STM32 New Year wish list

[gdp123a]

Posted on May 07, 2009 at 13:34

STM32 New Year wish list

[gdp123a]

Posted on May 17, 2011 at 12:57

JJ suggested people submit their top suggestions for improving any future versions of the STM32. Here are my five:

1. reduced current consumption for applications <= 8MHz (if such a chip has restricted memory size and frequency range that's fine by me) 2. much reduced sleep current for applications <= 8MHz 3. automatic CS for SPI transfers (so it can work seamlessly with DMA, ADCs, DACs etc.) 4. the ability to wake from STOP mode using the SECONDS RTC interrupt (rather than the RTC Alarm interrupt) (the SECONDS interrupt is a lower overhead, more robust/safer way of generating regular interrupts eg. 512Hz, 1024Hz,... ) 5. more options for the 32KHz clock to drive timers (if you are relying on the 32KHz crystal clock alone, then more timers linked to this clock would be useful.) So I don't sound ungrateful for what the STM32 has already provided, I should point out the things I really like about the STM32 and why I have chosen it for a number of my designs. - operation down to 2.0 volts (great for regulated/unregulated 3volt lithium battery applications) - small, but usable 36/48 pin packages - rich/powerful instruction set - single cycle 32 bit multiplication (great for higher precision digital filters) - single cycle 32 bit shifts - well suited for 32KHZ crystal based designs - 16 bit SPI frame option (great for those 16bit ADCs and DACs, pity about the CS though) - simple bootloader option - plenty of RAM, even on the low-end chips - good low power and shut down modes - speed selectable IOs to reduce EMI

[er2]

Posted on May 17, 2011 at 12:57

And again...

using USB and CAN at the same time in 48-pin devices. Happy new year Erwin

[16-32micros]

Posted on May 17, 2011 at 12:58

Hi Greg,

Thanks for valuable post, Could you please give me more details about your points (1), (2) & (3) : what target current & sleep consumption your applications need really ? some figures would be nice. Regarding (3) : is it possible to make me a synthesis :) and if possible some legacy or others MCUs are able to do it. Hi Erwin, Thanks for your post, Just to know why only 48-pin package is interesting for you ? not the others may be as well to have CAN & USB running at the same time like having a USB-to-CAN bridge ? Thank you. Cheers, STone-32.

[relaxe]

Posted on May 17, 2011 at 12:58

This thread is a good idea, imho.

My personnal story: I've used/programmed PIC16 in Assembly, then I was into PIC18 programming in C at my current job. Then, a new project came along, and I was looking to deal with a ARM core for a long time. The Cortex M3 had very high promises, and I choosed ST as the STM32 apps notes, datasheets, and firmware library were lokking very complete and well documented. Now, I'm a ARM develloper since 3 months only :D. The STM32 is very very great, lots of peripheral in lots of configurations. All around, my STM32 experience is extremelly positive, and I want to thank everybody at ST for making our ''thin-budget/extreme deadlines'' project a reality. Now, the things I've personnaly noted: 1) The firmware examples are so torough and well written, they deserve a better showcase, like a ''application library'', so one could load both firmware and applicatio libs and use all of these wonderfull functions already there in the various example. This would make really simple, every-day use of normal protocol easier [Like a WriteI2C(char * Buffer, u32 NumberOfBytes)]. 2) The MCU 2.0V minimal voltage is a pain. Bring it to 1.8V, or give us an option to disable the internal regulator and feed the core by our own regulated 1.8V. We are devellopping a battery power app, and most of the peripherals we use are designed for 1.8V interfaces. Translating from 1.8V to 2.0V add to part counts, and make me shouting ''Why! Oh Why!!'' in my office. Also, as we power the chip at 2.0V, and I suspect the core is at 1.8V, I know there is a 10% power loss just there. 3) The ADC power (VCCA) is a pain (too). We want to save energy, and run the core/peripherals at to lowest power possible. Then, the ADC requires 2.4V. I can live with that. Problem is, VCCA must be lower or equal to VCC! Making a requirement to raise the whole chip to 2.4V. If we could supply a clean, seperate 2.5V to the ADC and a small 2.0V(or less) to the chip, then it would be a great peripheral. 4) Make the ADC 16bits, it then would be a killer audio processor. There! Have a nice year to everybody at ST and those who cares to read and post on this forum :p -Relaxe

[janek]

Posted on May 17, 2011 at 12:58

My wishes are among others:

- 100MHz clock with faster flash - see NXP lpc1xxx - 32-bit timers instead of 16-bit ones, can be lower number of timers - again see NXP ARM7 devices - better pin assigment, FSMC data and addresses align with GPIO, FSMC doesn't need to have second alternative function (only FSMC/GPIO) - see Microchip PIC24xx with peripheral pin select New devices don't need to be pin compatible - USB OTG high speed - application library for RNDIS (ethernet over USB) - and please tell your friends from optical sensor division that VS6724 camera moule is needed, make it easily available in small volume and you will see people releasing their creativity. Jan

[obtronix]

Posted on May 17, 2011 at 12:58

I would just like to see a roadmap of what they are planning to give us and budgetary prices for all the chips on their website, like TI/AnalogDevices/Microchip does.

[16-32micros]

Posted on May 17, 2011 at 12:58

Hi obtronix,

Could please give me more details on what is your wish ? Do you like to see only the prices of our STM32 devices... Thanks for your post ! Cheers, STOne-32.

[picguy]

Posted on May 17, 2011 at 12:58

** DMA to/from GPIO pins. Some kind of strobe/ack will be needed.

** ADC regular channel conversions using scan mode results should go to dedicated registers. Likely best to save by position in ADC_SQRx and have 6 result halfwords for each sequence. Saving by ADC channel number would have worked for me. (I ended up using the confusing injected channels.) ** A timer mode where an external voltage level (via analog watchdog??) could control count enable. Easy enough using s/w interrupts but by no means fast enough to control input excursions only several clock periods long.

[muratterzi]

Posted on May 17, 2011 at 12:58

i wish st microelectronics develops a software like metrowerks visual bean design ide. if this occurs, usingwith this high performanced mcu series can be used more easily and efficiently without reading page to page datasheets:)