What is happening with Maxim?

[peter-h]

I've designed in a load of their chips over the last 20 years.

MAX232A, MAX3089, MAX3232, MAX489, MAX706...

The first few have gone from £0.40 (10k+) to over £2. And this isn't just the current chip bubble. Maxim have been raising prices continually.

The MAX232 is multi sourced of course but the "classic" MAX232 needs 1uF caps, while the Maxim one needs just 100nF. Should have though of that at the time

The MAX3089 is not multi sourced AFAICS, even if you just want 115k speed.

The MAX3232 is also made by TI and as a result the price has not gone up

The company has also cut itself off from normal means of communication. You have to raise a ticket on their website, and they reply to the ticket at some later point. It's weird. No "normal" comms.

[capt bullshot]

I've designed in a load of their chips over the last 20 years.

MAX232A, MAX3089, MAX3232, MAX489, MAX706...

The first few have gone from £0.40 (10k+) to over £2. And this isn't just the current chip bubble. Maxim have been raising prices continually.

The MAX232 is multi sourced of course but the "classic" MAX232 needs 1uF caps, while the Maxim one needs just 100nF. Should have though of that at the time

The MAX3089 is not multi sourced AFAICS, even if you just want 115k speed.

The MAX3232 is also made by TI and as a result the price has not gone up

The company has also cut itself off from normal means of communication. You have to raise a ticket on their website, and they reply to the ticket at some later point. It's weird. No "normal" comms.

For some bad experience reasons, up to 20yrs ago, Maxim isn't on the list of suppliers for me anyway. ADI has bought Maxim (got that official announcement from ADI a few days ago), so the consolidation to a few Mega-Suppliers is going on. I'd reckon with further increasing prices.

[mariush]

Yeah, I was looking at their parts for LED digit drivers ... but the prices were high and stock quantities low and people all around were complaining about availability.

A bit off topic but it makes me wonder, for chips like MAX232... why not put the ceramic capacitors on the package, i imagine the die size isn't as big as the SOIC package and 0201 or 0402 ceramics aren't that thick and I doubt a lot of people would complain about the part being thicker/taller.

Zeptobars has a die shot of TI's MAX3232 and it's 3.1mm by 2 mm : https://zeptobars.com/en/read/max3232-ti-rs232-transceiver-lvttl 

[Gyro]

Try complaining to AD, they seem to be convinced that the takeover is complete and all is well with their world. 

From an email a couple of days ago...

[Terry Bites]

Mismanagement as usual. I imagine AD will go in for some deep rationalisation of maxims product line up and a boot in the pants for max marketing.

[peter-h]

Some of the above is incorrect. Maxim used to be "cheap". I used to pay 40-50p (a few k +) for most of those parts, both via distis and direct from Maxim, and delivery was never a problem. The rot / customer screwing started about 5 years ago.

Maybe some of their less common parts were hard to get. I just never used them. I have bought some 0.02% references from them but only for prototyping / test gear, and anyway lots of people make references. I used their LED drivers many years ago (MAX7218 - used to be Siliconix) but they are now very expensive and for a new design I used the STM STLED316.

Re MAX232 caps in the package: you can't integrate 100nF caps. And much smaller caps need a much higher frequency if you want to get the required power, and you do need quite a lot of power because a MAX232 is driving a ~3k ohm load at the other end, with a -8V to +8V swing.

Actually a MAX3232 looks like a drop-in replacement for a MAX232, albeit with a slightly reduced (-5V to +5V) swing. And the TI 3232 is much cheaper than the Maxim one.

Maxim Marketing ought to be shot. Terribly arrogant.

[radar_macgyver]

MAX3089 is not multi-sourced, but the TI (SN65HVD33DR) and Intersil (ISL3176) have the same footprint similar functionality. I used the 3089 mostly for the half/full duplex switching, though I found that I rarely need to switch that often enough to justify the expense and constant lack of availability.

