Old microcontrollers: should I bother hanging on those?

[technix]

I still have a few LPC2103 (ARM7TDMI @70MHz) and CH563 (ARM926EJ-S @130MHz) in my parts bin. Should I keep hanging on to them? Should I keep a few of them on breadboard adapter boards?

LPC2103 doesn't have much special features other than a few UARTs, SPI, I2C and timers.

CH563 supports built-in Ethernet PHY, USB, audio ADC, external bus interface and parallel slave port, along with the standard set of UARTs, SPI, I2C and timers.

p.s. I bought a few TPS2375. Couple those to my PoE MagJack and 48V-5V isolated DC-DC modules I can implement PoE. Should I try to build a board with CH563 + PoE, or should I just go directly to STM32F107/217/756 + DP83848/RTL8201 + PoE?

[rstofer]

I have a couple of LPC2106 and LPC2148 prototype boards.  I think I'll hang on to them.

It seems to me that they are a great introduction to ARM processors without all the layers and layers of driver code (CubeMX, I'm talking about you!).  The 2106 is a perfect place to start with ARM and the 2148 adds USB capabilities.  Yes, the startup code needs to be written in assembly language but it isn't all that hard.

Then there is the demo code for the 2148 at:
http://www.jcwren.com/arm/

I'm a hoarder!  I still have my Altair 8800 (8080 CPU) and several incantations of Z80 machines one of which runs CP/M at 50 MHz - fast!.  Going back even further, I have an FPGA implementation of the IBM1130 just in case I want to write some Fortran (or assembly language).

They're all mine!  I bought/built them and I'm keeping them.

[technix]

I have a couple of LPC2106 and LPC2148 prototype boards.  I think I'll hang on to them.

I don't have the ready-made chips - bare chips on DIP-48 adapter only.

It seems to me that they are a great introduction to ARM processors without all the layers and layers of driver code (CubeMX, I'm talking about you!).  The 2106 is a perfect place to start with ARM and the 2148 adds USB capabilities.  Yes, the startup code needs to be written in assembly language but it isn't all that hard.

Then there is the demo code for the 2148 at:
http://www.jcwren.com/arm/

Are those ARM7/ARM9 a better starting place than Cortex-M3 like STM32F217? No I am not using CubeMX for my STM32 projects, it is plain old register manipulation and CMSIS.

I'm a hoarder!  I still have my Altair 8800 (8080 CPU) and several incantations of Z80 machines one of which runs CP/M at 50 MHz - fast!.  Going back even further, I have an FPGA implementation of the IBM1130 just in case I want to write some Fortran (or assembly language).

They're all mine!  I bought/built them and I'm keeping them.

Ahh.

I recently bought a full set of (mostly CMOS) chips for a DIY IBM PC XT computer: 80C88, 8087, 82C88, 82C84 & 74HC573 for processor cluster, AS7C4096 for memory (two of them makes the full 1MB supported by 80C88,) AT28C256 EEPROM for BIOS, EPM7128STC100-15N & 82C37A as the bus matrix, 82C59A PIC, 82C53 PIT, 82C55A GPIO, CH375 USB OTG & CH395 Ethernet for I/O devices. I am still unsure on how to construct the graphics hardware and secondary storage though. I do have some CY7C136-25NXC dual port SRAM that can function as VRAM though.

[ebastler]

I can understand hanging on to completed, working projects (or commercial intruments). I also see the point of keeping old, bare chips, which might come in handy to repair old kit at some point.

But I would not use discontinued chips in a new design, even for hobby purposes. (Unless they are really old, that is, and the point of the project is to recreate some historic design.) For a new design, I would like to have the ability to share the design with others and make it easy for them to recreate. Starting with a part that is no longer easily available may get in the way there. Hence, I'm not sure whether you would ever want to use those old ARM chips anymore?

