FT232RL: Real Or Fake?

Above are two FTDI FT232RL chips, an extremely common chip used to add a USB serial port to projects, builds, and products. The one on the left is a genuine part, while the chip on the right was purchased from a shady supplier and won’t work with the current FTDI drivers. Can you tell the difference?

[Zeptobars], the folks behind those great die shots of various ICs took a look at both versions of the FT232 and the differences are staggering. Compared to the real chip, the fake chip has two types of SRAM etched in the silicon – evidence this chip was pieced together from different layouts.

The conclusion [Zeptobars] reached indicated the fake chip is really just a microcontroller made protocol compatable with the addition of a mask ROM. If you’re wondering if the FTDI chips in your part drawers are genuine, the real chips have laser engraved markings, while the clone markings are usually printed.

65 thoughts on “FT232RL: Real Or Fake?

  1. Would be interesting to do something similar with the apparently more common PL2303. My experience with them is that they’re hit or miss, with some acting very buggy and others working very well. Maybe the buggy ones are fakes?

    Also would be interesting to see if the fake FT232 works properly with Linux drivers.

    1. I just had a look then at the three PL2303 devices I own. One is “legitimate”, the other two are chinese products. One of the Chinese ones is a blob IC, the other two are packaged but look absolutely nothing like each other. There’s a big warning about fakes on the driver site, but none of them seem to be getting kicked off by their new drivers. Pretty sure the blob top wouldn’t be real though.

    2. There’s an extensive thread here alleging exactly that — the bad pl2303’s are fake:

      Except that the allegation is a bit more complex — apparently the fake ones work fine with the older drivers, and then Prolific is claimed to have altered their drivers to only work reliably with the genuine chips and crash a lot with the fakes.

      I don’t know if that’s the case, but I’ve lost *literally weeks* to problems that were eventually traced to a whole lab full of Prolific adapters. Replaced everything with name-brand FTDI and all is well.

    3. If they instead make generic CDC class serial chips at the same cut throat prices as the fake one, they could easily corner the serial chip market. FTDI’s and Prolific’s propriety drivers for something as well defined as serial chip doesn’t really have any performance advantage at least for the average consumers.

      1. Well, the problem is they’d still need a USB VID/PID, and that’s really the problem here, since the USB-IF has said “no” to generic-type VID/PIDs.

        That being said, a VID’s a grand total of $5K, so we’re not talking about serious money here. It would be a huge help to have a generic, CDC class USB-to-serial chip, without an FT232’s extra functionality, that’s at the same cost as the fake chips.

        I’d still end up using FT232s because the extra functionality is fantastic, but heck, a ton of people who use FT232s have no idea that they can be used for anything else: they just see them used in existing USB->Serial implementations, and duplicate it.

  2. Aside from the negative implications (e.g. incompatibility with the drivers, ethical dimensions, etc), this is quite fascinating. These guys found it profitable to reverse engineer FTDI’s USB protocol and re-implement it using a mask-programmable off-the-shelf micro-controller. This is quite a hack. I wonder what the FTDI driver developers had to do to determine if the silicon they were talking to was “genuine-FTDI” or not. Probably the reverse engineer job did the bare minimum. More work and it may not be possible for FTDI to determine what silicon their drivers are talking too.

    1. They don’t even need to reverse engineering the protocols. Those are mostly documented these days and can be lifted from the open source project and/or official sources. The drivers people might have some undocumented features for chip testing stuff they could use.

      As for reverse engineering other products in general, it is not like the companies that do the outsourcing not send them all the necessary design/CAD files, part lists, firmware to be burnt and even artwork designs for company logos etc to manufacture their stuff in China. All that cost saving without passing onto the consumers. “Free market economy” have a funny way to re-balance things when the other guys don’t take as much profits or have the same overheads (R&D, marketing, legal, and CEO packages).

  3. Wasn’t the original FTDI stuff all based on a 8051/52 controller anyway?
    If they actually made a good go at emulating the specifics of their implementation I’m sure it would be VERY hard to tell one from the other!!

  4. Nobody (except IBM) seemed to mind it when Compaq did something similar back in 1982.
    Most people reading this are indirectly benefiting from their hack.

    That’s not to say that I endorse what these bogus chip makers are doing, I’m just saying that it’s a far more common practice than any of us would like to admit.

