Identifying and confirming fake 68040 CPUs (especially the L88M mask)

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
Some recent activities that I’m doing with my vintage Macs have prompted me to source Motorola / Freescale / NXT 68040 CPUs for use in Macintosh LC 475, Quadra 605, and other Centris/Quadra models.

I’ve placed this thread under the Quadra category because it deals with the 68040 processor. If anyone with Amiga experience can recommend a sub forum category that involves 68040 CPUs I’ll create a post there that references this thread. I’d appreciate someone reply with models for that.

This is just a placeholder / preliminary post that will be updated as I gather more information, photos, evidence, and specifications.

To summarize, there are a lot of fake 68040 and 68LC040 CPUs out there. I feel this is the forum that fakes on eBay can be discussed openly and freely, and sellers could be outed as a warning to others not to buy from those sellers known to selling fakes.

So, what is a fake? How can you fake a CPU?

Briefly, and I’ll go into more detail in the coming weeks and months, it involves removing the labeling on the top of the CPU as printed by Motorola or Freescale, and replacing those markings with ones making the CPU appear better than it is.

The 68040 CPU is usually marked as the following:

XC68040RC20
XC68040RC25
XC68LC040RC25
XC68040RC33
XC68040RC40

MC68040RC20
MC68040RC25
MC68LC040RC25
MC68040RC33
MC68LC040RC33
MC68040RC40

There are some early batches of 68040 which bear the PC68040RC25 and so on, which are prototype CPUs.

The above list is not exhaustive by any means but only indicates the most common ones found and used in Macintosh computers.

The first letter (X or M) refers to whether the part was fully qualified or not. When Motorola was working on the 68040 series, they had bugs and glitches (called errata) that results in them not being “fully qualified”. Those earlier batches had XC as their prefix. The MC are fully qualified parts, which shipped basically after Apple stopped using 68040 CPUs, so nearly all CPUs in Macintosh computers are the XC part.

The LC in the number refers to a CPU that lacks an embedded floating point math coprocessor inside. Most people don’t use software that takes advantage of the FPU, but some have found that it’s possible that Mac OS 8.1 utilizes the FPU in OS operations. I haven’t confirmed that.

When combining an XC part with an LC, you end up with software incompatibility with software needing an FPU, and those apps crash. The software “SoftFPU” did not properly function on those chips, because of the errata that those CPUs had. The MC part fixes this glitch and allows them to work properly.

The RC in the part number refers to the package itself. The ceramic ZIFF type CPU is RC, and used in all desktop Mac’s and Amigas (as far as I know). The FC part is one used in laptops and some upgrade cards, it required soldering the legs to surface Mount pads and is packaged in a smaller plastic type package.

These days, there are a lot of electronics being recycled, and sadly many of them are (or have been) vintage Macintosh, Amiga, and NeXT computers. Also, some telecommunications equipment, as businesses upgrade their phone systems. A lot of telecommunications equipment used varieties of 68040 CPUs, some compatible with Macintosh and some not.

Recyclers generally send their equipment overseas to Asia (China, India, etc) for component removal. That means the various 68040 CPUs are removed and retained by the sellers. They have no idea if they work or not, and some possibly are damaged during removal.

When Motorola was able to shrink the die manufacturing process of their 68040 CPUs, they went from .80 micron to .65 micron and finally to .57 micron. The various shrinks bring with them advantages. Quite possibly, die shrink results in a cooler running chip, or so the theory goes. I haven’t yet proven that, and I don’t know if anyone has definitively proven it either. More on that later.

The final iteration of the 68040 CPU was the infamous L88M (or 02L88M ) mask. This is the one most enthusiasts desire for the Mac and Amiga. Theoretically they run more efficient, and cooler, so they would have more overclocking potential. As such, recyclers have identified this CPU to be highly desirable, and so they mostly target this market.

Additionally, the LC processors (which lack FPU) are less desirable than the full 68040 CPUs and recyclers have noticed that, too.

Also, the higher speed-rated CPUs, clocked at 40mhz, are also more desirable even though there’s really no technical difference between a 20, 25, 33, and 40 MHz part. The parts themselves don’t dictate the speed, the logic board does. An oscillator on the board sets the speed of the machine and the CPU runs at that speed. Chips marked with a speed are certified by Motorola to run at those speeds.

So here you have the perfect conditions for maximum profit if you can sell MC68040RC40 CPUs with L88M mask. These are the most sought after of all of the 68040 CPUs. Recyclers and scammers know this, and so they commit fraud and relabel their batches of CPUs to this.

To dig further into this, we have to examine how the CPUs were marked by Motorola. The top left of the CPU has a symbol indicating the position it is mounted in the socket. To the right of that is the CPU specifications, for example MC68040RC40A, (the last character after the speed is a revision letter, if used), and to the right is the batch info.

