• Please review our updated Terms and Rules here

The curse of fake parts out of China.

I've not done a deep dive, but how does the USPS staff Sunday delivery? Does the USPS have extra staff for this service, or does it have to juggle workers around?

As to bankruptcy, perhaps it is trying to do everything Congress and federal rules say it needs to. Faced with a collection of bad to worse choices?

Ever notice that the cost shipping packages is getting closer to UPS and FedEx rates? Or so it seems to me... I use SpeeDee whenever I can when shipping packages [midwest only, unfortunately :( ]
 
VERAULT said:
Yeah?. .then why is the US postal service having such a hard time of it constantly crying bankruptcy?
Click to expand...
A large part of the problem is that "certain folks" who dreamed of privatizing mail delivery passed legislation that required USPS to pre-fund its pension plan for the next 50 years, a burden that no other employer in the US carries. Anyway, 'nuf politics.
 
Unfortunately, there's profit for the counterfeiters, and as long as that continues, the counterfeiting will continue.

My first introduction to counterfeit parts was the LM2596 regulator. They'll sell you one that says LM2596 on it, but the switching frequency is incorrect (and not due to some margin of error; it's clearly a different part). Thus all of the formulas in the datasheet are wrong, and you end up selecting the wrong inductor and capacitor. In the best case your project will run a little hot and/or a little inefficient. In the worst case you were actually expecting to maximize the performance of the part, and you'll end up melting something.

My latest experience was AS65C4008 static RAM chips. I needed 64 of them for a particular project, and Aliexpress was the way to keep the costs under control. They ended up working fine in my 4MHz project, but then this week I went to use them in something faster and had intermittent memory errors that got worse as clock speed increased. I swapped in 'known good Mouser parts' and all the trouble went away. Same identical part number. Same speed. The Aliexpress didn't match performance of the Mouser ones.

The profit in the above is clear -- the underperforming parts work for most people most of the time. Hobbyists don't usually measure the efficiency of their switching regulators (until we burn our thumb on an inductor) and so many people are building 4MHz Z80 projects that an underperforming static RAM will work just fine. But then there's the situation where you get a part that clearly will not work.

Take for example some D8008-1 CPUs that I bought on eBay. They didn't work at all. Based on what I see reported online they're probably 8284 clock generators that have been reprinted. In that case there's no chance they could work. You're not going to put them in a circuit and say "good enough". In this case, the profit comes by attrition. They argue and they delay. They question whether you're qualified to test them. They want to give you a 50% discount (why would I pay 50% for something that is 100% useless?). They'll begrudgingly accept the return, but are unable to create a shipping label. They run out ever clock eBay sets and respond at the last possible moment to reset the clock. It takes you weeks of persistence, but eventually you'll get your money. Here they make their profit because a lot of people just give up.

I think by now they're so contaminated their supply chain that there are a lot of things I just won't buy from China anymore because I don't want the hassle. Some of that contamination has reached back here to the states as well. I think a lot of sellers don't even know whether what they have is genuine or not.

As far as the USPS is concerned, that seems like an orthogonal discussion. I've always figured the shipping from China to be so cheap because it's subsidized at the source to get it across the sea, then they take advantage of our relatively inexpensive last mile at the USPS.
 
smbaker said:
I think by now they're so contaminated their supply chain that there are a lot of things I just won't buy from China anymore because I don't want the hassle. Some of that contamination has reached back here to the states as well. I think a lot of sellers don't even know whether what they have is genuine or not.
Click to expand...

One problem is, if you totally saturate the market with counterfeit parts, in the end nobody can spot the fakes. One example would be the ubiquitous 1N4004-1N4007 diode range. They invade supply chains including Military & Avionics ones.

If you examine the original 1N400X parts from Fairchild in the early 70's, not only was the epoxy casing much larger than modern ones, but the diode crystal inside the package was substantially bigger. As these got cloned/copied and faked over 30 to 40 years, they became more and more substandard with a smaller body & die (not dissimilar to what happened with the 2N3055 transistor), "working most of the time for most projects" but not lasting as long when pushed to max the spec of the original part.

There is also a "Specmanship" war out there, where somewhat outrageous claims are made about power handling and current handling ability of a lot of modern parts.

The designers from yesteryear were much more conservative. It is now like the design ethos is how many kilograms of material can you remove from a building's foundations, so that it just doesn't fall down ? It is about saving money.

It is very hard now to find original design 1N400X diodes, sometimes a year goes by before I see one. So I have abandoned them for most projects and instead use the BY448 to replace the entire series.

