It's worth mentioning that FTDI's response to this was to update their windows drivers [1] - which caused the fake devices to be bricked - and that affected a lot of companies who unknowingly had used the fake chips on their PCB designs. Eventually they reversed course but a lot of HW was essentially sent to the landfill.
Not quite, at least not necessarily. One feature of FTDI chips is an EEPROM that can be used to customize the enabled features, enumeration strings, and other things. That programming could probably be done in a separate jig, or in-circuit. So yes, 5 minutes of hardware work, but more potential rework at a higher level.
Not to mention the possibility of sealed cases/enclosures or potted PCBs.
Speaking as someone with a hot air rework station and no experience soldering SOICs, I can make a whole bunch of melted crap in 5 minutes all while burning my fingers.
5-minutes sounds insane to me, but I know it's accurate (the key being "experience").
Any tips for learning how to do this sort of thing with proficiency? How long does it take to become that proficient? What kind of coating/weirdness would present a problem? I'm asking for a bit OT reasons -- it's always been something I've wanted to learn/teach my kids.
For disassembly practice, I used to collect broken electronics, and desolder various components for fun. You learn the different heat requirements of eg 2 layer boards vs motherboards with many layers.
For assembly practice, you can buy "SOIC breakout boards", the cheapest chips you can find, and go to town. Either with a solder stencil and some paste, or a syringe of paste and a blunt needle (this gets old fast!)
Coatings tend to make lots of nasty fumes when heated. Some can also prevent solder from wetting the pads or legs. Depending on the coating, overheating/burning may change its electrical properties.
I have some experience soldering SOICs, but the difference between me and a pro is insane. I've seen someone remove and replace a half-dozen 0805 (2x1.2mm) SMD components with a fairly standard fine-tipped Metcal iron (not even a rework station) in under a minute.
Rework is extremely expensive, especially if you imported the devices and don't have manufacturing facilities of your own, plus the potential cost of retrieving units in the field. The soldering is the easy part. Labor and logistics are the hard part.
How to tell me that you understand zero shit about electronics and/or manufacturing without telling me that you understand zero shit about electronics and/or manufacturing
Some context about the what happened years ago with fake (but compatible and working) FT232 chips that were intentionally bricked by FTDI updated Windows driver, that is, hitting (often unaware) users instead of counterfeiters.
And amusingly, the bricking worked because again the clones worked better. The clones had implemented FTDI's EEPROM interface as designed, 16 bits at a time, while FTDI apparently had used a 32bit array for the internal FT232 EEPROM, and so would buffer 16 bits and only write all 32 bits when you issue writes to odd addresses. FTDI's bricking code uses writes to even addresses only (including computing a preimage attack on their checksum algorithm, as the checksum is at an odd address so they can't change it), which is why it doesn't affect the real ones.
In this particular case, the commonly-available FT232 clones supposedly worked better than the original - the bit banging mode actually worked reliably on the knock-offs, whereas not so much on the actual FTDI chips. (The difference that FTDI used to brick the clones was that writing to the EEPROM one byte at a time, which wasn't allowed according to the datasheet, worked on the clones but not their own chips.)
Unfortunately, the clones also weren't clock stable at all - they were fine at 115200 and such, but if you pushed them up to 3Mbaud or so, they mostly read/wrote garbage. I put a few of them on a good scope and the timing was just all over the place. The legit FTDIs are rock solid on their timing even up at high baud rates.
So I got some legit chips and just did the surface mount rework to replace the fakes with the legit ones and solved my problem. I had no idea when I got the adapters that they had fake chips on them. :/
Bit bang mode is the opposite. The clones work fine. The legit ones don't work at all (completely random timing) except at 3Mbaud (the official workaround), but at that speed you can't keep the buffer fed via USB so it's useless.
This is incredibly interesting. From the articles I've been able to dig up, it sounds as though they've outright admitted to breaking the counterfeit products and it was not an accident[0].
Rather than simply state "We've detected a counterfeit product. If our driver works at all with this product, it will probably be unreliable. It may also break it, entirely. We recommend buying a replacement from (xyz) and requesting a refund from the retailer your purchased it from" Maybe allow them to continue "at their peril", but intentionally break it? That's just stupid.
Consider that this is a device that is probably being used by a company, a developer or otherwise power-user who will likely purchase more than one, did intentionally breaking the one they had make them want to purchase a genuine one or any future product from you? Or did you just cause a poor sysadmin to make it their mission in life to cause your future bankruptcy? It'd make me investigate any alternative to an FTDI product for any future purchase, and they did that, not the counterfeit-maker.
This is such an awesome "punch the customer" response; it's like the RIAA/MPAA are moonlighting in management over there. I loved the FTDI official response, somewhere highlighted is how awful counterfeits are for their business. It's really difficult to feel sympathy for a company who so recklessly disregarded my business by destroying something I owned that they didn't make -- trademark/counterfeit law aside, the only person who got victimized twice was the person who thought they purchased an FTDI product.
Yeah, I used FTDI back when they first came out and there was nothing else like it on the market (that I know of, anyway). Would never use them again after that little stunt.
I used a fake arduino for my capstone project and just before we had to present, I needed to update the code. I didn't have my laptop, so I used my partner's which did not have the fake(?) driver. Just as I plugged it in I realized my mistake and bricked the arduino. We weren't able to demonstrate our project.
I can totally understand the driver not working, but to brick other people's hardware with an automatic driver update will forever make my blood boil.
If they had a case, it would have been against the Chinese manufacturers of the copycat chips, or against product manufacturers who (often inadvertently) used the mislabeled, fraudulently sold chips in designs that called for real FTDI chips.
If suing a Chinese company is hard, they could have also worked with law enforcement to conduct the same sort of "find and destroy inventory" sting operations that the police sometimes do for counterfeit handbags. Instead they went rogue and did the equivalent of personally barging into everybody's home and shooting up their appliances.
What if they had bricked life support equipment, or part of an airliner? It was so unbelievably reckless -- I don't understand how any of their lawyers could have possibly signed off on such a stunt. They should have been sued into oblivion for it.
