Counterfeit RTC modules

[gregebert]

Lesson learned: You get what you pay for.

I bought a $2 US realtime clock module on Ebay, imported from China, and the DS3231 chip appears to be counterfeit because it was not keeping accurate time. I was suspicious about the price, considering I paid almost $8 US just for the DS3231 from a reputable supplier. Over a few days, it lost about 1 hour of time.

After replacing the chip with a genuine Maxim DS3231, it's running correctly.

A quick web-search found there are several others who have experienced this.

Despite having to replace the RTC chip, the module itself is still worth the price because it included a rechargeable Li-ion coin battery and a serial EEPROM (no time yet to see if that's working correctly...)

[Terry S]

Greg, you really ought to inform Maxim. They may want the part.

Terry

[David Forbes]

Sorry, but I do not understand this race to the bottom, that produces
products that look like products that work. Why not just buy a working
chip for the price of a working chip, and save yourself the hassle and
disillusionment that comes with a part whose price is clearly too low to
be a working product?

The odds of the rest of the module working properly are not too high.
You might spend hours dealing with the fallout of saving ten dollars.

--
David Forbes, Tucson AZ

[Dan Hollis]

I don't understand why chinese bother to counterfeit LM2596s, but they do
all the same. In fact 99% of the dc buck converters you find on amazon and
ebay are counterfeits.

-Dan

> --
> You received this message because you are subscribed to the Google Groups
> "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to neonixie-l+...@googlegroups.com.
> To post to this group, send an email to neoni...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/neonixie-l/56B3F03C.6090903%40dakotacom.net.
> For more options, visit https://groups.google.com/d/optout.
>

[gregebert]

Another possibility is that these were scrapped/rejected units that escaped destruction at Maxim, and ended-up being sold, obviously illegally, and wound-up in products. 

It does seem odd to me that someone would go to the bother, not to mention the expense, of copying an inexpensive part and selling it.

------------------------------------------------------

You wouldn't believe the hassles I had to go thru to get an 8-inch 'souvenir' wafer of a chip I worked-on that was never taken into production.

[Oscilloclock]

I'd be concerned about the Li-ion button battery - leaking chemicals in a HV nixie clock, or possibly causing fire, is another level of risk.

[Dekatron42]

I agree on the battery, I bought some similar units a few years ago which were incorrectly designed so that they were trying to charge non-chargeable batteries!

I also had problems with the EEPROMs and about everything else on these and other modules. The modules were bought from a usually respectable Swedish company but they had made the poor decision to try to earn more money and also bought from the free Chinese market and not from their ordinary distributors. I got new modules when I could show how poor they were but they gave me a real headache before I figured out that they had poor components and were incorrectly designed.

I still find photos of the same modules from some resellers, but I don't know if they deliver the same poor modules.

/Martin

[GastonP]

Well, many times I have to go the low cost part risk way just because in my country the "reputable" sellers charge 5 to 10 times the US selling price and they carry the parts only when the final price is not outrageous after their uplift.

Said that, I only take the risk for hobby jobs. I would never risk a lab, prototype or production failure just because of being cheap.

On Thursday, February 4, 2016 at 9:43:43 PM UTC-3, nixiebunny wrote:

Why not just buy a working chip for the price of a working chip, and save yourself the hassle and
disillusionment that comes with a part whose price is clearly too low to
be a working product?

The odds of the rest of the module working properly are not too high.
You might spend hours dealing with the fallout of saving ten dollars.

[严泽远]

There're so many reconditioned DS3231SN chips in China market, most of them are disassembled from some old equipment, just like electricity meter... I can find some RTC module list price 0.5USD with DS3231SN on it, it must be reconditioned one since a new DS3231SN will over 1.5USD. But it's not difficult to recognize the new one or reconditioned one, the thickness and laser characters are different.

Laser characters of reconditioned have wider line than real one, looks rough, please check them:

在 2016年2月4日星期四 UTC+8下午11:04:45，gregebert写道：

[Terry S]

By "reconditioned", do you mean counterfeit?

