My chips were ordered direct from TI website (my post folowwed on from Simon7000s post about TI) and they work precisely as specified in datasheet. 😀
Maybe this will put a damper on the conspiracy theories and speculation for a bit. Just received this from Mouser. I placed the order in January.
These will be part of the next Modulus-686 build.
Tom
These will be part of the next Modulus-686 build.
Tom
Good things come to those who wait apparently. The order of 500 NXP micro controllers that I placed in April of last year looks to ship sometime this fall. NXP pulled the delivery date in considerably from the July 2023 they quoted a few months ago.
Tom
Tom
Yes I am also observing similar - that it has become a staple component.
While reading reviews on other magazines and forums I have stumbled upon he following using LM3886
1) Moon River
2) Peachtree
3) Leak 130
4) Cambridge AX35 series (I may be wrong)
5) Guitar amp manufacturers.
there would be more that I have not come across
I paid a total of $80 for 8pcs of LM3886 + shipping, bought them from Micheal Chua.
Using 2 for Mauro’s My ref design. But the 6 are for emergency and other project that is in my mind
Using 2 for Mauro’s My ref design. But the 6 are for emergency and other project that is in my mind
I can only wonder what "feature size" is used for making analog chips, and where they're made and such. Power transistors in power devices are obviously large to handle the heat.
About six months ago I saw a specific ST Arm Cortex M microcontroller priced at $3, but now it's $6 (both Mouser and Digikey), not that it matters, it hasn't been in stock for a year or two and apparently won't be for maybe another year. Oddly, the "Nucleo" board with the chip on it IS in stock. I guess they want to put some in the hands of designers and coders so when quantities do become available to "ordinary" customers they'll have new markets for the chips.
I've heard of chips being redesigned for newer, more modern processes (maybe it's cheaper to make since it uses a smaller chip area, or they need to redesign because the older fab/process is being phased out), reducing feature size while maintaining the same specs and logic functions. I've seen this cause a problem where a newer chip didn't work in a product, because the design (whether the designer realized it or not, often not and it took some work to discover it) relied on the chip pulling a certain amount of current and the new chip pulled substantially less current. Once that's discovered it's easy enough to fix.
All the info I've seen, from not being "in the industry," is on digital chips, as this video shows. I think it gives a good overview of how the current situation is unique in the industry's history. New fabs have always been built with the latest or next technology in mind (and older fabs continued making the larger feature size while there's still plenty of sales volume), but current chip demand apparently exceeds supply for everything, legacy/mature chips as well as the latest. According to this, ALL fabs, old and new, are running at or very near full capacity, and new fabs are being made to increase the manufacturing volume of chips using OLDER feature sizes (as well as the latest), something that has apparently never been done before. And of course making new fabs costs a LOT of money vs. just running older ones, so chip prices are going to rise to cover these costs.
About six months ago I saw a specific ST Arm Cortex M microcontroller priced at $3, but now it's $6 (both Mouser and Digikey), not that it matters, it hasn't been in stock for a year or two and apparently won't be for maybe another year. Oddly, the "Nucleo" board with the chip on it IS in stock. I guess they want to put some in the hands of designers and coders so when quantities do become available to "ordinary" customers they'll have new markets for the chips.
I've heard of chips being redesigned for newer, more modern processes (maybe it's cheaper to make since it uses a smaller chip area, or they need to redesign because the older fab/process is being phased out), reducing feature size while maintaining the same specs and logic functions. I've seen this cause a problem where a newer chip didn't work in a product, because the design (whether the designer realized it or not, often not and it took some work to discover it) relied on the chip pulling a certain amount of current and the new chip pulled substantially less current. Once that's discovered it's easy enough to fix.
All the info I've seen, from not being "in the industry," is on digital chips, as this video shows. I think it gives a good overview of how the current situation is unique in the industry's history. New fabs have always been built with the latest or next technology in mind (and older fabs continued making the larger feature size while there's still plenty of sales volume), but current chip demand apparently exceeds supply for everything, legacy/mature chips as well as the latest. According to this, ALL fabs, old and new, are running at or very near full capacity, and new fabs are being made to increase the manufacturing volume of chips using OLDER feature sizes (as well as the latest), something that has apparently never been done before. And of course making new fabs costs a LOT of money vs. just running older ones, so chip prices are going to rise to cover these costs.
I spent a decade of my life designing analog chips for National Semiconductor and Texas Instruments combined. The processes I worked in were SOI BiCMOS and SiGe BiCMOS processes optimized for analog performance (including RF performance) so the CMOS feature size was large (100-500 nm process nodes).
I never used the minimum size devices. First off because I usually needed much larger devices to keep the 1/f noise down and also because smaller devices are harder to manufacture. Just because the process can make a working 100 nm MOS device with one contact on each terminal 99.9999% of the time doesn't mean you should always push it to do so. I would certainly have red ears if one of my high-performance analog blocks would cause even a 0.01 % yield loss due to some minimum size logic gate used for power-down or something (and, yes, that does happen if you make enough chips!)
I apply the same philosophy in my Neurochrome designs. Just because the PCB manufacturer supports 4/4 mil trace/space and micro vias doesn't mean I should use those features everywhere (or at all). Reliability has value.
Tom
Interesting video. Note that some of the 'fixes' mentioned wouldn't apply to analog chips. You wouldn't be able to run the design files for an analog chip through a die shrink script and magically transfer it to a newer process node. Also, moving from 200 mm to 300 mm wafers isn't a universal answer. A 300 mm wafer lot would contain a lifetime supply of opamps, for example. Building a lifetime of inventory in one shot is extremely risky (and generally frowned upon).
The video didn't mention the shortages of lead frames, neon, and whatever else is used in the manufacturing. Or the backlog created when commercial flights were all but shut down. Some chips are made in the US and then shipped overseas for assembly and final test. That was hard to do when airplanes weren't flying.
Tom
The video didn't mention the shortages of lead frames, neon, and whatever else is used in the manufacturing. Or the backlog created when commercial flights were all but shut down. Some chips are made in the US and then shipped overseas for assembly and final test. That was hard to do when airplanes weren't flying.
Tom
The Asianometry channel (for some reason I've been watching a lot of videos in the last couple of years) has many such videos covering semiconductor technology and international trade. I recall one (or maybe more than one) that discussed the neon shortage. It's incredible how much technology goes into these things, the chips and the machines that make them.
I have a bunch of lm3876tf if anyone needs, pin compatible with the bigger brother lm3886, 56w vs 68w but a good temporary replacement
Are you helping? It's better to sell & make someone happy rather than keeping them in the drawer.
Last edited:
Aaaaaand here's a newly uploaded film from 1983 on semiconductor manufacturing. I'm amazed at how many steps were done manually! At least wire bonding was machine-done. I recall some earlier video, 1970s or 1960s, showing women looking through microscopes doing wire-bonding manually through some movement reduction mechanism.
Last edited:
Interesting that none of the folks handling the finished wafers are wearing ESD smocks, wrist straps, or any other ESD control measure. I suppose they could be wearing heel straps, but I doubt it. That certainly wouldn't fly today. 🙂
Tom
Tom