2012-03-13 08:47 AM
Hello! I have a question to STM32xx chip manufacturers.
Last yar I uses STM32F10xx chips in my applications. It's 8mHz HSI works perfect - stable, with small temperature sensistivity - works very well. CAN bus (slow) works perfect on these chips without crystals! Now I decide to change chips to modern 32F2xxx, but after testing I see it's HSI works pure - very temperature/voltage sensitive, and somthing else... In my application: VCAP1 = 2.2uF VCAP2 = 2.2uF VDD1...5, VREF+ connects to 3.3v with 100uF capacitor I have written the chip temp. monitoring loop in my application programm, but it helps to stabilize HSI frequency only for 60% chips (by manufacturer tape pack). 40% chips can't work anywhere.. Can someone say any way to make HSI stable same as STM32F10xx serie? #stm32f2-hsi-problems2012-03-13 11:22 AM
(CAN) bus, demands communication accuracy of less than ±1.0 percent because the microcomputers used are equipped with this high-speed communication function. Specifically, when a CAN-bus performs high-speed communications of 500Kbps, an accuracy of ±0.3 percent or less is required for a resonator used in the ECU.
The datasheet for the f1xx do not guarantee it to work within these limites.that it happened to was a matter of luck. ''design by test'' is guaranteed to bite you where you sit eventually. THUS, if the f2xx stay within the limits stated in the datasheet, you have no course of action. Erik
2012-03-13 02:51 PM
The STM32F1's HSI was specified as 1% typical, the STM32F4 is specified as 1%. It's trimmed, but at some arbitrary temperature, and is not temperature compensated.
The LSI on the F1's is 40 KHz, (30 - 60 KHz) @ 20-50% None of these are adequate for CAN if you expect 0.3% , especially across industrial/car temperature profiles. What's the problem with using a crystal or TCXO?2012-08-04 11:55 PM