Why would the markings be different? I could see if they were from an older batch... salvaged parts.

I got burned one time at work on counterfeit parts, bad op-amps from Burr-Brown. Except not really, fake parts made in China. Why would they counterfeit a 60 cent part?  Because they can make 55 cents on it and sell them by the millions.

In most applications it would never have been caught. But it was in mine, and caused quite a ripple back up the supply chain.

Terry

[Mich...@aol.com]

It's also possible it was simply a bad chip.

 

Michail

 

[threeneurons]

One thing about eBay sellers, is not properly handling the devices. I bought some power FETs from an eBay seller once, and he stuck them in plain white static generating styrofoam. Even Futurlec shipped chips to me in plain non-antistatic plastic bags. I now only buy semiconductors from vendors that static protection, like Mouser & Digi-key.

And those "counterfeits", can be recycled chips. They stick computer boards over open fires, to melt the solder, and let the ICs slide off. The ICs are collected, cleaned, and possibly remarked. Antistatic protection, is the least of your worries with those, even if they did handle them properly. I guess you can call those reconditioned. 

[Quixotic Nixotic]

Prosperity will eat itself.

I just bought a JYE tech mini dig-it scope to build into something. I know it's a pile of poo, but at about $13 USD inc. shipping, I really don't mind. I just wanted something that will view audio waveforms - the added noise is not an issue. But JYE tech are bleating that their own China-produced rock bottom priced items have been cloned and the software stolen. OK, I think I have a gen-U-whine one, but lots of people are getting the Sexton Blakes (fakes).

I bought a stick of 'new' CA3080s a while ago. They are obviously pulls - I can see the the legs have once had solder on them.

I believe the process of repainting and reprinting ICs is commonly known as 'black-topping'. They put a layer of black matt something over the top and then overprint the logo and the datestamp.

I bought a bunch of those little DS2321 boards with memory. So far they have all been okey dokey, but I expect I shall fall foul of the naughty people one day.

John S

[Dan Hollis]

Looks like a counterfeit to me.

I've been spending a lot of time lately analyzing chinese counterfeit
chips. They are everywhere.

-Dan

On Fri, 5 Feb 2016, Michail1 via neonixie-l wrote:

> It's also possible it was simply a bad chip.
>
> Michail
>
>

> --
> You received this message because you are subscribed to the Google Groups "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
> To post to this group, send an email to neoni...@googlegroups.com.

> To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/581b5f.761df886.43e64337%40aol.com.

[vfdclock]

It's very simple,a lot of this kind of chips are from older PCBs and de-solderd and recreate the surface to make it can be resell on the market, most of these kind of people are in GugangDong Prov. China,especially from the ShanTou city,it's a huge industry in GuangDong,China, you can find a lot of the fake/retread chips from taobao, most of the sellers are also from ShanTou,GuangDong City if you check the location of the sellers.
The price of on DS3231 retread chips are less than 0.1~0.2usd after been cleaned,plating the legs,and recreate the surface marks,and of course if the can send it to you via HONGKONG,the int-mail price can low down to started from ~1/6usd. That is why you can find a lot of the sellers who can give you the price like $2 even with the free shipping from HongKong.
So if the price is too low,and the sellers' location is GuangDong or written like HongKong, 90% are made by retread chips.But it does not mean all the stuffs are unsaleable.buy it or not depends on yourself.

在 2016年2月4日星期四 UTC+8下午11:04:45，gregebert写道：

Lesson learned: You get what you pay for.

[gregebert]

Happy New Year, unless you rely on the dirt-chip RTC module I bought several months ago, because today would be Sept 2, 2018.

WHAT ???  That's the date my RTC chip reported this morning: 09/02/2018

Last night, before midnight and right after I got tired of writing software, it was correct at 12/31/2018. 

The funny thing is, this counterfeit RTC module has kept the time very accurately over several months. My previous experience with a different fake chip was that it was losing about 2 hours per day; since this one was accurately keeping the time I assumed it was a genuine Maxim device. Nay, not so.......

