Counterfeit parts..

[EEEnthusiast]

I have had a nasty experience with fake ICs in my earlier company. We were building some boards in China (few thousand qty) and the contract manufacturer was supposed to source the PCB + BOM + ASSY and kit packing. Our BOM had all the parts which were from reputed manufacturers. But without our knowledge, the contract manufacturer would swap the components for equivalent parts which were 1/5th the price of original. Some of the easily swapped components were SMD capacitors, resistors, inductors and connectors. But they even swapped ICs from FTDI and some RF Filters which were custom built for this. We could figure out the FTDI was fake, as it functioned differently from the original ones on Windows10.

for some capacitors, we could recognize the difference in the colour. For others it was just our luck which made the boards work. But the long term reliability and derating were all in question.
We were never sure if the contract manufacturer did this or their supplier swapped the components. The element of risk is always there.

[OwO]

"Counterfeit" chips don't concern me. If it's a dud it will be caught by sample testing prior to assembly, and functional testing will catch remaining problems. I'm far more concerned about latent problems like ESD damage or moisture damage.

Never skimp on passives, and I recommend sourcing (and stockpiling) them yourself. They make up over 90% of your BOM by quantity but typically a small fraction by cost. Any failure rate in passives can greatly impact your yield. On the other hand expensive chips is where I hunt for the lowest possible price and even a 1-2% failure rate can be acceptable and work out far cheaper than buying from reputable sources.

[OwO]

FTDI is a sorry excuse of a company and if I ever have to specify their parts in a design for whatever reason, I'll make sure they are ordered from the shadiest source possible at the cheapest price possible to maximize the chance of getting a clone rather than the genuine thing. I don't want to give FTDI any money.

[EEEnthusiast]

why do you dislike FTDI.. Their parts are great in my experience..

[wraper]

"Counterfeit" chips don't concern me. If it's a dud it will be caught by sample testing prior to assembly, and functional testing will catch remaining problems. I'm far more concerned about latent problems like ESD damage or moisture damage.

. Functional testing after assembly usually won't catch anything. Then device will fail after some time. Also how about ESD and moisture damage in salvaged parts which are sold as new?

[OwO]

At the extortionate prices that most analog ICs sell at it's far cheaper even to do burn-in testing than to buy at list price. Example: ADF4350 - $10 on digikey, $2.5 new on the market, $0.4 NOS. Even just $1 of savings buys at least a month of burn-in testing, accounting for storage, power, labor, and test jig costs. The costs are per-unit of the finished product, not per-chip, so it's easily worth it.

[peter-h]

I would not slag off FTDI. They are a very good company. They are UK based and support their chips and support designers, both with good quality drivers which just work and with direct support. You get no, zero, zilch support from Prolific or any of the other chinese chip makers.

I have used tens of thousands of FTDI chips and never had a single failure.

They are entitled to scupper fake manufacturers by not supporting the differences in the fake chips, in the drivers.

I agree re passives. I free issue all components to the SMT assembly contractor. And buy all parts from proper distributors. Most are so cheap that it is not worth doing otherwise.

Just seen some interesting fake Hanrun RJ45 jacks with integrated magnetics. 1/3 of the Hanrun price. Easy to spot if you compare them side by side, but most people would not tell. The fake ones do appear to work.

[wraper]

Fake, OTOH, can be testes easily.

If by fake you mean something non functional inside. Otherwise you would need to do quite sophisticated testing for meeting electrical specs and X-ray them. Simply mounting on PCB and device passing tests means only there is no complete dud inside. Otherwise you can "successfully" use 7805 which can't output more than 500mA or transistors which eventually fail at 20% of rated current.

[peter-h]

That's true, but transistors (other than special mosfets etc) and 7805 type devices are so dirt cheap that why buy them from dodgy sources?

I think the "interesting" area for counterfeit parts, from the POV of the equipment builder, is expensive passives, like the RJ45 example I give above. It is very easy to make fakes of these, which may be way out of spec but will still function. And the incentive to buy fakes exists because if a part costs say $1 that is quite significant on a $30 (selling price) product, so a $0.30 fake is attractive.

