Is Your Cheap Pro-Mini REAL? Easy Way to Test! (ATMEGA328P)

[@langrock74]

Since you have the capability to test the low-power performance on these, why not try to establish a correlation between the register dump and the performance? If there isn’t any, then the register values aren’t good indicators for this being a genuine or fake chip. Also, have you contacted Atmel? What do they have to say? Maybe they’d recommend a way to tell these apart.

[@Kevindarrah]

yea, almost positive that the all 0xFF register dumps will have high sleep currents, but I will check that tomorrow and report back here, just. so. we all know for sure - that one that was sent to me was +100uA high, but will test the others. Also, have not reached out to Microchip(Atmel) about this yet. Maybe forum post?

[@Kevindarrah]

ok, so not worth making another video, bit all of those boards shown in the video just had their low power capabilities tested. Of course, first I had to remove the regulator and LED from the boards, then source externally with 3.3V. the pro-minis that failed the "dump" test with 0xFF's, read over 100uA in the lowest power test - the pro-minis that pass the dump test are in the nano amps, which lines up with the known good samples from Digikey

[@richardwicks4190]

@@Kevindarrah Are you related to Peter Darrah of AMD?

[@nicolasguichard9788]

I've been doing both tests (signature and lowpower) on a bunch of MEGA328P and PB chips: all of them are consuming in the range of nanoAmps, but surprisingly one of them has a weird signature, similar to the second dumb on the video. It's also marked as coming from Korea. I don't know what to think about it...

[@arduinocraftcorner6178]

I just bought 3 ATmega328P-PU (production year 2022) at Aliexpress and thought I test them. The register dump looks suspicious with a lot of FF's. However, deep-sleep current consumption is normal (around 150 nA), high-voltage programming works, and even debugWIRE is working. So, I guess it is the real thing. Here are the register dumps: boot sig dump 1E AA 95 FF F FF FF 26 FF FF FF FF FF FF 58 FF DF FF DF FF FF FF FF FF FF FF FF FF FF FF FF FF boot sig dump 1E B1 95 FF F FF FF 26 FF FF FF FF FF FF 58 FF DF FF DF FF FF FF FF FF FF FF FF FF FF FF FF FF boot sig dump 1E D4 95 FF F FF FF 26 FF FF FF FF FF FF 58 FF DF FF DF FF FF FF FF FF FF FF FF FF FF FF FF FF

[@antibrevity]

Important video and topic, so thanks for following this story and doing such a thorough investigation!

[@mingyu01]

Excellent video! Thank you for sharing an easy way to possibly identify a counterfeit atmega.

[@TT-it9gg]

Amazing! Appreciate for all these three related videos. Really help! Best~

[@Retro_Care]

for some projects its fine to use cheaper ones that are not real, good video to make us aware of it most importantly and very interesting video about the sleep current as well, thank you

[@JendaLinda]

Clones are OK as long as they're clearly labeled.

[@Retro_Care]

@@JendaLinda yeah, i agree, why not label them because for the price its kinda obvious anyways, would be better to not pretend to be original at all, they will still sell imo

[@Kevindarrah]

yep, exactly, they're not garbage, just not going to do anything fancy with them

[@Retro_Care]

@@Kevindarrah exaclty, its great you pointed out the difference so its even more clear where or for what project to use it or not, excellent video as always

[@PilotPlater]

I'm sure the counterfieters will up their game in the arms race and patch this, but great job, I'm sure this will work on a bunch of counterfiets already in the wild!

[@Kevindarrah]

yea, probably!

[@pyromen321]

Neat! I have a handful of 328P devices from different sources in a box somewhere. I’ll try this out when I have a chance!

[@Kevindarrah]

that would be awesome!

[@JendaLinda]

I've tested my Nano clones (with CH340 USB-serial chips) and they seem to use genuine Atmel MCUs.

[@aqib2000]

Decap

[@migry]

The extra information is likely to be manufacturing data, such as lot number (the way to record which wafer is which), possibly X and Y location on the wafer, and other test information. Many devices have this information programmed during testing, but the way it is stored is usually confidential and needs a special sequence to access. That might explain why the numbers from the genuine parts only differ slightly. This information is invaluable in the case that the manufacturer receives customer returns (parts which have failed in the field), as it allows analysis to see if the "failed" parts all came from the same wafer or were manufactured in the same month (for example).