And I agree, a lot of the sellers have no idea, so in many cases it is bad to name & shame them when they send fake parts, so I generally never do that.
 
Last edited:
I bought some 6522 from Aliexpress. They look fake / relabeled, but after measuring their behaviour on a breadboard, checking the current draw compared to a genuine one it looked ok, so I tried them in circuit. They both worked in UC2 of a 1541 short board.

Does anyone know if MOS produced ICs looking like this? One has "R6522" and "Mexico" on the bottom. I cannot tell about their long term reliability, but they passed the performance test of the 1541 diagnostic cartridge
 

Attachments

  • 6522.jpg
    6522.jpg
    835.7 KB · Views: 16
>>I bought some 6522 from Aliexpress. They look fake / relabeled
My experience is that if they work at all, they're most likely genuine. My own fakes have all been DOA,
probably a completely different type of chip to the type claimed by the (re)labelling.
 
So I think I may have gotten my first FAKE chips that have actually caused me problems.

I ran out of 74LVC245AN chips. I had some on order but needed a few quick for a board. So I bought this from Amazon: https://www.amazon.com/dp/B08GJDWK9R?psc=1&ref=ppx_yo2ov_dt_b_product_details

I built a couple of boards and one in particular wasnt working. Swapping these chips I noticed I could get it working.

Now If I test thee on a Minipro using the TTL checker as a 74ls245 they will check fine. But a 74LS245 is not a 74LVC245. I think these chips are rebranded 27LS245. With the qtip tons of black comes off. So these were rebranded.

Everyone seems to have these on backorder so I had to make an order from china. And I bought these from Amazon hoping they would work while Im waiting for my big order. They didnt, now I have real doubts about my china direct order of them.
 
Hugo Holden said:
Now you might want to trust them, maybe to hold the code for a nuclear reactor protocol that was running next door to your house, as Clint said: Go Ahead, make my day.
Click to expand...
If a system running a modern nuclear reactor is using 2732 UV EPROMs, whether they're fake or not is the least of my worries!

smbaker said:
As far as the USPS is concerned, that seems like an orthogonal discussion. I've always figured the shipping from China to be so cheap because it's subsidized at the source to get it across the sea, then they take advantage of our relatively inexpensive last mile at the USPS.
Click to expand...
At least until recently, it was not subsidised in China; the U.S. and similar countries were effectively subsidising packets shipped by mail from China.

For international mail the source country's postal service collects the postage and is responsible for getting the mail to the destination country. At that point, the destination country is responsible for the remainder of the delivery, and the source country pays terminal dues to the destination country based on weight, number and type of mail. Generally this is profitable for the source country as the cost of delivery to the destination country's postal service, minus terminal dues, is less than the source country's postal service received to mail the letter or package.

This wasn't a problem for decades while the U.S. shipped out more by post than it received; in 2010 the U.S. postal service made about a $275 million surplus on international mail. However, in the 2010s things started to switch around as people started order more and more things directly from China, which also happened to have low terminal dues due to being in one of the "developing country" categories. Combine that with the lower labour costs and thus lower postal rates, and now you could actually ship something from China to the U.S. more cheaply than from the U.S. itself, with the USPS picking up most of the cost of delivery.

This has been changing since then, especially since the 2019 Extraordinary Congress of the Universal Postal Union, but I don't really know the state of things now.
 
powerlot said:
I bought some 6522 from Aliexpress. They look fake / relabeled, but after measuring their behaviour on a breadboard, checking the current draw compared to a genuine one it looked ok, so I tried them in circuit. They both worked in UC2 of a 1541 short board.

Does anyone know if MOS produced ICs looking like this? One has "R6522" and "Mexico" on the bottom. I cannot tell about their long term reliability, but they passed the performance test of the 1541 diagnostic cartridge
Click to expand...
In my experience it's not unusual for vendors relabeling parts to relabel genuine parts. I picked up a bunch of "Rockwell 65C02" ICs from AliExpress a while back and half were CMOS, half were obviously NMOS parts. (Some drew under 10 mA, IIRC; the others drew well over 100 mA. I didn't check for the additional Rockwell instructions, but I'm willing to bet they weren't there on the high-current-draw parts.)

I suspect what's happening here is that the vendors know that hobbyists are looking for "Rockwell 65C02" parts, get in a load of "6502s" of various sorts from all manner of sources, and just relabel the entire lot, regardless of what they are. Some will be 65C02s of various types, others will be NMOS 6502s. They neither know nor care, and many purchasers won't see the difference because an NMOS 6502 will work well enough in their applications.
 
cjs said:
.....many purchasers won't see the difference because an NMOS 6502 will work well enough in their applications.
Click to expand...
This is the whole problem, Especially with re-labelled transistors diodes etc, especially power devices.

