Counterfeit products, or not?

[NaxFM]

Hello guys, it has been months that I thought of a project that I would like to sell, and today I decided finally to work on it. The main goal is to sell it, so I want components from reputable brands and reliable sources, I would really like to avoid components smoking and catching fire once in the house of the customer.
One of the most important one is a bipolar stepper driver, and I figured that an allegro A4985 would be perfect. On digikey, if I buy a thousand of them (which is really a LOT to start), the unit price is 1,52 dollars, fair enough.
The doubt is that on eBay and aliexpress, they sell entire pcb modules, already assembled with the required resistor and capacitors, at prices as low as one dollar or less, and they use the allegro a4988, which costs more than the A4985. I'm sure you saw them before, they're the typical stepper driver modules for 3d printers.
The price is so ridiculously low that I'm almost sure that they use counterfeit chips, but I want to ask you if they are really fake chips or they are actually original ones. I figured that maybe they have such a low price because they sell millions of them, I'm not really sure...
Obviously I want to build my product as cheap as possible, and if these modules aren't fake, then I don't see the point in spending 1.5 dollars on just the chip...

[janoc]

This has been debated to death already.

In short:

- A lot of those modules use fake/relabeled/compatible Chinese chips

- Thousand pieces of something is nothing. For A4985 it is not even a full reel! If you order 10-100k then the distributors and manufacturers start to be interested and you get very different prices.

- The module manufacturers aren't buying components for Digikey prices (Digikey has a very non-trivial margin!). If a factory somewhere in China buys a 100k pieces of something directly from the manufacturer (which often has the factory literally next door!) they aren't going to pay prices you see on Digikey but much less. And surplus/rejects then tend to wind up being sold for pennies on Shenzen markets (and then eBay/Ali/etc. inside these various modules). Some may work fine, some may be rejects that sort of work (e.g. had STM32F103s that worked inside a cheap STLink clone but wouldn't reliably work with custom firmware. A fresh chip bought from RS had no issues.)

- If you hope to compete with China on price, you will go bankrupt.

In that light penny-pinching on a motor driver for a small series production is silly, especially if you buy some who knows what quality parts from China and get your margin wiped out by returns. If you want to save no matter what and still keep the risk reasonable, look at e.g. LCSC and some of the Chinese components - there may be something similar to the Allegro driver and possibly quite a bit cheaper.

For example HR4988 - which seems to be an exact clone of your A4988 and costs $0.40@30 pieces ...
https://lcsc.com/product-detail/Motor-Drivers_HR4988_C128662.html

I hope that answers your question how is it possible that the eBay modules are so cheap.

[Seekonk]

I like to think of them as copies.  A lot of technology isn't hard to copy.  If it isn't a circuit designed right on the edge it may not matter. You may find it economic to test devices to come in.  That said, I just got a 100 TL431, a jellybean part, and all were not working except 5.  And those 5 were out of spec.  One had a lead defect which all leads me to believe these all came out of a sorting machine. Stuff is cheap if it doesn't have to be handled and stored. You got 3 cent microprocessors.

[NaxFM]

This has been debated to death already.

In short:

- A lot of those modules use fake/relabeled/compatible Chinese chips

- Thousand pieces of something is nothing. For A4985 it is not even a full reel! If you order 10-100k then the distributors and manufacturers start to be interested and you get very different prices.
...

Thank you a lot! That's all I needed to know!
I wanted to keep the production cost as low as possible because what I'm doing is a modular design, and can be comprised by many modules, so a saving of one dollar a module is really a big deal for me.
Of course I don't want to compete with China, I'm designing a product which needs different stepper drivers, but it's not for 3d printers, CNCs or something similar, I know that it's impossible to compete in a market ruled by China.
I'll check what lcsc has to offer, thank you again!

[NaxFM]

I like to think of them as copies.  A lot of technology isn't hard to copy.  If it isn't a circuit designed right on the edge it may not matter.
...

Of course, but can you trust cheap copies when it comes to safety and fault protection? For my 3d printers, I've been using for years stepper drivers which were basically free, but I know what I'm getting into and I take full responsibility for my choice.
But if I had to sell 3d printers that I made, I really wouldn't take any chance that someone could sue me because the power electronics went on fire and burned his beard after a short in the motor windings...

[OwO]

I've said this many times and I'll say it again: it is your engineering and QC that dictates your product reliability, not whether the parts you use are genuine. If you search this forum you will find many stories of product designers who have an unreliable product with high returns rate but used genuine parts from real distributors. Meanwhile I have used batches of taobao parts and shenzhen market parts in products and to this day haven't had a single warranty claim. That's because I employ several testing methods: statistical sample testing for incoming parts (to avoid a batch-wide failure); functional testing for the final product (all units); sample testing of the final product with more rigorous tests (shock & vibration, more detailed characterization, etc).

