Best and Worst Component Manufacturers

[IDEngineer]

There is a thread on here regarding an SMPS controller with a lockup mode. A subthread within has been bringing up component vendors that folks love and hate. I thought it might be useful to have a thread dedicated to the topic of Best and Worst component manufacturers.

I'll start by offering a short list, to which I may add later:

GOOD: Microchip. I know they get razzed a bit but I've found their MCU's to work very well. Their analog parts aren't the fastest/etc. but I've had good luck with their jellybean opamps, comparators, and VReg's. Their CAN PHY chips have worked well for me too. Personally, I find their spec sheets and app notes very readable and understandable. And I want to give a big shout out to their technical support... most (not all) of my tickets have been quickly answered, and in one case they set up a conference call with at least half a dozen chip engineers and app engineers that spent at least an hour with me on the phone coming up with solutions.

GOOD: Alpha and Omega. This is actually the one that made me think of starting this thread, because someone on the other thread said they'd never heard of AO. Not sure of their history but I've used their discrete transistors and smaller analog IC's with no problems.

QUESTIONABLE: Virtually every MLCC vendor. My biggest complaint is the inconsistency in their spec sheets. Pick a vendor... some spec sheets will include every parameter and variable you could want, while others will be outright missing things like the graph for capacitance vs. DC bias percentage (very important for MLCC's). I learned this painful lesson due to the industry shortage of MLCC's, which usually causes Production to rope me into qualifying substitutes, and it's hard to even start the process when it seems every spec sheet has its own special recipe of missing information.

GOOD: Linear Technology. Now owned by Analog Devices so we'll see how they fare in the next couple of years, but LT is a classic case of paying a bit more to make sure it works. Only downside is they tend to favor special-function parts (that's where the money is, and they know it) so second sources are often nonexistent. Just depends on how you like to gamble.

GOOD: Vishay for passives. Get out your wallet, but like Linear Technology, you can pretty much rely on their stuff working like they say it does. Vishay is also often a source for exotic parts that few others even bother to make. I remember Jim Williams mentioning a Vishay wet slug tantalum capacitor that cost $400 each (!!!). For fun, I spent some time looking for second sources but unlike the slug, I came up dry! They were also the only source for precision small bulk metal resistors years ago, a rare and almost unknown type that was considered nearly "noiseless" due to how the substrates were formed. I built some amazingly quiet audio preamps with them.

POOR: Texas Instruments. It kills me to give them a negative rating, because I've had great success with their parts in the past. But like that other thread pointed out, of late their components have seemed needlessly complex and, let's be honest, sometimes even difficult to use. And while their spec sheets and app notes used to be second to none - they sought not just to tell you about their parts but also to educate you on the technology and mindset behind them - lately they've deteriorated significantly. It's just not safe anymore to presume that a TI part will be bulletproof, and that's a shame.

OK, that's enough to get this thread started. What component vendors do you love or hate, and why?

[T3sl4co1l]

Putting any huge manufacturer in a list like this is sure to be problematic.

Such companies have a great many departments, of varying quality, between them, and over time.

TI makes a lot of expensive parts that are very good, too.

Conversely, Analog Devices makes a lot of expensive parts, and even some of those are straight up what the fuck.  Case in point, AD7793: read the sections on the SPI interface. https://www.analog.com/media/en/technical-documentation/frequently-asked-questions/AD779x_FAQ_Instru_Conv.pdf It doesn't reset bus state when /CS is deasserted!

What you won't hear about are the hundreds or thousands of parts that work to varying degrees of "well"; bad news travels faster.

Regarding Vishay: their current product line is built from dozens of absorbed brands.  There is no sense in attributing the quality of any one division to the whole.  They make extremely expensive, extremely good parts, like the precision foil stuff, and they also make competitive commodity parts.

Under that perspective, as far as I understand: Vishay as a whole, is mostly built from already well known quality brands, and it doesn't seem like they've sacrificed quality for profit, at least very much.

I suggest applying the same perspective to other large companies, too.

The drawback is, companies not built from brands, will not obviously suggest where their designs or products came from.  Regarding TI, there's Burr-Brown and Unitrode (old, probably not very meaningful experience anymore?), and National (huge portfolio; probably still fairly separate?) which come to mind.  Who knows what part of TI their TPSxxx's and such are designed at?

Tim

[hans]

In a recent board academia-related design, I've found that TI has also caught many "design wins". Especially if you select on criteria like low power they have plenty of analog parts available. I'm not too interested in their MSP430 series though. I think academia has some interest in them because of the learning tools/support, low complexity chips and reasonably low power consumption. There are also some interesting use cases to be thought of for FRAM, like intermittent computing, however you can also question how many computations you can run if you can't even power harvest 2uW of power to keep a modern MCU in standby.

