Vida CEM swapping

[Treur]

For 60€ I can buy so much beer that I'll pee myself.
...

I want to make j2534 and have your code on board that runs on demand.

But is it a good beer? =)

I see no reason why it can't be done. The hardest part will be converting to yet another CAN library. But it was done once already, MCP2515 -> FlexCAN.

The MCP2515 is absolute evil. It's impossible to operate at normal speed using this piece of junk. SPI will slow down any speed. We can implement anything, but for me personally, it's quite difficult to simultaneously design a USB protocol and implement it in both the library and the device. And that's not counting porting the CEM cracker, but that's a minor detail.

But technically, I have had the device itself ready for a long time.

It have on board 2xcan-fd, K-line, Bluetooth 5.1
The only thing missing from the implementation is power supply metering. But this isn't critical, and a software plug-in can be installed.

[Treur]

For 60€ I can buy so much beer that I'll pee myself.
...

I want to make j2534 and have your code on board that runs on demand.

But is it a good beer? =)

I see no reason why it can't be done. The hardest part will be converting to yet another CAN library. But it was done once already, MCP2515 -> FlexCAN.

I know you're not a fan of AI slop, but it can likely convert it in under 5 minutes.

Possibly, but only if AI doesn't notice access to the security system))))

A simple example: I decompiled a program that contained Security PIN search algorithms for the p3 platform. I fed it all to the AI. It went as far as implementing the algorithms themselves and started working. And then it suddenly dawned on it—security... 27 01..... No, I refuse to go any further. ))) I did the rest myself.

I won't even mention that this crap has led me to a complete misunderstanding of the process in the algorithms.

[Dudde]

But is it a good beer? =)

I see no reason why it can't be done. The hardest part will be converting to yet another CAN library. But it was done once already, MCP2515 -> FlexCAN.

I know you're not a fan of AI slop, but it can likely convert it in under 5 minutes.

Possibly, but only if AI doesn't notice access to the security system))))

A simple example: I decompiled a program that contained Security PIN search algorithms for the p3 platform. I fed it all to the AI. It went as far as implementing the algorithms themselves and started working. And then it suddenly dawned on it—security... 27 01..... No, I refuse to go any further. ))) I did the rest myself.

Then you just switch AI, there are plenty of them that don't care about security

[Treur]

I know you're not a fan of AI slop, but it can likely convert it in under 5 minutes.

Possibly, but only if AI doesn't notice access to the security system))))

A simple example: I decompiled a program that contained Security PIN search algorithms for the p3 platform. I fed it all to the AI. It went as far as implementing the algorithms themselves and started working. And then it suddenly dawned on it—security... 27 01..... No, I refuse to go any further. ))) I did the rest myself.

Then you just switch AI, there are plenty of them that don't care about security

I agree, but it's something like Deepseek or Grok. Both are simply terrible.

Well, there's a caveat: you need to understand what you're writing. Otherwise, if you encounter a problem in your code, you won't be able to solve it.

[Dudde]

Possibly, but only if AI doesn't notice access to the security system))))

A simple example: I decompiled a program that contained Security PIN search algorithms for the p3 platform. I fed it all to the AI. It went as far as implementing the algorithms themselves and started working. And then it suddenly dawned on it—security... 27 01..... No, I refuse to go any further. ))) I did the rest myself.

Then you just switch AI, there are plenty of them that don't care about security

I agree, but it's something like Deepseek or Grok. Both are simply terrible.

Claude has newer asked anything or refused to do anything about security.
ChatGpt started awhile back to refuse if it has anything to do with security or similar, but ChatGpt sucks at coding anyways so.

[Treur]

Then you just switch AI, there are plenty of them that don't care about security

I agree, but it's something like Deepseek or Grok. Both are simply terrible.

Claude has newer asked anything or refused to do anything about security.
ChatGpt started awhile back to refuse if it has anything to do with security or similar, but ChatGpt sucks at coding anyways so.

Each AI is good in its own way: Claude is good at starting a project from scratch, ChatGPT is more suitable for minor edits.

[Dudde]

I agree, but it's something like Deepseek or Grok. Both are simply terrible.

Claude has newer asked anything or refused to do anything about security.
ChatGpt started awhile back to refuse if it has anything to do with security or similar, but ChatGpt sucks at coding anyways so.

Each AI is good in its own way: Claude is good at starting a project from scratch, ChatGPT is more suitable for minor edits.

Agree on that, AI is only a helpful tool to save some time

[Treur]

Personally, I need to first understand what we're writing, and only then start writing. And not like this: write me who knows what, who knows how, as long as it works.

The logic there is simple: if nothing works, let's swap the bytes, and see if it works. And it can swap back and forth like that forever.

[vtl]

I know you're not a fan of AI slop, but it can likely convert it in under 5 minutes.

Yeah, I'm a fan of humans. Especially when they run git log =)

https://github.com/vtl/volvo-cem-cracke ... 782207c1fb

[vfred748]

Hi everybody,

What a great work on the CEM CRACKER,

I have a VOLVO S60 2.4D 2005 with a CEM L part number 30765646. Car won't start with multiple electronic issues.

Unfortunately, I have always same issue when launching CEM CRACKER on bench. Did someone try it on bench ?

PINOUT :
PIN BA BB : +12v
On grey connector :
PIN 6 : GND
PIN 33 : CANH
PIN 48 : CANL
Source IOTERMINAL

Mounted with display, says : CAN LS ERROR then unknow CEM exiting
On monitoring under arduino : Unknown CEM part number 0.

I've mounted the 120 resistors
I've triple checked wiring.
Read on the forum maybe to increase voltage on CAN modules to 5v

I have already read the original CEM with IOTERMINAL (the pin was recovered with modunlock tool) but I need the PIN for a used CEM.

On the bench, the CEM didn't seems to start, I don't hear the relay (like IOTERMINAL). The consumption of CEM was under 0.1 amp.

Any ideas ? Another way to virginize CEM ?

Will try on car.

Thanks a lot for your help.