The first line of the batch info is the mask (or process revision), middle line is the actual batch and date of manufacture, and the bottom line is the place of manufacture. Either Malaysia or USA.

Along the left side of the CPU, vertically are printed a series of digits which help validate the batch number (they coincide with each other).

I understand that pointing out how the CPUs are marked and pointing out how to spot a fake will garner attention of the fakers to “do better” , but it’s needed to educate those looking to buy. And by now I highly recommend anyone who is looking for the infamous 68040 40mhz L88M part to just avoid any being sold on the market. I’m jumping the gun before presenting any evidence, but I want people to know the likelihood of buying a real legitimate part in 2022 and beyond is basically zero. I’ll explain why and how in the coming weeks and months with evidence, but take my word for it for how.

If you have come here from a Google search to research the infamous L88M chip in order to find and buy, be warned, don’t bother. The reality is, being able to source a real legitimate chip in 2022 and beyond is virtually impossible.

I’ll update this post, and leave you with some images of a fake 68040 L88M I sacrificed $65 to see for myself.

Anyone who has real full insight into this subject, I welcome you add your own replies, comments, evidence, photos, and experiences below.
18000D6A-4735-489E-9E32-86EAD4E95BBB.jpeg
BE71E2ED-CD6B-4D2A-AB03-DBB12C72008F.jpeg
2197AAFC-9E7D-4803-ABC8-290D55D1C3B3.jpeg
669ADADF-F191-455C-96D7-338F9C78B0DD.jpeg
C170B845-E72E-468F-B768-71FBA1E372C3.jpeg
 
Last edited:

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,547
1,353
113
53
Japan
youtube.com
@Mr. Fahrenheit
Thank you for posting all of that for public eyes to see! I will try to make reference to this thread in my overclocking video so your excellent information garners the attention it deserves, in addition to putting a link in the text description too.

For those wanting more info on 68040 CPU types and known masks (not comprehensive), please check out the following web page and scroll down to the section named "MASKS" to examine Tables 1, 2 & 3:

 

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
@Mr. Fahrenheit
Thank you for posting all of that for public eyes to see! I will try to make reference to this thread in my overclocking video so your excellent information garners the attention it deserves, in addition to putting a link in the text description too.

For those wanting more info on 68040 CPU types and known masks (not comprehensive), please check out the following web page and scroll down to the section named "MASKS" to examine Tables 1, 2 & 3:


I plan to put together a one spot reference of good known CPU information. A good resource is the website “CPU-World” as well. There you can search mask versions, and even the string of digits I mentioned that run vertically on the left side of the CPU.

There is a lot of online documentation regarding these processors online, and a lot that has been lost to the hands of time.

You could almost write a book on the topic of 68040 CPUs and their varieties, along with fake “detection”.

For anyone serious about overclocking their 68040 CPU, knowing what’s out there is essential.

More to follow over the coming weeks as I gather and present information on this topic.

Again, I’m not a topic “guard” on this. I don’t “own” the post or information. Anyone with their own details, experiences, and information, is free to post a reply here for everyone to participate.

My hope is that with enough participants we can build the best set of information available online about these processors, and subsequent “outing” of fakes.
 
  • Love
  • Like
Reactions: Volvo242GT and JDW

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,547
1,353
113
53
Japan
youtube.com
...I’m not a topic “guard” on this. I don’t “own” the post or information. Anyone with their own details, experiences, and information, is free to post a reply here for everyone to participate.
I'm happy to play topic guard only to ensure freedom reigns. Obviously, it would be down right silly to start talking about vacations in Hawaii in this thread, but a little Wild West style deviation on a given topic is good for the brain. Even so, I look forward to all the information you can provide.

Honestly, I would prefer to briefly touch on 040 chips and masks and just direct people to an extensive write-up on the subject (which I myself have no time to write), rather than try to create an exhaustive explanation in my overclocking video. People need to know about MASK revisions, but I'm not the foremost expert on that, even after all the private emails you and I have shared on that topic!
 

alxlab

Active Tinkerer
Sep 23, 2021
287
312
63
www.alxlab.com
Yeah really sucks that's there's a lot of fakes out there now. Bought from this seller on ebay:

1643914907385.png


Basically I see it as a + if they're dirty and not cleaned. Higher chances that the're direct pulls that weren't tampered with. The one's I bought ended up being all XC.

The print on something like this looks suspiciously bright and bold with L88M:
1643915396379.png


Hard pass.


Coincidentally another post about fake 040s was just posted on 68KMLA with thermal images of different models.


Edit: I should really type more slowly... the amount is spelling and grammar errors I do is appalling.
 
Last edited:

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
Just to update this thread: I have been able to keep the fake CPUs I received from an eBay seller and they refunded me.