[Bassman59]

Try complaining to AD, they seem to be convinced that the takeover is complete and all is well with their world. 

The takeover is complete from the corporate-ownership perspective but none of the ordering and support systems have been merged.

I think that everyone's hope with this takeover was that Maxim's notoriously-shitty product availability policy would end.

For those who don't know: Maxim's catalog was always full of interesting and potentially-useful devices, with several variants and all that. But! Most of the product in the catalog had to be ordered in quantities of a fab run. They were simply never made available through distribution or at least in smaller quantities (reels, trays). So the lesson was that unless you were a Big Time OEM, Maxim didn't want your business.

That lesson was well learned, and many engineers and purchasing departments have a simple "No Maxim" maxim. The Maxim parts we use, like the RS-232 chips, are all widely second-sourced so we buy the TI versions.

The question ahead is what will Analog Devices do? Will they trim the product line and get rid of all of the "order by the pallet-load" variants? Will they make the variants more easily available? I don't think Linear Technology parts and availability suffered from ADI's ownership.

[SilverSolder]

If Maxim's limitations with respect to product availability was mostly due to them not having sufficient capital to put them in production, perhaps ADI can inject some new life into the range and make a success of it.  Here's hoping!

[SiliconWizard]

The question ahead is what will Analog Devices do? Will they trim the product line and get rid of all of the "order by the pallet-load" variants? Will they make the variants more easily available? I don't think Linear Technology parts and availability suffered from ADI's ownership.

Not so far, but LT didn't have the problems Maxim did, to begin with. I'm afraid AD is more likely to "clean things up" with Maxim...

[peter-h]

That's right; LT were overpriced to start with

LTC491? Take a look at the price now... and it's always been pricey. I went to the MAX489 way back, which was about 1/3 of the LT price.

LT was never a good choice if you could get the same part from TI or Nat Semi, etc.

[dmendesf]

Re MAX232 caps in the package: you can't integrate 100nF caps.

He talked about integrating into the package, not the die... And it was done. Look at the max233.

[peter-h]

That practice always produces expensive solutions though. You can get 100nF caps for a fraction of a penny.

Is there a cheap version of the MAX232A ("A" meaning 100nF caps rather than the "classic MAX232" 1uF)? EDIT: MAX202.

I reckon TI chose to make the MAX3232 because it replaces the MAX232A, as well as serving as a 3.3V or 5V VCC device (MAX232/A needs 5V).

[tszaboo]

The question ahead is what will Analog Devices do? Will they trim the product line and get rid of all of the "order by the pallet-load" variants? Will they make the variants more easily available? I don't think Linear Technology parts and availability suffered from ADI's ownership.

Not so far, but LT didn't have the problems Maxim did, to begin with. I'm afraid AD is more likely to "clean things up" with Maxim...

If they start by cleaning up their management, than everything will be OK. Honestly, I will wait a year or two before deciding on buying any parts that start with MAX, and ask a few distributors how it is going.

That's right; LT were overpriced to start with

LTC491? Take a look at the price now... and it's always been pricey. I went to the MAX489 way back, which was about 1/3 of the LT price.

LT was never a good choice if you could get the same part from TI or Nat Semi, etc.

Is it though? Most of the times LT parts are chosen, because nobody else even comes close in specification. And yes, they can ask 5 USD for a single opamp, but you only choose it when because of that opamp, you end up with a better product.
And yes, TI ADI was catching up in performance with a better price, sometimes. Like how their laser trimmed and chopper opamps can provide similar offset voltages, than a LT part 15 years ago.

[peter-h]

Certainly, high spec analog parts were never cheap.

But also that market was not really price sensitive. How many cheap products are there which need a 1ppm reference? Or even an autozeroing rail to rail op-amp? I've just done a design with a TLV23333 and that costs enough... One could use an OPA388 for an even higher spec. There are lots of choices...

