General > General Technical Chat
Why is the 741 op amp still produced?
<< < (16/24) > >>
Kleinstein:
There are still a few high voltage OPs available and even new one released, e.g. MCP6V51 is specified to 45 V, though not a direct 741 replacement but AZ and only smaller cases.
At the lower cost end, when the LM358 and related are not an option (e.g. higher supply voltage, speed, cross over) the 741 is still a viable option, but there are alternatives besides the LM301.
coppice:

--- Quote from: james_s on September 28, 2020, 04:08:16 am ---
--- Quote from: coppice on September 27, 2020, 11:57:57 pm ---It will be interesting to see what happens to these higher voltage parts as less and less fabs remain functional which can produce them. People do not seem to see enough market size to justify developing new high voltage parts right now, although I guess they will be forced to as the last ancient fabs close. Some people will always need high voltage operation.

--- End quote ---

What about fabs making HV parts like mosfet drivers, SMPS ICs and that sort of thing? Couldn't they make HV op amps? The market for things like motor drives and power supplies isn't going anywhere.

--- End quote ---
Of course there is a need for. and will always be, high voltage parts. However, with many of the fabs that make the ancient higher voltage parts reaching their end of life there are going to be some significant changes in the industry. Will the people scrapping ancient fabs replace them? Will new players take up the work? It will be interesting to see.
exe:
Why a new fab cannot produce an old part? (I can see why this is uneconomical == doesn't make sense on current market).
T3sl4co1l:
For one, the part probably has to be redesigned, because all the characteristics are different.

For another, no one's making the high voltage BJTs like the old way.  A 10 or 20V part would be easy enough to do.  But that's nowhere near the original spec.

TLV2372 is probably an example of this, using a high voltage CMOS process (18V max).  Though I don't know offhand if that's metal or poly gate HVCMOS -- there's poly HVCMOS, right?

Which is not to say there aren't any fabs doing that sort of thing.  Discrete BJTs aren't going away any time soon.  I'm not sure why they couldn't simply run a different mask set there (and insert a couple more passes for doping as applicable, and metallization; of course, doping can't be done willy-nilly, that's a whole process refinement step).  Perhaps access to the fabs is poor (who's making BJTs anymore? It's not all China now is it..?).

There are plenty of HV processes, ranging from ~20V amps like the above example, to 50-60V regulators and switchers, to 600V+ gate drivers.  Probably these are optimized for CMOS though, and BJTs are rough on a CMOS process.  (Also, not that the 600V transistors in HVICs are necessarily any good -- they do actually exist, and the HVIC process isn't just a high side isolation thing, transistors are included and that's indeed how they couple signals to the high side -- but not much performance is needed for that, either.)

I suppose LT (now ADI)'s internal fab is one of the most likely candidates; they've been making interesting BJT-based parts for ever.  There aren't many ways you can make a buck regulator operate from 2-30V, but BJTs are a good one.  Datasheets don't usually say much of anything about internal design or process, so it's not obvious from here how much they do that, but obviously it's what they're doing when they show an NPN switch on the block diagram.


Going back to direct substitutes, there are plenty of >30V op-amps out there, some with quite impressive specs.  I don't know how many are bipolar, BiCMOS or what.  The biggest downsides will be: cost, knock-on changes (because the GBW, etc. are different), and approval.

Tim
coppice:

--- Quote from: T3sl4co1l on September 28, 2020, 01:26:07 pm ---I suppose LT (now ADI)'s internal fab is one of the most likely candidates; they've been making interesting BJT-based parts for ever.

--- End quote ---
Making them forever is exactly the problem. Ancient analogue fabs are shutting down because they are so old people can't get spares for them any more. It looks like the remaining ones are only being kept alive through the reuse of parts taken from the ones closing down. Nobody is going to invest in new plant for tiny wafers, so a redesign for a process that exists for large wafers is usually needed. The original designers have probably retired, and the original design data has rarely been kept intact, so it will be a redesign from scratch.

--- Quote from: T3sl4co1l on September 28, 2020, 01:26:07 pm ---There aren't many ways you can make a buck regulator operate from 2-30V, but BJTs are a good one.  Datasheets don't usually say much of anything about internal design or process, so it's not obvious from here how much they do that, but obviously it's what they're doing when they show an NPN switch on the block diagram.

--- End quote ---
Nobody wants to tell customers about the processes used to make their devices, unless it has marketing value, like "uses TSMC's latest super duper 10pm process, with less than one atom per transistor". It can affect their flexibility in re-engineering parts for lower cost.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod