Embedded Developer

This post continues the low level use of the STM8S registers and shows you how to configure the STM8S system clock. In particular we will configure the system clock to use the internal high speed clock.

So let us start with a little background. The STM8S can be driven by three clock sources:

• High Speed Internal (HSI) RC oscillator running at 16MHz
• Low Speed Internal (LSI) RC oscillator running at 128KHz
• External clock (1-24 MHz) which can be either an external crystal or user supplied clock.

We will be using the internal oscillator as this will be satisfactory for our purposes. An external clock may be more accurate but for most of the work I am doing I am happy to sacrifice the accuracy for the lower component count and greater simplicity of my designs.

At start up, the STM8S will be using HSI with a prescalar to bring the initial clock speed down to 2MHz. We will be resetting the clock prescalar to allow the chip to run at the full 16MHz.

The switching algorithm we will use is as follows:

1. Reset all of the clock registers to the power on state
2. Select the clock we wish to use
3. Enable the switching mechanism
4. Wait until the STM8S indicates that the clock has stabilised

Clock Registers

The description of the clock registers start around page 89 of the STM8S Reference Manual. In this article we will run through the registers concentrating on the values we will be setting. For a fuller description you should refer to the STM8S Reference Manual.

CLK_ICKR – Internal Clock Register

The only bits we will really be interested in for this example bits 1 and 0, this allows us to select the HSI and see if it is ready.

Our example here will be resetting the clock source on the STM8S chip to use the internal oscillator using HSIEN and using the HSIRDY bit to determine when the clock has been setup and is stable.

CLK_ECKR – External Clock Register

This register is used to provide information about the state of the external clock. We will not be using this register for anything other than resetting the register to the power on state.

CLK_CMSR – Clock Master Status Register

This register is set and cleared by hardware and indicates which clock source is currently selected as the master clock source.

CLK_SWR – Clock Master Switch Register

This register is written to by the user program and is used to select the master clock source. The register cannot be written to whilst a clock source switch is on-going. You can test for the switch using the SWBSY in the CLK_SWCR register.

Clock Switch Control Register – CLK_SWCR

The Clock Switch Control Register is used to control how and when the clock switch is performed.

In our case we will be using this register to start the clock switch process.

CLK_CKDIVR – Clock Divider Register

The system clock can be divided down to allow for lower clock frequencies to be used. This register allows for the HSI prescalar (divider) and the CPU prescalar to be set.

The following values can be used for the HSI prescalar:

The following values are allowed for the CPU prescalar:

In this case we will want the CPU to be running at full speed and so no prescalars will be applied to either the master clock or the HSI.

CLK_PCKENR1 & CLK_ PCKENR2 – Peripheral Clock Gating Registers

These two registers enable (set to 1 and disable (set to 0) the clocks used to control the peripherals used in the application. I have fallen foul of these values a few times. If you set these registers or zero then you will turn off the peripheral clocks. This has the effect of turning off the feature you may be trying to use. If in doubt set both of these registers to 0xff. You can always refine the values later by restricting the clock signals to only those which you are using.

Enabling the service will use the frequency of the master clock. Disabling the clock to the service will turn it off.

The <iosstm8s103f3.h> file does not contain definitions for these bits, you need to define your own constants for these. I have recently taken to just turning them all on as I’m not too worried about the power requirements for the applications I am working on.

CLK_CSSR – Clock Security System Register

We will not be using this register in this example apart for setting the default power on value. Please see the STM8S Reference Manual for more information.

CLK_CCOR – Configurable Clock Output Register

We will not be using this register in this example apart for setting the default power on value. Please see the STM8S Reference Manual or future examples for more information.

CLK_HSITRIMR – HSI Clock Calibration Trimming Register

We will not be using this register in this example apart for setting the default power on value. Please see the STM8S Reference Manual for more information.

CLK_SWIMCCR – SWIM Clock Control Register

We will not be using this register in this example apart for setting the default power on value. Please see the STM8S Reference Manual for more information.

Software

As previously mentioned, this example will simply set the system clock to the maximum speed the HSI can deliver. We will then toggle a GPIO pin and hook this up to a scope to get an idea of the output. We will never see the full 16MHz clock speed of the HSI but we should be able to see that the output frequency changes when we change the clock dividers.

So first thing, let’s get the system set up and running at the full 16 MHz and see what we the scope produces.

