• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

A 'once and for all' declaration regarding L88M mask MC68040 CPUs

jessenator

Well-known member
^^^^ For transparency: the CPU previously tested is a XC68040RC33M. so thought I'd make that clear.


E31F XC68040RC25M
p0SxQm0.png

:|
 

jessenator

Well-known member
D50D XC68040RC25B
CXEABIW.png

As should be clear, at 40 MHz the Mac refused to POST. I didn't get a sad mac and I didn't have a speaker plugged in. And with that, I didn't feel the need to try to test at 44 MHz. I can't say whether it's beyond the thermal design or what.
 

MrFahrenheit

Well-known member
I just purchased another two 68040 40mhz CPUs off eBay. They arrived and are L88M mask. These chips look and feel different than all of my other 68040 CPUs.

In fact, the surface of the package, where the silkscreen is done, feels quite rough and plastic. Examining the silkscreening closely, it appears to be authentic.

I have not yet tested them. I note they appear to be brand new and never used. I had bought 2 from this seller prior, advertised as new, and they were a bit banged up and some pins slightly bent. I mentioned this to the seller and received a 50% discount and a reorder opportunity for new ones, which I took.

I wonder what others with L88M mask experience as far as surface texture on top of the CPU package itself. I’ll examine with a microscope later when I can.
 

Attachments

  • 2CA33B48-04E4-41C3-96B0-D90190CFAD52.jpeg
    2CA33B48-04E4-41C3-96B0-D90190CFAD52.jpeg
    1.2 MB · Views: 28
  • 9EFE1997-FB0E-45EF-9FCE-141BFD70B236.jpeg
    9EFE1997-FB0E-45EF-9FCE-141BFD70B236.jpeg
    1.3 MB · Views: 29

MrFahrenheit

Well-known member
Well, this may be a sign of a fake or a change in labeling process. I am unsure at this point.

See attached photo. The CPU has a plastic type coating on the surface and it can be easily removed with a knife.
5A7617F5-2720-4D14-921B-C7B38EAC256C.jpeg
 

MrFahrenheit

Well-known member
Well, one of the two CPUs is missing a pin. Ugh. This pin seems to be required because it will not POST with this pin missing.

The surface is almost like a plastic, and can be removed with a knife edge. The lettering and branding and batch ID are all laser etched into the plastic layer. Removing it near letters reveals nothing underneath.

If the seller refunds me and let’s me keep the CPU I will remove the entire surface to find out what is going on here.
 

Attachments

  • 2233603A-087E-48D6-8394-97FF8FEF5018.jpeg
    2233603A-087E-48D6-8394-97FF8FEF5018.jpeg
    1 MB · Views: 14
  • E026F670-01D3-4A1F-8A31-FD5F99B4AF6B.jpeg
    E026F670-01D3-4A1F-8A31-FD5F99B4AF6B.jpeg
    3.4 MB · Views: 22
  • C3B24BB0-F7EE-49CE-8315-3C2B811C43C3.jpeg
    C3B24BB0-F7EE-49CE-8315-3C2B811C43C3.jpeg
    2.5 MB · Views: 25

MrFahrenheit

Well-known member
I’ve had an idea regarding the 68040 CPUs and various revisions: could you not sort them out based upon the size of the gold part on the pin side?

Theoretically a smaller process like .57 micron would result in the bottom area to be smaller, because the CPU die is smaller. Conversely, an older process, XC part would have a larger area, wouldn’t it?
 

Rustynutt

New member
Hey all, have not posted much on 68kMLA. It's 4:45am PST just finished reading Motorolas Designers Handbook for the 68040 (for the umpteenth time).
Somehow in thos haze, came across thos thread concerning 040 mask, CPUs on eBay, and personal experience using a few different mask, including the elusive L88M at 46MHz clock. Interesting note, have ran the CPU at up to 55MHz without fast ram enabled, with performance taking a hit. Have also adapted the Sonnet Quad doubler to test as well.
It's late, or early depending on how the wife sees it, will be back tomorrow, today, read through all post and surely learn a lot, ask one of a hundred questions, and possibly contribute a little from my own experiences.
So that's my intro.
Nite :)
 

MrFahrenheit

Well-known member
Hey all, have not posted much on 68kMLA. It's 4:45am PST just finished reading Motorolas Designers Handbook for the 68040 (for the umpteenth time).
Somehow in thos haze, came across thos thread concerning 040 mask, CPUs on eBay, and personal experience using a few different mask, including the elusive L88M at 46MHz clock. Interesting note, have ran the CPU at up to 55MHz without fast ram enabled, with performance taking a hit. Have also adapted the Sonnet Quad doubler to test as well.
It's late, or early depending on how the wife sees it, will be back tomorrow, today, read through all post and surely learn a lot, ask one of a hundred questions, and possibly contribute a little from my own experiences.
So that's my intro.
Nite :)

Good morning and welcome to the forum.