That's not to say I don't see bad parts; for example one batch of a high speed ADC had a fault rate of around 2%, but since i got it for an extremely low price ($0.5 vs $6 on mouser) and all faulty ones can be caught by functional testing it was worth it.

[chris_leyson]

^^ What janoc said

Obviously I want to build my product as cheap as possible

Then spend the time to do some research, make contact with far eastern manufacturers of stepper driver chips and then evaluate the parts. You really think Allegro A4988 is second sourced in the far east ? Have you made any effort to get samples and evaluate the HR4988, just saying it's a lot of work save less than $100.

What happens when you design in the HR4988, and later on you find that lets say 10% of them fail for whatever reason. Maybe you saved $10 on ten chips, but that is going to cost you in the long run. Customer refunds, shipping, rework and loss of reputation.

[OwO]

Addon: If your product is in a position where failure can cause dangers (e.g. fire), tread carefully. Just because you followed all design "best practices" and used only parts from digikey, it does NOT mean you have a safe design. You could easily overlook aspects of your circuit like inrush current, transient voltages, etc that overstress a certain component, for example exceeding the SOA of a mosfet with a turn-on transient current (something most people neglect to think about), you could have failures after a few months or only at high ambient temperature. If said mosfet is part of a motor driver and fails short, you can have fault currents that starts a fire. If you rely on components not failing in order to not start a fire, you are possibly in for some expensive lawsuits. I only gave a very simple example and reality is far more complex, but these are the kinds of things you must think about.

[OwO]

What happens when you design in the HR4988, and later on you find that lets say 10% of them fail for whatever reason. Maybe you saved $10 on ten chips, but that is going to cost you in the long run. Customer refunds, shipping, rework and loss of reputation.

If your functional testing doesn't catch those failures and you let them fail in the field, then you probably want to rethink your testing strategy.

[soldar]

A chain is as strong as its weakest link.

You can buy cheap parts and test them yourself or you can buy parts that you can trust are reliable.

As has been said, quality parts do not make a product reliable and guaranteed not to fail. A good design is also essential.  Give me the best parts and I can easily start a fire with them.

And, even the best designs built with the best parts will sometimes surprise you with unexpected failures. Before I bet the farm on a product I would need to have used and tested it for at least a couple years.

I have come across a type of watering timer that might have looked OK on paper but the power supply had a resistor that just got too hot for the neighboring components and would lead to failure.

IMHO, it is best to over-design a product at first and later tighten it as price constraints become stronger.

I would say a new product needs to start out with a very healthy profit margin which can later be adjusted. I think those who start trying to sell a product with the narrowest of margins often go belly-up because they just do not foresee or forecast warranty and other costs.

[amyk]

Printer manufacturers use these ICs in huge quantities and the price does go down significantly if you buy millions or more.

There's also the negotiation and guanxi factor that the Chinese are very keen on; I've heard it phrased thus: "If you know the right people and have the right connections, 1 million or 10 million parts may cost the same."

[thm_w]

(e.g. had STM32F103s that worked inside a cheap STLink clone but wouldn't reliably work with custom firmware. A fresh chip bought from RS had no issues.)

What kind of reliability issues out of curiousity?

Also possible:
- Remarked GD32F103's or other compatible (forget the name)
- Remarked STM32F103 in a lower grade (ie less flash memory, or some section of the flash was bad), since its not possible to actually tell which device you have unless you trust the silkscreen. AFAIK device ID is the same for quite a few parts. But it seems rare to have this problem.
- Earlier revision (maybe, if it was ever fixed in eratta)

[CatalinaWOW]

You also have to understand your user base well.  Those stepper drivers used in 3D printers apparently have a pretty low failure rate.  I am not seeing screams of anguish from users and these are sold in fairly large quantities.  Certainly tens of thousands, perhaps hundreds of thousands.  But that application has pretty good control on the environment.

If your customers are likely to short outputs, stall motors, run in extremely dusty or hot or wet environments and otherwise abuse your equipment you will need to spend far more effort on protection, fault tolerance and idiot proofing than goes into the dirt cheap modules.  It still might pay to buy them and then plug them into a protection wrapper.

[janoc]

(e.g. had STM32F103s that worked inside a cheap STLink clone but wouldn't reliably work with custom firmware. A fresh chip bought from RS had no issues.)

What kind of reliability issues out of curiousity?

It has been a while but I recall the chip had issues with accepting code over SWD (verification failures and what not) and was dead as a dodo when connected to the PC.

Most likely a reject with defective flash where the original firmware has just fit into the good portion of it and mine didn't.