#include <intrinsics.h>
#include <iostm8s103f3.h>

//
//  Setup the system clock to run at 16MHz using the internal oscillator.
//
void InitialiseSystemClock()
{
    CLK_ICKR = 0;                       //  Reset the Internal Clock Register.
    CLK_ICKR_HSIEN = 1;                 //  Enable the HSI.
    CLK_ECKR = 0;                       //  Disable the external clock.
    while (CLK_ICKR_HSIRDY == 0);       //  Wait for the HSI to be ready for use.
    CLK_CKDIVR = 0;                     //  Ensure the clocks are running at full speed.
    CLK_PCKENR1 = 0xff;                 //  Enable all peripheral clocks.
    CLK_PCKENR2 = 0xff;                 //  Ditto.
    CLK_CCOR = 0;                       //  Turn off CCO.
    CLK_HSITRIMR = 0;                   //  Turn off any HSIU trimming.
    CLK_SWIMCCR = 0;                    //  Set SWIM to run at clock / 2.
    CLK_SWR = 0xe1;                     //  Use HSI as the clock source.
    CLK_SWCR = 0;                       //  Reset the clock switch control register.
    CLK_SWCR_SWEN = 1;                  //  Enable switching.
    while (CLK_SWCR_SWBSY != 0);        //  Pause while the clock switch is busy.
}

//
//  Main program loop.
//
int main(void)
{
    __disable_interrupt();
    //
    //  Initialise Port D.
    //
    PD_ODR = 0;             //  All pins are turned off.
    PD_DDR_DDR4 = 1;        //  Port D, bit 4 is output.
    PD_CR1_C14 = 1;         //  Pin is set to Push-Pull mode.
    PD_CR2_C24 = 1;         //  Pin can run up to 10 MHz.
    //
    //  Now setup the system clock
    //
    InitialiseSystemClock();
    __enable_interrupt();
    while (1)
    {
        PD_ODR_ODR4 = 1;    // Turn Port D, Pin 4 on.
        PD_ODR_ODR4 = 0;    // Turn Port D, Pin 4 off.
    }
}

Running this code resulted in the following trace on the oscilloscope:

GPIO Output when the System Clock is set to HSI with no Divider.

The frequency of the out was 2.693Mhz.

So now let’s see what happens when we set the prescalar for HSI to be 2 (i.e. half the clock speed. To do this we need to edit the code slightly, so change this:

CLK_CKDIVR = 0;                     //  Ensure the clocks are running at full speed.

to this:

CLK_CKDIVR = 0;                     //  Ensure the clocks are running at full speed.
CLK_CKDIVR_HSIDEV = 1;				//	Set the HSI divider to 2.

The result is the following trace:

GPIO Output when System Clock is set to HSI divided by 2

This looks remarkably like the trace above but the oscilloscope shows that the frequency of the signal is now 1.335MHz, or approximately half of the frequency of the first trace.

The full project for this article can be downloaded from here. This code has been tested with my reference platform, the Variable Labs Protomodule and the STM8S Discovery board.

Compatibility

System	Compatible?
STM8S103F3 (Breadboard)
Variable Lab Protomodule
STM8S Discovery

Komodex Labs recently released a couple of new modules for the Netduino GO! platform. The first is a seven segment display module and as we will see later this can be used to show sensor data. The second module is a breakout module for the GO! bus and this is probably more interesting to module developers.

As the postman was kind to me and managed to get these modules to me this week I thought I’d hook them up to the light sensor I have been playing with to see what they can do.

A Little Background

I have recently been working with a simple sensor to teach myself how to work with the STM8S. It has been a long slow task as the normal competing pressures of work, the house and general life stuff keeps calling for my attention.

The project uses a TSL235R sensor which converts the current ambient light level into a square wave with 50% duty cycle. The STM8S periodically reads the frequency of the signal generated by the sensor. This value is then made available to the Netduino GO! via the GO! Bus.

This project is going to read the value from the STM8S and display it on the seven segment display. The value is a 16-bit value and so can range from 0 to 65535 so we will be showing the value in hex.

Seven Segment Module

The seven segment display is a nice little module very professionally finished. It feels a solid – sort of odd to say that about a module. I was kind of surprised by the weight but then I have only been working with small modules so far.

I’d write more but I think the module is best described my Matt’s blog post. This shows some photos of the module along with a description of the API used to control it.

GO! Bus Breakout Module

This is a handy little module which only had a short production run. I was lucky enough to have ordered one before stock ran out. The module is aimed at module developers. It has a couple of sockets for module cables along with some handy breakout points.

Komodex Labs GO! Bus Breakout Module

The idea is simple, you connect your Netduino GO! to one side of the breakout board and your module to the other side of the breakout board. You can then hook up test equipment (scope / logic analyser etc.) to a row of headers on the board. This allows you to debug your module by snooping on the data going down the GO! Bus.

