Maxim love 'em or hate 'em?

[grouchobyte]

Is it just me or do others also have a love-hate relationship with Maxim.

Don't get me wrong I think Maxim makes some great stuff. But everytime I spec in one of their parts, I get samples and protos built and then the problems begin. You often can't get their stuff unless you buy huge quantities or wait for the incredible long  lead times sometimes dragging on for years after product ship release. I have been burned time and time again over the past 30 years and so I rarely consider them in production designs.

Also, it seems many other engineers  and those responsible for obtaining their parts moan when I even suggest Maxim. I usually get a response like...."are we desparate? Isn't there an alternative?.."

Is this an international phenomenon or are those Jack Gifford days over? I can rely on Digikey and Mouser if they somehow have the supply issues under control?  Also, it seems that unless you are a big OEM, the local sales engineers for Maxim wont give you the time of day. In my experience  TI, Linear Tech and ADI are much easier to do business with. Opinions and rants welcome! 

[floobydust]

I have been burned by Maxim's lead-times, and they won't advise product life-cycle or lead-times if you ask them. I see them as high risk to design in.

Once I had to phone them and beg, support asks me "Are you line-down?", I'm like uh yup and they sold me a reel to get the (SMT pick'n'place) line going and the build done. Otherwise 42 weeks lead-time on the part. Brutal.

Lately, I used their thermocouple-to-digital part in a design and it was a disaster. It is superb at picking up RF with readings all over the place. It also has silly fault checking, like a thermocouple shorted to Vcc. That happens none of the time. The parts are not practical in real-world applications, but look nifty.

[Psi]

I use one of there automotive reluctance sensor amp ICs in a product. Its not cheap but works so damn well and means i dont have to stuff around making and testing an equivalent which probably wouldn't work as well anyway.

About 3 months into product release the price doubled to $9 a chip

Thankfully the margin on the product is huge so it didnt matter so much.

[exmadscientist]

Yep, I can't stand those guys either. Maxim are definitely a supplier of last resort for me, even though I only deal in prototype/very low volume designs these days and don't face the challenges of production sourcing. I'll only design them in in one of two cases:

• They're the only game in town
• It's a jellybean part with multiple drop-in replacements readily available

Case #2 covers things like op-amps in standard packages. Maxim does have some nice low-voltage analog parts, and if I grab one in a standard package, the worst that can happen is I'd have to design-in the runner-up I'd have used anyway if I blacklist Maxim.

A couple examples of case #1: as far as I know, the DS1374 is the only binary RTC readily available. (Anyone know of a substitute? I'd love to drop it! I²C interface and battery backup are mandatory, square wave output is nice-to-have.) I have also used their ModelGauge battery meters (MAX17048) in a design with external battery packs, for which most battery meters are unsuitable (Coulomb counters + hot-swappable batteries + mixes of different cells = non-working fuel gauges). That design revealed some classic Maxim... trademarks:

• Silly selection of available packages
• Sillier prices and distributor lead times/availability of the preferred compromise-choice package, leading to picking the second-choice package
• Slightly optimistic spec sheet (though not absurdly so... just a little bit)
• Stupid design choices (why have three signal pins with two mandatory connections to power and ground? Why not two power pins and one signal pin, if the design really needs them? Why have multiple power pins at all for this "micropower" device? Why are the 1-cell and 2-cell models completely opposite in how they're powered? Why can't I power this device externally, only through the battery it's supposed to be measuring? That battery might not be present! Why do I need an I²C multiplexer to combine two of these chips on one bus? You had spare pins! Why did you allocate a pin to a mostly-entirely-useless feature that can destroy device accuracy if misused (QUICKSTART, which the datasheet basically says never to use...)?

I could go on, but I think my feelings are clear by now

Maxim: only if I really, really have to.

[AndyC_772]

I'll stick up for Maxim here.... credit where it's due.

The MAX16910 is by far my favourite linear regulator. I do a lot of automotive designs, so the low (20uA) quiescent current and ability to withstand 45V on the input are incredibly useful. It's rated to a junction temperature of 150C, and the output voltage is accurate to 2%.