After resetting the correct date, it's showing that now. So I wonder....if someone went to such lengths to make such a time-wise accurate fake, why would they bungle so poorly on a simple Dec31 --> Jan01 rollover ?

Sure enough, when I forced the RTC back just before midnite and let it roll-over, I got another wrong date (09/01/2018), so this RTC is even more messed-up than I thought.

[John Smout]

Which IC is this? I have a couple of clocks that are gaining time in a similar way to each other. I’ve checked my software and think it must be down to rogue DS3231s. Funny thing is I think the ICs are samples direct from Maxim.

John S

[gregebert]

Label says DS3231, but it's definitely fake.

I have 2 other clocks with genuine DS3231's (purchased from DigiKey), and they have behaved as expected for years.

[Dekatron42]

Any chance that you can ask Maxim about this behavior, it could be a bug in a batch of these ic's.

Someone here on the forum had the capability of etching away the case on ic's and looking at the chip if I remember correctly, it would be interesting if these "fake" ic's could be sent to that person and have this procedure done to them so we could learn what's inside!

There are some websites saying these ic's are fakes and some saying they are true original ic's and having proof of one that behaves badly and then seeing what is inside would be very interesting, I can't remember ever having seen such a comparison anywhere.

/Martin

[Paul Andrews]

What happens if you let it roll over from a different date? Say 12/31/2017? My guess in that case would be 08/01/2017.

[gregebert]

It comes up as 09/01/2017. I'll keep this RTC online for a few more weeks, as the system debug is almost done.

There was another rollover a few months ago where the date was messed-up, but I traced that to a software bug that was my fault.

[Terry S]

I've been thru the process a dozen times -- but using an independent lab that specializes in the process.

You should have several samples -- at least 3 should go to a lab, and 3 (unused if possible) to the manufacturer. The manufacturer will also want to see the independent labs' results.

The lab will use a Scanning Electron Microscope to verify the origin of the die as real or counterfeit. It can also scan for other types of failures such as ESD or EOS.

The manufacturer will run a full parametric test that can definitively ID the part as real or fake -- before decapping. That is, if you are a meaningful volume client for the product. If not, you may just get a run-around. Get the local FAE on your team first.

It's expensive, typically the lab charge is $1k for 1 to 5 samples.

Terry

[Frank Bemelman]

But if your first rollover problem was a software bug, then why would this particular rollover problem not be  another software bug?

Op wo 2 jan. 2019 om 15:58 schreef gregebert <greg...@hotmail.com>

It comes up as 09/01/2017. I'll keep this RTC online for a few more weeks, as the system debug is almost done.

There was another rollover a few months ago where the date was messed-up, but I traced that to a software bug that was my fault.

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/84ed03ec-1fff-491e-ba9f-cead91660273%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Vriendelijke groeten,

Frank Bemelman

Frank Techniek

Vennestraat 11X

2161LE  Lisse

Telefoon +31 252 769 178

[gregebert]

When I looked at the raw data from the RTC, which is in packed BCD format,  I could clearly see that my code was not properly parsing it. I believe it was Sep 30--> Oct 01 where I calculated the wrong month.

The Jan 01 rollover bug has incorrect BCD data from the RTC itself.

[Dekatron42]

What happens if you set the alarm registers to 00:00:00 at 1/1, is the alarm triggered at rollover?

/Martin

[gregebert]

For now I'm going to hold off on any experiments that require code-changes. I'll keep this RTC chip online for additional experiments; known-good RTC will go into the second board-set.

It appears the alarm registers are testable with software; I dont have the INT/SQW pin connected in my system.

I'm really itching to get the NIMO lit-up and running. So close to actually firing it up, but I wont attempt that unless everything is 100% stable. Mindset is that I have only 1 chance to get it right and it must be right the first time. Anything that can go wrong and remain uncorrected for more than a few milliseconds is assumed to result in a dead NIMO tube, and I only have 1. 

[Dekatron42]

Ok, but if you feel like going back to the faulty one, setting the alarm register will show you if the RTC is doing what it should internally and the data read from it is wrong (unless you make an error in programming the alarm registers that is ;) ).

/Martin

[Frank Bemelman]

Ok, I see. Perhaps it is just a single defective RTC chip you have there. You could write a software patch to set the day to 1, at every rollover :) And then wait for the next weird behaviour :)

Op do 3 jan. 2019 om 00:27 schreef gregebert <greg...@hotmail.com>

When I looked at the raw data from the RTC, which is in packed BCD format,  I could clearly see that my code was not properly parsing it. I believe it was Sep 30--> Oct 01 where I calculated the wrong month.

The Jan 01 rollover bug has incorrect BCD data from the RTC itself.

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/e92236f5-0fbe-46f3-ab1b-efbf339aa91e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Luka C]

I have written previously about a similar incident with ATmega328p chips ordered from Aliexpress. They did not arrived in a reel, but just packed in a small plastic bag. Two of them worked fine, but Atmel Studio was unable to even read the ID of the third one, thus rendering it unprogrammable and useless..and since it's an SMD, it was not as easy as just popping it out of the socket. I guess they are chips which failed QC and somehow, these companies got them, performed a somewhat "basic check" and sold them as new ones.

I have also ordered a lot of 10 DS3231 chips from some other Aliexpress seller (they had considerable 5* ratings so I decided to give it a try). They were like 14$/10pcs. The chips worked, the time is still very precise after two years of operation of the older clock I made. However, I did notice one "bug" in the chip. If you keep querying the chip too often (like in an unconditional loop) for seconds, it will eventually hang the I2C line. The solution was to set a timeout for I2C read operation, and if timeout does occur, the master (microcontroller) has to keep clocking the SCL line until the DS3231 releases the bus. However, I've read some people reporting the same "bug" on multiple sites, so I'm not sure if it's a "bug" specific to possibly counterfeit chips or just a general bug that affects genuine ones as well.

[Bill van Dijk]

There can be a few different reasons for the i2c buss to hang, and may not be the fault of the DS3231. For instance, if there is a software interrupt during an i2c read or write, that can do it. Are you using a mpu based hardware i2c engine, or do you bit-bang the buss?

 

If the DS3231 is battery backed up, than resetting the power will not reset the buss either. In cases where I use a battery backup, I use a small routine to test the buss at every startup, including brownout and watch dog events:

 

void I2C_Test() //test to see if I2C line is in a valid (stopped) state

{

if                            //test I2C line

   (!input(i2c_SDA))           //if SDA stuck low - Fault Condition

   {   

   I2CEN=0;                    //temporary disable I2C Hardware Module

   output_drive(i2c_SCL);      //set SCL as output

 while

   (!input(i2c_SDA))           //Continue loop until SDA is released (high)

    {                           

    output_toggle(i2c_SCL);    //Toggle I2C clock line till fault clears

    delay_us(10);

    }

   output_float(i2c_SCL);      // set SCL back as input

   I2CEN=1;                    // re-enable I2C Hardware Module

   }

}

 

Bill

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/9529a437-2084-41b9-af82-158aa95ff505%40googlegroups.com.

[David Pye]

It seems even things barely worth faking are being faked also.

I have a bag of 500 74HC595 shift registers, that are actually rebadged 74HC596s (as in, open collector, SINK, not SOURCE-capable).

Which are useless for my application :-(

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/003a01d4b4f0%24aaac8250%24000586f0%24%40gmail.com.

[Bill van Dijk]

That is really weird. Please understand I am not questioning what you are saying, but perhaps there is another explanation. The 74LS596 (I have never seen an HC version) is indeed an open collector chip similar to the 74HC595, which is a tri-state device. On the 74HC595 the inv G (pin 13) should be held low for normal operation. If it goes high for any reason, the output will float in tri-state mode, similar to what an open collector would look like. As you say, I can’t for the life of me not figure why anyone would bother to rebadge those chips especially since there does not seem to be an economic one (which is usually the motivation).