> What if they had bricked life support equipment, or part of an airliner
While I agree that was a bad strategy, even more so from a PR standpoint, I'd like to believe critical industries like medical electronics or avionics would do a better job not only validating components (even at the individual level) but also subsequent software updates. Electronics are less a wear part than mechanical parts where counterfeits have an easier time sneaking in (the occasional counterfeit screw let's say).
Any scenario involving any kind of failure looks more dire when you attach it to "life support", "airplane", or "nuclear facility".
That's a bad failure for sure but it's a very different kind. Checking for counterfeit parts is a supply chain issue, you know what you're looking for (the original part is very well described) and you check for specific markers.
Design flaws are different. You can fail using an original part because you picked the wrong one for the job. Validating something for (long term) use is far trickier than checking an already validated part for a set of markers. The higher the stakes, the better the checks should be. An airplane should get better validation than a portable blood pressure monitor.
Imagine something more relatable: picking the perfect person for the job is far harder than verifying that the person you chose is the one coming into the office each day.
The FTDI debacle may be old news, but lots of companies are probably faced with similar risks now, having been forced to turn to unfamiliar Asian brokers and other unofficial sources for parts that used to be commonly available but now have lead times of a year or more.
That includes me. :( Do I recommend that the company order thousands of STM32 MCUs from a place in Shenzhen that they've never heard of, and that our EMS facility won't have anything to do with? Assuming the parts even show up, will they be counterfeits? Recycled? Handled and stored inappropriately?
Is it OK to ship hardware built with dubiously-sourced parts to customers? Can we rationalize the quality risk away? Or is it more ethical to sit around and starve because we can't ship products for a year...?
Being in the exact same boat as you, wondering whether to order STM32 MCUs from who knows who in China, we punted the questions you ask directly to our customers. Do they want to risk the quality issues or wait a year for their product?
TIL FTDI converters use a 600-800 nm node... that's almost a micron!
The fact that they sell their chips for five fucking dollars a pop has always been kind of obscene, but on such an old node, they really shouldn't cost more than a dollar, tops. No wonder they're so heavily cloned.
I wasn't complaining about the process. Use 10 micron for all I care. This particular process was perfected in the late '80s, and the costs reflect that. From TFA:
> Die manufacturing cost is roughly the same for both dies (~10-15 cents) .
Again, this chip should never cost more than a dollar, even in single unit quantities.
> Do you want to say we should abolish free market?
Don't build a straw man.
FTDI has every right to charge whatever they want, just as I have every right to call them greedy bastards who would rather play dirty games with their drivers than acknowledge that their products haven't been priced competitively for years.
If you really want to play this game... One could argue that in a truly free market, the copycat chips would be legal.
But it's not even about quality vs. price. FTDI's legitimate competitors often have better designs while also selling them for less. FTDI's only advantage is that they were early, and people are often reluctant to change an existing design that already works.
process is relevant to pricing which is the point of GP's complaint. FTDI is not pricing their products competitively. That's why clones are eating their lunch.
> FTDI is not pricing their products competitively.
There are alternatives to FTDI that are not clones to FTDI. For whatever reason people prefer to use FTDI in their design.
FTDI does quite well on their chips which means it is stupid to say their products are not priced competitively -- they have competition and are still winning at it.
> FTDI does quite well on their chips [...] and are still winning at it.
You are making a guess.
FTDI is a private company, and unless you are an insider or have done original research, you have no way of knowing whether the company is doing well financially.
The price-per-reel on Digi-Key and how many chips you see in teardowns of legacy products and on cost-is-no-object FPGA boards isn't necessarily going to paint an accurate picture of this for you.
In my experience, when chips are expensive (for non-supply reasons), it's usually for one of three reasons:
1. There's nothing else like it on the market, so the manufacturer can charge whatever price they want, and designers will still buy those chips (that last part is critical). (e.g. certain specialized chips from Analog Devices)
2. The manufacturer's costs are out of control. This is rare today, but was common in the '80s and '90s. Most of those companies went bust or were acquired, and now the semiconductor design industry is extremely focused on cost efficiency.
3. The manufacturer can't sell enough chips at a competitive price to pay back the NRE costs of the design + a reasonable margin, so they're forced to sell at a higher price. If they don't have other products, and are not able to diversify or sell the company, this is a death spiral.
For FTDI, (1) is clearly not the case -- they have lots of competitors, and the task that their chips perform (shoveling bytes around) is a basic one. FTDI's primary office is in Glasgow, Scotland, and they have satellite offices in China, Taiwan, Singapore, and the US. Their offices are tiny. (2) seems unlikely.
That only leaves option (3). The semiconductor business is famously reliant on volume, and this is especially true for low-cost, small-die-area designs. You need to sell a metric fuckton of chips if you're selling them for a few dollars each. And the only way to get that kind of volume is to secure design wins in high-volume consumer electronic products that sell by the metric fuckton.
Most of FTDI's volume appears to be in legacy products. Products that used to have a serial port, but that now need a USB port because computers don't have serial ports anymore. They make a few higher performance parts that seem to be designed specifically for FPGA vendors, because FPGAs often need a way to get as many Gbps as possible over a USB port, and FPGA vendors don't particularly care about an extra $20 chip, because their FPGA already costs $1000 and they don't want to muck-up their low-voltage process or waste pins and die area to support USB. Also the FPGA vendors picked FTDI ages ago, and don't want to rewrite their shitty, shitty software.
But you'll never see that FPGA USB-serdes chip designed into a laptop or a phone or a camera, because there are cheaper options available, and because the functionality can be had essentially for free when integrated onto the high performance SoC that requires it.
FTDI probably has enough volume on their legacy chips to cover their costs, but that's because the design is ancient and was paid off years ago, and because there are a lot of legacy industrial products that aren't going away anytime soon. The fact that they're still overpriced relative to competitors means they're not a product of choice for new designs, and are a particularly obvious candidate for cost-reductions. I don't know anyone who willingly designs in an FT2232 when a CP2104 will do the trick, especially after FTDI flushed their reputation away with that driver fiasco. So this part of their business is likely to shrink.
