4
\$\begingroup\$

Very common ICs are sold by many manufacturers, for example let's say the TL07x range (which is vintage, but does demonstrate the case).

Inspecting datasheets of two manufacturers, TI and Diodes Inc, I find that they spec the same typical THD (0.003%) and noise (18nV / sqrt(Hz)).

What differentiates these devices? Is the silicon design simply licensed to both of these companies? Or do they purchase the device from the same fab house and simply stamp their individual brands on them?

\$\endgroup\$
  • 1
    \$\begingroup\$ Probably a bit of Column A and a bit of Column B. \$\endgroup\$ – DKNguyen Oct 13 '19 at 23:54
  • 3
    \$\begingroup\$ It's not against patents to make a device that achieves the same results as the patented device. Imagine if noone could make headphones, because someone patented something like "a device which uses electric current to produce audible sound waves". Patents are supposed to be about very technical and specific things (though they are often abused) - e.g. a manufacturing process to make a particular kind of transistor for an integrated circuit, not about things like "AND gate". Keep in mind the justification for patents used to be "give the inventor time to build a factory to make those things". \$\endgroup\$ – Luaan Oct 14 '19 at 9:35
8
\$\begingroup\$

Most likely one company came up with the design first.

Then other companies reverse-engineered the design to sell copycat parts.

Some customers might even insist that a second source of supply be available for such parts (linear regulators, jelly bean transistors, basic logic gates, etc) so it wouldn't help the original vendor to try to stop the copy-cat from selling their version of the part.

Is the silicon design simply licensed to both of these companies? Or do they purchase the device from the same fab house and simply stamp their individual brands on them?

Neither of these are very likely.

These are normally very low-cost parts, so it's unlikely anybody is paying license fees to anybody else for them.

And the margin on these parts is so low that they are really only profitable for integrated device manufacturers who can produce them at very low cost, rather than fabless companies.

What differentiates these devices?

There is likely to be some difference in performance between the parts from different vendors, if you examine the behavior beyond the guaranteed datasheet specifications. For example, the distribution of some parameter over 100's or 1000's of units may be different, or the way some parameter drifts with temperature may be different.

But the main things that drives most purchasing decisions between cross-referenced parts are non-technical issues such as:

  • Price
  • Vendor-customer relationship
  • Availability
  • Most of all, price
\$\endgroup\$
  • 2
    \$\begingroup\$ The behavior distribution curves may very well be different since they are likely using different FABs and different means to their madness in making the parts. It's very unlikely that the distribution shape and spread over 100k parts would be the same between two manufacturers. I've seen instrumentation that was designed, after extensive testing and discussions with supplier companies, to only work with one manufacturer and only for the one FAB they are using at the time to make them. Sometimes, it's worth the trouble. (A case for a specific reed relay part from COTO springs to mind, too.) \$\endgroup\$ – jonk Oct 14 '19 at 2:07
  • \$\begingroup\$ Not just for very low cost parts but some complex ICs as well. ARMv2 instruction set for example has been fully reversed engineered and all patents relating to it has expired so there are several open source designs of it. Not to mention China has developed several closed source designs of patent-free ARM CPUs for domestic use especially for their military/space applications. And then there's AMD (and formerly Cyrix) and other Intel x86 cloners such as DM&P, Advantech etc. \$\endgroup\$ – slebetman Oct 14 '19 at 8:50
  • \$\begingroup\$ Note that there's two things "reverse engineering" may mean. One is designing your own device that follows the specifications of the original device (having to follow the actual behaviour of the device if the specs aren't available or accurate - otherwise it's not reverse engineering). This is fine, and indeed many countries give specific protections to people doing this. The other is analysing e.g. the actual layout of the device and copying it in your own. This happens a lot more rarely, though one rather glaring example was the Eastern block copying Western chips verbatim back in the day. \$\endgroup\$ – Luaan Oct 14 '19 at 9:38
  • \$\begingroup\$ Note that there can be slight technical differences in the part behaviors, as well. Not just price. See this answer on another question, for example: electronics.stackexchange.com/a/242654/107479 \$\endgroup\$ – dim lost faith in SE Oct 14 '19 at 9:43
  • \$\begingroup\$ @slebetman, I know lots of examples of MCUs with the same core. I don't know any that are truly form-fit-function replacements with the same pin assignments, same peripherals, same memory, etc. Can you cite any specific examples? (I won't be surprised at all if there's some Chinese vendor with a knock-off of a popular MCU from a western vendor, but I've never gone looking for one myself) \$\endgroup\$ – The Photon Oct 14 '19 at 16:08

Not the answer you're looking for? Browse other questions tagged or ask your own question.