[@ArnaudMEURET]

Aren’t they simply LGT8F328P chips with a “custom” laser etching ?

[@berriebean]

I ran this sketch on a LGT8F328P green Nano clone and it gives a completely different return boot sig dump C 94 0 3A C 94 5D 0 C 94 5D 0 C 94 5D 0 C 94 5D 0 C 94 5D 0 C 94 35 0 C 94 5D 0

[@avejst]

Great video and walkthrough 👍 Thanks for sharing 👍😀

[@iotan09]

Awesome, thank you so much to spare your time to sharing many think about electronics

[@ytfp]

Great like we don't have enough work to do lol, now we have to go through all our pro minis:) Appreciate the work you have done here. Btw what is the RTC part number on the trigboard 8, hardware info on the wiki seems to be missing.

[@Kevindarrah]

lol I know! Yea, RTC is RV-8263-C7 and on the hardware page on the docs: trigboard-docs.readthedocs.io/en/latest/Hardware.html

[@ytfp]

@@Kevindarrah oh awesome I must have mssed this page, thank you!

[@troybaird404]

Great catch, only problem is this will also expose to those who create these what else to change to hide within the mass on legit chips...

[@Kevindarrah]

yep, I know...

[@dave_dennis]

Yes but not so easy to change. Requires at least one metal mask change. Maybe more. Pretty expensive.

[@gtcollection6933]

All chips (13) I got between 2017~now turned up to be genuine parts. Thanks for your quick test-check. Worth thinking of changing your eBay & Amazon cheap'er part supplier perhaps? Recommend the folks on Aliexpress for genuine "and affordable" parts since they buy in bulk these guys, as opposed to those hiding behind the credibility of American ecommerce platforms selling plain and obvious fakes.

[@JBDAC]

It's not just the low power that's different - I just ran some extensive calculations on both genuine and non-genuine chips. The fake ones took 1.6 times longer to execute the same code..

[@thegreyfuzz]

I just received an order of 5V Pro Mini's via slow rowboat (initiated via eBay). At ~$2/ea I really expected them to be counterfeit. Pleasantly surprised to see they checked out to have values close to the results expected for valid units. Now I'd feel better ordering a few more lots, who could ever have enough on hand???

[@Kevindarrah]

good to hear!

[@Abcdefghijklmnopqrstuvwxyz121]

Great work Kevin👍

[@reseausalsadream2691]

Just received my first Pro Mini by mail two weeks ago (I want to try that radar thing you did - got the same Bushnell radar btw). I'll test the Pro Mini with your above code soon.

[@Kevindarrah]

yea, and that project required pretty good ADC readings, so would be a good test. I don't know if this test will work on the 32u4 though, unless you're trying to get it to work with the 328P

[@DjleonLeon]

Good job!

[@ralmslb]

I feel that Atmel should release an oficial way to identify counterfeits

[@Kevindarrah]

yea, that would be nice

[@crasbee]

It wouldn't be too smart to do that as a manufacturer, because then the fakers can adapt to the test strategy. The manufacurers however usually do answer e-mails when you ask if they can tell if the chip is real or not. At least ST does that with counterfeit STM32F103 chips.

[@ralmslb]

@@crasbee not if the system is only on their website, where you can insert these random hex values and it tells you.

[@Asyss_Complex]

It does seem that there is a pattern between the fake and real ones. I will keep this in mind in the future, thanks again!

[@Peter_S_]

Excellent sleuthing! 🏆

[@luizcassettari]

A lot of FF...! Looks link has a new serial in there, different manufacture kinda make sense! The byte 0x0E to 0x17 by my tests is a serial number. On the suspicious chips has a lot of 0xFF but is not the same if you compare the two chips has one byte different. (if you share more samples could be cool). The other thing is the byte 0x18 to 0x1d, always looks like a date. On the suspicious chips some 0xFF... Suspicious... Thanks for sharing.

[@ascuber4650]

Yes, 0x0e to 0x17 is the serial according to (page 322) ww1.microchip.com/downloads/en/DeviceDoc/40001906A.pdf

[@luizcassettari]

​@@ascuber4650 This datasheet is from Atmega328PB (version B), the version Atmega328P datasheet doesn't have an official serial number.

[@almostanengineer]

Yo know the Internet is gonna reverse engineer those bytes and figure out what they are used for now, my guess would be a serial number, batch number, factory ID and date.