It's also available in either an SO-8, or a tiny 3x3mm leadless package.

Yes, it's expensive, but worth it IMHO. I've never seen one fail, and that's worth a lot.

If anyone can suggest an alternative that's as well specified, I'd love to hear about it - otherwise it's Maxim all the way, at least for this application.

[tszaboo]

Err, no. Sorry but you can only screw up so many times before it bites back. I had components going NRND and obsolete is 2 months, the same 2 months the design was made. And if you designed in something, there is zero guarantee that the chip is going to be available years from now. There is something they make, which would be a perfect fit, I rather re-design it just to avoid maxim.  I was genuinely surprised when I see that the uCurrent uses maxim chip. Even though it is a low offset jelly bean opamp, that can be replaced when they decide to screw up.
I've heard that they change their behaviour if you order seven digit numbers. Sorry I dont.

[nctnico]

I guess the trick is to use parts they sell a lot. I'm also not inclined to design in parts made by Maxim but that also had to do with the not-so-good distributor they had in the NL. Nowadays I check part availability from Farnell, RS, Digikey and Mouser to see whether a part is somewhat popular or not before deciding to design it into a circuit. BTW I have designed in a reference made by Maxim recently.

[Kjelt]

Recognize the story, ten years ago that was common in our company also only use if no other choice or replacements are available.
However in the last 4 years my company has designed in some Maxim chips in products that run in huge quantities and that part has no replacement, to my big surprise.
I asked the HW architect and HW designers why and how, because I was curious if something changed.
They told me that now they set up a contract in front about quantities to buy and that should be delivered just in time for production or Maxim faced huge penalties (cost of not being able to produce and sell).
I guess if you can make a contract in front because you want to buy large enough quantities of their chips, much more is possible than if you are in a smaller market.

[Gyro]

Just had a strange feeling of deja vu...

https://www.eevblog.com/forum/projects/do-you-avoid-maxim/msg677525/

[grouchobyte]

I must say, it sure is nice to know I'm in good company, especially from the international opinions thus far.

FullRantMode=ON

Maxim is in fact the last resort for me and I only spec them in when its a jelly bean part in a jelly bean package and available in bizzilion quantities, for a very long time by Digikey and Mouser and can be substituted in a new york minute by the likes of LTC, ADI or TI

Engineers that use Maxim as a first choice are just not old enough to have been fu**ed over by them.
Tempting as it is to use their unique, proprietary parts, I often just walk away shaking my head mumbling the silent curse.....no way, don't do it......

Of course, if you are a huge OEM like General Motors or Apple that gives them lots of money, you can be assured a lifetime of unwavering support.

The little guy gets the fuzzy end of the broomstick when it comes to Maxim everytime, and I hear these stories and warnings constantly from the community. Indeed, even the Maxim rep has acknowleged that the company takes the high road when it comes to premature obsolescence or sudden availability issues.

When is a decent company like TI or ADI, NXP, Microchip, Micrel or, heaven-forbid LTC,  gonna cash out Maxim and make our lives a little easier? 

FullRantMode= OFF

@grouchobyte;

[Kjelt]

What I did for my own hobby project is after the testboard, I buy the ic's first in 50's quantity (enough to last your need) then start the pcb design etc. etc. In that way you know you are covered.

[mikeselectricstuff]

Unless there is plenty of stock at multiple distribitors, or it's a one-time build that there is stock for, Maxim is pretty much bottom of the list. They've been terrible for low-mid volume users for as long as I can remember. Just not worth the risk.

[Philfreeze]

Pretty much the same problems for me.
Though I have to say their application notes are usually pretty good and they can give you some nice ideas.
Another minor problems is that I always forget that I can't just google "Maxim" and should use "Maxim integrated" instead.
I hate it when I do it wrong and Google shows me the magazin Maxim... 

[Wilksey]

I love Maxim for their list of incredibly useful products and useful featured products, but hate them for not being able to bloody get the parts from the standard suppliers, or, as many have pointed out, want to buy 2 million and wait 4 months.

