 

[Paralel]

Probably not enough room. The screws to hold the heatsink in place are pretty short, a socket + PGA chip would be quite an offset.

 

[techknight]

Overall this seems like alot of effort for a lowly little blackbird. 

NetBSD? Whats so special about that? out of curiosity. It wouldnt seem to be any more supported on that ancient hardware than Mac OS would. 

 

[Paralel]

Some of us just love or blackbirds. I love my pimped out blackbird. Same with my Classic II.

Plus, people love a good mystery, even an old one. When we know a 550c CPU daughterboard has a functional FPU, and then a transplanted 68040 on a 540c CPU daughterboard doesn't appear to have a functional FPU, people want to solve the issue.

Same as Alaska and I building the FPU/ROM board for the Classic II. It was to see if it could be done (Since no company had appeared to successfully build an FPU/ROM board with the proper socket for maxing out the additional ROM space)

 

[techknight]

Yea I hear ya. I have been trying to piece together a good blackbird for a long time, and just cant get the parts I need reasonably. 

I have 2 or 3 carcasses, 2 of which each house a PPC CPU. 

I wanted a 540c LCD and I ended up getting one, but the backlight is nice and pink, the LCD has lots of hours on it. and the bottom peice on all of mine are broken. hehe. 

Maybe some day. 

 

[quarterturn]

Hey there! I'm OP from the hackaday.com article. I appreciate everyone's effort to look into the CPU issue for me. I am pretty sure, based on discussion on hackaday, that I got screwed with a counterfeit CPU. To be clear, the on the first try I screwed myself by not looking carefully enough at the docs, which stated that 'V' series isn't 'LC' but nevertheless does not have an FPU.

I'm a long-time Mac user from way back in the Mac Plus days. I'd picked up the 520c at a hamfest a few years back for just $10, and thought it would be fun to see if it could be like a portable Quadra, if the CPU could be upgraded. Quadra series was my favorite - a full 32-bit architecture, not much in the way of artificial limitations, decent A/V capabilities, and of course a CD-ROM drive, which was very cool at the time.

I don't think of the CPU upgrade project as a completely wasted effort, in that I didn't kill the powerbook and I learned that it is possible to unsolder and solder tight-pitch QFP parts.

For the person who asked about "why netBSD", it has a certain nostalgia for me as it was the first unix I ever installed for myself - back in '94 I think - on a IIsi. Of course, it didn't run very much, as at the time, netBSD required an FPU, and the IIsi didn't have one, so eventually it'd do something where it expected it and it would kernel panic.

 

[Paralel]

I can tell you from the benchmarks I have done, a blackbird running a full 68040 is indeed equivalent to a Quadra. The only 68k system that really beat the crap out of it was the 840AV, but, that's to be expected.

Also, I totally understand about trying to get one blackbird from many. Mine is probably at least 5 blackbirds total... But I was also pursuing a true 540 screen, and all of them had the tunnelvision problem, so that ate up a few.

 

[CC_333]

I guess having "frankenstein" PowerBooks 500 isn't unheard of, then?

I have made two decent machines out of four: a 520c in fair condition, and an otherwise very nice 540c that needs a new LCD/lid assembly. I had it sort of working with a 520 LCD, but it was no fun to use (the 520 LCD's quality is quite low, even when compared to the 520c, although out of fairness, mine is probably broken).

That being said, if anyone has a need for extra parts, I have a decent heap. A logic board or two, some keyboards, CPU cards, cables, screws, and various other odd bits. Oh, and I have three 2.5" SCSI hard drives, all fully functional (so far; I realize failure is inevitable).

And if anyone happens to have a real 540c LCD and would like to trade for something, let me know!

c

 

[reukiodo]

I am still interested in replacing the 68LC040 in my PowerBook 520c with a proper 68040 for the hardware FPU. Was there any update other than finding out both replacements were fakes? Are there any places to get a real MC68040FE33A?

What about getting a MC68040FE40A ? Would that require getting a 20MHz oscillator also?

 

[rsolberg]

If you can find a genuine MC68040FE40A, then that might be a good option.