[@LeoDDJ]

That would be amazing. We didn't have that capability on "Arduino" AVRs yet. It's a feature I find most useful on the STM32 so I can tell them apart even when they run the same code. E.g. MAC address cretion, etc without changing it before you upload it to each board.

[@gudimetlakowshik3617]

Hey kevin, I landed at this video when I was trying to upload code into my MCU on custom boards of mine. The exact marking on the part is as following MEGA328P U - TH 354730 2009YG0 I am having lot of trouble uploading code into this. It has fuse bits set to 00 for all (L,H,E and lock bits too) I am not understanding what to do and how to recover it. Please suggest.

[@billiejean3748]

What are your thoughts on the relationship between solder flux and sleep currents? Do unclean boards do worse?

[@noonabomber]

Just tested with a $1.50 aliexpress clone and got all FF's

[@EddieGooch]

F

[@Kevindarrah]

dang! well just be careful if something isn't quite working right. For most projects, you probably won't even notice

[@patprop74]

I had bought 16 of them from ebay for $2.00 each and they also give me mostly FF's

[@Ed19601]

Have 15 prominis (aliexpress) 2 of those are 3v3 8Mhz rest 5V 16mhz. All seem to be genuine. Bought them i think 5 yrs ago at least

[@BruceMardle]

Just tested my Arduino Uno-compatible. Not a real ATMEGA328, it seems, but then it was only £7.27 incl. P&P!

[@jerenejavelosa735]

Sir how can you program arduino atmega328 pro mini as auto shutdown hardware?

[@googleuser6875]

Great Stuff!!

[@Arvinthhh]

√9 = 3 , got it👍

[@Kevindarrah]

that's nothing, you should see my calculator history :)

[@CXensation]

Whats the fuzz if they work as expected by the avarage user ? IMHO there is no need to test if they can do extreme specs like their origins, as long as they do the work.

[@FlapjackFritz]

How do you know that the ones that are close but slightly different aren't just better fakes?

[@demoncloud6147]

You don't and it does not matter for hobbyist, just don't use these cheap chips on life support or critical application.

[@FlapjackFritz]

@@demoncloud6147 it matters if you want to support the company who made it, or just out of curiosity. Knowing is good even if you still use the chip.

[@nxxxxzn]

does the boot sig look the same on the same genuine chip on different boots?

[@emmalf5496]

i have one of these boards. they are shitty... high sleep current and the ADC is very inaccurate... 0.3 volt error

[@fuckthisksksjjksdfjd]

Is there a similar test for esp8266/32?

[@bonnome2]

Hmm could you maybe ask them to compare the counterfeit chips with LGT8F328P? What is an clone chip of the atmega328p

[@Kevindarrah]

yea good point!

[@jacquesvermaak373]

I found the LGT8F328P to be a useless chip. No 1) You cannot use the EEPROM function like with normal NANO. I found that only every 2nd and 3rd byte of memory works. (Eg, Addr 0,1 bad, 2,3 good. 4,5 bad, 6,7 good etc). Also, using interrupts does not work. It only allows for external interrupts. Using I/O change interrupts that all Arduino chips support, does NOT work with the LGT8F328P. I bought 10 of these clones because of the 12bit AI (0..4096) and the slightly faster clock frequency. But apart from that, I found the clone completely useless for the projects I've successfully with the Nano

[@bonnome2]

@@jacquesvermaak373 Is that according to the official specifications for these chips or did you get bad chips?

[@bonnome2]

@@Kevindarrah There is also another clone found on the WAVGAT pro mini with labeling "WAVGAT AVGA328P". Probably just a re-labling of the logic green one.

[@jacquesvermaak373]

@@bonnome2 I have not read anything about the chips. Maybe I got a bad bunch. Tested about four of them and all had the same result.

[@ticktock8751]

how to test arduino mega 2560 to see if it fake??????

[@TOMTOM-nh3nl]

Thank You

[@dave_dennis]

I have had exactly the same issue with and STM32F103C8T6. Same chips as on a Blue Pill. I worked for days trying to get USB bootloader to work. Finally bought a few from mouser and they worked the first time! Ugh 🤦🏻‍♂️.

[@Kevindarrah]

yea, that's brutal debugging right there

[@ticktock8751]

Will it work for the nano as i just got about 20

[@stevetobias4890]

Would this work for the nano v3 Scratch that, just seen its for any 328P

[@JendaLinda]

It works for Nano and Uno as well.