Contrary, I've also looked at AD/LT for low power opamps, and have found some weird parts for sure. Take the LTC6255 for example. Their GBW, voltage noise and other specs are really the best you're going to get under 70 micro amps. The catcha: the input impedance is only 1-10M ohm, with lots of current noise. Obviously that is due to being a bipolar input opamp (instead of the more common FET types), but nonetheless it's interesting to see such a part in the low power domain spectrum.

[mrpackethead]

GREAT: TI.   Amazing customer support and community. Rock solid parts that have not caused me any issues in tens of thousands of power supplies.   Great doco and examples, eval boards.  Easy to source and pricing is realistics.

Good and getting better. Microchip.  Very good at keeping things in Stock.

Poor. Kamakooza whoit.

[Synthtech]

TI unleashed millions of sub-standard 4000 series IC’s and some op-amps onto an unsuspecting world some time ago. They were failing so badly that one company that I do servicing for had to recall huge numbers of large circuit boards and send them back to the board manufacturer in the USA to laboriously have all of the TI chips removed by hand and replaced with chips from NXP. I won’t purchase any TI DIL 4000 series IC’s, particularly 4050, 4051, 4053s because there must still be millions of defective ICs out there in supplier warehouses.

[mrpackethead]

TI unleashed millions of sub-standard 4000 series IC’s and some op-amps onto an unsuspecting world some time ago. They were failing so badly that one company that I do servicing for had to recall huge numbers of large circuit boards and send them back to the board manufacturer in the USA to laboriously have all of the TI chips removed by hand and replaced with chips from NXP. I won’t purchase any TI DIL 4000 series IC’s, particularly 4050, 4051, 4053s because there must still be millions of defective ICs out there in supplier warehouses.

Do you have a link to this event?

[ali_asadzadeh]

I think it would be nice to tell which parts form a brand did you encounter fake parts,at least from china sources.

I have seen fake parts specially from TI, R485 phy's like ADM3485, SN65HV75 , SN65HVD78.

LPC1768 from NXP and ATXMEGA128A1 from ATMEL.

Also I have seen countess fakes of jelly been parts like op-amps OP07,LM358,LM324 from ST and TI mainly. or at least not fulfilling the Datasheet specs.

So I think if you can find some fake parts from a well know supplier in china, it means that part is not recommended for new designs or maybe because it's very easy to fake it in china.

Also I Found FPGA parts from Xilinx and Altera are very good, at least you can not find fakes easily in china

Also STM32 from ST and LPC series from NXP are very good.
TI,NXP,ST,Xilinx, vishay,Yaego, Microchip, analog are my favorite brands. And I like ST because of it's cheapness

[tsman]

Maxim - The parts are generally good but availability can be awful. If you don't want the 1 variant that is stocked by the distributors then you're going to need to order direct in massive quantities. You also can't rely on that 1 variant actually being in stock next time you order.

[palpurul]

I think Linear Technology was very good because they got the best application notes. You can learn a lot from Jim Williams' application notes.

[IDEngineer]

Putting any huge manufacturer in a list like this is sure to be problematic. Such companies have a great many departments, of varying quality, between them, and over time.

Agreed, but I like how this thread is developing. The participants are giving enough detail to differentiate when it's the "whole vendor" vs. "certain good or bad parts". That says something positive about the quality of the folks that frequent this site. Hopefully this thread will evolve to become a useful source of information for all of us when we're considering designing in a new (to us) part.

[mrpackethead]

TI unleashed millions of sub-standard 4000 series IC’s and some op-amps onto an unsuspecting world some time ago. They were failing so badly that one company that I do servicing for had to recall huge numbers of large circuit boards and send them back to the board manufacturer in the USA to laboriously have all of the TI chips removed by hand and replaced with chips from NXP. I won’t purchase any TI DIL 4000 series IC’s, particularly 4050, 4051, 4053s because there must still be millions of defective ICs out there in supplier warehouses.

I did a bit of a google and could not find anything about this.   

[David Hess]

In the past TI has be pretty scuzzy with datasheets which were wrong or outright lied.

[TrickyNekro]

POOR: Texas Instruments. It kills me to give them a negative rating, because I've had great success with their parts in the past. But like that other thread pointed out, of late their components have seemed needlessly complex and, let's be honest, sometimes even difficult to use. And while their spec sheets and app notes used to be second to none - they sought not just to tell you about their parts but also to educate you on the technology and mindset behind them - lately they've deteriorated significantly. It's just not safe anymore to presume that a TI part will be bulletproof, and that's a shame.