But most of the Maxim product line is not high spec analog parts. It was cheap ordinary parts, on which they are now blatently screwing everybody who designed them in. I don't know why but it could be to fatten-up the P&L accounts for a takeover.

[Doctorandus_P]

I refuse to design in any maxim (They don't get a capital M from me) products, unless there are readily alternative pin compatible alternatives, (such as for their analog switches). And even then I don't like to spend any money that may end up in the company, and rather use that other chip, which is probably also cheaper.

They have a long history of designing chips that fit well in niche markets and sell them at inflated prices until they're so expensive nobody wants to buy them anymore and then they obsolete the chips.
This has happened so often that I have strong suspicions it is a deliberate policy, although I must admit: If you design chips for small niche markets, then those markets won't last long.

I bought a few of the 8*8 matrix LED's with chinese clones of the MAX7219. It's a very simple chip, it does not even do proper dimming, (What you'd expect from a decent LED driver) so automatic adjusting to lighting conditions is out. And for the price they sell the original chips for, you can get a dedicated microcontroller and some extra circuitry inclusive assembly costs.

--- 8<------ 8<------ 8<------ 8<------ 8<------ 8<------ 8<---
About the Chinese MAX7219 clones. They're fun cheap displays to toy with, but the chip seems very sensitive to "latch up". When you add an external P-MOSfet for their power, then at startup, drive all data connections to the chip low and only then open the power MOSfet to apply power to the IC it works a lot better.

[Hydron]

General rule of thumb for myself and most other EEs I know is to only use Maxim where there is either no choice, or where you can drop in an alternative with little/no effort.

Unfortunately I've recently had the displeasure of being forced to use a poorly documented proprietary interface IC from them that doesn't even exist according to their website - had to have the distributor going into bat for us to even get a datasheet, and then again to be allowed to buy it (taking a couple of months total). Then of course after accepting the order (with a 3 month lead-time) they told us that it would be an extra 6 months to get the parts - will be nearly a year total to get them in-hand, and we had to pay double to source some through alternative (thankfully reliable) means for initial production.

Fuck Maxim.

[peter-h]

Yes that's all true today. Nobody would design-in a Maxim device today. It's just completely pointless. Well, unless you are (a) working for somebody else and (b) you totally don't care about the company.

But why did Maxim change?

MAX232A, MAX3089, MAX3232, MAX489 etc were GBP 0.40 or so for many years. Now look at them.

[CaptDon]

Very sadly, a lot of those very nice functional packages that were very innovative but only available when ordered by
the thousands with long lead times will simply vanish because narrow minded 'business managers' will look at prior
sales and come to the mis-guided conclusion that the item was not a hot seller and delete it from the line-up. That
'Business Management Degree' insures the decision makers will be to stupid to realize the item would be a hot seller
if available to the general population by onesies, twosies, or 10 at a time. The U.S. tax laws punish manufacturers for
having surplus unsold stock on hand!!! I have personally seen hundreds of thousands of dollars of test equipment,
parts and circuit boards declared obsolete and crushed or ground up and destroyed as a way to 'write down' tax
burdens. You can't spell garbage without GE and they pay almost nothing to the communities where they operate,
even if they operate under the name Wab...  A perfect example of tax corruption, they tore down several buildings
which were taxable dwellings and now rent storage trailers in place of where the buildings stood!!! I doubt the Maxim
part availability situation will change for the better.

[peter-h]

Certainly anybody who designs in something like this
https://www.maximintegrated.com/en/products/power/led-drivers/MAX25606.html
needs their head examined, because that function is trivial to do with a bit of hardware and you won't be dependent on a totally obscure single sourced chip.

[peter-h]

This has been quite an interesting exercise.

The Maxim MAX232A is rather pointless. TI do a MAX202ID which is similar price to what Maxim used to sell at. They also do a TRS202 which seems a very similar part - hard to tell the difference and probably at/below 115k there isn't any.