The 68040 series operates at half the frequency of its input clock. That means a 20MHz oscillator would run the CPU and bus at only 10MHz.  In the case of the 520c, a 50MHz oscillator is used to set the 68LC040 and its bus at 25MHz.  If you were to replace the original CPU with a 33 or 40MHz-rated version, the new CPU would still operate at 25MHz.  Replacing the oscillator with a 66MHz unit would yield 33MHz CPU and bus.  80MHz oscillator would give you 40MHz.  Keep in mind the CPU and bus are at 1:1 and some logic board components may not work reliably or at all at the higher speeds.  RAM and VRAM are just some of the usual suspects.  It seems pretty likely to me that the 520c would work at 33MHz with a 66MHz oscillator, but I'm doubtful that 40MHz is achievable without issues.

Remember that heat dissipation climbs significantly with a small frequency increase.  Also note that the "full" 68040 with FPU dissipates significantly more heat than an equivalent 68LC040 at the same clock.

If you can find a genuine MC68040FE40A for a reasonable price, it's likely to run cooler with a 66MHz oscillator (at 33MHz) than a MC68040FE33A would.

 

[rsolberg]

Whoops I'm wrong!  The 5xx series CPU daughtercards use a 4x frequency multiplier IC between the oscillator and CPU.  This means that a 12.5MHz oscillator's clock is multiplied x4 to 50MHz for the CPU's external clock input, running the CPU at 25MHz.

I would still aim for 33MHz for stability's sake, so that would require a 16.7MHz oscillator.

 

[reukiodo]

Where did you find the oscillator info for the 5x0s?

Is this not the part we are seeking?

http://www.kynix.com/Detail/681375/MC68040FE40A.html

Does anyone have experience with kynix?

 

[Franklinstein]

Whoops I'm wrong!  The 5xx series CPU daughtercards use a 4x frequency multiplier IC between the oscillator and CPU.  This means that a 12.5MHz oscillator's clock is multiplied x4 to 50MHz for the CPU's external clock input, running the CPU at 25MHz.

So this is incorrect, as is Everymac's disclaimer that the '040 is not clock-doubled. According to both Apple (in pretty much any Developer Note concerning an '040 sytem) and Motorola/Freescale/NXP documentation (68040 User Manual) the 68040 requires two clock inputs: system and 2X system for internal clocks. Thus an 040 marked as 25MHz runs on a 25MHz bus with a 50MHz internal clock. Why Motorola chose not to advertise this on most systems is beyond me, but they did for the QFP chips used in PowerBooks: those usually have 25/50MHz or 33/66MHz stenciled on top.

The '040 Macs usually have several oscillators, with the system bus oscillator running at the same or half frequency of the desired system bus (so a 25MHz system would use a 25 or 12.5MHz oscillator), though PowerBooks may use a single oscillator and a clock generator circuit somewhere to reduce component count and power draw. I haven't really investigated too in-depth because these things run warm enough as it is so overclocking isn't a great idea. Plus I don't have the specialist equipment to attempt this sort of surgery just yet.

As for not recognizing the FPU on a properly replaced full '040, I can't imagine why it wouldn't, unless there's some difference between the ROM in the PB 550 and others (it assumes there's no FPU so doesn't check? Simply marks it as absent?) or a gestalt ID setting that blocks/ignores it. Has anyone run benchmarks to try? Assuming nobody finds a difference in physical configuration between the daughter cards, it may be worth it to compare ROM dumps. Has anyone bothered to check the PowerBook 190? It uses most of the same chipset as the PowerBook 5x0 series and would also be awesome to swap a full '040 into.

Unfortunately I'm not sure where to get a full '040 in a QFP: as is noted in the UM above, the 68040V has no FPU (it's a newer, lower voltage part developed well after Apple was putting them in PowerBooks) and the QFP version of the full 68040 is no longer listed as available.

 

[Paralel]

I honestly wouldn't try the a full 040 @ 40MHz in the blackbird. The 550's are known for shutting down due to thermal issues on warm days in Japan. If you wanted to go to 40 MHz you's likely have to implement some kind of active cooling system, and the blackbird was never designed with an active cooling system in mind, as such the airflow through the system was not optimized for that, so its hard to say how well it would even work.

If anyone ever wants a dump of the 550c CPU daughtercard ROMs to disassemble or mess around with let me know, I'll dump them for you. I can't guarantee it would be prompt, but I would eventually get it done.