The above video shows a lot of Yamaha parts, last made in (IIRC) 1995, and they would have been quite simple chips to countefeit and end up with something functional.

If you make non functional fakes (pretty much the only option for stuff like ARM processors, or even a Z80 which is quite a significant logic design project) then you will achieve just one hit on somebody and then you have to empty out that bank account, disappear, and start again.

A bigger problem is hacking the email accounts of chinese companies (seems to be very easy) and when you see a conversation where the context is a new order, and bank details, you fake an email with "updated" bank details. I've just lost $2k this way, and it is *very* hard to guard against. This is another topic

[OwO]

Of course it has to be judged on a case by case basis, but any part that costs more than $1 and isn't arguably very good value is generally a candidate for (1) design-around (2) substitution (3) grey market sourcing. An Ethernet jack is an example where $1 is justified and reasonable value, but expensive chips can be safely skimped on with some QA measures, and you can get total BOM costs down to 1/5th what it would be with digikey prices, which makes the difference between a nonviable product and a competitive one.

[peter-h]

Nobody would buy from Digikey (or Mouser, Farnell, etc) in production. Their prices for say 1k are roughly 2x higher than from proper distributors.

I use Mouser a lot but only where price doesn't matter.

However the "proper" distis (Avnet Future etc) have zero service. It is quite impossible to ask them for an 0805 resistor 100R 1%. They simply have no idea. You have to give them a P/N and then they may be able to find that one, or equivalents. So every time we buy passives from any of these, we have this silly game where they have no idea whether they sell resistors or chocolate bars.

Mouser still do have a clue, but importantly they have a website which shows the actual parts and prices.

[Mr Evil]

The worst fakes I have had were fake UK plugs from eBay. Among other dangerous flaws, they came with the fuses shown in the photo, which have poorly fitting end-caps, and you can even see a bit of fuse wire sticking out from the one on the bottom left. I reported the seller to eBay and trading standards, but sadly they are still selling them.

[FTDI] are entitled to scupper fake manufacturers by not supporting the differences in the fake chips, in the drivers.

If all they did was not support the fakes then that might be acceptable, but they effectively destroyed completely innocent end-users' property. They make ncie hardware, but after they did that I decided to never buy anything from them ever again.

[Cerebus]

They are entitled to scupper fake manufacturers by not supporting the differences in the fake chips, in the drivers.

Yes, but they are not entitled to write drivers that deliberately disable parts on other people's products. That's what they did, and people who had, unbeknownst to them,  fake FTDI chips make it into the supply chain had their boards disabled by FTDI's drivers. Not simply, "no that's a fake, I'll print an error message and abort" but "that's a fake, I'll deliberately go out of my way to alter its registers to make it never work again".  They deliberately damaged other people's property with malice aforethought. It is literally criminal behaviour, and FTDI were very lucky not to get their collars felt. See Criminal damage in English law.

Certainly no sane person would choose to use products from a company who think it's OK to commit the criminal act of trashing their customer's property. I used to specify FTDI's parts, I have not bought or specified an FTDI part since the whole sorry affair came to light even though they would often be the best, if not the cheapest, part to specify. They could have chosen to handle the issue fairly by simply getting their drivers to fail gracefully, they did not and opted to take a course that any reasonable person would have known was wrong.  I do not believe in encouraging their sort of arrogant, selfish behaviour by specifying their parts.

[vwestlife]

I got an Intel 8087 chip from China on eBay that is obviously fake because it shows a copyright date of 1978, when the real 8087 chip wasn't introduced until 1980. But it seems to work identically to a real one.

[peter-h]

I really doubt somebody designed a functional 8087 replica. It's a massive digital design project.
https://en.wikipedia.org/wiki/Intel_8087

Far more likely these fake 8087s are real ones, recovered from scrap boards, or subspec ones marked with a higher spec number, or some old stock somebody found and maybe "repainted".

I recall seeing fake Intersil 7107s in 1980 but a chip like that is pretty easy to design.

