it seems to me that LPC3180 (or similar) recent ARM processors offer all computing power and IO one might need for any “reasonable” project even at a verly low price. However, it’s quite painful for “hobby” people to design a board for an LFBGA package (we probably need at least 4 layer, maybe 6…) — and how to solder that is another question.
I am aware of development boards, but these boards often come with a huge set of extra stuff (like LCD, pushbuttons, etc) - which I don’t want on my small robot or other electronics — and they tend to be quite expensive. So I couldn’t just get 4 or 5 of them at 300 US$ each.
Wouldn’t it be great to have a “breakout” board for LPC3180, that includes all “basic stuff”, such as power regulation, crystal, extra flash and extra SRAM (32MB)! That’ll allow us to start projects with larger software / memory requirements beyond those typical few KB on-board memory of microcontrollers. Adding an LCD, or whatever you need, should be straight forward (in contrast to adding high-speed SRAM with 100++ connections). Having SRAM/FLASH on-board would save most of the 320 pins for external IO; we might be left with a 100 or so… That’s absolutely do-able for a breakout board.
Looking at individual prices, LPC3180 is about 10 US$, whereas 32MB SRAM and 32MB flash is probably another 10 -15US$. So I think it’d be possible to sell a small (tiny?) board with CPU and memory for maybe 50 US$. You provide power and software, and get IO-pins that fit on your standard 2-layer PCB. No need to worry about external SRAM, soldering tiny chip packages, etc…
Any thoughts about that? Any ideas?
I am not absolutely focused on LPC3180, other “fast” microcontroller (other ARM9? Blackfin? X-Scale?) are also fine. But I really like the low price of LPC3180
Does anyone know of a board similar to the specs above for other microprocessors?
jconradt:
I am aware of development boards, but these boards often come with a huge set of extra stuff (like LCD, pushbuttons, etc) - which I don’t want on my small robot or other electronics — and they tend to be quite expensive. So I couldn’t just get 4 or 5 of them at 300 US$ each.
What you are really asking for is a header board like Olimex makes for several different NXP MCUs. I am becoming quite fond of the NXP ARM MCUs, and have two LPC-2148 ARM7 boards now (Header and Proto).
jconradt:
Wouldn’t it be great to have a “breakout” board for LPC3180, that includes all “basic stuff”, such as power regulation, crystal, extra flash and extra SRAM (32MB)! That’ll allow us to start projects with larger software / memory requirements beyond those typical few KB on-board memory of microcontrollers.
It would be AWESOME to have this in a header board as well as prototype board form. Yeah, an inexpensive LPC-3180 ARM9 board that can run full blown Linux! Add a usb host/client port, a standard serial port, and a couple rows of pins on the bottom to make it solderless breadboard friendly, for a very nice header board. :D Socket headers could be added on top of the board to allow connecting to the rest of the MCU I/O pins.
jconradt:
Adding an LCD, or whatever you need, should be straight forward (in contrast to adding high-speed SRAM with 100++ connections). Having SRAM/FLASH on-board would save most of the 320 pins for external IO; we might be left with a 100 or so… That’s absolutely do-able for a breakout board.
Yes, there would have to be ample Flash and RAM added ON the board. I'd love to see enough of each to allow comfortably running real Linux (not uCLinux). 64 Meg RAM and at least 16 Meg Flash would work nicely. :) I would get in line VERY quick for a header board such as this. :D:D
linuxguy:
Yes, there would have to be ample Flash and RAM added ON the board. I’d love to see enough of each to allow comfortably running real Linux (not uCLinux).
The main reason that uCLinux came to fruition was actually to allow Linux to run on microcontrollers that don't have a memory management unit (MMU). Unless the LPC3180 has an MMU, you will be confined to using uCLinux on it. As far as the user or programmer is concerned, though, there's not a tremendous difference in the two, except (obviously) in memory management and access.
brennen:
Unless the LPC3180 has an MMU, you will be confined to using uCLinux on it. As far as the user or programmer is concerned, though, there’s not a tremendous difference in the two, except (obviously) in memory management and access.
The LPC-3180 is an ARM9 MCU based on the ARM926EJ-S core.
Interestingly enough, the chip actually has an MMU. In theory, you could run a full-blown Linux distro on the chip with a proper port and the right hardware. Good call on that one.