An alternative configuration and one I am using for development on breadboard, is to connect the breakout module to breadboard with the Netduino GO! connected to one side of the breakout board. This allows you to feed the GO! Bus signals into your circuit whilst at the prototype stage.

The breakout also has a lovely feature in that you can connect the ST-LInk/V2 module to the board and at the flick of a switch you can disconnect the SWIM/NRST signals from the Netduino GO! and connect them to the ST-Link/V2 instead. This allows you to re-program the STM8S on the fly without having to modify your circuit – NEAT !

For more information about this module I’d head over to Matt’s blog post.

Hooking it all up

Connecting the Seven Digit Module up to the go was easy, it simply needed a standard 5cm GO! Bus cable and that’s it.

The Breakout module was a little more difficult as it had to be hooked into an existing prototype circuit. From the Netduino side it was easy, just another 5cm GO! Bus cable. The ST-Link/V2 connection was also easy as the module has a connector already on the board and it is designed to ensure that the connection can only be made one way. The only complication was then ensuring that the connections between the GO! Bus and the prototype circuit were correct. Adding this module has in fact made my prototype circuit a little cleaner.

Here’s what the final circuit looks like:

Komodex Seven Segment in Light Sensor Prototype Circuit

Installing the Drivers

The Seven Segment Module is supplied with a set of drivers in the form of an executable. The drivers can be found on the Komodex Labs web site. Installation was quick, simple and painless.

Komodex Labs also provide the source code for the drivers along with a sample application. This is a separate download and as you will see later, this came in handy.

As you would expect, the breakout module does not require any drivers as there is no interaction between the module and the PC.

Writing the Software

The API provided with the module is comprehensive but in this case we are going to use only a small part of it. We need to display a hex representation of a 16-bit number.

So the first step is to create a new project and add a reference to the module. You will have to browse to the directory in which you installed the drivers and add the Komodex.NETMF.SevenSegmentDisplay.dll file. On my machine this was installed in C:Program Files (x86)KomodexModule DriversAssembliesv4.2

The next step is to add a using statement

using Komodex.NETMF;

and then declare and instantiate an instance of the module:

SevenSegmentDisplay display = new SevenSegmentDisplay();
display.SetValue(0);

The above shows the module being instantiated and the display value set to 0.

The remainder of the program continuously polls the sensor for a reading. This reading is converted into a string containing the hexadecimal representation of the reading. This reading is then displayed on the Seven Segment Module. The code looks something like this:

while (true)
{
	short diff = module.GetReading();
	string output = "";
	char c;
	for (int index = 3; index >= 0; index--)
	{
		int nibble = (diff >> (4 * index)) & 0xF;
		if (nibble < 10)
		{
			output += nibble.ToString();
		}
		else
		{
			c = (char) ('A' + nibble - 10);
			output += c.ToString();
		}
	}
	display.SetValue(output);
	Thread.Sleep(1000);
}

The line short diff = module.GetReading(); obtains a reading from the sensor module which I am prototyping. This value could come from any sensor or from any source / calculation.

Hitting F5 deployed the code to the Netduino GO! and values quickly started appearing on the display. Success!

No… Wait…

I’m getting some unexpected output. Every now and then the module displays a number with some spaces in place of digits. Hmmmmm… what is going on!

Time for a quick test program. I know that the module works OK as the test projects ran fine. Adding a Debug.Print of the output variable gave me a clue. There was a pattern. Any numbers with the digit 9 in them showed a space instead of a digit. A quick test program showed that displaying 2929 where the variable of type int displayed the number correctly. Doing the same where the value was stored in a string showed any digit except the digit 9, these were replaced with spaces. Now we know where the problem is.

As noted earlier, Komodex Labs supply the source code for the drivers. So opening this project and some poking around the code for the display resulted in the problem being found. A quick modification to the source, recompile and then reference the newly compiled DLL in my code and Success !

Edit: Komodex labs have confirmed that the driver has been patched and as of 27/07/2012 the driver download contains the fix for the problem noted.

Here is a short video of the sensor and display working together:

If you have been following this blog for a while you will be aware that I have recently changed the way I write code for the STM8S from using the STD Peripheral Library to using direct register access. This has required that I go back to basics learning a new way to perform tasks I had only just mastered. The Way of the Register posts will show you have to control the STM8 using register access.

This first post will look at a simple task, controlling GPIO lines. The aim of this first exercise is to simply toggle one of the GPIO lines.

