ST / stm-spirit1-rf-driver

Prototype RF driver for STM Sub-1 GHz RF expansion board based on the SPSGRF-868 module for STM32 Nucleo.

Prototype RF Driver for STM Sub-1 GHz RF Expansion Boards based on the SPSGRF-868 and SPSGRF-915 Modules for STM32 Nucleo

Currently supported boards:

Note, in order to use expansion board X-NUCLEO-IDS01A4 in mbed you need to perform the following HW modifications on the board:

  • Unmount resistor R4
  • Mount resistor R7

Furthermore, on some Nucleo development boards (e.g. the NUCLEO_F429ZI), in order to be able to use Ethernet together with these Sub-1 GHz RF expansion boards, you need to compile this driver with macro SPIRIT1_SPI_MOSI=PB_5 defined, while the development board typically requires some HW modification as e.g. described here!

This driver can be used together with the 6LoWPAN stack (a.k.a. Nanostack).

Files at this revision

API Documentation at this revision

Revision 44:aedd63cb0ce3, committed 2016-12-16

Comitter:
Wolfgang Betz
Date:
Fri Dec 16 13:46:02 2016 +0100
Parent:
43:a512f909514a
Child:
45:2d01cc9bc761
Commit message:
Set 'MAX_PACKET_LEN' to 128

Changed in this revision

source/SimpleSpirit1.cpp Show annotated file Show diff for this revision Revisions of this file
source/libs/Contiki_STM32_Library/spirit1-config.h Show annotated file Show diff for this revision Revisions of this file 
--- a/source/SimpleSpirit1.cpp	Fri Dec 16 12:06:21 2016 +0100
+++ b/source/SimpleSpirit1.cpp	Fri Dec 16 13:46:02 2016 +0100
@@ -138,8 +138,6 @@
 	};
 	spirit_gpio_init(&x_gpio_init);
 
-// #ifdef RX_FIFO_THR_AO_CSMA_WA
-	// betzw - TODO: enabling CSMA/CA seems to be incompatible with TX FIFO usage (to be investigated)
 	/* Setup CSMA/CA */
 	CsmaInit x_csma_init = {
 			S_ENABLE,         // enable persistent mode
@@ -150,12 +148,9 @@
 			8                 // BU prescaler
 	};
 	csma_ca_init(&x_csma_init);
-// #endif
 
-#ifdef RX_FIFO_THR_AO_CSMA_WA
+#ifdef RX_FIFO_THR_WA
 	linear_fifo_set_almost_full_thr_rx(SPIRIT_MAX_FIFO_LEN-(MAX_PACKET_LEN+1));
-#else
-	linear_fifo_set_almost_full_thr_rx(20); // betzw: heuristic value
 #endif
 
 #ifdef USE_STANDBY_STATE
@@ -190,10 +185,7 @@
 
 	pkt_basic_set_payload_length(payload_len); // set desired payload len
 
-// #ifdef RX_FIFO_THR_AO_CSMA_WA
-	// betzw - TODO: enabling CSMA/CA seems to be incompatible with TX FIFO usage (to be investigated)
 	csma_ca_state(S_ENABLE); // enable CSMA/CA
-// #endif
 
 	int i = 0;
 	int remaining = payload_len;
@@ -229,10 +221,7 @@
 
 	_spirit_tx_started = false; // in case of state timeout
 
-// #ifdef RX_FIFO_THR_AO_CSMA_WA
-	// betzw - TODO: enabling CSMA/CA seems to be incompatible with TX FIFO usage (to be investigated)
 	csma_ca_state(S_DISABLE); // disable CSMA/CA
-// #endif
 
 	cmd_strobe(SPIRIT1_STROBE_RX); // Return to RX state
 
@@ -518,8 +507,8 @@
 		debug_if(!((*tmp) & IRQ_RX_DATA_READY_MASK), "\n\rAssert failed in: %s (%d)\n\r", __func__, __LINE__);
 #endif
 
-		if(!_is_receiving) { // spurious irq?!?
-#ifdef DEBUG_IRQ
+		if(!_is_receiving) { // spurious irq?!? (betzw: see comments on macro 'RX_FIFO_THR_WA'!)
+#ifdef HEAVY_DEBUG
 			debug("\n\r%s (%d): irq=%x\n\r", __func__, __LINE__, *tmp);
 #endif
 		} else {

--- a/source/libs/Contiki_STM32_Library/spirit1-config.h	Fri Dec 16 12:06:21 2016 +0100
+++ b/source/libs/Contiki_STM32_Library/spirit1-config.h	Fri Dec 16 13:46:02 2016 +0100
@@ -42,14 +42,23 @@
 #define SPIRIT_MAX_FIFO_LEN         (96) // betzw - WAS: 600
 /*---------------------------------------------------------------------------*/
 
-/* Define beyond macro when your network includes a platform - like the K64F from
- * Freescale - which suffers from a not yet analyzed (HW) bug in delivering/receiving
- * further interrupts by/from Spirit after having reached the RX FIFO threshold and
- * elaborated the corresponding 'IRQ_RX_FIFO_ALMOST_FULL' interrupt.
- * Furthermore, enable this macro if you want to use CSMA/CA.
- * NOTE: this enables just a workaround!!!
+/* Sometimes Spirit1 seems to NOT deliver (correctly) the 'IRQ_RX_DATA_READY'
+ * event.
+ * This can be avoided by reducing the maximum packet length to a value which
+ * is lower than the RX FIFO size.
+ * The mbed driver currently implements another workaround to this which allows
+ * stable packet delivery of packets with maximum length up to 128 bytes
+ * (which is the recommended 6LoWPAN payload length).
+ * Enable beyond macro if you want to use the version of the driver which avoids
+ * FIFO overflows by reducing packet length.
+ *
+ * NOTE: the non delivery of event 'IRQ_RX_DATA_READY' MUST still be
+ *       investigated further deeply (both on HW & SW level)!
+ *       Furthermore, the current limit of 128 bytes should also be overcome,
+ *       which again requires a further analysis of the RX process, which
+ *       currently seems to run into RX FIFO (most likely) overflow situations.
  */
-// #define RX_FIFO_THR_AO_CSMA_WA
+// #define RX_FIFO_THR_WA
 
 /**    
  * The MAX_PACKET_LEN is an arbitrary value used to define the two array
@@ -57,10 +66,10 @@
  * The SPIRIT1 supports with its packet handler a length of 65,535 bytes,
  * and in direct mode (without packet handler) there is no limit of data.
  */
-#ifdef RX_FIFO_THR_AO_CSMA_WA
+#ifdef RX_FIFO_THR_WA
 #define MAX_PACKET_LEN              (SPIRIT_MAX_FIFO_LEN-1)
 #else
-#define MAX_PACKET_LEN              (255) // betzw - WAS: SPIRIT_MAX_FIFO_LEN, but LEN_WIDTH is set to 7 so the variable payload length is theoretically from 0 to 255 bytes
+#define MAX_PACKET_LEN              (128) // betzw - WAS: SPIRIT_MAX_FIFO_LEN, but LEN_WIDTH is set to 7 so the variable payload length is theoretically from 0 to 255 bytes
 #endif
 
 /*---------------------------------------------------------------------------*/