Hi everyone,

I purchased ATmega328P-MU from a certain site, then I received ATmega328P U-TH.
The MU is a 5x5mm trapezoidal VQFN package, but my U-TH appears to be a square 5x5mm QFN package.
Their pin spacing is the same, so it would seem that they are physically interchangeable.

However, this model type is not listed in the datasheet. A google search finds a picture of a U-TH in a 7x7mm TQFP package.
Usually, if the packages are different, the type cannot be the same. Is this a new fake?

Thanks,
tyro

 

Have a look at post #6 in this thread:
https://www.avrfreaks.net/forum/atmega328p-u-th-vs-au

In February microchip wrote:
""Thanks for contacting Microchip. The ATMEGA328PU-TH is not a Valid Microchip catalog part number. We do have the ATMEGA328-PU available on our website. Also, we do not recommend purchasing devices from anyone besides Microchip's Authorized Distributors. If you did not purchase from an Authorized Distributor there's nothing we can do. You can try to look this part number up in Microchip's Product Change Notification database: https://www.microchip.com/product-change-notifications/#/ "

To the best of my knowledge the PU version is a full speed (20 MHz) version of the chip in a DIP package.

 

Thank you, I understood that this IC is not a legitimate product.

Following thread starter looks solved Invalid Device Signature issue at ATmega328P U-TH.
https://forum.allaboutcircuits.com/threads/problem-flashing-atmega328p-u-th-mcu.178853/post-1628235

If this report is correct, it appears that this IC is not empty inside, but has some circuitry. I suspect that ICs with illegally copied internal circuits may be on the market. However, not many people will melt down the package to see what is inside. This is a very annoying fake product.

 

If you can’t get a refund I would chalk it up as a lesson learned and move on.

 

The global semiconductor shortage eased for a time, but now seems to be worsening again. For all SMT packages of the ATmega328P, which are essential for Arduino compatible production, there is no stock available from reliable semiconductor distributors such as Mouser and Digikey.

In a situation where genuine products are not available in abundance, it is easy to fall back on unreliable vendors. It is not surprising that fake ICs are sold.

The unit prices of genuine ICs (with the exception of FPGAs) are distributed at prices that make sense, and there is no incentive to look for parts that are a little cheaper. To get rid of fake ICs, ample distribution of legitimate products is necessary.

 

My regular distributor has 981 of the surface mount parts in stock. They are an authorized Microchip distributor and I have been using them regularly since 2008. This distributor is very good and only once out of hundreds of kinds of parts ordered did they send the wrong part. I received the correct part the next day.


The price in U.S.$ is approximately US$6.05 each.

If you buy this TQFP packaged part from Microchip directly, the 1-24 piece price is only US$2.49 but you have to wait until next June (2023).

You might be pleased to see that the price for a few parts is like that of very large quantities when you buy it from one of the big distributors.

 

Two Thousand Twenty-Three!
Oh well. It happens all the time, so I don't mind.

 

I was prepared for the loss of other SMT components and made one unit for a trial run.
ATmega328P U-TH (5x5mm QFN) + CSTNE16M0V53L000R0 (16MHz ceramic resonator) works normally.

I used an Arduino Uno R3 as ISP and burned a bootloader with Uno as target.
As a result, the The Blink sketch worked fine.

 

That is a positive note! Maybe you have a more complex sketch that exercises a larger part of the chip.

 

Can we conclude that the OP got an apparently working clone?

 

It is unlikely that anybody is making a clone of an AVR controller because making even one that actually worked would be difficult with modern CPU speeds and the large expanse of silicon that would need to be flaw-free, more likely he was sent pull-outs that were cleaned up.

I think the slogan from the land of fake parts is "Why create when you can steal and lie about it?"

 

The behavior when writing large sketches must be further verified.

Various Web sites have pointed out that there is no evidence that P U-TH is also a fake. But the question is what to call a fake.

It is possible that a third party simply obtained the ATmega328P mold and produced it without permission. In any case, it cannot be used for mass-produced products because there is no quality assurance.

 

I am just curious and ignorant on these matters. How does a semiconductor fab clone an MCU?
What sort or expertise and fab facility would be required to pull this off?

 

It is unlikely that anybody is making a clone of an AVR controller because making even one that actually worked would be difficult with modern CPU speeds and the large expanse of silicon that would need to be flaw-free, more likely he was sent pull-outs that were cleaned up.

I think the slogan from the land of fake parts is "Why create when you can steal and lie about it?"

There are die that fail final test that are sold as bare die for use in toys and throw-away applications to recover the production costs. The chip might not be 'fake' but it's likely not a 'good' chip either.
http://mmt-technology.com/

 

For example, Rochester Electronics holds a license to store and reproduce molds of discontinued ICs.

It is technically possible for a third party with production facilities to duplicate ICs. This may be done on a larger scale and in a more organized manner than vendors who produce ICs that are empty inside or simply scrape the surface of the package and retype it with a laser.

 

I can see cloning an op-amp or 555-timer circuit, etc. But cloning an MCU is a totally different level of complexity.

 

Licenses for second sourced parts probably come with enough training, documentation, and tools to make quality second sourced parts.

There are several Chinese companies whose components I've used over the years and can say that I've never spotted a bad component from those companies. Like MrChips mentioned, there are some simple old integrated circuits that are not that hard to make well for a company that knows how to make them.

 

There is a theory circulating on the Internet that U-TH refers to a factory in Thailand and U-KR to a factory in Korea. There is another U-PH, which is thought to refer to the Philippines.

In images of ICs found in internet searches, we can see a legitimate Die ID 35473D.

Microchip has locations in these countries. It is possible that lots for internal use made at factories around the world were leaked.

 

It remains that the -TH suffix is not a valid catalog number, from that I infer that those chips either cleaned up pull-outs, or with you new insight not intended for sale,

I have noticed that some wafers have a mixture of designs on them, for example a standard product and product with a minor improvement or bug fix. Process an entire wafer if only a few good chips are needed for initial testing.

 

I have written a sketch that includes a large library, but I do not see any problem with its operation.
This is just a report that such a thing exists in the market. Looking forward to further information.

Thank you all.