Fake MAX3232, any additional details?

[heypete]

Hi all,

I recently purchased some RS232-to-TTL interface boards from a vendor in China. They purported to use a MAX3232 chip to do the conversion, and the chip packages were marked as MAX3232s. However, I was suspicious: these chips failed after a few weeks by garbling the serial data and then shorting out and getting dangerously hot.

Clearly, some investigation was in order. I got some free sample MAX3232s directly from Maxim so I have a known-good reference, desoldered the failed Chinese chips, then tossed all of them into nitric acid to dissolve the epoxy packaging and metal legs, leaving only the dies.

Here's what I've got (click on images to enlarge). Sorry for the odd images: the microscope-mounted camera was kaput, so I aimed a handheld camera through the eyepiece:

Legitimate MAX3232 package. Note the quality of the laser markings and the use of an alignment stripe rather than a dot to mark pin 1. This poor guy is less than a month old and is about to get dissolved in my lab. Muwhahaha!


Here's the suspected fakes. Note the lower-quality laser markings, different typeface, and the use of the alignment stripe and a dot. Still, it's possible these chips could have been made in a different fab, so the markings don't really prove that they're fake. Please ignore the leftover solder from my attempt at de-soldering using a handheld iron.




Here's the die of the legitimate MAX3232. It's roughly twice as large as the suspected fakes. Note the use of gold bond wires that were not affected by the nitric acid bath. All the following die photos are using the same magnification, so you can directly compare relative sizes.


Let's zoom in on the black patch in the upper-left corner, immediately to the right of the bonding wire. It's hard to see with my camera, but it clearly says "Maxim".


Now let's look at the suspected fakes. They both had identical dies. The dies were quite a bit smaller than the Maxim one and had a distinctly different layout. The bond wires were clearly not gold, as they were dissolved away in the nitric acid.


The left side looks to have some text, so let's see what it says. This is the best quality photo I have of the markings, but it was much clearer by eye. The top line says 2009.11 (date?) and the bottom says "WWW01" (or possibly the letters O and I instead of the numbers 0 and 1).


In conclusion, I can safely state that those chips are fake. Although they did seem to function identically to the MAX3232 in that they ran on 3.3V DC power just fine, had charge pumps to pump the voltage up to +/- 5.5V, and correctly converted serial data between TTL and RS232 polarities, they're clearly different, much less reliable (though several have worked fine for months, quite a few have died horrible deaths in normal operation), and have none of the Maxim markings on the die.

Also, dissolving things in nitric acid is fun.

That said, does anyone have any idea who makes these chips, where, and under what original brand names? Are they manufactured specifically to be seemingly-legitimate MAX3232s or was it just unscrupulous middlemen who remarked chips with similar functionality to the MAX3232?

Can anyone help me in identifying the different sections and components on both the real and fake dies? I'm curious about what sections of the dies make up the charge pumps and other functional parts.

Any insight would be very much appreciated.

[georges80]

My insight: Stop buying chips from the Chinese grey market and you'll be able to get on with building stuff (and have it work) versus playing with acid

Remarking IC's is an active business, especially when they can sell some cheap/garbage component at the higher premium of real components (even at marked down prices you see on the various asian sites).

By purchasing components from these sources we feed the folk that will continue to do this.

cheers,
george.

[langwadt]

I think max232 and max3232 has become a sorta generic part numbers for a rs232 transceivers, there are several manufactures, I know TI makes a max3232

[AndyC_772]

It's definitely a device that gets faked, though. I once spent ages trying to figure out why the serial interface on a brand new PCB didn't work, but the identical circuit on the previous design which I'd copied worked perfectly.

It wasn't until I swapped the MAX3232 from a working board onto the new one that it became apparent the chip was a fake... apparently (it later transpired) purchased from a grey market source to meet the schedule.

The fake chips weren't completely dead, but there were some odd waveforms on the charge pump pins, and the chip couldn't sustain a negative voltage into a receiver. QC rejects from somewhere.

[bktemp]

Not buying from China does not help unless you buy from a big distributor where you can be sure it comes directly from the manufactorer.
I recently had a similar problem with a MAX232 bought in Germany. The one at the top works fine, the other one draws ~100mA and gets hot.
If you look at the marking it is clearly a fake one, because I could not find any other TI part with a dent and a line for pin 1 marking. The dent looks like Maxim parts, but the marking has clearly a TI logo. The logo has a much lower quality than any other TI logos I have seen.
If I connect only +5V and the 1uF caps for the positive charegpump the ic works. But with all capacitors connected, it gets hot. The chargepump frequency is 380kHz. The other part uses 45kHz. The waveform looks odd, but I found a appnote from Sipex describing a 4 phase chargepump. Maybe this is a factory reject from Sipex or a bad copy.