[technix]

But I would not use discontinued chips in a new design, even for hobby purposes. (Unless they are really old, that is, and the point of the project is to recreate some historic design.) For a new design, I would like to have the ability to share the design with others and make it easy for them to recreate. Starting with a part that is no longer easily available may get in the way there. Hence, I'm not sure whether you would ever want to use those old ARM chips anymore?

I don't think LPC2103 is out of production.

It is an old chip but for educational purposes this chip does have some special benefits as it uses the same full sized ARM instruction set and ARM exception vector mechanism as the modern Cortex-A cores, unlike the Cortex-M which received a vastly simplified vectored interrupt controller, an incomplete instruction set and assembler-free startup.

Speaking of really old parts,I believe my 80C88 kit can fall into this category (as well as my 68HC000 which I have yet to figure out how to support.) It is crazy to realize that 80C88 is still second sourced by many manufacturers like Intersil and OKI along with its peripherals chips. Maybe using a CPLD to condense the motherboard logic will ruin some fun though.

[ebastler]

I don't think LPC2103 is out of production.

Ah, right. Only the VFQFN package seems to be obsolete, while the LQFP is still active.

[technix]

I don't think LPC2103 is out of production.

Ah, right. Only the VFQFN package seems to be obsolete, while the LQFP is still active.

VQFN isn't useable to me either: no way to solder.

[nctnico]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

[legacy]

I recently bought a full set of (mostly CMOS) chips for a DIY IBM PC XT computer: 80C88, 8087, 82C88, 82C84 & 74HC573 for processor cluster, AS7C4096 for memory (two of them makes the full 1MB supported by 80C88,) AT28C256 EEPROM for BIOS, EPM7128STC100-15N & 82C37A as the bus matrix, 82C59A PIC, 82C53 PIT, 82C55A GPIO, CH375 USB OTG & CH395 Ethernet for I/O devices. I am still unsure on how to construct the graphics hardware and secondary storage though. I do have some CY7C136-25NXC dual port SRAM that can function as VRAM though.

and

90 Percent of the time I have no idea what the fuck I'm talking about

[technix]

I recently bought a full set of (mostly CMOS) chips for a DIY IBM PC XT computer: 80C88, 8087, 82C88, 82C84 & 74HC573 for processor cluster, AS7C4096 for memory (two of them makes the full 1MB supported by 80C88,) AT28C256 EEPROM for BIOS, EPM7128STC100-15N & 82C37A as the bus matrix, 82C59A PIC, 82C53 PIT, 82C55A GPIO, CH375 USB OTG & CH395 Ethernet for I/O devices. I am still unsure on how to construct the graphics hardware and secondary storage though. I do have some CY7C136-25NXC dual port SRAM that can function as VRAM though.

and

90 Percent of the time I have no idea what the fuck I'm talking about

F***ing yeah.

[technix]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

[KL27x]

Well..... how many of these do you have?

Hell, I got hundreds of leftover microcontrollers. Thousands of a couple very old PICs.

I will build a (project specific) dev board for these devices.... if I ever come across a project that is suitable for one of these AND will FOR SURE be covered with whatever quantity I have lying around.

Other than that, they just sit in storage along with every other piece of garbage which I can't bear to throw out.

[technix]

Well..... how many of these do you have?

10x LPC2103, 3x CH395.

Hell, I got hundreds of leftover microcontrollers. Thousands of a couple very old PICs.

I will build a (project specific) dev board for these devices.... if I ever come across a project that is suitable for one of these AND will FOR SURE be covered with whatever quantity I have lying around.

Other than that, they just sit in storage along with every other piece of garbage which I can't bear to throw out.

Do you have a list of them? I may be interested in some slightly larger chips.

[KL27x]

Unused, stored in sealed faraday plastic bags in a dry climate.

18LF4550 QFP  due to the packaging in stacking trays, I don't even know how to split this up into smaller packages.

16LF1828 SSOP in tubes and some in cut tape. Probably some 16F1828, too, but mostly LF

18LF2550 SOIC in tubes.  These are cool because you can make PICKit2's with this chip.