In the good old days when designers had solid & detailed data sheets to work with and parts that were described by that data, they could design with components and keep them inside the manufacturers parameters, with a sensible margin, but still push them to their full specs. However, the people who are finding equivalent parts do it with very limited data. Just a few basic parameters and often they use suggested equivalents from transistor equivalence manuals. Once I ordered a Siliconix high voltage Nch Jfet, the supplier sent re-labelled NPN silicon transistors. After I investigated it, I found that one of the transistor equivalence manual had incorrectly listed it as a silicon NPN device.

But the thing is the suppliers have no idea about the particular application, how the device is being used, especially where the part is pushed near its margins and where the part requires all of its original properties to work. The are many fakes out there for Horizontal deflection transistors for VDU's, the people supplying the equivalents have no idea of the combination of special properties the particular application requires to work reliably, if at all.

(PS: I'd still prefer a vintage mil spec 2716 to hold the reactor code than a modern uP with forgettory flash memory, just a personal preference mind you, main reason being we know that these can still be good 50 years after the original programming, we won't find this out for another 30 years or more for modern parts made in the last two decades. Vintage fuse & mask ROMs appear very good too. I'd also prefer if it was controlled by vintage mil spec TTL than modern cmos, the junctions in the vintage IC's are much bigger and more robust in TTL devices and more immune to impacts from high energy particles and any kind of ESD damage too. Also ceramic body IC's , which most of the mil spec ones are, are radio opaque to X-rays).
 
Last edited:
We vintage hobbyists in 2024 can take something from the late 1970s or early 1980s and with pretty good probability it'll work. Component quality and consistency was pretty good. If a component has failed, then you can generally find a replacement because we made standard parts and we documented their behavior well.

Will vintage hobbyists in 2064 be able to use our 2020-era tech with such good luck? I don't think so. I would either expect our tech to not work at all, have its storage media and/or firmware corrupted, or in the off chance that it was capable of working, it would probably stubbornly refuse to work.
 
Of course not--stuff from the 1970s was not designed to be disposable. In a sense, e-recycling has encouraged the "throw it away" mentality. Things are no longer designed to be serviceable at a component level.
 
One thing I have found, it is much easier to rehabilitate vintage computers, because the RAM & ROM are mostly separate from the CPU. Usually the ROM can be dumped easily, if the file is corrupt, often someone else has dumped it or it can be repaired.

The problem with a lot of appliances filled with modern uP's, the manufacturers never release the firmware. It often gets locked in with various security methods. It will become a futile and uneconomic prospect to repair many computers, appliances of all kinds, in the future. So it creates a machine, that is near impossible to economically repair, because it contains what is effectively a unique component.

But this started out in the 80's, Motorola made a MC1468705G2 with the UVeprom inside the CPU package (my nemesis IC). Easy to program one for the manufacturer, but trying to get the ROM data out of it is a whole other story. So if you own a machine with one in it, in the systems control (I have a VDU with one), and it fails, it is game over. If only they had used a separate CPU and UVeprom.

I made the Motorola programming pcb for this IC, but so far, with modifications, I have been unable to extract ROM data.
 

Attachments

  • prog.jpg
    prog.jpg
    285.7 KB · Views: 4
There are a number of methods for extracting MCU program memory contents that were effective up until the mid-2000s when the manufacturers started designing against them. Have you tried any of these tricks? I'm not too savvy on them, but I know many involve special power/reset timings and lowering or raising supply rails to get the read protect transistors to turn off, but still have the memory array active.
 
cjs said:
For those who are interested (as I am) in the MC1468705G2 saga, there are three threads discussing this in more detail, including several different ideas for getting the program data out:
Click to expand...
As you can see this project has tormented me for some time.

I was interested to see recently somebody de-capped a PAL or similar and was trying to resolve the fuse arrays. They must be equally tormented by the hidden data.
 
Chinese did not invent selling scams.

If you use a site whose review system you can't trust, not from it being inherently untrustworthy but from another language/culture where you can't easily verify or disprove the comments, just use common sense.
DIP package chips, easy scam, like USB sticks, plastic boxes that can be relabeled. VLSI chips? Hard.

Ebay is like our fish market/flea market, Aliexpress is more like an oriental bazaar, there's a bit of difference in quality and approach but its the same league.
 
You must log in or register to reply here.
Back
Top