If you could post good pictures of your CPU (front and back) that would be a great help. Also if you can give some sort of indication as to where and when you bought your CPU.
 

Bolle

Well-known member
I’ve got a cpuple of D50D and D98D ones and those have the bigger golden cap at the bottom.
Everything from E31F and newer has the slightly smaller cap. Size doesn’t differ between the E31Fs, E42Ks, K63Hs and L88Ms I got… they seem to be all the same.
 

MrFahrenheit

Well-known member
I’ve got a cpuple of D50D and D98D ones and those have the bigger golden cap at the bottom.
Everything from E31F and newer has the slightly smaller cap. Size doesn’t differ between the E31Fs, E42Ks, K63Hs and L88Ms I got… they seem to be all the same.

Based on your observations, it would validate my suspicion that the larger cap on this CPU, and the fact it appears remarked, and it runs hot at he double hockey sticks, it’s likely a very early production and not L88M 2009 as marked on it.

Would you agree that seems like a valid conclusion? I can’t understand how a 2009 L88M would feel so fake and also run so hot. Like above 80c hot at only 33mhz (when it’s rated for 40mhz- supposedly).
 

jessenator

Well-known member
While recognizing fakes is a worthwhile effort, it really should have its own breakout thread. I'd like to steer this one back onto thermal performance of the L88M 040s (and comparing it to other versions). Also, in general, the whole point of the exercise is to eliminate conjecture as well as "feel," as both Fizzbin and myself have done. Actual measurements are wanted and much appreciated.

personal experience using a few different mask, including the elusive L88M at 46MHz clock. Interesting note, have ran the CPU at up to 55MHz without fast ram enabled, with performance taking a hit. Have also adapted the Sonnet Quad doubler to test as well.
Thanks for chiming in! I'm curious what your experiences and measurements are :) Do you have a way to measure the thermals from your tests? Also, I'm curious which machine/device you're clocking to 55 MHz, and would be most obliged if you could share those findings in this thread here. :) just so we can keep this one centered on the topic at hand.
 

MrFahrenheit

Well-known member
While recognizing fakes is a worthwhile effort, it really should have its own breakout thread. I'd like to steer this one back onto thermal performance of the L88M 040s (and comparing it to other versions). Also, in general, the whole point of the exercise is to eliminate conjecture as well as "feel," as both Fizzbin and myself have done. Actual measurements are wanted and much appreciated.


Thanks for chiming in! I'm curious what your experiences and measurements are :) Do you have a way to measure the thermals from your tests? Also, I'm curious which machine/device you're clocking to 55 MHz, and would be most obliged if you could share those findings in this thread here. :) just so we can keep this one centered on the topic at hand.
I get it. However it’s also important to note that some chips are likely fakes and therefore the temperature readings from those chips should be scrutinized and possibly eliminated.

The idea behind the thread is to determine if L88M chips run cooler than others. If the L88M chips being examined are fakes, but not differentiated as such, they will affect the result of this test. Therefore, if fakes are not excluded, and you determine that L88M chips are no cooler than others due to this, the thread is meaningless.

I purchased a basic infrared thermometer off Amazon to do some basic testing on my own chips. Do you have a link to the device you used for these tests so that I can attempt to replicate some of them with the variety of chips and mask versions that I have ? I presently have 80 micron, 65 micron, and 57 micron chips. I currently can’t clock any to above 33mhz but it would give some more data points to examine.

It is my opinion that nearly all L88M chips sold through markets such as eBay and other sites over the past 8+ years are likely fakes if not personally extracted from a device that contained it. Meaning, if it’s a chip coming a recycler, it very well may be remarked and not what it says it is.

I’ve learned an awful lot just reading through this single thread, apart from the temperature readings.