Thanks for the replies so far. Just to be clear: What I have in mind is a “minimalistic” board - just what everybody needs (which I guess is regulated power supply, an external crystal and SRAM / FLASH).
I eg. have no plans of using USB, so I’d prefer that people using USB add their own connector/circuit. That’ll keep cost minimal. Otherwise we’ll quickly end up close to a “standard” development board.
reklipz: thanks for the link. I have seen your post before, but also you were asking for a prototyping board. I guess these are painful to develop; that’s why I am trying to reduce on-board components to a minimum. I have the impression that someone with experience in ARM9-design should be able to design such a board in almost no time: CPU and memory. All other stuff is left for hobby people (like adding IO, LCD, buttons, etc).
Anyone else interested? I guess we’d need a few people to get OLIMEX and/or Sparkfun awake
I’ll oder at least 10 boards immediately. Seriously!
For those who can afford it: Phytec does have a prototyping board, similar to what I suggest:
It features 4 rows of 50 pins each on the bottom - not suited for breadboarding, sorry linuxguy. Plus they have a lot of unused pins in their connectors. Strange…
AND… it’s 250 US$. Don’t understand what makes it so expensive. The components will be cheap, I assume. Well, and they sell it as a “one time prototype module”.
What I have in mind is a different concept: Sell the boards as “use once and trash” or “use and keep on your device (robot?) forever”. Make it so cheap that it’s worth purchasing a new board rather than reusing the old board.
jconradt:
AND… it’s 250 US$. Don’t understand what makes it so expensive. The components will be cheap, I assume. Well, and they sell it as a “one time prototype module”.
After looking at the chip, I saw that it's only available in a BGA package (or at least the datasheet led me to believe that). One of the reasons the board is so expensive is that BGA boards are almost always at least 4 layers, but can go to 8 or even more depending on the layout. Also, since you can't visibly inspect the solders on a BGA device, you have to use really expensive means (like X-ray) to verify the balls all got soldered properly.
In general, any chip that is a BGA is going to have an expensive breakout. You could always try to roll your own in Eagle, Protel, etc. and possibly get the board for cheaper, but you’re still likely going to need a 4+ layer board, and the free version of Eagle will only do 2 layers.
In all honestly, $250 for that board isn’t really a terrible price once everything is considered. I don’t even think that Olimex could do much better (if at all), especially considering that their EP9301 boards (which aren’t BGA) cost around $150.
jconradt:
Thanks for the replies so far. Just to be clear: What I have in mind is a “minimalistic” board - just what everybody needs (which I guess is regulated power supply, an external crystal and SRAM / FLASH).
The clock circuit with the crystal should be with the MCU. If it is socketed, it can be replaced. :) The RTC also would need to be setup on the board.
jconradt:
I eg. have no plans of using USB, so I’d prefer that people using USB add their own connector/circuit. That’ll keep cost minimal. Otherwise we’ll quickly end up close to a “standard” development board.
USB would have to be on the board, or a standard serial port. It's needed as a backup to program the MCU for those who can't use JTAG for some reason. If a company is going to go to the trouble of making a board like this, it has to be done such that it appeals to the widest audience.
jconradt:
Anyone else interested? I guess we’d need a few people to get OLIMEX and/or Sparkfun awake
I am definitely interested, but cost is also an issue for me.
jconradt:
For those who can afford it: Phytec does have a prototyping board, similar to what I suggest:
I've checked this out in my search for an affordable LPC-3180 board. I'll probably end up going with an i.MX COG though, based on the Freescale i.MX21. I'd much rather have an LPC-3180 based board though.
jconradt:
What I have in mind is a different concept: Sell the boards as “use once and trash” or “use and keep on your device (robot?) forever”. Make it so cheap that it’s worth purchasing a new board rather than reusing the old board.
I think the use once and trash concept is just plain silly.
I’m not going to pay for any board that is not a quality product and won’t last. I will trade dollars for higher quality.
linuxguy: I was not suggesting making it so cheap that it’s not functioning. I was trying to suggest: Make it so cheap that people are willng to buy a few copies and use them (maybe also in permanent setups), rather than buy one copy and move this between designs.
I assume everyone here has several “projects” running - or ideas that you occasionally work on. If a board is 300 US$ I’ll buy one and move it from project to project. If it’s only 50 US$ I’ll buy several and keep one in each of my projects, possibly for good.