^^^These guys I have enough of, if anyone successfully sells stuff like this on eBay/Ali or something, I might have enough of these to bother with if you want to make an offer. Or if you have any tips for selling stuff like this, I'm all ears.


DIP chips... in various states. I have at least a tube or 3 of each, new. I have some pulls. I even have some with the legs cut off where I got tired of pulling them properly:

12F508

12F629

12F505

12F683

12F629

[nctnico]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

There are probably lots of dev boards out there you can buy for a few dollars. I'd go for a project specific board.

[technix]

Unused, stored in sealed faraday plastic bags in a dry climate.

18LF4550 QFP  due to the packaging in stacking trays, I don't even know how to split this up into smaller packages.

16LF1828 SSOP in tubes and some in cut tape. Probably some 16F1828, too, but mostly LF

18LF2550 SOIC in tubes.  These are cool because you can make PICKit2's with this chip.

^^^These guys I have enough of, if anyone successfully sells stuff like this on eBay/Ali or something, I might have enough of these to bother with if you want to make an offer. Or if you have any tips for selling stuff like this, I'm all ears.

Do you mind parting with some of those PIC18LF2550 and '4550?

I am slightly intrigued with the idea of making my own PICKit 2. Having one of those does mean I can have something to repair my PICkit 3 with though, and I'd rather build my own clone programmer than buying someone else's untrusted hardware. And I can throw in my own features too (e.g. isolated programming port.)

DIP chips... in various states. I have at least a tube or 3 of each, new. I have some pulls. I even have some with the legs cut off where I got tired of pulling them properly:

12F508

12F629

12F505

12F683

12F629

I am a bit torn between those 8-pin microcontrollers: IAP15W105 (fast 8051 core,) ATtiny85, or PIC12? I have samples of all of those yet I am still undecided.

[technix]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

There are probably lots of dev boards out there you can buy for a few dollars. I'd go for a project specific board.

So any ideas? I do want to make a NeoPixel RGBW light controller for my server rack, but that thing needs wired Ethernet (no PoE though) and SSL (for iPhone compatibility) so I would prefer using STM32F217 or STM32F107 (if iOS accepts ECDSA-ECDHE-ChaCha20-Poly1305-SHA256 cipher suite) for it.

[nctnico]

If you need SSL then I'd use a Raspberry Pi. Getting SSL to work on a microcontroller is a world of pain. You'll need at least 128kB of SRAM and an OS. SSL and microcontrollers don't go well together.

[technix]

If you need SSL then I'd use a Raspberry Pi. Getting SSL to work on a microcontroller is a world of pain. You'll need at least 128kB of SRAM and an OS. SSL and microcontrollers don't go well together.

I don't think Raspberry Pi is warranted for my use case, as it simply pipes the HTTP POST message to NeoPixel. All the NeoPixel protocol handling is done on the phone.

[legacy]

F***ing yeah.

Exactly: find it, instead of proving you still need to find it

it, as a meaning for your continuous and compulsory chip listing
which is
-1- usually no-sense
-2- so freaking annoying after a while

[schmitt trigger]

The difference between old junk and a collectible item is time.

Since I started playing with electronics 46 years ago, I've disposed of so much stuff that I now regret.
All the way to punched cards programmed in FORTRAN.
I wish I had kept at the very least a single one of those, and have it framed today. It would make a real conversation piece.

My advice: keep a single item of the most representative devices. If you are lucky enough to have an 8080 or an original Z80, by all means, put it in a nice display.

[technix]

The difference between old junk and a collectible item is time.

Since I started playing with electronics 46 years ago, I've disposed of so much stuff that I now regret.
All the way to punched cards programmed in FORTRAN.
I wish I had kept at the very least a single one of those, and have it framed today. It would make a real conversation piece.

My advice: keep a single item of the most representative devices. If you are lucky enough to have an 8080 or an original Z80, by all means, put it in a nice display.