I have started the process of removing the top “skin” that they applied to them on one of the CPUs so that I can get a better temperature measurement without that layer impeding or magnifying temperatures.

I installed one of the fakes in my LC475 overclocked at 33mhz. My ambient room temperature is 16’c. After running for a while I ran several benchmarks in a row while measuring surface temperatures using a FLIROne for my iPhone.

I then ran the same test on a known food K63H, which is supposed to share the same lithography of 0.57 microns with L88M and ran the same tests.

This was a casual test just to see if I could identify differences using the FLIR. I can.

The fake marked CPU measured above 80’c and the real K63H measured around 48’c. That is significant.

It proves that high temperature measurements of alleged L88M likely mean they are not truly L88M CPUs and likely have been remarked. To include those temperature measurements in an official list as L88M temperatures would be haphazard at best, negligent at worst.

I don’t have time to dedicate to testing at the moment so I’ll just leave this update here as of today. Enjoy the two pictures.
6BF1F3C9-699F-44BA-8118-1E7B2B147E8C.jpeg
22758660-CAB8-41BD-846F-E005ED3051B4.jpeg
 
Last edited:

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,547
1,353
113
53
Japan
youtube.com
These are the markings on the LC040 that came with the LC575 in untouched condition which I purchased on Yahoo Auctions Japan at the end of 2021...

TOP
1645072587049.png


LEFT SIDE
1645072611527.png


Those markings are originals make by Motorola and do not come off even after repeated application and removal of thermal tape.

Here's a photo of two 040 CPUs I bought from UTsource...

1645072771259.png


One didn't work. @Kay K.M.Mods kindly confirmed it too. I then shipped it back for a refund. Both chips have the same markings, which were clearly added by the seller (whoever sold them to UTsource). I wish I had taken a photo of them immediately when they arrived, but the above is the best photo I have. I sent the good one to Kay Koba too, and he confirmed it works. But when he removed a heatsink on it, the markings in the upper right corner became this...

1645073026385.png


Even the side markings became faint...

1645073324751.png


From what I understand, the original, authentic Motorola markings are much more durable and will not start to come off like that after a heatsink is removed.

The big problem is that I don't know what my working 040 from UTsource was originally. It may not even be a 40MHz rated chip. I haven't the faintest idea what the actual Mask version is. But that is the 040 I am using in my Overclocking video, along with the LC040.

Here is a set of 3 photos showing 3 040 chips owned by @Kay K.M.Mods . I don't think he knows which are true to what the markings say...

1645073187418.png
1645073204003.png
1645073218570.png
 

Bolle

Tinkerer
Nov 1, 2021
46
74
18
Here is a set of 3 photos showing 3 040 chips owned by @Kay K.M.Mods . I don't think he knows which are true to what the markings say...

1645073187418.png
1645073204003.png

At least they all have consistent mask revisions and matching package identifiers.
I haven't seen those correctly matched on fakes yet.
 
  • Like
Reactions: JDW

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,547
1,353
113
53
Japan
youtube.com
At least they all have consistent mask revisions and matching package identifiers.
I haven't seen those correctly matched on fakes yet.
Those chips are owned by @Kay K.M.Mods . Maybe when he has time he can tell us how hot they get when overclocked. That might tell us a lot about the printed numbers being real or fake.
 

Garrett

Tinkerer
Oct 31, 2021
103
108
43
MN, USA
I recently bought one of these chips from an eBay seller out of China as an "experiment" I guess you could say. I took a couple of photos from my microscope to show some differences between a known genuine XC versus this questionable MC chip. I scrubbed the top of the MC chip with IPA, and none of the markings lifted or showed signs of fading.

I'm trying to decipher the mask to determine what's what here. Any tips for verifying the markings along the side of the chip align with the date it was supposedly manufactured?
 

Attachments

  • MC1.jpg
    MC1.jpg
    431.3 KB · Views: 131
  • MC2.jpg
    MC2.jpg
    494.4 KB · Views: 141
  • MC3.jpg
    MC3.jpg
    245.5 KB · Views: 139
  • XC1.jpg
    XC1.jpg
    330.9 KB · Views: 140
  • XC2.jpg
    XC2.jpg
    443.7 KB · Views: 165

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
I recently bought one of these chips from an eBay seller out of China as an "experiment" I guess you could say. I took a couple of photos from my microscope to show some differences between a known genuine XC versus this questionable MC chip. I scrubbed the top of the MC chip with IPA, and none of the markings lifted or showed signs of fading.

I'm trying to decipher the mask to determine what's what here. Any tips for verifying the markings along the side of the chip align with the date it was supposedly manufactured?