[Kasper]

Last time I talked to Maxim, a few months ago, to get a datasheet, I couldn't even get the NDA that is required to get the datasheet.

[SilverSolder]

Are we going back to designing things with reliably available discrete components instead of modern ICs?

[peter-h]

Well, perhaps not use Maxim if TI make a popular part which does the job.

But I would always have said the same of Linear Tech, for anything other than precision analog. The LTC491 comes to mind.

[E-Design]

Are we going back to designing things with reliably available discrete components instead of modern ICs?

Sure as long as it doesn't include a JFET...

I'm hoping Maxim parts that remain in their product offerings, will get the LT and ADI obsolescence policies which can support low volume customers.

[tszaboo]

Are we going back to designing things with reliably available discrete components instead of modern ICs?

Sure. 4 years ago I couldnt buy TVS diodes somehow. And a bunch of MOS gets obsoleted every year, because they have a betterererer package, which is not footprint compatible, but smaller, and no second source. JFETs are impossible to buy. If you designed in a dual transistor few years back, you are also out of luck. I dont think I can design a circuit with BC817 and 1N4001 (SOT-23 package, the other might be out of stock)and 10 KOhm resistors to be honest.

[SilverSolder]

[...]  If you designed in a dual transistor few years back, you are also out of luck. [...]

Yeah, I wonder why they disappeared?  - actually rather handy...

[capt bullshot]

[...]  If you designed in a dual transistor few years back, you are also out of luck. [...]

Yeah, I wonder why they disappeared?  - actually rather handy...

The "good stuff" gets all replaced by more or less "one special purpose" IC, that the manufacturers can sell with higher margins to inexperienced (younger, fresh from university) engineers. I was told this statement (not by the words, but in this sense) by some TI (afair) FAE / marketing guys when I asked why the heck they have that many very specialized IC for similar purposes. Yes, they want to you to get lost in their selection lists, then ask them for help, which they'll gladly offer to you, and then they'll select a device that is good for their revenue.

[peter-h]

Discrete components are however usually multi multi sourced. Also there is no shortage of most of them - because the end product demand has hardly increased and the current bubble is just stock hoarding by those who can afford it.

JFETs are rarely used today because most people don't know what a JFET does To most designers, it's a bit like a krytron

"the manufacturers can sell with higher margins to inexperienced (younger, fresh from university) engineers. "

Spot on. Nobody designing a product for their own business would use most of those chips. For a start, you want to minimise your stock list. So you use a 4k7 instead of a 3k9

[tszaboo]

[...]  If you designed in a dual transistor few years back, you are also out of luck. [...]

Yeah, I wonder why they disappeared?  - actually rather handy...

The "good stuff" gets all replaced by more or less "one special purpose" IC, that the manufacturers can sell with higher margins to inexperienced (younger, fresh from university) engineers. I was told this statement (not by the words, but in this sense) by some TI (afair) FAE / marketing guys when I asked why the heck they have that many very specialized IC for similar purposes. Yes, they want to you to get lost in their selection lists, then ask them for help, which they'll gladly offer to you, and then they'll select a device that is good for their revenue.

Inexperienced? I wouldn't say so. I design in quite a lot of special purpose ICs, like load switches instead of passive parts and a tiny FET. Or a shunt amplifier instead of a general purpose opamp and the passive parts. It's not like I cannot design in the older general purpose part. I don't have the time to do so.

I would say it is inexperienced (people who call themselves) managers that are responsible. In the past 10 years, I never had the luxury of testing a board completely, or to investigate more than one solutions to a problem. Everything is needed yesterday, the client is waiting for it. I cannot say, yeah, I used an LM358 and some resistors and transistors, but I have to spend 2-4 days characterizing the circuit, maybe place a few in a thermal chamber to make sure that it works in worst case temperature conditions.