The FPGA vendors sell maybe a few thousand units of their $1000 - $10,000 FPGA boards. That's probably not enough volume to cover the costs of FTDI's higher performance chips, even at $20 a pop.
The sad thing is that (1) used to be true, back when USB was new, and there were a lot of buggy implementations. FTDI got it right, and they even had a reputation for a driver that "just worked." The answer to the question "How do I do USB?" used to be "Just shut up and buy an FTDI chip." An enviable position at the time, but as USB became more ubiquitous and better understood, their competitors caught up. Other vendors have good implementations and good drivers now. Good USB IP for chip designs is also readily available, so many microcontroller designs can do without FTDI now.
The FTDI driver fiasco was the worst thing they could have possibly done with the one thing they still had going for them -- they trashed the reputation of the quality of their driver, and now everyone looks at them with suspicion. Normally you would expect a semiconductor company with a limited, overpriced product line to be an acquisition target, but the fact that this hasn't happened despite the fact that they're a relatively small target suggests that the bigger players either looked at their books or did the math and realized that their market is shrinking and they have no growth prospects.
tl;dr
FTDI may not be doing as well as you think they are. Rather than doing well because their chips "command a high price" -- every indication suggests that they are resting on their laurels, they know their long-term prospects are grim, and are trying to squeeze out as much as they can before they go bust or are forced to sell themselves for scrap.
It's not about the chips. They aren't in the business of selling chips, for better or worse. They are in the business of selling access to Windows drivers that work.
First, you need to talk about whole cost of production and quality of the product. When you reduce your node your development and production costs grow exponentially. Size of the wafer is only small part of the production cost.
Second, the chip area can't be arbitrarily small. You still need area to connect it with outside, area for larger components, etc. When you reduce your transistor size at some point the chip area will stop getting reduced because of packaging requirements.
> First, you need to talk about whole cost of production and quality of the product. When you reduce your node your development and production costs grow exponentially. Size of the wafer is only small part of the production cost.
What exponent?
The number of mask steps is similar or a few more from 600nm to 180nm. Isn't that the best way to estimate cost per wafer, in combination with the cost to make the masks?
> Second, the chip area can't be arbitrarily small. You still need area to connect it with outside, area for larger components, etc. When you reduce your transistor size at some point the chip area will stop getting reduced because of packaging requirements.
That's why I said 5x instead of 20x. I looked at the size of the pads and the circuitry around them, assumed they wouldn't shrink at all, then combined that with a somewhat pessimistic estimate of how much the logic could shrink.
As it turns out, apparently it's easier to develop an exact protocol-level clone of an existing USB to serial chip which already has working drivers than it is to write reliable, high-quality serial drivers for a new chip. This may say something about the state of software these days.
It's presumably for the same reason that people make knock-offs of chips like the MPU-401 UART and the ELM327. There's a large installed base of applications for them.
The drivers themselves can't possibly be that complex. But getting a USB-IF vendor ID as well as drivers in Windows is probably a bit harder.
Not only does Linux have its own drivers, they can unbrick the damaged devices.
> Not only does Linux have its own drivers, they can unbrick the damaged devices.
Thanks for that. I haven't used my counterfeit FTDI cable in probably 2 years, but I was trying to figure out how I had not encountered any issues with it.
I only happened to find out it was counterfeit due to the fact that I purchased it via a link in a forum recommended by "some user", and later, when I had this stupid looking cable and no idea where to plug the ends into my Pi, I revisited the forum, scrolled down a few replies, and found out that all of the "FTDI" products from the manufacturer of my cable were fakes.
Once I read that, I had intended on replacing it, but decided to wire it up anyway to see if it would work. I run OpenSUSE Tumbleweed on my main computer, and I don't remember having to do anything unusual, or encountering any warnings/errors at the time trying to use it[0], I'm guessing that's why...
[0] I was trying to connect into a headless RPi and that was the most convenient way sometimes.
These days, Windows 10 finally supports the standard CDC ACM protocol for USB over serial as do most other operating systems, but as I recall even Microsoft seemed to struggle getting the drivers for that to work reliably and not trip over various gotchas on the Windows side of things. It just genuinely seems to be difficult to develop good drivers for these devices.
Not just that - if customers are already asking for an existing part, it's easier to sell them something that pretends to be that part than it is to convince them to change to something else. They'll notice if you suddenly start shipping chips that need a different driver, if you're pin compatible and work with the existing builds there's a much better chance nobody will notice and you can pocket the difference in price.
Not necessarily. I'll take "doesn't work at all" over "works great but counterfeiter didn't copy any safety controls so at some point it's gonna burst into flames".
The fake FT232RL device in this article is actually a lot better than some of the devices on the market now.
I bought an FT232RL breakout on eBay a year or so back, planning to use it in bit bang mode, and it turned out that while it worked for talking to UARTs, the bit bang mode was completely broken - and the EEPROM (needed to configure it the way I wanted it) was not even writable at all (probably not real EEPROM).
In my case, it was a counterfeit - the product was explicitly advertised as FTDI FT232 but wasn't. I was fortunately able to get a refund for it through eBay. Sadly it is very hard to know what you are actually getting due to the prevalence of identically branded clones, and beyond refunding me eBay doesn't seem to have done much despite me pointing out that these devices are being advertised as FTDI FT232 but are actually significantly below the advertised specs.
As discussed elsewhere here, FTDI pushed an update that deliberately tries to brick FT232 clones by writing bad data to their EEPROM - which obviously non-configurable devices would be immune from and might be one reason (alongside cost) which the people creating clones have cut out the real EEPROM now.
I agree with those here the bricking was a very bad idea - not all clone chips were necessarily even counterfeits (it bricks based on PID / VID, and can't see what is actually written on the face of the chip; reusing a USB PID/VID registered by someone else to enable compatibility without using another company's trademark isn't counterfeiting). Given they deliberately destroyed other people's property because it contains a chip that might not even be counterfeit deeply embedded in it by the manufacturer, I think they are lucky not to have faced any lawsuits or prosecution.
