Well, I'm going to have to bail, guys. Feel free to carry on without me. Thanks all for joining in the chat. It's been one of the best we've had for awhile IMHO.
And I agree Java for embedded is interesting. I've blogged about a few times.
I cannot say today if you will see large parts with real cheap prices, as don't forget producing at 40nm or 55nm (sorry not 50nm) is still quite expensive. But yeah a MCU with a couple of MB on-board flash is likely to become accessible
but today smaller geometry are mainly of advantage if you take the opportunity to increase the features, as the production price partly offset the smaller chip size
Thomas , currently most microcontrollers are implemented in 90nm , because there's no flash memory for lower nodes, and as you said analog and memory determines the price of the chip . Assuming the possibility to design mcu's using 40nm , how do you see that affect micrcontrollers ?
Thomas, i heard renesas is doing an ARM at 40nm including their 40nm flash, it will include something like 10Mbyte memory. I wonder what 40nm flash means for lower end ARM micrcontrollers ?
Third try to answer the # of M0+ products: today available from Freescale and NXP. I believe around 30 devices before Christmas, it might have changed since then. For reference there are 2.000 Cortex-M based MCU currently available from a dozens of vendors
Nishant, that would have me starting work at 7am, unless I can schedule the email for automatic delivery. I'm not functional at 7am. But I'll look into the possibility. How about a tweet, instead?
Yes Rich but then the thing is eventually the cost rises I want to eliminate this for always , I can order like 10 or 20 only at a time so it becomes too costly
Rich - I had thought that would be the case, but some of the key pins aren't re-mappable, like power and ground. However, if the LPC800 were available in chip form today with the right peripheral set, that would be my choice. I'd probably only need to re-rout a third or fewer pins.
hi andyk. Is my time zone conversion wrong in my blogs? We started an hour or so ago. Just wondering if I am giving bogus advice to the world outside the US timezone
Rich, can I buy Very small quantities for this in India! I doubt that, while getting them from US etc through DIGI, Farnell or others I have to pay very hefty cost in Tax, shipping etc and here vendors only sell what is very widely sold not everything
Rich - That would have been necessary a year ago, but with NXP having several thru-hole varients, I wouldn't need tro do that now. I've got a rhu-hole LPC1114 in a solderless breadboard right now and as soon as I get my current PIC-based MCU, with an SSOP28 package, board working right, I'm going to plop an SSOP28 LPC1114 in and just re-wire the pins to match ther peripherals up.
Nishant. M0+ chips are as low as 39 cents - http://www2.electronicproducts.com/Low_cost_simple_MCU_uses_Cortex_M0-article-ICDJH02_NXP_LPC800_Jan2013-html.aspx
I've been learning ARM, but for me, the real ability to see if it can reasonably replace 8-bit will happen when I pull a 28 pin SSOP or SOIC PIC chip and replace it in the same design with an ARM of the same form factor.
and then the cost comes hard if you become stubborn and just think that you have to adopt ARM only because you are fed up of 8bitter's and would like to become updated with the tech.
The main problem with always adopting ARM is with prototyping simpler function's, like I just made a Bike controller with AVR I chose it because all that processing power in an ARM isn't just needed
@ Curt: for he ISA we try to keep backwar compatibility with our Thumb/Thumb2 technology for the MCU world, so it will be more evolutions than revolutions
It's an interesting idea to balance performance and power using two cores with the same ISA but with different implementation of the pipeline. For example Cortex-A15 is big and Cortex-A7 LITTLE
Intel getting into the MCU space would probably be as sucessful as Intel getting into the digital watch business. They tried that once but who has ever seen or heard of an Intel watch?
If you have a BIG intellectual investment in, say, 8051 + support tools proprietary code libraries, it's not hard to understand why you might want not to switch to anything else.
@Curt: good question on the ISA ;-)! For the MCU space we still very constraint by the number of gates we can use making it difficult to extend capabilities. On the Application processor side it's a different story. BTW has any of you looked at the big.LITTLE architecture from ARM?
The actuall difference between David and Goliath might be even bigger than just 100MM vs. 890MM - One is a sales price/market value, whe other is revenue which might very wel be lower than market value
Proprietary architectures have their followings. Is there enough advantage to be had to convince these folks to switch? Seems some MCU vendors are hedging their bets by introducing ARM parts now that never had them before.
Even if Intel works their way down into the low-watt SOC space for phones and tablets, I can't in my wildest imagination see them doing anything in the MCU space and they're really the only company big enough to attemt to unseat ARM.
With the other architectures being proprietary and MIPs being marginalized, I can't see an architecture really competing with ARM in the MCU space any time soon.