No, what I'm going to do is place there an INA193, that costs 3x the price, that has 100% characterized performance. And my testing will be 5 minutes. I honestly couldn't care less if the product cost more because of this. I drew the triange of cost-time-quality every single project, and they always choose all three, which doesn't work that way. Since the only performance indicator of my work that they are measuring is time, I'm just going to choose whatever is the most convenient for myself.

[peter-h]

Well, yes, in a premium price area one can use premium price chips. Especially in low volume stuff, where the price of silicon is likely to be almost immaterial.

But then one should not complain when the said chip (which probably never saw decent volumes) gets discontinued. There is no solution to that.

I looked up the INA193. Clever chip, and cheap for what it does. They don't give the CM spec as % and the "80db" sounds like the two resistors are 0.1% or better on ratio.

[Terry Bites]

Shitty, surely you mean dynamic and "agile", "what our customers demand". 
Or just degenerate gamblers?

[tooki]

The first few have gone from £0.40 (10k+) to over £2. And this isn't just the current chip bubble. Maxim have been raising prices continually.

A lot of things have changed over 20 years, one of them being the GBP’s decline in value relative to other currencies.

The other thing is that for old parts, we may be seeing reduced economies of scale. Old parts are made on old processes, and if the newer stuff is made on newer, more efficient processes, the volume for the old stuff goes down, effectively increasing the cost of each production run on old processes.

[peter-h]

Yeah - I remember the $2.40 = £1 too But I am talking of 50p 5 years ago.

And TI are selling much cheaper versions of some Maxim parts for much less - see foregoing posts. It is as if Maxim are extracting what they can get from inertia-bound large companies.

[tooki]

They probably don’t charge the large companies any more than TI would. They probably just want to discourage small customers. :/

[Cervisia]

Are we going back to designing things with reliably available discrete components instead of modern ICs?

Sure as long as it doesn't include a JFET...

So what about a discrete JFET as an IC?
https://www.ti.com/product/JFE150

[E-Design]

Are we going back to designing things with reliably available discrete components instead of modern ICs?

Sure as long as it doesn't include a JFET...

So what about a discrete JFET as an IC?
https://www.ti.com/product/JFE150

Well.. no problem.. its fine to use.


Until it isnt.

[peter-h]

I am looking at replacing the MAX3089 with the ISL3173 (question mark re 5V VCC operation) and with the SN65HVD55DR (costs even more than the MAX3089!) - thanks for the tip earlier.

The ISL has a gotcha in that it draws way more current on the TXE pin, so a 10k pullup is no good. Somebody else on the internet had got burnt with that... It actually takes a very careful examination of the various chip functions to work out a circuit into which any of these can be simply dropped.

But neither is available from distribution. Only the overpriced outlets like Mouser have stock, at about 3x the distribution price I've been quoted today. That's a similar price to the cowboy sellers in the US.

It's gonna be interesting how far this situation collapses, when it does.

[SilverSolder]

I am looking at replacing the MAX3089 with the ISL3173 (question mark re 5V VCC operation) and with the SN65HVD55DR (costs even more than the MAX3089!) - thanks for the tip earlier.

The ISL has a gotcha in that it draws way more current on the TXE pin, so a 10k pullup is no good. Somebody else on the internet had got burnt with that... It actually takes a very careful examination of the various chip functions to work out a circuit into which any of these can be simply dropped.

But neither is available from distribution. Only the overpriced outlets like Mouser have stock, at about 3x the distribution price I've been quoted today. That's a similar price to the cowboy sellers in the US.

It's gonna be interesting how far this situation collapses, when it does.

When you begin designing UARTs around vacuum tubes, you know the situation has collapsed! 

[peter-h]

Replacing Maxim parts with TI parts is quite a fun exercise. I've spent the last few days on it Once the bigger customers wisen up to this, Maxim will lose big chunks of their sales, totally, and will never get them back.

Renesas won't comment on the ISL3173 at 5V. They say it is below the absolute max (7V) but aren't saying it will function. Quite strange.