Buying all kinds of stuff on ebay, I accidentally stumbled upon the OPA627. This is a rather old, popular and high-quality operational amplifier, which audiophiles sometimes use in amplifiers (and not only). However, I was embarrassed by the price - from the manufacturer (Texas Instruments / Burr Brown) they were priced at $ 16-80 (depending on the version), and on ebay they were sold at $ 2.7, including shipping.

It is immediately clear that there is something wrong, but it was interesting to know what it was. Ordered 1pc, and for comparison - OPA627 in a metal case for $ 5, clearly from somewhere dropped out. The results of comparing the viscera - under the cut.

Received ICs

')
The plastic case had to be poisoned , and the metal could just be cut down:

Compare

First the original. You can see at least 4 resistors cut by laser - it becomes clear where this cost comes from. Laser-tuned resistors are needed because in the production of the chip, the parameters of the components “float” a little, and in the operational amplifiers they need to be coordinated very precisely.


"Chinese" chip for $ 2.7. The laser-tuned resistor is only 1, but the AD (Analog Devices) and B744 inscriptions are confusing - probably the model number. If we open the datasheet on the AD744 , then we will see a complete match in the photo of the crystal shown there.

What happened?

The Chinese thought it over, and found a cheaper replacement for the OPA627 - that turned out to be the AD744. The AD744 has a similar operating speed (500 ns up to 0.01% accuracy), the same type (* FET), and an external offset voltage correction is supported. The AD744 also has the possibility of external frequency correction - this is necessary for high-frequency applications with high gain, but this conclusion is simply not brought out (since it has no analog in OPA627).

At the same time, the AD744 is worse in noise (3 times) and bias voltage (0.5mV vs. 0.1mV).

The Chinese were not lazy, bought the AD744 in the form of crystals, packed and labeled as OPA627. I'm not sure that they made a lot of money on it - this is more like an economic sabotage. But they could have thrust something at the level of LM358 - then it would be much easier to notice a dirty trick ...

So be careful and carefully monitor the suppliers of the component - otherwise everything will be “optimized” for you :-)

Ps. See also the previous series with the fake FT232RL .