LPC43XX and "end of life"

[henrik_glader]

Hi again : )

This one is maybe posted in the wrong forum so you're welcome to redirect me.

I've read about NXP longevity statement and notices the LPC43XX series chips are about to hit the end.

I'd speculate a LTB in 2022 or so, in a worst case.

Is that correct?

Are there any planned uppgrade paths?

Cheers

/Henrik 

[msimons]

Hi Henrik & others,

We use the LPC4333 (BGA256) a lot. Consuming 6k / year now.

So it is not likely that it will go out of production. NxP has lots of worries now and the only thing they want is to sell good products to their base customers, like we are. Until now (may 2020) no signs of going obsolete. 

About the performance: I only can agree. I am the hardware guy, but I hear the software guy here programming all in large state machines, super fast and slim, very strict timing and great handling of our samples. For lots of products the LPC5xxx is much much too heavy. There certainly is a market for this LPC4333 as I have chosen it due to the '80C552'-feeling it gave me when I studied all the peripherals. Back then it seemed to be an industry standard to me.

So only if a blind stupid marketing man will come around within NxP to take it out... Well then we have a big problem !

Changing to other variants is not easy. One thing is for sure: If they are going to mess around with it @ NxP we will definitely switch over to ST.

[hlsamts]

In the meantime, NXP has extended the longevity period from 10 to 15 years. Thus I expect the LPC43xx series being available at least until 2029. This is pretty good news.

[bernhardfink]

I can't give a statement about the possible lifetime of the LPC4300. But many MCU parts in the past of NXP and Freescale entered a pro-longed lifetime period, so hopefully the LPC4300 has a long and prosper life as well :-)

Successors with internal flash are there, please have a look at the existing LPC54xxx family. Brandnew is the LPC5500 family, that's bleeding edge in terms of features and technology.

Regards,

Bernhard.

[hlsamts]

No: the LPC5xxx series is not a successor of the LPC43xx series.

There is no combination of a ~200MHz dual core + Ethernet and the LPC43xx has a faster UART interface.

Hopefully the LPC43xx will be "alive" much longer!

[bernhardfink]

Well, that's right. The LPC4300 ist still a kind of flagship in our portfolio.

An alternative could be the i.MX RT family. The software eco system is from the microcontroller world, the performance more from the apps processor world. The price by the way is also from the MCU world :-)

For example the i.MX RT1050:  One Cortex-M7 with 528MHz, Ethernet and a fast UART

Even higher performance would be delivered by the i.MX6ULL.

If you really insist on dual core, the there is the i.MX 7ULP, an interesting architecture when it comes to dual core.

Anyway, if you plan to go for 204MHz operation with the LPC4300, please go for the 256-pin BGA package, this offers the highest performance in combination with SDRAM. For the LQFP packages I think we made some restrictions with regards to SDRAM frequency.

Regards,

Bernhard.

[hlsamts]

The i.MX RT is a nice product, but it is a single core.

All i.MX have a very poor UART performance.

In my case, I need one core for very accurate timing purposes and a second core doing communication stuff. With the M0/M4 of the LPC43xx this can be perfectly splitted.

Further I extensively use the SCT with several states & events + three UART interfaces at 8 MBit/s each. 

Currently I cannot find any other product from NXP, which fulfills all these requirements. The only single-chip alternative currently available is the TI Sitara family.

By the way: we do use the LPC4337 with the BGA package and 32Bit SDRAM. This is a very good combination.

I hope that NXP will keep up supplying their flagship until they have an appropriate successor.

For the customer, replacing a low-end MCU is much easier than replacing a flagship MCU.

Regards,

Holger

[henrik_glader]

Hello guys : )

Been very busy with other things but this is interesting = )

I like hard concurrency but dislikes FPGA:s - I have tested the Zync platform and nearly changed opinion, but that combo too powerful and hungry.

The LPC43XX is a very stong candidate, the only real competitor (my application) in the ARM space might be the Vybrid family. However then you loose on-chip flash.

The Sitara is nice with its micro engine at your finger tips, although I've not programmed that beast yet.

Have a couple of BeagleBone Black in my trash-can if I get very boored = )

I did dead-bug the LPC4337 QFP144 I recently got from NXP and the few things I tested works brilliant.

I work in the Keil tool chain for now.

Like Holger I clearly see a benefit with separated cores, one with real-timey work and the other with the ethernet stack asf.

My initial strategy was the LPC4337/BGA256 (I'd rather go with the BGA100 but theres no MCPWM there).

Now I've wondered if there might be possible to run an advanced PWM on the third core in the LPC4367 chip.

Well see how things pan out.

Thank you for your input = )

/Henrik

[bernhardfink]

Hi Henrik & Holger,

two real fans of this MCU architecture, like me  :smileyhappy:

It's a pitty that we didn't really follow up on this, especially with SCT and the SGPIO feature.

The second M0 in the LPC4367 is behind a sync bridge, so if you have some in/out activities you normally work with about 4 waitstates. However, there are direct connections to SGPIO and some other internal blocks. There is also a limited amount of memory exclusive for this M0 (which means it is also behind this bridge), so that this core could run on 204MHz without waitstates.

For PWM the SCT and the SGPIO features could be used, running PWM with the ARM cores causes jitter.

Regards,

Bernhard.

[henrik_glader]

Indeed : )

The idea using the M0SUB for the PWM is that I might be able to go for the BGA100 package.

I don't use full blown OS like Linux so theres not as important with external memories and if ETH is problematic in this case I can go for a separate controller for that.

What I need (and wish for) from a PWM is

1. Good resolution like 10ns or better (204MHz would give 5ns I'd guess).
2. Synchronous update.
3. Hard abort.
4. Automagic dead-band up to 1µs in small incremental steps (at least sub 100ns).
5. Edge/center alignment option.
6. In H-bridge, support for automagic forward/reverse.
7. Blanking up to 100ns or so.

The MCPWM does not support all of this I suppose but the SCT feature maybe does with some effort, I have no clue at the moment.

Thank you Bernhard = )

[bernhardfink]

Just in case you don't have that App Note package.

[henrik_glader]

Joey, Bernhard : )

I have a bad habit of collecting all information others have created too avoid mistakes already made.

There is a good chunk of various examples in the LPC Open/Xpresso zips too = )

BTW, I forgot one good PWM option, synchronous ADC-sampling.

It's good there are two ADC modules in this chip for motor control however some applications requires lot higher resolution so external trigger channels would be handy.

BTW, thank you for your consideration.

/Henrik

[starblue]

There is a newer version of AN11538 from 2016-03-21:

https://community.nxp.com/external-link.jspa?url=http%3A%2F%2Fcache.nxp.com%2Fdocs%2Fen%2Fapplicatio... 

I also like the SCT and SGPIO very much.

I used the SGPIO to do 2-bit parallel SPI from an ADC, defining the waveforms for the signals by patterns in the shift registers, a kind of free-form bit-banging at 200MHz. Unfortunately it seems the SGPIO wasn't considered a winner, it didn't reappear in any later family. My guess is it is too complicated to program for most people, though I liked the puzzling (lots of constraints due to pins blocked by other needed signals, registers being chained in a certain way, etc.).

Fortunately the SCT is in almost all recent devices.

Regards

Jürgen

[henrik_glader]

Yes, indeed : )

You can still buy a 68000, but then again history never reveals the future.

I'll have a go on your suggestions, thank you = )

/Henrik