rcooke:
I wasn’t getting much response from this forum so I looked around the 'net a while and I found this article: http://www.embedded-control-europe.com/ … t05p22.pdf
Which compares the MAXQ part to the TI MSP430, ATmega16L, PIC16F87 and the EM6812 (which I’d never heard of before). If I’m reading it correctly the MAXQ part is fast but not particularly stingy with the battery power. My app needs as much battery life as I can get.
I sent a message to the Maxim Tech Support guys and received a nice note back saying that for my app the MAXQ probably isn’t the best choice. It was actually refreshing to get something like that from the manufacturer - they didn’t try and “sell” me on their part.
For me I’d like to use the MSP430 but it is $6.21 (qty 100) vs. the Atmega32L at $4.75 (qty 100). I think I can run it in a low power mode for most of the time so the 1.1ma current drain (@1MHz) should be OK.
First, the article you are referring to was written by the manufacturer of the EM6812 device, so I would be very leary of any conclusions/test cases it presents.
Second, for the MSP430/AVR/MAXQ/PIC type processors battery consumption depends greatly on the specific application, each processor has certain modes where they are very efficient and others which aren’t. It’s very easy to come up with tests that shows one processor is more power efficient then the other.
For example, you say you can run the MSP430 at 1Mhz at 1.1ma (I thought it’s much lower then that), but the MAXQ2000, can run at 1Mhz at .85mA, if you run the code out of SRAM the current drops to .14mA (see page 2 of http://pdfserv.maxim-ic.com/en/ds/MAXQ2 … 00-RBX.pdf), I am sure the EM guys didn’t run the MAXQ2000 in this mode. Maybe you can, maybe you can’t but if your code is small enough you can copy it out of FLASH into SRAM and run it there and save mucho power.
Another example, the MAXQ2000 and MSP430 can both perform a 16 bit multiply in 1 cycle, but it takes the MSP430 5 cycles to grab a word indirectly from SRAM, the MAXQ2000 can do it in 1 cycle, so that’s a big factor IF you are multiplying two huge tables in SRAM (most tests assume all the data is in registers to begin with).
All I’m saying is that you have to look at your specific application to determine what is the lowest power processor for your task. I’m curious what you exact question was to Maxim that they said their device wasn’t the best solution.