I managed to find two D8087 from Intel in ceramic DIP package, when I was hunting down chips for a semi-modern 80C88 DIY computer - runs FreeDOS or MS-DOS, but also have support for USB and Ethernet. I think that should be considered one of those period chips, and I should put it in a working board and frame one. I do have a full set of chips for that computer project. I just have not figured out how to build the graphics hardware though.

[ebclr]

I still have 80C35 in stock, Probably will be a heritage for my kids .....

Engineers tend to be junk accumulators, hoping that someday it might be useful, 0.01% can finally be used, the rest their children will throw away after their death

[technix]

I still have 80C35 in stock, Probably will be a heritage for my kids .....

Engineers tend to be junk accumulators, hoping that someday it might be useful, 0.01% can finally be used, the rest their children will throw away after their death

If you can make your kids realize that those are heirloom pieces they would cherish it. I would allow my future kids play on that 5MHz 80C88 machine unrestricted and unsupervised on internal battery power, with a stack of bilingual reference manuals next to it. Then I would teach them BASIC on it. Then move to an ARM architecture for the UNIX experience.

[rstofer]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

In my view, yes.

Somewhere along the path comes Arduino where there is quite a bit of hand-holding and huge provided libraries and tons of examples.  At the other end of the scale there are the Cortex M series ARMs that usually come with a flood of driver code you didn't (need to) write but you now need to understand.  Everything is hidden in magic STRUCTs fed into somebody else's code.

In the middle, with no hand-holding and no driver code, are the ARM 7TDMIs.  It seems to me that these are a great transition between Arduino and ARM Cortex M series because you will write all your own code and all of it will be based on an understanding of the User Manual (plus the Datasheet in the case of NXP).  You will deal directly with the hardware and, if you use STRUCTs, they will be your STRUCTs.  The good news is that the chips like the LPC2138 are easy to code; the peripherals are pretty straightforward and well documented.  I like the brevity of coding for these chips.  Set up a few registers and watch it rip!

[technix]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

In my view, yes.

Somewhere along the path comes Arduino where there is quite a bit of hand-holding and huge provided libraries and tons of examples.  At the other end of the scale there are the Cortex M series ARMs that usually come with a flood of driver code you didn't (need to) write but you now need to understand.  Everything is hidden in magic STRUCTs fed into somebody else's code.

In the middle, with no hand-holding and no driver code, are the ARM 7TDMIs.  It seems to me that these are a great transition between Arduino and ARM Cortex M series because you will write all your own code and all of it will be based on an understanding of the User Manual (plus the Datasheet in the case of NXP).  You will deal directly with the hardware and, if you use STRUCTs, they will be your STRUCTs.  The good news is that the chips like the LPC2138 are easy to code; the peripherals are pretty straightforward and well documented.  I like the brevity of coding for these chips.  Set up a few registers and watch it rip!

For such a dev board, is it a good idea to keep the Arduino form factor?

Speaking of Cortex-M, I never use their driver code except the vendor-agnostic CMSIS (which almost never uses the structs)

[rstofer]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

In my view, yes.

Somewhere along the path comes Arduino where there is quite a bit of hand-holding and huge provided libraries and tons of examples.  At the other end of the scale there are the Cortex M series ARMs that usually come with a flood of driver code you didn't (need to) write but you now need to understand.  Everything is hidden in magic STRUCTs fed into somebody else's code.

In the middle, with no hand-holding and no driver code, are the ARM 7TDMIs.  It seems to me that these are a great transition between Arduino and ARM Cortex M series because you will write all your own code and all of it will be based on an understanding of the User Manual (plus the Datasheet in the case of NXP).  You will deal directly with the hardware and, if you use STRUCTs, they will be your STRUCTs.  The good news is that the chips like the LPC2138 are easy to code; the peripherals are pretty straightforward and well documented.  I like the brevity of coding for these chips.  Set up a few registers and watch it rip!