Is this FTDI stuff still of interest to hardware people, now that every jellybean microcontroller has USB and UARTs built in? I see that an FTDI usb-to-serial cable is still around $20 but a Raspberry Pi Pico is $4 and and this board ( https://www.seeedstudio.com/CH551G-Development-board-p-4764.... ) (also available from Digikey if you don't want to order from China) is $1.49. Can I do the same things with those, easily enough?
FTDI-style chips are still extremely relevant to hardware development, because they're plug-and-play. They just go. No buggy software to work around or kLOC to write yourself. The time and effort saved there is real and valuable.
That's for one-offs and debugging connections, though. If you're shipping a product, then suddenly the economics change and the software-heavy solutions start to look good again.
Thanks, what I'm wondering is whether you can program one of those boards as a USB converter one time and be done with it. It really shouldn't take KLOC, I hope, and it even seems likely that such code is already out there. The idea then is that you can use a board with that code instead of an FTDI cable from then on.
Unfortunately USB stacks are pretty heavyweight. On a Cortex-M3 class part they end up around 6kB compiled. You have three basic options for getting them going:
1. Trust your vendor's libraries/ROM drivers/example code. This is the fastest to bring up, the least flexible, the buggiest (it's vendor code!) and the hardest to debug. (Trust me on the debugging part.)
2. Use a library like tinyUSB. This can actually go pretty quickly, since you've got full source control. If your part is popular enough that someone else has already done most of the work, this is by far the best choice. Even if you have to get personal with the hardware, I still like this option. (tinyUSB is much better than what came before... I wish it'd been around last time I had to do this.)
3. Write your own. Don't do this. (You might have to do parts of this if you tried #1 and it isn't panning out.)
All of these are doable. Any of them are enough work that if you're only making one, just put the FTDI down and call it done.
FT232RL is unbrickable, microcontrollers can be and its USB might not work without the firmware blob in the Flash ROM, that’s why engineers put a CH340G on — I mean an FT232RL on a board.
Lots of ARM chips come with USB DFU mode firmware baked in ROM, those don’t need one.
Probably an unpopular opinion, but if you have a bricked device with counterfeit components you should go to the manufacturer of the device and demand they make you whole. Also call into question their quality control mechanisms and supply chain security. They bought defective components.
It is insane to blame the chip marker. They have a reputation to uphold. How can they be expected to continue operating if they not only have to compete with shoddy knockoffs, but basically support them with QA as well?
If China can't afford to play with closed source, maybe they should invest in open source.
The genuine chips can also be bricked by a very similar command sequence. Are they also defective?
> It is insane to blame the chip marker. They have a reputation to uphold. How can they be expected to continue operating if they not only have to compete with shoddy knockoffs, but basically support them with QA as well?
So refuse to operate with the fake chips. Bricking is unnecessary and deserving of blame.
And as the article points out, a chip that's merely compatible and not counterfeit might also get bricked.
You use a driver of a manufacturer on a device not manufactured by this manufacturer.
This driver is licensed to you only to be used with such devices.
"The licence only allows use of the Software with, and the Software will only work with Genuine FTDI Components (as defined in the Licence Terms). Use of the Software as a driver for a component that is not a Genuine FTDI Component MAY IRRETRIEVABLY DAMAGE THAT COMPONENT.
It is your responsibility to make sure that all chips you use the Software as a driver for are Genuine FTDI Components. If in doubt then contact FTDI.
The Software is provided “as is”.
There are no warranties (or similar) in relation to the quality of the Software."
Linux kernel? Really? The whole point of that is to reuse code as much as possible.
Even for Windows, the point of getting things into Windows Update is to make hardware just work, not to advantage one company over another.
But sure, they can make a reasonable argument toward making the driver incompatible. (If they're also willing to hurt people that tried to buy genuine parts.) It still doesn't justify the bricking.
Assuming the bricking of FTDI devices as mentioned in the update isn't "We sent out code to intentionally destroy counterfeits" (i.e. it served no other purpose, wasn't an accident[0])
Did they QA it with counterfeit devices to ensure they were bricked? Or did they just write a new driver for their devices that happened to result in (some/all) counterfeits being bricked?
If it's the latter, then it's not FTDI's fault. It's unreasonable to expect FTDI to QA their drivers against products that aren't theirs.
It is, however, their problem. The last activity that user did was install their driver, and the result is broken stuff. They've just accidentally pissed off a bunch of non-but-possible-future-customers. Taking "fairness" away: a percentage of those affected will assume it was intentional, even if it is very obvious that it is not[0]. And even of those who don't blame FTDI, a percentage of those will feel FTDI owes them something, anyway simply because "big bad corporation"[1]. It's probably in their best interest[2] to detect non-genuine products, inform the user of the potential risks and either prevent the driver from loading or allow the user to accept the risks at their peril if there's a chance the driver can brick the device.
[0] Say, the counterfeit has half the storage/ram of the geniune, and reports it has the same capacity, etc, and a driver update passes the "reality barrier"
[1] Forgetting how often "big bad corporation" is really, really, tiny -- and I have no idea whether or not FTDI is big; it's not relevant.
[2] Albeit, it may not be from a legal perspective...
Apple would have no problem bricking a fake iPhone that ran a cracked ios. They regularly try to lock in hardware components with software and firmware.
And in the case of a counterfeit low level component then they should.
If China wants to compete, then compete. If they are allowed to do this, legitimate companies will suffer and legitimate progress will stagnate. Just release the same chip open source with instructions for using it in this use case. Then when the legitimate company bricks the device, they look like the bad guy stifling competition.
It's a little unpopular with me, yes, but it also depends:
(1) Was the "bricking" intentional[0], as in, did FDTI's driver look for the counterfeit, and upon finding it, issue instructions that would make the device useless, even when used with drivers/configurations that previously worked. It also matters, a little, whether or not this is easily reversed.
(2) Did the FDTI take similarly extreme measures to prevent/stop unauthorized chips that weren't impacting customers/non-customers.