    1. Sure, but Compaq didn’t label their computers with a big IBM badge on them. Reverse engineering or building work-a-like’s is one thing, but this is outright misrepresentation.

    2. Not even close. Compaq created a computer that was compatible with the IBM PC. Not that hard really since all the chips where off the self on the PC. The only thing they had to do was write a BIOS that was compatible with the IBM bios.
      What they did not do is put an IBM PC label on their machine.

  5. FTDI chips comes in at a premium price and I am surprised that it took so long for the “fake” ones to show up. I saw some PL2303 chips 10 for $1 with free shipping. I don’t even understand where the margins are.

    The PL2303 chips might even be the real chips that sat in a warehouse or their chip mask got “found in the trash” when the newer rev. came out.

    Don’t forget chips dies get packaged into plastic packages in China too and certainly those cheap laser engraving machines are also helping mark fake chips which strangle enough wasn’t involved in these serial chips.

    1. I’m having trouble with 5 units Arduino Duemilanove and a USB to serial converter. All purchased on Ebay.

      FTDI FT232RL

      date code
      1116-C GN051461 Works
      1211-C GN410601 Works

      1213-C CN480661 Does not work
      1403-C GN480661 Does not work
      1403-C GN480661 Does not work
      1403-C GN480661 Does not work
      1403-C GN480661 Does not work
      1403-C GN480661 Does not work

      It seems that there is no driver solution but that the chips must be replaced.

      1. K…. can you get the ‘serial numbers’ from the non-working ones.

        Here is an example

        Product ID: 0x6001
        Vendor ID: 0x0403 (Future Technology Devices International Limited)
        Version: 6.00
        Serial Number: A800GKEC
        Speed: Up to 12 Mb/sec
        Manufacturer: FTDI
        Current Available (mA): 500
        Current Required (mA): 90

      2. It sounds like your FTDI chips got bricked by the “bad” Windows driver, by reprogramming the chip’s PID to 0. There is a solution to un-brick, but it will require running Linux.

        I’m not familiar with the process personally, but I’d guess there’s some bash command/script that just dumps raw data into the USB without regard for device identification, that causes it to rewrite the PID to the appropriate number. Once this is done, the device should work with a non-bricking version of the Windows driver.

    1. Massive pile of yes here, looking to buy optical toslink receivers and transmitters, my options are £15 each from digikey or £0.50 each with £15 postage from alibaba for a part that’ll do the job, but won’t do all the fancy 7.1 channels of 384KHz stuff. Given I’m buying 10 or so it’s a no brainer, though working out why on earth they cost so much from official sources is definitely a brainer

    2. I won’t go directly to ebay or alibaba directly if DX or other reshipping companies are selling the parts I want as they are much better to deal with for shipping, quality issues or even refunds. The selection is very limited, but they are likely buying from the local sources that other made in China stuff are using.

      As for shipping charges *outside* of US, digikey have a high enough shipping charge which can be a large percentage when you are ordering only $20-$30 worth of parts.
      I have done electronics for a long time that I do have lots of parts especially the general stuff and passives, but just missing a few chips that I can’t get locally nor design without. (I don’t want to abuse the samplings from vendors.)

      If you are putting for weeks etc until you have a large enough list of parts to order, might as well wait for someone that have “free shipping” and not pay sale tax on purchases. I still have a list of stuff I yet to order from them for a couple of years, but DX & others filled my needs for some of the short term stuff I wanted.

      1. I regularly buy from a handful of chinese ebay vendors. At least one of them will combine shipping on multiple items, so I can get a pile of stuff shipped for $2-3. Compared to dx, the prices are typically lower and the shipping faster; I’ve never had a problem with quality or needed to get a refund- YMMV. I do buy from DX occasionally as well.

    3. I use Aliexpress extensively. Sure there is some wacky on there, but if you are realistic about your expectations you can get quantity deals and a LOT of stuff you’ll never find in a digikey catalog.
      The best reason for me though is that being in Tokyo, I rarely have to pay shipping.

    4. I go with Newark, Mouser, and such for the majority of passives, discretes, and bare ICs. All electrolytic caps get ordered here, no exceptions! Plus any LEDs for which I need high power, reliability, and/or a positively known manufacturer. And countless other items, as long as they’re reasonably competitive.

      But for some connectors, most electronic modules (power management, RF), and component assortments (ceramic caps, resistors) to quickly fill out a parts collection, ordering direct from the Orient is the way to go. Often you get the exact same parts domestic suppliers are reselling for hefty markups.