For such a dev board, is it a good idea to keep the Arduino form factor?

Speaking of Cortex-M, I never use their driver code except the vendor-agnostic CMSIS (which almost never uses the structs)

I hadn't thought about using the Arduino form factor but if the LPC2138 has the right IO configuration, why not?  The pins need to be brought out somehow, it might as well be a 'standard'.  You could even use the same USB->serial chip (I think) and it might as well have the optional ability to provide power to the board via USB power.

[technix]

The LP2103 has been around for a very long time. It has some nice features which still make it an interesting & usefull controller. AFAIK this is quite a popular model. FWIW Farnell has 350 on order for delivery this month.

Is it worth it for me to build a new dev board for it?

In my view, yes.

Somewhere along the path comes Arduino where there is quite a bit of hand-holding and huge provided libraries and tons of examples.  At the other end of the scale there are the Cortex M series ARMs that usually come with a flood of driver code you didn't (need to) write but you now need to understand.  Everything is hidden in magic STRUCTs fed into somebody else's code.

In the middle, with no hand-holding and no driver code, are the ARM 7TDMIs.  It seems to me that these are a great transition between Arduino and ARM Cortex M series because you will write all your own code and all of it will be based on an understanding of the User Manual (plus the Datasheet in the case of NXP).  You will deal directly with the hardware and, if you use STRUCTs, they will be your STRUCTs.  The good news is that the chips like the LPC2138 are easy to code; the peripherals are pretty straightforward and well documented.  I like the brevity of coding for these chips.  Set up a few registers and watch it rip!

For such a dev board, is it a good idea to keep the Arduino form factor?

Speaking of Cortex-M, I never use their driver code except the vendor-agnostic CMSIS (which almost never uses the structs)

I hadn't thought about using the Arduino form factor but if the LPC2138 has the right IO configuration, why not?  The pins need to be brought out somehow, it might as well be a 'standard'.  You could even use the same USB->serial chip (I think) and it might as well have the optional ability to provide power to the board via USB power.

I am using LPC2103, not 2138. It does have the correct types and amount of I/O for an Arduino Uno sized board. However I do have to put some consideration into the JTAG interface, as it calls for a full 20-pin J-Link style connector (which will make the board slightly longer than Arduino Uno.) Maybe I can pack an additonal ESP8266 module on there if I have two extra pins?

[rstofer]

I am using LPC2103, not 2138. It does have the correct types and amount of I/O for an Arduino Uno sized board. However I do have to put some consideration into the JTAG interface, as it calls for a full 20-pin J-Link style connector (which will make the board slightly longer than Arduino Uno.) Maybe I can pack an additonal ESP8266 module on there if I have two extra pins?

Sorry about the 2138 bit, I got this thread confused with the other one.  You could just use serial programming and skip the JTAG but you would lose debugging.  Adding the ESP8266 would be sweet!

[technix]

I am using LPC2103, not 2138. It does have the correct types and amount of I/O for an Arduino Uno sized board. However I do have to put some consideration into the JTAG interface, as it calls for a full 20-pin J-Link style connector (which will make the board slightly longer than Arduino Uno.) Maybe I can pack an additonal ESP8266 module on there if I have two extra pins?

Sorry about the 2138 bit, I got this thread confused with the other one.  You could just use serial programming and skip the JTAG but you would lose debugging.  Adding the ESP8266 would be sweet!

The chip came with two hardware UARTs: one for ISP and communication to the computer, and the other for ESP8266. Now I need to figure out how to build the (slightly complicated) bootloader switching circuit.

[rstofer]

The chip came with two hardware UARTs: one for ISP and communication to the computer, and the other for ESP8266. Now I need to figure out how to build the (slightly complicated) bootloader switching circuit.

I think you just need 2 pushbuttons:  One to do the reset and the other to signal you want the device to go into ISP mode.  Of course, that puts some usage limitations on the ISP pin - P0-14.  Hold the bootloader button pressed while doing a reset.