The main thing that would leave me with little sympathy for FDTI[0] is if it was intentional, i.e. the update had no other purpose except to break these devices. Past counterfeit situations come to mind where an update "just happened to brick" counterfeits, but it was due to the counterfeit having some limitation[1] not present in the genuine product.
But that second point is beginning to matter a whole lot more, these days. If FTDI makes it extremely difficult for me to determine what is a real FTDI chip and what is not a real FTDI chip -- or, put another way, if the counterfeits are that difficult to tell from the real thing -- and FTDI decides to breaks the one I thought was real...? I'm going to be swearing. I'm going to be at least as pissed off at the FTDI folks as I am at the outfit responsible for the counterfeit, but FTDI is forever associated in my head with "I can't do what I need to do, now". It's terribly unfair that FTDI has to go through trouble to keep a$$holes from counterfeiting their products, but it is the world we live in.
There are also companies that specialize in this sort of thing for brands/products of all types[2].
At the end of the day, the worst FTDI should do is throw up a large notice "This is not an FTDI device. If it indicates that it is an FTDI device, it is a counterfeit. This driver most likely will not work with your device because it isn't ours." ... here's a link to buy a real one from places that we know won't sell you a counterfeit, etc.
That puts blame where it belongs and even if the driver worked would probably result in me purchasing a replacement from the link, knowing that I'd never have problems in the future.
[0] And hope that some legal issues would come their way
[1] Maybe less RAM/storage than it's supposed to have, and maybe it lies about it, too.
[2] I'm doing work for one, right now... it's an interesting space; I used to think it was as pointless as trying to stop media piracy. There's, surprisingly, a lot of effective (and creative) measures to greatly disrupt/stop these larger scale rip-offs, and a lot that can be done which legitimately protects the customer -- but it's a thousand-pronged attack and you'll never completely eliminate counterfeits, especially for designer brands and such.
> What's the economic reason of making software fake of well-known chip instead of making new one under your own name? This way they don't need to buy USB VID, sign drivers in Microsoft, no expenses on advertisement.
I thought this was a great example of second order effects at work.
This would have been nice in the title of the post here.
I was getting slightly worried about some of my tools until I saw the "Update 25th of October 2014:" at the bottom (Cause I missed the date header at the top).
Queue the entitled prats lining up to punch FTDI because they feel entitled to FTDI drivers and software be free to them and not damage their Chinese knockoff chips.
But their time and labor? Nobody can understand how valuable it is!
Ah yes, the "entitled prats" such as the hundreds of organisations that had no idea their products were even using fake chips or the student in the comments who wasn't able to demo their project because they had to use a cheaper Arduino board that had a fake chip on it.
Just like the chinese nuclear plant that is reaching failure stage, the strategy one is bog enough and large enough can get away of everything. No one can stop it. Instead all should help it to continue to run unsafely or illegally.
Just like the virus, we cannot investigate, cannot stop it, cannot even name (but can say for others abd even now x.x mostly from Uk etc.).
Humanity not just your serial driver is actually at stake. Have to stop or start somewhere.
Wrong is wrong. Say it, do it abd actively buy only real and legal one.
[1] https://arstechnica.com/information-technology/2014/10/ftdis...
https://github.com/torvalds/linux/blob/master/drivers/usb/se...
I mean, you could replace fake with the original quite easily as a drop-in replacement? 5-minute soldering job.
Not to mention the possibility of sealed cases/enclosures or potted PCBs.
5-minutes sounds insane to me, but I know it's accurate (the key being "experience").
Any tips for learning how to do this sort of thing with proficiency? How long does it take to become that proficient? What kind of coating/weirdness would present a problem? I'm asking for a bit OT reasons -- it's always been something I've wanted to learn/teach my kids.
For assembly practice, you can buy "SOIC breakout boards", the cheapest chips you can find, and go to town. Either with a solder stencil and some paste, or a syringe of paste and a blunt needle (this gets old fast!)
Coatings tend to make lots of nasty fumes when heated. Some can also prevent solder from wetting the pads or legs. Depending on the coating, overheating/burning may change its electrical properties.
[edit] I misremembered the component size
https://www.microchip.com/forums/m828027.aspx
Luckily Open Source drivers were unaffected. One more point in favor of Open Source, and a good argument to stay away from products by FTDI.
So I got some legit chips and just did the surface mount rework to replace the fakes with the legit ones and solved my problem. I had no idea when I got the adapters that they had fake chips on them. :/
https://twitter.com/marcan42/status/695292366639378433?s=19
Rather than simply state "We've detected a counterfeit product. If our driver works at all with this product, it will probably be unreliable. It may also break it, entirely. We recommend buying a replacement from (xyz) and requesting a refund from the retailer your purchased it from" Maybe allow them to continue "at their peril", but intentionally break it? That's just stupid.
Consider that this is a device that is probably being used by a company, a developer or otherwise power-user who will likely purchase more than one, did intentionally breaking the one they had make them want to purchase a genuine one or any future product from you? Or did you just cause a poor sysadmin to make it their mission in life to cause your future bankruptcy? It'd make me investigate any alternative to an FTDI product for any future purchase, and they did that, not the counterfeit-maker.
This is such an awesome "punch the customer" response; it's like the RIAA/MPAA are moonlighting in management over there. I loved the FTDI official response, somewhere highlighted is how awful counterfeits are for their business. It's really difficult to feel sympathy for a company who so recklessly disregarded my business by destroying something I owned that they didn't make -- trademark/counterfeit law aside, the only person who got victimized twice was the person who thought they purchased an FTDI product.
[0] https://arstechnica.com/information-technology/2014/10/ftdis...
I can totally understand the driver not working, but to brick other people's hardware with an automatic driver update will forever make my blood boil.
If suing a Chinese company is hard, they could have also worked with law enforcement to conduct the same sort of "find and destroy inventory" sting operations that the police sometimes do for counterfeit handbags. Instead they went rogue and did the equivalent of personally barging into everybody's home and shooting up their appliances.