There isn't a complete list of side markings but there's a general guideline that @Bolle put together:

@Bolle
D98D ends with -010
E31F ends with -001
E42K ends with -101
K63H have seen them with both -104 and -103
L88M ends with -104

Do you have a photo of the bottom side? Temperature readings using something like a FLIR would be ideal to verify at least whether it's a remarked much earlier revision or whether it's a newer manufacture.

What's strange is the date of manufacture is 2008, and from what I've found online, that would be a Freescale part, not a Motorola part. So theoretically, if legitimate, it should have a Freescale logo. Additionally, and I'm unsure of this part as well, being a 2008-made chip made by Freescale, it should say 'USA' for country of manufacture, at least in theory. Due to too many fakes existing, it's difficult to validate this.

Edit: The side-code on your 'L88M' chip (S22899-010-F) appears to correspond to D50D mask, which is very old:


If it is really D50D, that would be a remarked 25Mhz chip.
 
Last edited:
  • Like
Reactions: JDW

Garrett

Tinkerer
Oct 31, 2021
103
108
43
MN, USA
Thank you for the information! To be honest I haven't tested the processor yet, but if I understand correctly, it would show up as 25MHz in my Q700 and 33MHz in my LC575? Who knows if it even has an FPU! I wish we had a tool to determine "rated" speed. I suppose heat output could be used to gauge some of that, but I don't have a FLIR camera at the moment unfortunately.

Bottom side appears to have nothing out of the ordinary going on.
 

Attachments

  • IMG_7931.jpg
    IMG_7931.jpg
    471.6 KB · Views: 111
  • IMG_7932.jpg
    IMG_7932.jpg
    396.6 KB · Views: 107

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
Thank you for the information! To be honest I haven't tested the processor yet, but if I understand correctly, it would show up as 25MHz in my Q700 and 33MHz in my LC575? Who knows if it even has an FPU! I wish we had a tool to determine "rated" speed. I suppose heat output could be used to gauge some of that, but I don't have a FLIR camera at the moment unfortunately.

Bottom side appears to have nothing out of the ordinary going on.

The bottom gold cap covers the CPU die. It is my expectation that a smaller die-shrunk variant of 68040 would have a smaller gold cap.

That cap is as large as it could possibly be, which tells me that is likely an early mask CPU.

The 68040 CPU speeds are set by the oscillator on the logic board. The printed speed is just the speed it was certified at by Motorola at the factory.
 

Garrett

Tinkerer
Oct 31, 2021
103
108
43
MN, USA
Interesting! So something like the below image would be indicative of a later revision chip?

1646164109328.png


I wonder if the above evidence is enough to go back to the seller and question the authenticity. It's a fine line between giving away everything we know about fakes versus obscuring it and simply saying "I know it's fake". We don't want them to get any better at faking!
 

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
Interesting! So something like the below image would be indicative of a later revision chip?

View attachment 3318

I wonder if the above evidence is enough to go back to the seller and question the authenticity. It's a fine line between giving away everything we know about fakes versus obscuring it and simply saying "I know it's fake". We don't want them to get any better at faking!

I would confirm by installing it and trying it out. If you can even put your fingers on it to see if it’s too hot.

If it is an early mask as I suspect, it will run really really hot. Too hot to comfortably touch. I would then introduce that as proof.
 

Garrett

Tinkerer
Oct 31, 2021
103
108
43
MN, USA
As was suspected, the L88M chip runs very hot. During benchmarking in my 575, ~3 seconds was the longest I could rest my finger on the surface.

I also have a 1994 68040RC33M and the stock 1993 68LC040RC33 I tested. During benchmarking, I could rest my finger for ~8 seconds before these became too hot. Keep in mind all three were running at the stock 33MHz speed. I know the finger test isn't scientific by any stretch, but I think it illustrates the point.

All that being said, I agree that this L88M I have is probably a 68040RC25 chip, and would work great in a stock Q700, but not so much if I wanted to overclock the 575 to 40MHz. Sad!
 

Attachments

  • 1.jpg
    1.jpg
    385.1 KB · Views: 144
  • 3.jpg
    3.jpg
    440.9 KB · Views: 120
  • 2.jpg
    2.jpg
    491 KB · Views: 120

Mr. Fahrenheit

Tinkerer
Founder
Sep 2, 2021
141
230
43
I purchased two x 68040 40mhz CPUs, supposedly L88M from this eBay seller:


They are not legitimate. They are remarked fakes. See attached photos.

Buyer Beware!!

2B2BF4E0-8599-44B9-81C9-8399AAA72181.jpeg
8A157CF5-24AE-468E-BF77-8DDB0616B6D8.jpeg
4F79B87D-8E62-400B-A078-CFC1832636B4.jpeg
88A100A2-C93C-4486-B77B-359FDC609360.jpeg