Does anybody have a recent MAX232 in SO16 package from TI and could post a picture?
Despite working fine, the marking on the top one looks a bit unusual to me. All other TI parts I have seen had the pin 1 bar completely filled, the individual lines from the laser were not visible. And the TI logo also looks a bit different to other ones.

[heypete]

My insight: Stop buying chips from the Chinese grey market and you'll be able to get on with building stuff (and have it work) versus playing with acid

Good advice. Still, I have a fully-stocked chemistry lab at my work (non-electronics-related scientific research), so I might as well take advantage of it.

This was simply for hobbyist/prototyping stuff, not for production purposes. If I ever was lucky enough to get enough interest in things for production purposes then pinching pennies on counterfeit stuff simply isn't worth it and I'd buy from proper distributors.

By purchasing components from these sources we feed the folk that will continue to do this.

Absolutely. Unfortunately, it's difficult for the average person, particularly hobbyists buying one or two of a component, to determine if a component is legitimate or not short of testing them individually or simply avoiding sites like AliExpress and DealExtreme entirely (which might not be a bad idea).

In my case, I was curious to see if the chips were legit (they're not) and to see if they functioned correctly (they did, at least until they failed). Combined with wanting to do view some chip dies, it was some of the least-expensive fun I've had in a while.

[amyk]

This is what the die of a TI one looks like:
http://zeptobars.ru/en/read/max3232-ti-rs232-transceiver-lvttl

It doesn't appear to have any brand markings, unlike this TI MAX232:

https://en.wikipedia.org/wiki/File:TI_2BC121M_MAX232N.jpg

Someone else appears to have come across the same fake die with "WWW01" marking:

http://affable-lurking.org/2015/07/31/de-capping-ics-for-fun-and-profit-but-mostly-fun/

Maxim's datasheet has a die image, although it doesn't match either of yours. It claims the part only has 339 transistors, so no doubt cloners will find it easy to reproduce.

[heypete]

Someone else appears to have come across the same fake die with "WWW01" marking:

http://affable-lurking.org/2015/07/31/de-capping-ics-for-fun-and-profit-but-mostly-fun/

Good find! That's definitely the same die that I'm seeing on these chips.

As for the legit dies, it's interesting to see so much variation over the years that the chips were made. I would have thought that they'd want to minimize changes, but maybe not?

[Tomorokoshi]

Some insight on the counterfeit industry in China:

http://www.asq.org/asd/2009/03/compliance/counterfeit-parts.pdf

[Skashkash]

I also got bit by some fake max3232s earlier this year.

 Random failures and chips running very hot. Did some research,  and a post in another forums pointed to the specific problem.

 Turns out some of the clone chips are missing pull up resistors on the logic level input pins. If you leave any pins floating you'll pick up stray signals,  start oscillating and get the observed heating.

It's also pretty easy to check for the presence of that resistor with a meter.

All in all, a royal pain to deal with.       

[Gyro]

Some insight on the counterfeit industry in China:

http://www.asq.org/asd/2009/03/compliance/counterfeit-parts.pdf

That pdf is both scary and bizarre... ICs drying on the pavement? a pile of 0402 caps for goodness sake! 

[PedroDaGr8]

Hi all,

I recently purchased some RS232-to-TTL interface boards from a vendor in China. They purported to use a MAX3232 chip to do the conversion, and the chip packages were marked as MAX3232s. However, I was suspicious: these chips failed after a few weeks by garbling the serial data and then shorting out and getting dangerously hot.

Clearly, some investigation was in order. I got some free sample MAX3232s directly from Maxim so I have a known-good reference, desoldered the failed Chinese chips, then tossed all of them into nitric acid to dissolve the epoxy packaging and metal legs, leaving only the dies.

Here's what I've got (click on images to enlarge). Sorry for the odd images: the microscope-mounted camera was kaput, so I aimed a handheld camera through the eyepiece:

Legitimate MAX3232 package. Note the quality of the laser markings and the use of an alignment stripe rather than a dot to mark pin 1. This poor guy is less than a month old and is about to get dissolved in my lab. Muwhahaha!


Here's the suspected fakes. Note the lower-quality laser markings, different typeface, and the use of the alignment stripe and a dot. Still, it's possible these chips could have been made in a different fab, so the markings don't really prove that they're fake. Please ignore the leftover solder from my attempt at de-soldering using a handheld iron.




Here's the die of the legitimate MAX3232. It's roughly twice as large as the suspected fakes. Note the use of gold bond wires that were not affected by the nitric acid bath. All the following die photos are using the same magnification, so you can directly compare relative sizes.


Let's zoom in on the black patch in the upper-left corner, immediately to the right of the bonding wire. It's hard to see with my camera, but it clearly says "Maxim".