[@stevetobias4890]

@@JendaLinda thank you, very much appreciated,

[@rhysthurin6098]

Wait...my pro Mini from Kafka and Camus Electronics isn't REAL?! How absurd!

[@lasersbee]

Thanks for this..

[@jaredjones6570]

Will this work as a general test to distinguish genuine and counterfeit boards for all the Arduino models?

[@jaredjones6570]

This is what I got on an Arduino 2560: Arduino 2560 R3 genuine: boot sig dump 1E 4F 98 FF 1 FF FF FF ED FF FF F7 FF FF 57 32 39 30 38 39 FF 7 15 F 17 8 12 7 13 7 FF FF Arduino 2560 pro mini clone: boot sig dump 1E A9 98 FF 1 FF FF FF ED FF FF F7 FF FF 6E 75 6E 6B 77 6F FF 0 4 3 17 1 12 8 13 8 FF FF I don't know if that simply means that the clone has good components - or if anything can be concluded since the boards do not use the ATMEGA328 that you wrote the code for. Edit: I believe this should work on any Arduino models with Atmel AVR microcontrollers - does not work with teensy, from PJRC, for instance - I did test myself. The library used in the sketch: avr/boot.h should work with all AVR microcontrollers. It would be interesting to know what performance deficits exist for clones of Atmega 2560 in addition to what you've shown for the Atmega 328P.

[@Kevindarrah]

Well that looks pretty good, so should give you a little warm and fuzzy about it. I think I might have one of those cheap boards around here too, maybe I'll test out on my end.

[@jaredjones6570]

@@Kevindarrah I tested with two Arduino Uno clones (328P) that have ch340g chips onboard - test returned with no issues in the unique id's (I don't have a proper Arduino Uno to compare against, though). So perhaps the test does not catch all counterfeit 328p (A negative cannot prove a chip is not counterfeit, but a positive can prove that the chip is counterfeit). I'll keep an eye out for your future videos - You have one more subscriber. This is super interesting.

[@rowifi]

Have you contacted Atmel / Microchip for their comments.? They may be fake, but there may be another reason.

[@aslPlayer]

It's open source; how can it be "counterfeit"?

[@BrandonDrury]

Arduino is open source. The Atmega328p is not.

[@liquidsonly]

Interesting. Out of a five recently purchased five nanos from the same seller, two appear genuine, two appear fake. The genuine have a date code 2005, the fakes are 1930 ? and 1903 with really crap laser engraving. The other nano is running a light show so I've no idea if it's genuine or not. No sleeping during the show though so I'm sure it'll be fine. Edit: It's possible to use avrdude to get the same data as the sketch. If you are using avrdude to program your chips then this may be simpler. You need to edit avrdude.conf though. To do so find the Atmega328 section and just above the last "part parent "m328"" insert the following: memory "serial" size = 32; read_hi = "0 0 1 1 1 0 0 0", "0 0 0 x x x a9 a8", "a7 a6 a5 a4 a3 a2 a1 a0", "o o o o o o o o"; read_lo = "0 0 1 1 0 0 0 0", "0 0 0 x x x a9 a8", "a7 a6 a5 a4 a3 a2 a1 a0", "o o o o o o o o"; ; Save it and run avrdude in terminal mode. You should see an extra option to dump serial. As an example: avrdude -C/etc/avrdude/avrdude.conf -v -patmega328p -cstk200 -t avrdude> dump ser >>> dump ser 0000 1e 92 95 ff 0f bc 00 26 ff 09 ff 17 ff ff 57 39 |.......&. ....W9| 0010 33 33 35 34 ff 14 12 09 17 05 12 09 13 09 ff ff |3354... ... . ..| Gives the same output as the sketch. Found by a lot of trial and copy/paste, I still have no idea how the config file works. My apologies for not formatting in monospace. I wrote a little more at www.millstream-computing.co.uk/pages/posts/is-your-atmel-chip-genuine-17.php

[@MikeBramm]

Cheap Chinese knock-offs are not uncommon. About 5 years ago, FTDI got tired of all of the fake USB to serial converters marked as FTDI devices, so they changed their driver to purposely brick the fake devices rendering them inoperable. This obviously bricked all of the boards that the fake devices were on. So, instead of going after the manufacturers (which would have been an impossible fight), the consumers of the boards that contained these chips (who had no idea the "FTDI" device was fake) were the ones that suffered. I remember seeing an X-Ray of a fake chip that actually contained a repurposed micro that was programmed to act like the FTDI device. It's incredible that the Chinese companies go through all the trouble of making the fake devices but they totally miss simple things like memory locations or functions that don't operate like the original. It's obvious that they don't really care how it works, they just want to make the sale and not worry about what happens next (because the Chinese government will defend them).