Hardware

For this task we need very little hardware, only the STM8S set up on breadboard. For this we need a minimum of two capacitors and the chip itself. A 1uF capacitor is placed between V_SS and V_CAP and a 100 nF capacitor is placed between V_DD and V_SS. An additional 1uF capacitor is also placed across the +3.3V and ground of the power supply. The following shows this set up on breadboard along with the breadboard power supply, connections to the oscilloscope and the ST-Link programmer:

STM8S103 set up on Breadboard

The Registers

Before we can start coding we need to have a look at the registers we will be using. The GPIO pins have five registers used to control the GPIO function and set/read the value of a pin. These are:

The following sections provide a brief description of the registers. For more information you should consult RM0016 – STM8S Reference Manual.

Px_DDR – Data Direction Register

This register determines which or the GPIO pins are outputs and which are inputs. A value of one in a bit sets a port line to output and zero sets the port line to input. You can set the value of the whole register or each individual bit in the register.

To set the whole register in one go you would execute something like the following:

PD_DDR = 0x0f;

This sets bits 0-3 inclusive of port D to output whilst bit 4-7 are inputs.

An alternative to this is to set up each GPIO pin individually, so the equivalent of the above is:

PD_ODR = 0;
PD_DDR_DDR0 = 1;
PD_DDR_DDR1 = 1;
PD_DDR_DDR2 = 1;
PD_DDR_DDR3 = 1;

Px_CR1 & PxCR2 – Control Register 1 and 2

These two registers determine the properties of the pins depending upon the operating mode of the pin. The following table summarises the operating modes:

In our case, we need the output pin to operate in Push-Pull mode and as we do not know the final speed of the output signal we will set this to the maximum, 10 MHz.

As with the data direction register, each bit in the registers maps to a specific GPIO on each port. In the case of these registers the bits are labelled C10-C17 (for Control Register 1) and C20-C27 (for control Register 2). The format appears to be CRB, where C stands for Control Register; R is an integer representing the register number and B is an integer representing the bit to be set.

In our case, this results in the following code:

PD_CR1_C14 = 1;
PD_CR2_C24 = 1;

Px_ODR – Input/Output Data Registers

These two registers allow you to read the value of an input pin and set the value of an output pin. As with the previous registers you can set/read the register in its entirety or bit by bit. So to turn all of the outputs off you would execute something like:

PD_ODR = 0;

In the case we have been using we will be outputting data on pin 4 of port D. So to set the pin high you would execute the following statement:

PD_ODR_ODR4 = 1;

Software

Now we have the theory in place lets put together a simple application which uses it. The following code sets up port D to have pin 4 configured as an output operating to 10 MHz. The main program loop then simply keeps flipping the bit. The result should be a square wave signal.

#include <iostm8S103f3.h>

int main( void )
{
    //
    //  Initialise Port D.
    //
    PD_ODR = 0;             //  All pins are turned off.
    PD_DDR_DDR4 = 1;        //  Port D, bit 4 is output.
    PD_CR1_C14 = 1;         //  Pin is set to Push-Pull mode.
    PD_CR2_C24 = 1;         //  Pin can run upto 10 MHz.
    //
    //  Now lets toggle to IO line.
    //
    while (1)
    {
        PD_ODR_ODR4 = 1;    // Turn Port D, Pin 4 on.
        PD_ODR_ODR4 = 0;    // Turn Port D, Pin 4 off.
    }
}

If we hook an oscilloscope up to PD4 on the STM8 chip we see the following:

Simple GPIO Output

As you can see, we have a square wave. Note that the wave is not symmetrical. This is due to the branch instruction being executed in the while loop to go back to the start of the loop.

Source Code

Here is the IAR project and source code for this post. This code has been tested on my reference platform, Variable Lab Protomodule and the STM8S Discovery board.

Compatibility

System	Compatible?
STM8S103F3 (Breadboard)
Variable Lab Protomodule
STM8S Discovery

A few weeks ago I had a rant about the STM8S Standard Peripheral library after it cost me a fair amount of time tracking down what appeared to be a bug in the library. As a result of this I have moved over to accessing the registers on the chip directly to control the operation of the chip. A recent question ion one of the forums I haunt has prompted this post. Here I am going to present a few different ways of achieving the same task, one using the STD Peripheral Library and two examples using the registers directly but in different ways.

The task we are going to be looking at is one I perform as part of my initialisation of the STM8S, namely I set the clock up to a known state. In this case we will be setting the system to use the HSI clock running at 16 MHz with no dividers.

It is important to note that you will need to have a copy of the reference manual for the chip available when using direct register access to control the microcontroller. In this case you should be looking for document RM0016 on ST’s web site.