      I tried to stay loyal to domestic businesses, and shied away from the potential risk of ordering overseas, until I realized I’d been paying Sparkfun an 800% markup on connectors over what I could order direct from their supplier in small quantity. Sure they’re entitled to some profit, I would have found 100-200% reasonable, maybe even a bit more. But 800%? It reminds me of the brick-and-mortar stores that charge $30 for a short USB cable because they know you need that cable right away. If not downright unethical, it’s certainly offensive, and I won’t support such practices and businesses out of loyalty or charity.

      1. I also think that adafruit re-brands Chinese products & resells ’em for few times more bucks. But they test what works and what not before put it on the market.
        I don’t mind against cheap fake FTDI as long as the code of the diver does not brick it for a revenge. Under Linux I think the code won’t kill them.
        Beware of fake opamps. I’ve got fake AD823 for ridiculous cheap price. When tested, this thing managed to still have gain at 3 MHz but offsets were higher and the power consumption was different from the AD datasheet. Since I don’t have $14 for a genuine opamp, I don’t know how well the genuines stick to the datasheet.

    5. Just for fun, I check how much will it cost me to order one chip from Mouser (I live in Poland). The chip itself is 5 EUR (not so bad), but…

      40,00 EUR

      And free delivery for orders above 150 EUR. This is *g ridiculous. For a hobby this is absolutely insane, I am not even sure if I bought 150 worth of electronic parts IN WHOLE LIFE (yeah, I’m just starting).

      Sorry, I will stick to China based supply for now.

  6. I was I could get hold of stuff to do decapping at home. I have a few parts bought from Chinese suppliers that are just wrong. Like I have some parts with markings that are for a different pitch (the part number I ordered) than the chip. They all seem to work so I wonder if they just changed a few letters on the markings to make them match what I ordered. It bugs the hell out of me not knowing for sure.

  7. Those die images are really something. My father used to work at an ATMEL factory in France and I get to tour the facility and see the wafers. I find them to be absolutely beautiful and intriguing at the same time. For many people, heck, even for me, at this level it almost looks like alien technology ;)

  8. Hah! Very interesting to read this. I had EXACTLY the same problem last year with my temperature controller (www.eyejayinstruments.com). It uses a FT232RL, and I found that a batch of controller from the board assembler wouldn’t work with an updated driver version. Sent a chip to FTDI, and they confirmed it was fake. Fortunately only a few of the controllers were actually sold, and I was able to swap the chips quickly.

    The assemblers, although I told them to use RS or Farnell parts, had seemingly got the chip from “their own sources” – probably meaning Joe’s Cheap Chips around the corner.

  9. Digikey a joke. They used to be able to charge me in USD at digikey us price. Just last night, they kept forcing the local currency pricing on my order and their exchange rate is too steep. So I wrote them an email about the problem, gave my order to mouser. And digikey came back and asked what part — still no clue what is not going on.

  10. For similar reasons, I chose Mouser over Digikey now. As for Alibaba… I only see it as super “shady”, I’m not sure I would use it as a “last resort” even. I trust what I buy at Mouser to be legit and good quality,

  11. I stick to digikey, farnell, newark, and mouser. Ebay, DX, alibaba etc for assemblies, but never for raw components. And always with the expectation that the assemblies won’t work. Like a 3.3V-5V (usb) boost converter. Sure it works under light loads (although it outputs closer to 6V)… but with my cell phone attempting to charge the inductor saturates and the whole thing draws a ridiculous amount of current. Oh well.

  12. I’m not surprised there are fakes out there. I had gotten one of the PL usb-serial adapters and it would only work with windows, even then not quite. fortunately the parts that made it up were worth more to me than I paid, so it wasn’t a loss. I have a FTDI FT232RL adapter I got from ebay for a few bucks shipped. It works fine, but I figured even if it didn’t work right (because of a counterfeit chip), I could still buy a legit chip and replace it on the board (I have a hot air station) and still have spent less than buying a board from other suppliers.

    Looking at the chip though, if the counterfeits really do only have printed labels and not laser etched, than mine is legit (it’s laser etched).

    I look at anything coming from ebay as at the least, a cheap source of premade boards. I’ll generally only buy it if I know that is where everyone else sources that same board/component, since the odds of it not working are pretty darn low in those cases. Otherwise I’ll source the module from where it is actually manufactured. Stuff like passive components, such as jumpers and headers and whatnot I’ll get from china as well.