What if they had bricked life support equipment, or part of an airliner? It was so unbelievably reckless -- I don't understand how any of their lawyers could have possibly signed off on such a stunt. They should have been sued into oblivion for it.
While I agree that was a bad strategy, even more so from a PR standpoint, I'd like to believe critical industries like medical electronics or avionics would do a better job not only validating components (even at the individual level) but also subsequent software updates. Electronics are less a wear part than mechanical parts where counterfeits have an easier time sneaking in (the occasional counterfeit screw let's say).
Any scenario involving any kind of failure looks more dire when you attach it to "life support", "airplane", or "nuclear facility".
Design flaws are different. You can fail using an original part because you picked the wrong one for the job. Validating something for (long term) use is far trickier than checking an already validated part for a set of markers. The higher the stakes, the better the checks should be. An airplane should get better validation than a portable blood pressure monitor.
Imagine something more relatable: picking the perfect person for the job is far harder than verifying that the person you chose is the one coming into the office each day.
That includes me. :( Do I recommend that the company order thousands of STM32 MCUs from a place in Shenzhen that they've never heard of, and that our EMS facility won't have anything to do with? Assuming the parts even show up, will they be counterfeits? Recycled? Handled and stored inappropriately?
Is it OK to ship hardware built with dubiously-sourced parts to customers? Can we rationalize the quality risk away? Or is it more ethical to sit around and starve because we can't ship products for a year...?
The fact that they sell their chips for five fucking dollars a pop has always been kind of obscene, but on such an old node, they really shouldn't cost more than a dollar, tops. No wonder they're so heavily cloned.
Most chips are not general purpose CPUs and don't need and would not benefit from 5nm transistors.
Larger components are actually preferable when talking signal noise, ESD susceptibility, etc.
> Die manufacturing cost is roughly the same for both dies (~10-15 cents) .
Again, this chip should never cost more than a dollar, even in single unit quantities.
Do you want to say we should abolish free market?
There are alternatives to FTDI in its niche but for some reason people prefer FTDIs chips even though the cost much more than alternatives.
It is practically definition of fair competition. Some try to produce cheap and some try to produce quality.
Don't build a straw man.
FTDI has every right to charge whatever they want, just as I have every right to call them greedy bastards who would rather play dirty games with their drivers than acknowledge that their products haven't been priced competitively for years.
If you really want to play this game... One could argue that in a truly free market, the copycat chips would be legal.
But it's not even about quality vs. price. FTDI's legitimate competitors often have better designs while also selling them for less. FTDI's only advantage is that they were early, and people are often reluctant to change an existing design that already works.
There are alternatives to FTDI that are not clones to FTDI. For whatever reason people prefer to use FTDI in their design.
FTDI does quite well on their chips which means it is stupid to say their products are not priced competitively -- they have competition and are still winning at it.
You are making a guess.
FTDI is a private company, and unless you are an insider or have done original research, you have no way of knowing whether the company is doing well financially.
The price-per-reel on Digi-Key and how many chips you see in teardowns of legacy products and on cost-is-no-object FPGA boards isn't necessarily going to paint an accurate picture of this for you.
In my experience, when chips are expensive (for non-supply reasons), it's usually for one of three reasons:
1. There's nothing else like it on the market, so the manufacturer can charge whatever price they want, and designers will still buy those chips (that last part is critical). (e.g. certain specialized chips from Analog Devices)
2. The manufacturer's costs are out of control. This is rare today, but was common in the '80s and '90s. Most of those companies went bust or were acquired, and now the semiconductor design industry is extremely focused on cost efficiency.
3. The manufacturer can't sell enough chips at a competitive price to pay back the NRE costs of the design + a reasonable margin, so they're forced to sell at a higher price. If they don't have other products, and are not able to diversify or sell the company, this is a death spiral.
For FTDI, (1) is clearly not the case -- they have lots of competitors, and the task that their chips perform (shoveling bytes around) is a basic one. FTDI's primary office is in Glasgow, Scotland, and they have satellite offices in China, Taiwan, Singapore, and the US. Their offices are tiny. (2) seems unlikely.
That only leaves option (3). The semiconductor business is famously reliant on volume, and this is especially true for low-cost, small-die-area designs. You need to sell a metric fuckton of chips if you're selling them for a few dollars each. And the only way to get that kind of volume is to secure design wins in high-volume consumer electronic products that sell by the metric fuckton.
Most of FTDI's volume appears to be in legacy products. Products that used to have a serial port, but that now need a USB port because computers don't have serial ports anymore. They make a few higher performance parts that seem to be designed specifically for FPGA vendors, because FPGAs often need a way to get as many Gbps as possible over a USB port, and FPGA vendors don't particularly care about an extra $20 chip, because their FPGA already costs $1000 and they don't want to muck-up their low-voltage process or waste pins and die area to support USB. Also the FPGA vendors picked FTDI ages ago, and don't want to rewrite their shitty, shitty software.
But you'll never see that FPGA USB-serdes chip designed into a laptop or a phone or a camera, because there are cheaper options available, and because the functionality can be had essentially for free when integrated onto the high performance SoC that requires it.
FTDI probably has enough volume on their legacy chips to cover their costs, but that's because the design is ancient and was paid off years ago, and because there are a lot of legacy industrial products that aren't going away anytime soon. The fact that they're still overpriced relative to competitors means they're not a product of choice for new designs, and are a particularly obvious candidate for cost-reductions. I don't know anyone who willingly designs in an FT2232 when a CP2104 will do the trick, especially after FTDI flushed their reputation away with that driver fiasco. So this part of their business is likely to shrink.
The FPGA vendors sell maybe a few thousand units of their $1000 - $10,000 FPGA boards. That's probably not enough volume to cover the costs of FTDI's higher performance chips, even at $20 a pop.
The sad thing is that (1) used to be true, back when USB was new, and there were a lot of buggy implementations. FTDI got it right, and they even had a reputation for a driver that "just worked." The answer to the question "How do I do USB?" used to be "Just shut up and buy an FTDI chip." An enviable position at the time, but as USB became more ubiquitous and better understood, their competitors caught up. Other vendors have good implementations and good drivers now. Good USB IP for chip designs is also readily available, so many microcontroller designs can do without FTDI now.