I got to agree on that, but really partially. Sometimes information digging happens in their datasheets but that´s up to an extent normal.
What totally frustrated me though was trying to use the ADS1274. The information on the datasheet would really leave you wondering for minutes,
and we had to actually find ourselves how to reset the chip. We had the problem that the power-rails wouldn´t come up with the expected order (they would come up pretty much simultaneously),
and it would output junk data. Found how to reset it and then it was a green light. But nowhere in the darn datasheet would tell you how to do all that.

Apart from that though, all their other datasheets are relatively good and on spot, haven´t used TI parts for a while now.

About their parts, some are top like the best, some eeehhh, let´s say better not have to encounter them. I got cancer trying to repair a BQ24735 based laptop, but like serious cancer...
Probably not the parts fault but datasheet information was downright insufficient for debugging. And the laptop implementation was pretty much textbook also.

Microchip started having very good datasheets, especially their package cads are really top and their library support is good to be there.
Although older parts did not get the treatment, some package layout data are totally unreadable. You can still find them from their site but pfff... It doesn´t take that much to copy paste some pdf files...
Another thing I love about Microchip is how I see that they combine AVR and PIC technology. You see that more in peripherals but darn they try to make the best of it.
Although their datasheets can be very massive. Have not tried many application note up to now, so I will not complain.

AD is winner winner chicken dinner, their analog is top. Never had to think twice when using them. I also used FET drivers and isolators, again no complains at all.
And I like their datasheets, they are not overwhelming, accurate and on point. Not much else to say.

Maxim is also pretty solid. Their datasheets can be a bit overwhelming but nothing out of nightmares. Their part portfolio is a bit of an odd ball sometimes, but if you have the application for them,
well... they can fit in quite nicely. Their 1-W stuff are nice, I don´t know if they progressed as far as they would have liked though. And one thing that pisses me off is that some of their best for
the job ICs do not come in a SPI variant. I2C is relatively slow, especially if you build an RTS. Or else you got to break the execution time between repetitions and that´s a nightmare.

I have used many other manufacturers, but didn´t spent all that much time using their products, or didn´t have to spend a lot of time designing around their products, so I will refrain from saying
anything else.

[David Hess]

AD is winner winner chicken dinner, their analog is top. Never had to think twice when using them. I also used FET drivers and isolators, again no complains at all.
And I like their datasheets, they are not overwhelming, accurate and on point. Not much else to say.

I like the older AD, National, and LT datasheets which showed equivalent or complete schematics or at least input and output structures because some characteristics are just not listed.

Which way do bias currents flow?  I have had to disqualify several TI parts because I had no idea, no way to find out from the datasheet, and TI was more interested in why I wanted to know than telling me and I would have had to verify it anyway.

AD's web site really pisses me off; why would I possibly want to sort by part number?  I guess it is a good thing that I cannot.  TI's site hides low cost parts in favor of high cost ones.  Microchip does not properly list parts from their acquired businesses; who knew they make depletion mode MOSFETs now?

[TrickyNekro]

AD's web site really pisses me off; why would I possibly want to sort by part number?  I guess it is a good thing that I cannot.  TI's site hides low cost parts in favor of high cost ones.  Microchip does not properly list parts from their acquired businesses; who knew they make depletion mode MOSFETs now?

Every site has it´s quirks, sometimes it´s a thing, sometimes not so much... But I accept I got to do some digging when I go to a manufacturer´s page.
What I can not accept in any way, shape or form, is that when I go buy something from suppliers, like Mouser for example, they do NOT have price listing in quantities!

Like I am going to buy 1 diode or 1 IC?! FFS really... unless I am missing something really badly. But you can not select price listing in the, for example, 100 parts quantity.
Un-fffffing-acceptable. Unless I am wake-sleeping or something, checked some days ago, still nothing in the obvious field of view.

[hans]

I just remembered 1 part manufacturer..

Broadcom

You won't get their top spec parts without signing a NDA first.

But IMO it gets worse. Turns out they EOL'ed some very niche neat parts that originated from Agilent semiconductors. In particular I'm talking about zero bias schottky diodes, which were very commonly used in RF energy harvesting and detectors, as they had excellent biasing characteristics but above all the datasheets showed what kind of about amplitude you can expect given a certain signal power input. This is what I find particularly lacking in other RF schottky diodes that are advertised for detector applications.

[MD Kowshik]

POOR: Texas Instruments. It kills me to give them a negative rating, because I've had great success with their parts in the past. But like that other thread pointed out, of late their components have seemed needlessly complex and, let's be honest, sometimes even difficult to use. And while their spec sheets and app notes used to be second to none - they sought not just to tell you about their parts but also to educate you on the technology and mindset behind them - lately they've deteriorated significantly. It's just not safe anymore to presume that a TI part will be bulletproof, and that's a shame.