[@pyromen321]

That’s such a dick move from whatever executive or manager decided to “solve” that issue. I’m sure they could have come up with a better solution

[@bonnome2]

Yes but this is different, they had to make the chip only compatible with the ftdi driver. These counterfeits run the same avr instructionset. This requires considerable more resources to pull off.

[@Kevindarrah]

oh I remember that one - FTDI GATE!

[@station240]

The annoying part was the reason for all the fake FTDI serial chips in the first place was the difficulty in getting their chips in the first place. When they are available, they are way more expensive than they should be, costing more than the micro they are attached to. I had to order some FTDI chips this year, the first order from one supplier still hasn't shipped (restocking date keeps shifting), had to order more from another supplier to actually get something.

[@jontech_Lab]

Hello mr KEVIN,would you make a video on how to solve"fatal error"on esp32 camthanks in advance

[@ursmii_]

nice, but what's the real IMPACT having a non-genuine 328P ??? Did you find any errors in executing code, timing faults or other suspicious things?

[@JohnGodwin777]

Watch the first video in this series

[@paulcohen1555]

Ha, Ha, Ha, They will fix that on the next wafer iteration.

[@Kevindarrah]

lol, I know!

[@jimstand]

They are retooling right now.

[@Spritetm]

Possibly they don't even need to change their masks... this data usually is in eFuses or (more likely here) some bits of flash that are programmed and made read-only during the wafer testing process. Should only take them a few tweaks to the automated test equipment scripts to fix this.

[@habiks]

Why on earth would you buy original overpriced arduino??

[@Kevindarrah]

you don't have to do that - just buy a clone that uses genuine parts - you can still get those pretty cheap. Even I sell a breakout with the boot loader for $10 and my parts are all sourced through Digikey. You might want to check out the first video linked to in the description on reasons why genuine parts matter. lol, nice user name btw!

[@antibrevity]

To support the organization that made all of this possible and still developers the IDE? I buy generics, but I think that everyone should buy /something/ genuine from Arduino to support their work. No overpriced Arduino sales = no Arduino.

[@habiks]

@@antibrevity ppl actually use the shitty arduino IDE? Damn.. I feel their pain. No overpriced arduino = no arduin = finally getting ppl to move over to proper IDEs.

[@jimstand]

@@habiks Why do you care what other people use for an IDE? Do you also care whether they like Coke or Pepsi?

[@habiks]

@@jimstand I don't care what they use. I was trying to find a reason why would you support overpriced piece of crap that does nothing.. the compiler is not from arduino, the mcu is not from arduino, most of the libs are not from arduino... not to mention the whole aruino project is open source and most of the push forward was made by ppl that are not part of Arduino.. not to mention the "clones" (what exactly are they cloning?! ; it's not the IDE) are many times upgraded with fixes arudino never put out (beffier lin. regulators, extra holes next to pins, etc).

[@demoncloud6147]

avrdude was my grandpa !

[@Ncky]

Now Chinese chip manufacturers will start putting random bytes instead of 0xff after this vid 😂. I just don't get how much they earn for fake chip.

[@Kevindarrah]

ha, I know that's what everyone is saying... oh well

[@startobytes]

Hi, could someone Help me?! I want to include my own Boards in the Arduino IDE. I want it Like with the esp8266 boards, but I should Not be published. I hope you can Help me.

[@paulcohen1555]

They will cheat you for few cents savings in the product.

[@nitinkumar29]

I think, not only manufacturing cost but license fee too add up.

[@paulcohen1555]

@@nitinkumar29 Right. And I hope Microchip/Atmel will fight against the clones using their name and/or part numbers.

[@RobNO]

Honestly, who cares as long as it works. If you want/have to work with genuine Arduino you certainly don't buy on Aliexpress

[@normcaissie5598]

you know if it cost 2$ LOL

[@Kevindarrah]

yea, maybe that's easier :)