The FTDI driver fiasco was the worst thing they could have possibly done with the one thing they still had going for them -- they trashed the reputation of the quality of their driver, and now everyone looks at them with suspicion. Normally you would expect a semiconductor company with a limited, overpriced product line to be an acquisition target, but the fact that this hasn't happened despite the fact that they're a relatively small target suggests that the bigger players either looked at their books or did the math and realized that their market is shrinking and they have no growth prospects.
tl;dr
FTDI may not be doing as well as you think they are. Rather than doing well because their chips "command a high price" -- every indication suggests that they are resting on their laurels, they know their long-term prospects are grim, and are trying to squeeze out as much as they can before they go bust or are forced to sell themselves for scrap.
But I'm also surprised something so large is being used. Isn't 130/180nm cheap for any kind of significant volume?
Edit: maybe not, I found a die shot: https://zeptobars.com/en/read/FTDI-FT232RL-real-vs-fake-supe... - even on the genuine 2005 die there's lots of empty space. Could probably squash that by 50% with only moderate effort without even a process shrink.
Now I'm wondering what volume FTDI see. Can't be too high or it would be a more competitive market.
Hence their knee-jerk reaction to the clones.
First, you need to talk about whole cost of production and quality of the product. When you reduce your node your development and production costs grow exponentially. Size of the wafer is only small part of the production cost.
Second, the chip area can't be arbitrarily small. You still need area to connect it with outside, area for larger components, etc. When you reduce your transistor size at some point the chip area will stop getting reduced because of packaging requirements.
What exponent?
The number of mask steps is similar or a few more from 600nm to 180nm. Isn't that the best way to estimate cost per wafer, in combination with the cost to make the masks?
> Second, the chip area can't be arbitrarily small. You still need area to connect it with outside, area for larger components, etc. When you reduce your transistor size at some point the chip area will stop getting reduced because of packaging requirements.
That's why I said 5x instead of 20x. I looked at the size of the pads and the circuitry around them, assumed they wouldn't shrink at all, then combined that with a somewhat pessimistic estimate of how much the logic could shrink.
https://arstechnica.com/gadgets/2021/06/chip-shortages-lead-...
(which was posted to HN here: https://news.ycombinator.com/item?id=27506662)
It’s interesting that they went to the bother to mimic the chip without simply duplicating the entire chip and layout.
The drivers themselves can't possibly be that complex. But getting a USB-IF vendor ID as well as drivers in Windows is probably a bit harder.
Not only does Linux have its own drivers, they can unbrick the damaged devices.
https://www.mail-archive.com/linux-usb@vger.kernel.org/msg50...
I only happened to find out it was counterfeit due to the fact that I purchased it via a link in a forum recommended by "some user", and later, when I had this stupid looking cable and no idea where to plug the ends into my Pi, I revisited the forum, scrolled down a few replies, and found out that all of the "FTDI" products from the manufacturer of my cable were fakes.
Once I read that, I had intended on replacing it, but decided to wire it up anyway to see if it would work. I run OpenSUSE Tumbleweed on my main computer, and I don't remember having to do anything unusual, or encountering any warnings/errors at the time trying to use it[0], I'm guessing that's why...
[0] I was trying to connect into a headless RPi and that was the most convenient way sometimes.
I bought an FT232RL breakout on eBay a year or so back, planning to use it in bit bang mode, and it turned out that while it worked for talking to UARTs, the bit bang mode was completely broken - and the EEPROM (needed to configure it the way I wanted it) was not even writable at all (probably not real EEPROM).
In my case, it was a counterfeit - the product was explicitly advertised as FTDI FT232 but wasn't. I was fortunately able to get a refund for it through eBay. Sadly it is very hard to know what you are actually getting due to the prevalence of identically branded clones, and beyond refunding me eBay doesn't seem to have done much despite me pointing out that these devices are being advertised as FTDI FT232 but are actually significantly below the advertised specs.
As discussed elsewhere here, FTDI pushed an update that deliberately tries to brick FT232 clones by writing bad data to their EEPROM - which obviously non-configurable devices would be immune from and might be one reason (alongside cost) which the people creating clones have cut out the real EEPROM now.
I agree with those here the bricking was a very bad idea - not all clone chips were necessarily even counterfeits (it bricks based on PID / VID, and can't see what is actually written on the face of the chip; reusing a USB PID/VID registered by someone else to enable compatibility without using another company's trademark isn't counterfeiting). Given they deliberately destroyed other people's property because it contains a chip that might not even be counterfeit deeply embedded in it by the manufacturer, I think they are lucky not to have faced any lawsuits or prosecution.
That's for one-offs and debugging connections, though. If you're shipping a product, then suddenly the economics change and the software-heavy solutions start to look good again.
1. Trust your vendor's libraries/ROM drivers/example code. This is the fastest to bring up, the least flexible, the buggiest (it's vendor code!) and the hardest to debug. (Trust me on the debugging part.)
2. Use a library like tinyUSB. This can actually go pretty quickly, since you've got full source control. If your part is popular enough that someone else has already done most of the work, this is by far the best choice. Even if you have to get personal with the hardware, I still like this option. (tinyUSB is much better than what came before... I wish it'd been around last time I had to do this.)
3. Write your own. Don't do this. (You might have to do parts of this if you tried #1 and it isn't panning out.)
All of these are doable. Any of them are enough work that if you're only making one, just put the FTDI down and call it done.
Lots of ARM chips come with USB DFU mode firmware baked in ROM, those don’t need one.
It is insane to blame the chip marker. They have a reputation to uphold. How can they be expected to continue operating if they not only have to compete with shoddy knockoffs, but basically support them with QA as well?
If China can't afford to play with closed source, maybe they should invest in open source.
The genuine chips can also be bricked by a very similar command sequence. Are they also defective?
> It is insane to blame the chip marker. They have a reputation to uphold. How can they be expected to continue operating if they not only have to compete with shoddy knockoffs, but basically support them with QA as well?