    I don’t have the money to be spending on strictly US stuff (that generally is made in China anyway) for my hobbies. If we were talking about something I was making to sell, or if I had a lot more disposable income the story will likely be different. But that is my opinion. Not everything has to be or is better if it’s made in the USA. We make plenty of garbage here.

  13. I used to work for a company called America2. They bought bulk components, sorted, vetted, and resold them on the secondary market. My job consisted of deciphering part numbers, date codes, lot #s, etc, as well as making sure our products were not counterfeit or tinned. We used to take a sample part from a lot we had bought and scrape the top of it to see if there was anything underneath. We would look for things like missing date codes/lot numbers or anything that didn’t follow the normal cryptic patterns we were used to seeing from a particular manufacturer. We had stacks and stacks of books with all this information.

      1. A fake chip could subtly weaken an encryption key, thereby making it easier for a third party to decrypt comms.
        Eg, if RSA is used for key exchange, and a dynamically generated aes-256 key is used for encrypting the payload, then weakening the aes-256 key generation would plausibly go unnoticed.

    1. From the article:
      “If you’re wondering if the FTDI chips in your part drawers are genuine, the real chips have laser engraved markings, while the clone markings are usually printed.”

      A further clue may be where you got it and what you paid for it. As FTDI chips are often sought after they tend to be relatively pricey (hence the market for fakes). Digikey currently lists the FT232RL at $4.88 in single quantities. If you bought yours for significantly less than that or bought it on a large online marketplace (ex. ebay) it may well be fake.

      You can’t tell by the numbers on the chip, the knockoff manufacturers can print whatever they want on them and you can bet they’ll print legit codes on.

        1. If your wondering if the FT232 on your Arduino clone is genuine, then there’s really no other way to tell than asking robosoft systems where they get their parts or decapping the chip yourself (very dangerous and would probably ruin it). Fortunately if the board is working with your computer already, then either the chip on it is genuine or it doesn’t really matter to you. The fake chips are mostly a problem for consumers because the new FTDI drivers don’t work with them.

          Ultimately when buying any piece of hardware you have to trust that the manufacturer did their homework and used good parts. Clones of popular products may be a lot cheaper than the ‘real thing’, but they can be lower quality to make up for it. ‘You get what you pay for’ is an often true adage, though considering the genuine Arduino is so overpriced it may be less applicable unless the clone is really inexpensive.

  14. Might be why my PIC programmer mysteriously went dodgy, failing to program chips even though voltage etc was fine.
    I thought it was a batch of bad chips but checked the FTDI and it looks like a bad chip as the markings are obviously off centre while the (genuine) A variant on a TV decoder board has laser etched markings.

  15. Most of the Windows gurus strongly recommend NEVER letting Windows Update install drivers for hardware! If you just keep using the FTDI drivers you have (working) you should have no “bricking” issues. Right?

  16. All you have to do is change 6001 to 0000 in the .INF install file and the driver will install to the chip having a PID of 0000.

    Remember to do this for both ftdibus.inf and ftdiport.inf.

    This is what the lines in the .INF file look like that need modified”

    Change 6001 to 0000. 0000 is what was written to the chip to “-BRICK-‘ it. So if that’s what they want to write to it all we need is a driver for a PID of 0000.

    The modified driver looks like this:


    Now they can send all the 000’s they want and it won’t do anything as you now have a chip and a driver for a PID of 0000.

  17. Works on x32 W7, not 64.
    Chip is recognised but won’t work with even the x64 “old” driver, but
    it still works fine on my x32 W7 netbook once VID/PID is changed.

    I can only conclude that the x64 version is hardcoded to not accept 0000 as a valid PID so this also explains why my two serial adaptors also don’t work.

    1. I just bought Six “FTDI” leads from a chinese outlet. They all have the markings of the right hand of the three chips. Two of them register on Windows as FTDI FT-232 chips, and the other four come up as Prolific PL2303 Rev. 4 (despite the chip saying FTDI) That chip on the right must be fake, ad the PL2303s can be faked with the same chip.

  18. Isn’t that interesting? You have identified the two chips with the laser engraving as fake and the one with printed markings as real… Exactly the opposite of what the OP says.

    Does anybody know of a way to determine a fake chip through software?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.