Yet another question about the STM32 clones...

[ataradov]

I looked at the "smallest CM4F", that MAX32660

That device is really weird. It probably does not have mass appeal, but I assume there are applications where it fits precisely. I've used it as a test debug target because it is indeed very small and reasonably cheap. But I see no actual uses for it.

all peripherals are behind an AHB-APB bridge

There is distinction between the way main matrix is designed on the D21 vs STM32. D21 has all 3 AHB-APB bridges as individual slaves, so in theory CPU can access PORT registers while DMA does SPI transfer, since they are on two different bridges. On STM32 DMA access would preempt any other APB access. But then there split transfers and other complex stuff like this, so how much this matters in practice is hard to tell. My guess would be - not at all, except in some pathological cases.

[wek]

MAX32660

That device is really weird. It probably does not have mass appeal

No Maxim IC has mass appeal. Well, maybe except MAX232... :-) It's most likely a custom product offered to public.

[...] how much [details of DMA connection and consequences] matters in practice is hard to tell. My guess would be - not at all, except in some pathological cases.

Replace "pathological" for "extreme/exceptional" and I fully agree. :-)

Jan

[thm_w]

We were just discussing this, you'll want to put some effort in and read a bit for yourself:

https://www.eevblog.com/forum/microcontrollers/which-of-the-many-stm32f103-clones-are-actually-usable/
https://www.eevblog.com/forum/microcontrollers/stm32f103-the-security-of-drop-in-replacement-and-counterfeit-microcontrollers/
https://www.richis-lab.de/STM32.htm
https://www.eevblog.com/forum/other-blog-specific/*32f103-comparison/msg2830314/#msg2830314

[dietert1]

...
Probably Artery AT32F403

But those have 256K flash or more. My clone doesn't have the FPU either.
The two images show clone and original for comparison, assuming ST don't use clones for the STLink.
Third image is an "honest" clone in a STLink USB stick.

Regards, Dieter

[peter-h]

The 1st image has bad marking.
The 2nd image looks like ST marking I see on the 32F4.

And why the extra "F"?

If I had even half a brain I would make the fake look right.

[ataradov]

If I had even half a brain I would make the fake look right.

I've seen quite a few fakes and they all seem to not have access to a decent laser or they don't know how to setup it correctly. Even if they change the font (9 is especially non-ST-like), and figured out where the "F" in the name goes, there width of the line would be too wide and too wobbly. Or may be it has something to do with the type of plastic.

Fakes also seem to use raster scan for bold elements and most real ICs use vector mode. You can see it in the ST logo. Legit devices draw the outline and then fill it as a CNC mill would. This probably also contributes to the visuals. Thin lines are essentially made out of dots, which makes them look thicker.

[peter-h]

Firms should have a contact which can be used to verify fakes.

I have just got 3k x 28C256 TSOP-28, Atmel (now Microchip), from a US based "opportunist excess stock dealer" but can't find a contact. The chips do work (tested functionally and over temperature) and the marking looks exactly right, but one can never be 100% sure.

And these dealers are vulnerable; I got a few k fake H8/300s a few years ago - fortunately the package didn't quite fit the programmer

[ataradov]

Firms should have a contact which can be used to verify fakes.

I do that for Microchip ARM MCUs. But you would need to get though the support ticket. Requests to verify happen from time to time.

But most of the time all we can do is check that the lot code is in the database and belongs to the claimed device. But if you are creating a fake, you can just use the data from a real device. But I've seen faker even fail to do that. This is inexcusable.

And I do compare visual quality with the reference devices, of course.

For MCUs specifically, having Unique ID values from the device also helps, as they encode potentially useful information, which may not be copied correctly by the fake. It would be actually interesting to see what UID values are programmed into the fakes.

[dietert1]

Meanwhile i found that in other places this fake is labeled CKS CS32F103C8T6 and the package label "991KA 93 MYS 807" has been confirmed by STM to be fake. The chip is said to report 128K flash and the unexpected CM4 core id. And one finds reports about problems with DMA controller and USB connection of that chip. There is chinese datasheet, yet no reference manual.

Regards, Dieter

[peter-h]

I am expecting more fakes now, because distis are imposing price increases, after the order was placed. The customer is supposedly contractually bound to pay the higher price - without limit. It is a disgusting development, done totally without regard for a future business relationship. I've had a barrister's opinion on a current case (where, notably, the Ts and Cs were not supplied until the order confirmation) and it is that we can't get the price back to the contracted price (short of suing them for a breach of contract, once the goods fail to arrive) but we can return the goods (which they shipped despite correspondence making it clear we do not accept the price increase). It is a very big order and I will probably just return the items.

Another disti, for ST actually, had to climb down because they could not prove the price increase email was sent out. In fact an employee mistakenly emailed me saying it failed to go out

And returning items for credit/refund to a disti is the favoured way of injecting fake stock into the franchised disti pipeline, which most people would regard as secure.

This totally unethical behaviour by Western companies is going to drive business to chinese clones. China never claimed any ethics, indeed doesn't have any, and everybody knows and accepts that, as the cost of getting a lower price

[hans]

That's just a fake, I would not consider using that at all. "Cloned" devices released under their own name with their own datasheet are fine.

In theory yes. Even though it's ethetically-questionable that GD copies all part numbers, peripherals and documentation of ST, they're building their own brand. Clones do not build their own brand. Only e-waste
I do wonder if some clones will, at some point in future as I've no indication it happens right now, will re-badge themselves as GD devices?  Or maybe there is too little margin to be had on an already copied design. Or can chinese companies murder each other for copyright infringement?

I mean, how reliably are you going to source these parts. These are plenty of enough reasons for me to source as much as possible from western big name companies. I rarely even order PCBs from Asia anymore because there are plenty of local alternatives that are slightly more expensive, but at least keeps someone in my home country/EU in a job. (But I digress, that's a different topic)