Personally, if the manufacturers adopt ARM and steer their investments toward richer peripheral spaces, better tools, and better multicore, I'll say "great!" But I don't want to see them close the patent office :-)
I'd watch Intel and the university labs. But no -- I don't know of a candidate to push ARM off the hill right now. Any one else have a candidate? MIPS?
Maybe someone will invent an ISA that incorporates more probabalistic computing capability -- which is, perhaps, the key to machine cognition? But meantime, I agree -- multicore is where the architecture action seems to be now.
@curt, where I am seeing innovation in the MCU core is in the area of multicore and in particular redundancy with lockstep operation for safety. I think there could be a lot more done with security as well. But these are specialized areas for now, although I think security will become increasingly important
Things like instruction pipe-lining, anticipatory execution and that sort of thing are pretty transparent to an IDE -- at least until it's time to use the debugger... But who knows what other breakthhroughs are still waiting to happen in ISA?
So, the question is do we need different MCU cores?
Need? probably not. There is a confusing array out there and certainly for my part (here on the trailing edge) I could lve happily with the 8051, especially when I have great pepriperals.
But we will always strive to improve, and segments of the market drive changes. It is up to the chip designers to decide whether the market justifies the expennse and whther to genearlise the design to expand the market to other users.
There used to be a big debate about CMOS vs Bipolar vs ECL and that pretty much went away. Wonder if the architecture question will follow the same trajectory
If we're talking about Harvard vs Von Newman architecture, the code would make a difference. bit depth in memeory addressing, registers and accumulators would, and, of course, speed. Other than that, it's pretty much just the peripheral set. Like you said, Rich, an ARM could have any of those PIC features I mentioned.
So, the question is do we need different MCU cores? If everyone uses the ARM core, the tools become easier to use for the basic programming, right? Then the only variations are in how to use the peripherals
AD - I would egree with that point. I would say that if I took the IAR ARM tool, the Code Red ARM tool and Micrchip MPlabX, it wouldn't be that difficult to see that the biggest variation is in the IDE, not the chips.
@duane, true these are improvements, but not in the processor core of the MCU, only in its surroundings. An ARM chip with all these features should be quite feasible
Microchip, for example: For quite some time, every chip looked pretty much the same, but with just more or less of the same features. Recently though, they've added in things like programmable logic, crystalless USB and hardware multipliers to their 8-bit lines
Depending on the IDE the micro is almost irrelevant. I am currently using the Cypress PSoC5 which is ARM M3. But the user interface is almost exactly the same as the PSoC3 which is 8051. Most of the stuff you develop for the one is the same on the other (except for execurtion speed). In this case all the peripherals are almost the same and so the processor is almost irrelevant. I certainly don;t feel any special performance because it's an ARM
@duane, that would be interesting to know. Need to talk with an analyst to see what can be figured out. I do see more companies embracing ARM along with their proprietary MCU designs
Yes, I think so. But again, there doesn't seem to be much innovation going on in the basic MCU architecture, so maybe a defacto standard is a good thing, net-net
I would be interested to know how many people are switching to an ARM product vs. how many ARM people are just increasing the number of deigns and products they're putting out.
A "standard" ISA would open the door to more efficient development -- but close the door some on innovation in MCU architecture. That seems to be how standards work.
When I first started looking at the ARM seriously, I liked the IAR tool better. However, as I've gotten deeper in, I've found that the Code Red has been easier for me to get to know in more detail
Yes - It'll be a plug in to the power driver boards. I'm going to have a raneg of drivers and probably a few different MCU boards. I'm almost done with the first driver board - for a pair of half amp motors
This board is loosley based on some other I have and I just about stoppped and put a 28 pin SSOP PLC1114 in it, but decided to finish up as is with the PIC and then make an ARM version.
Just a reminder - the chat on ARM's potential for domination is scheduled for Thursday the 17th at 11am EST (1600 gmt). Please join in for what promises to be a spirited discussion.
I'm looking forward to this topic sans server issues. I'd also be curious to hear from any MIPs experts on their thoughts about how MIPs and ARM compare.
We tried this topic before, but a server failure cut our chat short. So, we'll try once again. the ARM architecture is increasingly being adopted by MCU vendors as support for proprietary MCU architectures becomes increasingly difficult. How far will ARM be able to go in the market?
To save this item to your list of favorite Microcontroller Central content so you can find it later in your Profile page, click the "Save It" button next to the item.
If you found this interesting or useful, please use the links to the services below to share it with other readers. You will need a free account with each service to share an item via that service.
Tweet This