[magic]

You bet

https://en.wikipedia.org/wiki/K1810VM86

Wouldn't be surprised if China also had something at one point. As for 7106/7107, sure, they have been cloned many times and many years ago all over the commie countries. To this day every $5 multimeter ships with a 7106 clone.

[Karel]

They deliberately damaged other people's property with malice aforethought.

They didn't damage the fake chip. They wrote a value to a non-volatile memory inside the fake chip so that the OS didn't recognize the fake chip anymore.

Btw, if you write back the old value into the fake chip (using Linux and a small piece of software), the fake chip starts to work again. So, no physical damage.

If you ask me, it was brilliant 

[iMo]

You bet
https://en.wikipedia.org/wiki/K1810VM86

Soviets or DDR did 2.5mm pitch DIL40 afaik..

[Mr Evil]

They didn't damage the fake chip. They wrote a value to a non-volatile memory inside the fake chip so that the OS didn't recognize the fake chip anymore.

Btw, if you write back the old value into the fake chip (using Linux and a small piece of software), the fake chip starts to work again. So, no physical damage.

If you ask me, it was brilliant 

Technically the damage could be reversed, yes, but you're thinking about it like an EE. From an ordinary end-user's perspective, their device simply stops working, they have no way of knowing why, and it's as good as dead. Totally unfair.

[Cerebus]

They deliberately damaged other people's property with malice aforethought.

They didn't damage the fake chip. They wrote a value to a non-volatile memory inside the fake chip so that the OS didn't recognize the fake chip anymore.

Btw, if you write back the old value into the fake chip (using Linux and a small piece of software), the fake chip starts to work again. So, no physical damage.

If you ask me, it was brilliant 

Of course that is damage, it's not physical damage, but it's still damage. Deliberately turning off the electricity to someone else's dwelling doesn't physically damage anything, but I know someone who was threatened with prosecution for criminal damage for doing exactly that. (It was some university friends whose landlord would turn off their electricity supply regularly to try and control them. They were law students, their professor set their landlord straight with a letter that promised to have him in criminal court so fast his feet wouldn't touch the ground if he did it again. The professor also arranged a visit by the police. The landlord, sensibly, stopped.)

[peter-h]

Yes of course the USSR could clone an 8086 and 8087. They cloned the VAX11/780 too.

But that is the technical might of a whole country.

Some chinese fake chip operation is not going to be cloning an 8087.

[Karel]

They didn't damage the fake chip. They wrote a value to a non-volatile memory inside the fake chip so that the OS didn't recognize the fake chip anymore.

Btw, if you write back the old value into the fake chip (using Linux and a small piece of software), the fake chip starts to work again. So, no physical damage.

If you ask me, it was brilliant 

Technically the damage could be reversed, yes, but you're thinking about it like an EE. From an ordinary end-user's perspective, their device simply stops working, they have no way of knowing why, and it's as good as dead. Totally unfair.

Unfair from the counterfeiters, yes.

This discussion has already been done here: https://www.eevblog.com/forum/reviews/ftdi-driver-kills-fake-ftdi-ft232/

[all_repair]

Should put a stop to the discussion on FDTI gate thing.  FDTI already a history, people are actively avoiding and their plan to dominate the USB to serial have failed miserably, almost all China gears are not using the FDTI anymore.  FDTI was not better, it was just most convenient, even then there was much cheaper alternative at that time.  The margin in those cheap gears can hardly pay for what the FDTI wanted.

[wraper]

FDTI already a history, people are actively avoiding and their plan to dominate the USB to serial have failed miserably, almost all China gears are not using the FDTI anymore.

Was there any significant amount of real FTDI to begin with? There were a lot of cheap "FTDI" adapters simply because they all used counterfeit ICs. That's after Prolific started fighting counterfeits with drivers. Waiting when counterfeit WCH will appear if there aren't already.

[magic]

Some chinese fake chip operation is not going to be cloning an 8087.

They will stamp "Intel inside" on a random mix of recycled Intel parts, second sources and unauthorized clones from the 80s