 

Bill

 

From: neoni...@googlegroups.com [mailto:neoni...@googlegroups.com] On Behalf Of David Pye
Sent: Friday, January 25, 2019 4:12 PM
To: neoni...@googlegroups.com
Subject: Re: [neonixie-l] Re: Counterfeit RTC modules

 

It seems even things barely worth faking are being faked also.

[David Pye]

Hi Bill,

Indeed, it seems bonkers.  I assume it's because someone has a huge stash of open collector output 74 series shift registers, and either thinks they're the same, or that most people won't notice the difference.

I bought two batches of 500 of them, from two different chinese suppliers, and they are all exactly the same, having the same laser engraved batch number even.  To be fair, the prices were suspiciously good, but I wasn't banking on fake shift register ICs....

They are badged as TI SN74HC595N, and batch GM1807FSF.

I have some 'normal' SN74HC595N, and swapping these out for the real ones generates the expected behaviour ( and yes,  inv G is indeed pulled low).

With the 'fake' ones, they will appear to work OK with a pull up resistor present (as you'd expect with an open collector output), but are unable to source any current.

I thought it was just me, but then I googled GM1807FSF and found someone else (in German) having the same problem! 

https://www.mikrocontroller.net/topic/463936

David

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/008501d4b4fc%24bb636a80%24322a3f80%24%40gmail.com.

[Tony]

There are indeed some strange 'counterfeits' around. Maybe they were a
deliberate attempt to confuse reverse engineering of some product, but
never used? or it could have been a simple mistake.

I have a few thousand MPSA92 which have been remarked from MPSA42,
with the 4 and 9 superimposed. They work perfectly as PNP HV
transistors so it's possible somebody just forgot to change the
engraver text and the mistake wasn't noticed until they had a large
pile of mismarked PNP MPSA42s.

Tony.

>> *From:* neoni...@googlegroups.com [mailto:neoni...@googlegroups.com] *On
>> Behalf Of *David Pye
>> *Sent:* Friday, January 25, 2019 4:12 PM
>> *To:* neoni...@googlegroups.com
>> *Subject:* Re: [neonixie-l] Re: Counterfeit RTC modules

>>
>>
>>
>> It seems even things barely worth faking are being faked also.
>>
>>
>>
>> I have a bag of 500 74HC595 shift registers, that are actually rebadged
>> 74HC596s (as in, open collector, SINK, not SOURCE-capable).
>>
>>
>>
>> Which are useless for my application :-(
>>
>>
>>
>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "neonixie-l" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to neonixie-l+...@googlegroups.com.
>> To post to this group, send email to neoni...@googlegroups.com.
>> To view this discussion on the web, visit
>> https://groups.google.com/d/msgid/neonixie-l/008501d4b4fc%24bb636a80%24322a3f80%24%40gmail.com

>> <https://groups.google.com/d/msgid/neonixie-l/008501d4b4fc%24bb636a80%24322a3f80%24%40gmail.com?utm_medium=email&utm_source=footer>
>> .

[gregebert]

Happy New Year, everyone. If you built a new clock last year with a used or cheap DS3231, check the date to make sure it's correct.

I posted about this 2 years ago when I found another fake clock chip.

[Robert G. Schaffrath]

Over a year ago I purchased three of the four-digit Chinese IN-12 clocks with the LED background lights. Of those, two of them were complete junk and AliExpress refunded my money so I only paid USD$20 for the one good clock (I offered to return the bad ones but they did not want them - surprise). One of the bad ones just stopped working (appears to be a CPU issue though one tube lights up so the tiny quarter sized HV power supply is good and salvageable) and the other loses around 10 minutes every few hours. It cannot keep accurate time at all. Unlike another Chinese board I have for IN-14's that is rock solid and does not lose any time at all over several weeks of operation. I figure the clock that cannot keep time can eventually be fixed by replacing the DS3231M chip. I picked one up on eBay for the day when I can find someone who has an SMD rework setup to that I can remove the old chip and install the new one. But this is definitely a case of a worthless DS3231M being used in products. I assume the manufacturer wound up with a bunch of counterfeit or salvaged DS3231M chips and just passed them off. Lucky for me, one of the chips does work properly.