See page 239 here:
http://www.nxp.com/documents/user_manual/UM10161.pdf

[technix]

The chip came with two hardware UARTs: one for ISP and communication to the computer, and the other for ESP8266. Now I need to figure out how to build the (slightly complicated) bootloader switching circuit.

I think you just need 2 pushbuttons:  One to do the reset and the other to signal you want the device to go into ISP mode.  Of course, that puts some usage limitations on the ISP pin - P0-14.  Hold the bootloader button pressed while doing a reset.

See page 239 here:
http://www.nxp.com/documents/user_manual/UM10161.pdf

I actually wanted to implement the 1200 baud or DTR line trick used by the Arduino folks, but the big question here is how. Two buttons seemed like an acceptable solution though.

[rstofer]

https://www.sparkfun.com/products/retired/714
https://www.sparkfun.com/datasheets/LPC/LPC-ICP.pdf

This isn't totally automated but it could be.

[nctnico]

The chip came with two hardware UARTs: one for ISP and communication to the computer, and the other for ESP8266. Now I need to figure out how to build the (slightly complicated) bootloader switching circuit.

I think you just need 2 pushbuttons:  One to do the reset and the other to signal you want the device to go into ISP mode.  Of course, that puts some usage limitations on the ISP pin - P0-14.  Hold the bootloader button pressed while doing a reset.

See page 239 here:
http://www.nxp.com/documents/user_manual/UM10161.pdf

I actually wanted to implement the 1200 baud or DTR line trick used by the Arduino folks, but the big question here is how. Two buttons seemed like an acceptable solution though.

If you buy a USB to serial converter with RTS and CTS lines then Flashmagic can control the reset and bootloader entry pins directly on any NXP LPC ARM controller. Put a switch in between and you have a convenient programming solution.

[rstofer]

The chip came with two hardware UARTs: one for ISP and communication to the computer, and the other for ESP8266. Now I need to figure out how to build the (slightly complicated) bootloader switching circuit.

I think you just need 2 pushbuttons:  One to do the reset and the other to signal you want the device to go into ISP mode.  Of course, that puts some usage limitations on the ISP pin - P0-14.  Hold the bootloader button pressed while doing a reset.

See page 239 here:
http://www.nxp.com/documents/user_manual/UM10161.pdf

I actually wanted to implement the 1200 baud or DTR line trick used by the Arduino folks, but the big question here is how. Two buttons seemed like an acceptable solution though.

If you buy a USB to serial converter with RTS and CTS lines then Flashmagic can control the reset and bootloader entry pins directly on any NXP LPC ARM controller. Put a switch in between and you have a convenient programming solution.

Yes!  I had forgotten about FlashMagic.  Somewhere along the line, I migrated to JTAG and forgot my roots...

It's all right here:
http://www.flashmagictool.com/assets/resources/PoUSB12LPCProg.pdf

[technix]

The chip came with two hardware UARTs: one for ISP and communication to the computer, and the other for ESP8266. Now I need to figure out how to build the (slightly complicated) bootloader switching circuit.

I think you just need 2 pushbuttons:  One to do the reset and the other to signal you want the device to go into ISP mode.  Of course, that puts some usage limitations on the ISP pin - P0-14.  Hold the bootloader button pressed while doing a reset.

See page 239 here:
http://www.nxp.com/documents/user_manual/UM10161.pdf

I actually wanted to implement the 1200 baud or DTR line trick used by the Arduino folks, but the big question here is how. Two buttons seemed like an acceptable solution though.

If you buy a USB to serial converter with RTS and CTS lines then Flashmagic can control the reset and bootloader entry pins directly on any NXP LPC ARM controller. Put a switch in between and you have a convenient programming solution.

CH340B have those lines... Maybe I can use a diode and a pull-up to make this work without affecting this pin being used as digital I/O.