So refuse to operate with the fake chips. Bricking is unnecessary and deserving of blame.
And as the article points out, a chip that's merely compatible and not counterfeit might also get bricked.
This driver is licensed to you only to be used with such devices.
"The licence only allows use of the Software with, and the Software will only work with Genuine FTDI Components (as defined in the Licence Terms). Use of the Software as a driver for a component that is not a Genuine FTDI Component MAY IRRETRIEVABLY DAMAGE THAT COMPONENT.
It is your responsibility to make sure that all chips you use the Software as a driver for are Genuine FTDI Components. If in doubt then contact FTDI. The Software is provided “as is”. There are no warranties (or similar) in relation to the quality of the Software."
Seems pretty clear to me.
I still get to blame them for designing a puppy kick module!
Edit: Someone made a reply quoting this comment https://www.eevblog.com/forum/microcontrollers/ftdi-gate-2-0... and I already wrote a response so I'm going to put it here:
Linux kernel? Really? The whole point of that is to reuse code as much as possible.
Even for Windows, the point of getting things into Windows Update is to make hardware just work, not to advantage one company over another.
But sure, they can make a reasonable argument toward making the driver incompatible. (If they're also willing to hurt people that tried to buy genuine parts.) It still doesn't justify the bricking.
> slightly higher sales price
Ha! I'm pretty sure it isn't "slightly".
Should Apple be responsible if these devices stop working after flashing the latest iOS update on them ?
Here you try to use a FTDI driver on a device that is not made by FTDI.
Seems like using a nVidia driver on a nVidio card. It's not the problem of nVidia to make sure the driver is compatible with nVidio.
Assuming the bricking of FTDI devices as mentioned in the update isn't "We sent out code to intentionally destroy counterfeits" (i.e. it served no other purpose, wasn't an accident[0])
Did they QA it with counterfeit devices to ensure they were bricked? Or did they just write a new driver for their devices that happened to result in (some/all) counterfeits being bricked?
If it's the latter, then it's not FTDI's fault. It's unreasonable to expect FTDI to QA their drivers against products that aren't theirs.
It is, however, their problem. The last activity that user did was install their driver, and the result is broken stuff. They've just accidentally pissed off a bunch of non-but-possible-future-customers. Taking "fairness" away: a percentage of those affected will assume it was intentional, even if it is very obvious that it is not[0]. And even of those who don't blame FTDI, a percentage of those will feel FTDI owes them something, anyway simply because "big bad corporation"[1]. It's probably in their best interest[2] to detect non-genuine products, inform the user of the potential risks and either prevent the driver from loading or allow the user to accept the risks at their peril if there's a chance the driver can brick the device.
[0] Say, the counterfeit has half the storage/ram of the geniune, and reports it has the same capacity, etc, and a driver update passes the "reality barrier"
[1] Forgetting how often "big bad corporation" is really, really, tiny -- and I have no idea whether or not FTDI is big; it's not relevant.
[2] Albeit, it may not be from a legal perspective...
It wasn't an accident.
And in the case of a counterfeit low level component then they should.
If China wants to compete, then compete. If they are allowed to do this, legitimate companies will suffer and legitimate progress will stagnate. Just release the same chip open source with instructions for using it in this use case. Then when the legitimate company bricks the device, they look like the bad guy stifling competition.
(1) Was the "bricking" intentional[0], as in, did FDTI's driver look for the counterfeit, and upon finding it, issue instructions that would make the device useless, even when used with drivers/configurations that previously worked. It also matters, a little, whether or not this is easily reversed.
(2) Did the FDTI take similarly extreme measures to prevent/stop unauthorized chips that weren't impacting customers/non-customers.
The main thing that would leave me with little sympathy for FDTI[0] is if it was intentional, i.e. the update had no other purpose except to break these devices. Past counterfeit situations come to mind where an update "just happened to brick" counterfeits, but it was due to the counterfeit having some limitation[1] not present in the genuine product.
But that second point is beginning to matter a whole lot more, these days. If FTDI makes it extremely difficult for me to determine what is a real FTDI chip and what is not a real FTDI chip -- or, put another way, if the counterfeits are that difficult to tell from the real thing -- and FTDI decides to breaks the one I thought was real...? I'm going to be swearing. I'm going to be at least as pissed off at the FTDI folks as I am at the outfit responsible for the counterfeit, but FTDI is forever associated in my head with "I can't do what I need to do, now". It's terribly unfair that FTDI has to go through trouble to keep a$$holes from counterfeiting their products, but it is the world we live in.
There are also companies that specialize in this sort of thing for brands/products of all types[2].
At the end of the day, the worst FTDI should do is throw up a large notice "This is not an FTDI device. If it indicates that it is an FTDI device, it is a counterfeit. This driver most likely will not work with your device because it isn't ours." ... here's a link to buy a real one from places that we know won't sell you a counterfeit, etc.
That puts blame where it belongs and even if the driver worked would probably result in me purchasing a replacement from the link, knowing that I'd never have problems in the future.
[0] And hope that some legal issues would come their way
[1] Maybe less RAM/storage than it's supposed to have, and maybe it lies about it, too.
[2] I'm doing work for one, right now... it's an interesting space; I used to think it was as pointless as trying to stop media piracy. There's, surprisingly, a lot of effective (and creative) measures to greatly disrupt/stop these larger scale rip-offs, and a lot that can be done which legitimately protects the customer -- but it's a thousand-pronged attack and you'll never completely eliminate counterfeits, especially for designer brands and such.
I thought this was a great example of second order effects at work.
I was getting slightly worried about some of my tools until I saw the "Update 25th of October 2014:" at the bottom (Cause I missed the date header at the top).
But their time and labor? Nobody can understand how valuable it is!
Yeah, really "entitled". /s
Just like the virus, we cannot investigate, cannot stop it, cannot even name (but can say for others abd even now x.x mostly from Uk etc.).
Humanity not just your serial driver is actually at stake. Have to stop or start somewhere.
Wrong is wrong. Say it, do it abd actively buy only real and legal one.
Please.