Now let's look at the suspected fakes. They both had identical dies. The dies were quite a bit smaller than the Maxim one and had a distinctly different layout. The bond wires were clearly not gold, as they were dissolved away in the nitric acid.


The left side looks to have some text, so let's see what it says. This is the best quality photo I have of the markings, but it was much clearer by eye. The top line says 2009.11 (date?) and the bottom says "WWW01" (or possibly the letters O and I instead of the numbers 0 and 1).


In conclusion, I can safely state that those chips are fake. Although they did seem to function identically to the MAX3232 in that they ran on 3.3V DC power just fine, had charge pumps to pump the voltage up to +/- 5.5V, and correctly converted serial data between TTL and RS232 polarities, they're clearly different, much less reliable (though several have worked fine for months, quite a few have died horrible deaths in normal operation), and have none of the Maxim markings on the die.

Also, dissolving things in nitric acid is fun.

That said, does anyone have any idea who makes these chips, where, and under what original brand names? Are they manufactured specifically to be seemingly-legitimate MAX3232s or was it just unscrupulous middlemen who remarked chips with similar functionality to the MAX3232?

Can anyone help me in identifying the different sections and components on both the real and fake dies? I'm curious about what sections of the dies make up the charge pumps and other functional parts.

Any insight would be very much appreciated.

Nice to see someone else getting into Nitric Acid, though I must admit you did a MUCH better job than I did!

[nctnico]

That said, does anyone have any idea who makes these chips, where, and under what original brand names?

Probably Sipex.

[BarsMonster]

Gold/non-gold bondwires - not a big deal, it's only question of technology, gold is not universally better for this application. For example, gold could have reliability issues when fused with aluminum.
The mentioned photo on the zeptobars could also be fake one branded as Ti.
The problem of fakes is getting real.

Soon I will reveal comparison of fake and real Ti NE5532 :-)

[heypete]

Nice to see someone else getting into Nitric Acid, though I must admit you did a MUCH better job than I did!

Thanks!

My philosophy for this test was "more is better" (particularly since the fake chips were non-functional and I didn't plan to do any live testing) and I immersed the whole chip in acid rather than dropping it on the package. This ended up dissolving the whole package and leadframe, which was fine for me and left the dies relatively clean. An acetone bath followed by a few minutes in the ultrasonic cleaner in a vial containing acetone got rid of all the crud.

Your method looks to be a lot more efficient in terms of acid use.

Thank goodness for fume hoods and thick sashes.

[micro23]

Recently bought 4 RS232 to TTL Converter Module at Digikey, all of they fail. I asked myself, "such a simple circuit and is not working", I tried many things, change caps, check pcb connectivities, remove the IC and try to make it working on a protoboard. Result: nothing works. After that, I found this post 
So taking a look under the microscope I found the same chip than heypete.
Here the pictures:

[stj]

maybe the caps are wrong for the chip.
some chips use 1uf and some use 100nf
what would a charge-pump do with the wrong caps??

[micro23]

maybe the caps are wrong for the chip.
some chips use 1uf and some use 100nf
what would a charge-pump do with the wrong caps??

Caps were 100nF. It means the power supply needed is 3.3V. Also tried the following configurations taked from the datasheet (nothing worked)
Note: I had in mind that the caps labels in the pcb are not matching with the labels from the datasheet.
After soldering a MAX232 with 1uF caps, that's works, so the PCB layout is ok.

[stj]

i didnt mean that,
i meant chips from maxim,sippex,harris,ti etc dont use the same cap values even for the same part numbers.

[TK]

It might be safer to go with ICL3232 as probably they are not cloned as much as the MAX3232.  There are other similar chips that are MAX3232 compatible from other manufacturers.

[NivagSwerdna]

Sounds familiar https://www.eevblog.com/forum/chat/fake-chips/msg1552951/#msg1552951

[tsman]

The markings on that MAX3232 are at an odd angle as well.

[Dave_g8]

Hi All,

I recently purchased a few of the RS232 to TTL (DB9) modules from Amazon.
The boards are marked “HW-044”
The chips are marked “MAX3232 ESE+2228”.

I tied off the unused TTL input “T2IN” on pin 10 which is a general recommendation in all MAX3232 datasheets.

The issue I have observed is that the device fails to function correctly (charge pumps become erratic) if the Receiver RS232 “RIN” voltage is driven higher than about 7.2V. I had it connected to a unit that was driving +/- 9.4V on the RS232 interface and this caused the problem.

I added a simple resistive attenuation pad on the RIN, and everything worked fine. I suspect that there is some form of protection diode on the RS232 inputs. The MAX3232 specification states that the receiver inputs should tolerate +/-25V.

As stated previously, it is best not to use these devices, but there may be simple workarounds to use them if desired.