In the past I have found that the Maxim parts will do exactly what I want, but had to go elsewhere as the parts were not available, even Maxim don't know, but they will sell me 500,000 direct and sample me some. 

[chris_leyson]

Did a brownout test on a small PCB to check that the microcontroller was doing the right thing and the Maxim Bimos or Cmos op amps let the magic smoke out.
Couldn't get their low noise ultrasound amplifier chips to work either, not a great fan of Maxim but still use MAX232s occasionally.

[technix]

I usually avoid MAXIM except a few parts:

* DS18B20: There are a few corner cases where a digital temperature sensor with as little lines as possible make sense. Otherwise I will always stick to LM35.
* DS3231: Those ultra high precision RTC chips are really good, software compatible with the popular DS1307, and there are an absurd amount of NOS available in China. Once those NOS run out I may have to look somewhere else. (DS1374 users, why not switch to this, at least this will survive longer, and you can design your board accepting DS1307 + crystal, PCF8561 + crystal and DS3231 at the same time)
* MAX3232: When you need RS232 this is indispensible. Although I am currently looking into using signal isolators and isolated power supply to replace them.

[nctnico]

Fortunately the MAX232 and MAX3232 have lots of second sources!

[Wilksey]

I quite often substitute for SIPEX / EXAR 232 driver chips as they are a lot cheaper and work just as good, and ST RS485 driver chips.

[uncle_bob]

Hi

I really wish I had video of each time I've seen:

Them: Hello, my name is XXXXX and I just hired on with Maxim. I'm here to tell you what a wonderful outfit they are.

Me: How about last month when you *still* had us line down because after 12 months you *still* could not source the YYYYY.

Them: Those problems are all behind us and we have learned our lesson. I can look into the YYYYY's for you. I'm sure we can get them to you next week.

Me: No need now, it' took us 10 months to complete the re-qual on the part and the new design just went into production with another part.

Them: Can I show you all our great new parts for you to design in?

Me: Care to pay us the $20,000 the re-qual cost?

====

The Maxim guy is showing up at the plant tomorrow ...

Bob

[Wilksey]

Yeah, that's the problem I get with some of my work, can't wait around months for the manufacturer to "cave in" and finally supply you with parts, sometimes you are better off sourcing perhaps a slightly worse specced alternative that is at least available from the major stockists, and it's all well and good some sales rep saying "oh we now have product x, y and z that will work 10 x better than the product you are currently using, why not swap it out", yeah, not that simple matey, I'm sure they would love to blow their sales budged on re qualifying products! Not! 

[exmadscientist]

(DS1374 users, why not switch to this, at least this will survive longer, and you can design your board accepting DS1307 + crystal, PCF8561 + crystal and DS3231 at the same time)

Uh, because they're not compatible? DS1374 is a binary RTC, typically used for storing something like a Unix timestamp. (As everyone who's done enough working with dates and times will know, that's about the only format robust enough to survive the myriad insanities of timestamping things in the real world.) Almost all RTCs store a BCD-encoded version of human-friendly local time. This is great for displaying to humans or for interfacing with a brain-damaged 8-bit CPU, but really rather dreadful for doing computations. Not to mention the indignity of time zones -- some devices don't stay put in just one time zone.

The DS1374 and friends allow one to simply read out the 32-bit count of seconds since the epoch, neatly sidestepping much of the mess. They're rather more difficult to substitute than I'd like.

* MAX3232: When you need RS232 this is indispensible.

Hardly -- last I looked, ADI had a full line of RS232/422/485 parts, with better specs to boot.

[coppice]

* MAX3232: When you need RS232 this is indispensible. Although I am currently looking into using signal isolators and isolated power supply to replace them.

Huh? I can't imagine a part more easy to replace, from numerous sources, than that part.

[sarepairman2]

keep in mind this company is named after the man who made WWI ultra lethal and set the stage for modern warfare.

[coppice]

keep in mind this company is named after the man who made WWI ultra lethal and set the stage for modern warfare.

I expect our local Maxim's bakery was named after him too.

[Wilksey]

Well, if you're a real cheap skate you could replace the 232 with a few trannys!