To break out some of this important information into a separate thread would do those seeking answers a disservice. Additionally, testing methodology should include ways to determine whether a chip is legitimate or remarked. Like the scientific method.

Or would you rather this thread contain results from supposed L88M chips (thus being misleading)?

I once helped Gamers Nexus scientifically verify their thoughts about a particular water cooler for PCs (failing), which resulted in a recall by the manufacturer. They didn’t tell me their thoughts until I had done my testing, and we shared our results. Our combined information validated the claims.

Would it make sense if a “wiki” page of sorts was setup on here to document the various masks and versions, photos of what people have of them, and temperature readings at given speeds on given hardware, so that it could all exist in one easy to locate space ? There is a wealth of information present in these 4 pages (and growing) that some of which is hard to find online.
 

jessenator

Well-known member
To me, the die looks bigger in this image. So this seems like one way to determine what version the chip is.
GIF and side-by-side :) So I think you're correct. The jump is noticeable!

I’ve got a cpuple of D50D and D98D ones and those have the bigger golden cap at the bottom.
Everything from E31F and newer has the slightly smaller cap. Size doesn’t differ between the E31Fs, E42Ks, K63Hs and L88Ms I got… they seem to be all the same.
I agree here. I think the bottom cap isn't a sure-fire method. A part of me wishes I could thermocouple it :p that's well outside pracitcality. The fortune I had in accessing a $,$$$ camera for family pricing (i.e. $0) in itself a miracle. I'm not dropping those figures in the shape of a new instrument just for this.

D50D XC68040RC25B
I'd like to test a D98D, as that was the first 100*C-tested chip. It might have a better chance of posting and/or booting

…so I would have expected a real L88M to run cooler than the E31F, not sure there is another explanation. Perhaps the measurements under load would be better.
Sorry to necro this comment, Fizzbinn, but @demik brought up a good point to consider: a die-shrunk CPU might run hotter. How this relates to older dies and their thermal performance, like the D-series might be other inefficiencies in the die... many factors in this equation. Apologies if you mentioned this earlier up the thread, but did you test all of your 040s with your FLIR?

Also (apologies for the reiteration), from what I've gathered, the LRC on the chips I procured might indicate an "extended temperature range" —and with a lot of this minutia locked in a vault, I can't say for certain right now if that means both low and high temp ambient operational environments, thus an extended tolerance for internal operational temperature. Again, haven't seen this in documentation, so I'd call it "plausible" for now.

So far in the grand scheme of things, the adage "L88M 040s run cool to the touch" isn't looking so ...hot... right now. I might ping a few people in PM to see if I can arrange to test their less-dubious RC versions, just to increase the sample size.
 

VWGuy

Member
Recently purchased a fleabay 40mhz chip. Marked as:

MC68040RC40A
02L88M
S23159-001

I suspect it is really an E31F or E23G. Obviously a fake, trying to claim back through Ebay now. Do not buy from this seller:


Looks like someone else left interesting feedback after my purchase. I have yet to leave feedback, waiting to see how the seller handles my return first.

1652828176162.png
 

MrFahrenheit

Well-known member
Recently purchased a fleabay 40mhz chip. Marked as:

MC68040RC40A
02L88M
S23159-001

I suspect it is really an E31F or E23G. Obviously a fake, trying to claim back through Ebay now. Do not buy from this seller:


Looks like someone else left interesting feedback after my purchase. I have yet to leave feedback, waiting to see how the seller handles my return first.

View attachment 41835

I purchased from this seller and messaged them about their chips being remarked. The seller offered a full refund and to keep it.

I did not post about it here because I was discouraged from sharing information about fake (remarked) 68040 chips in this thread.
 

rplacd

Well-known member
If you're interested in real '040s, I think Herb Johnson has some used-and-pin-repaired 40 MHz 68040s in stock! 40 MHz models seem unobtanium these days...
 

VWGuy

Member
If anyone has a spare they'd kindly let go? I am doing a scratch built Amiga 4000 from bare motherboard and accelerator card.
 

cheesestraws

Well-known member
If you look for L88M masks directly, you will probably get fakes: demand for them far outstrips supply (on probably largely mythical grounds), and eBay doesn't have any meaningful protection against faking unless there's a current IP holder, nor do they really have any incentive to.
 
Top