I personally hate the cheapo Chinese capacitor brand ChongX ...man their capacitors are flooding the markets and spreading like cancer.
There just bad they blow up at 60% of their rated voltage and Bulges at 70~80C temperature...
And Yes Ti is slowly but surely starting to move away from usability by the general public to complicated and difficult to operate and often requiring many additional components than other chips which do the same thing but with way little components and way easily.

[SiliconWizard]

Regarding TI, of course they have bought off many companies over the years, so it would also depend on what components you're talking about. Always had good success with former Burr-Brown devices for instance, and their direct TI successors. TI is not quite a "monolithic" company.

I've already had troubles with some TI LDOs though, some of which happened to have pretty poor line and load regulation, certainly worse than what was "advertised" in the datasheets.

Some of the GOOD: Analog Devices and LT, Microchip, SiLabs, Nordic, ST, On Semi. Absolutely not exhaustive of course.
GOOD SURPRISES: for power supply ICs: Torex, MPS.
The UNEVEN: ALPS. Can be very good or shitty. Hard to know beforehand unless you already know the product.

[coppercone2]

The horrific: unitrode

Why? They are called unitrode.

My biggest fustration is with atmel 8bit mcu.


I really cant complain aboutelectronics parts. So long they dont need code its groovy. If it fries you replace it. At least a capacitor error wakes you up... unlike the compiler which will put you into your eternal dirt nap with its errors

I find the deficencies and manufacturing tradeoffs very interesting.

I have yet to feel the urge to tune up a breadboard unlike a fucking pc. With my fists

[Gibson486]

I think they all have their fair share of good and bad. My fav was always Linear Tech because they had fabulous sales support. That has changed with AD buying them. Now everything is done through their online forums which is a mess, but I guess it was only appropriate that an expensive IC company bought another expensive IC company. TI, on the other hand, has a great forum and the people who respond do a good job.

[Gibson486]

AD is winner winner chicken dinner, their analog is top. Never had to think twice when using them. I also used FET drivers and isolators, again no complains at all.
And I like their datasheets, they are not overwhelming, accurate and on point. Not much else to say.

I like the older AD, National, and LT datasheets which showed equivalent or complete schematics or at least input and output structures because some characteristics are just not listed.

Which way do bias currents flow?  I have had to disqualify several TI parts because I had no idea, no way to find out from the datasheet, and TI was more interested in why I wanted to know than telling me and I would have had to verify it anyway.

AD's web site really pisses me off; why would I possibly want to sort by part number?  I guess it is a good thing that I cannot.  TI's site hides low cost parts in favor of high cost ones.  Microchip does not properly list parts from their acquired businesses; who knew they make depletion mode MOSFETs now?

Microchip makes a lot. They actually make good op amps, for half the price. Unfortunately, everything is a 5V or less part.

[Gibson486]

Maxim - The parts are generally good but availability can be awful. If you don't want the 1 variant that is stocked by the distributors then you're going to need to order direct in massive quantities. You also can't rely on that 1 variant actually being in stock next time you order.

When I worked for a large headphone company, I was instructed to never use Maxim parts because of that very issue.

[T3sl4co1l]

TI, on the other hand, has a great forum and the people who respond do a good job.

Either you've gotten lucky, or you asked about a fairly common problem, with a common answer.

I've never had a single question that was answered satisfactorily, typically taking three or four back-and-forth responses to insist on what it is I'm after.  Regardless of how much background, scope, detail or analysis I put in the original post.

So they can't read.  I doubt they are paid well enough to, anyway.  (If they were worth it, they'd be in the design department.)

Tim

[SiliconWizard]

TI, on the other hand, has a great forum and the people who respond do a good job.

I've never had a single question that was answered satisfactorily, typically taking three or four back-and-forth responses to insist on what it is I'm after.  Regardless of how much background, scope, detail or analysis I put in the original post.

Same here.
And I've seen countless threads ending up that way, some even not answered at all.

The best support I've seen on the other hand was with On Semi. Very qualified application engineers, and you didn't even need to be a big company to benefit from that.

[bloguetronica]

Personally, I had a few bad experiences with TI. One was when using the TLV76050. This regulator tends to fail when outputting a bit too much current. The other one I had was when using TPS55010. That one would fail even when idle, starting to consume too much current and eventually burning out. Lets say that I've bought the evaluation board to no avail. In total damaged two parts. Even though, TI chips are generally good.

With Analog Devices, I had several issues and questions when reading the datasheets. I had to suggest corrections regarding the SPI clock phase and polarity in the AD5932 datasheet, and also did the same regarding other two products (but those were never accepted). The most recent problem I had was with the AD9102 datasheet: it is terse and rather cryptic.

On the good side is Linear Technology. Their components are very good and the datasheets quite complete. I didn't used them before because, before the merge with Analog Devices, their components only existed on some distributors.

Kind regards, Samuel Lourenço