Using the Standard Peripheral Library

Using the STD Peripheral Library makes this a relatively simple task. We only need to call four methods:

CLK_DeInit();
CLK_SYSCLKConfig(CLK_PRESCALAR_CPUDIV1);    // CPU Prescalar = 1.
CLK_SYSCLKConfig(CLK_PRESCALAR_HSIDIV1);    // Prescalar = 1, 16 MHz.
CLK_ClockSwitchConfig(CLK_SWITCHMODE_AUTO,  // Automatically switch
                      CLK_SOURCE_HSI,       // Switch to internal timer.
                      DISABLE,              // Disable the clock switch interrupt.
                      CLK_CURRENTCLOCKSTATE_DISABLE);   // Disable the previous clock.

The headers for the above methods can be found in the file stm8s_clk.h and the source code for the methods can be found in the file stm8s_clk.c. These source files can be found in the source code folder of the STD Peripheral Library.

Direct Register Access – Method 1

This first method of accessing the registers continues to use the STD Peripheral Library files but we do not make any calls into the library. Instead we use the definitions for the registers to access the chip directly.

So let’s start by breaking the above four methods down into their direct register equivalents.

CLK_DeInit

The first thing we need to do is to reset all of the registers to the default values:

CLK->ICKR = CLK_ICKR_RESET_VALUE;
CLK->ECKR = CLK_ECKR_RESET_VALUE;
CLK->SWR  = CLK_SWR_RESET_VALUE;
CLK->SWCR = CLK_SWCR_RESET_VALUE;
CLK->CKDIVR = CLK_CKDIVR_RESET_VALUE;
CLK->PCKENR1 = CLK_PCKENR1_RESET_VALUE;
CLK->PCKENR2 = CLK_PCKENR2_RESET_VALUE;
CLK->CSSR = CLK_CSSR_RESET_VALUE;
//
//  The following set has to be performed twice.
//
CLK->CCOR = CLK_CCOR_RESET_VALUE;
while ((CLK->CCOR & CLK_CCOR_CCOEN) != 0);
CLK->CCOR = CLK_CCOR_RESET_VALUE;
CLK->HSITRIMR = CLK_HSITRIMR_RESET_VALUE;
CLK->SWIMCCR = CLK_SWIMCCR_RESET_VALUE;

As you can see, the library certainly hides a large amount of work from you. Most of the above code is simply a case of setting up the registers to default values.

To find out how this all works you need to start looking in the stm8s.h file. A quick search for CLK soon leads you to the following type declaration:

typedef struct CLK_struct
{
  __IO uint8_t ICKR;     /*!> Internal Clocks Control Register */
  __IO uint8_t ECKR;     /*!> External Clocks Control Register */
  uint8_t RESERVED;      /*!> Reserved byte */
  __IO uint8_t CMSR;     /*!> Clock Master Status Register */
  __IO uint8_t SWR;      /*!> Clock Master Switch Register */
  __IO uint8_t SWCR;     /*!> Switch Control Register */
  __IO uint8_t CKDIVR;   /*!> Clock Divider Register */
  __IO uint8_t PCKENR1;  /*!> Peripheral Clock Gating Register 1 */
  __IO uint8_t CSSR;     /*!> Clock Security System Register */
  __IO uint8_t CCOR;     /*!> Configurable Clock Output Register */
  __IO uint8_t PCKENR2;  /*!> Peripheral Clock Gating Register 2 */
  uint8_t RESERVED1;     /*!> Reserved byte */
  __IO uint8_t HSITRIMR; /*!> HSI Calibration Trimmer Register */
  __IO uint8_t SWIMCCR;  /*!> SWIM clock control register */
}
CLK_TypeDef;

If you have a look at the technical reference sheet for the chip, you will find section which shows the memory layout for the registers. These are at fixed locations in memory and should map to the above layout. The __IO is defined as volatile and will prevent the compiler from optimising out any references to the variables.

The next thing to note is that we still do not have a definition for CLK. A little more searching in the same file will lead you to the following statement:

#define CLK ((CLK_TypeDef *) CLK_BaseAddress)

So this defines CLK for us as a pointer to a location in memory. Some more searching (again in the same file) leads us to the following line of code:

#define CLK_BaseAddress         0x50C0

So the code CLK->ICKR = 0; will set the register at location 0x50C0 to zero.

One point to note is the statement while ((CLK->CCOR & CLK_CCOR_CCOEN) != 0);. This illustrates the use of another type of declaration you will find in stm8s.h, namely, CLK_CCOR_CCOEN. This declaration allows you to mask off certain bits within a register and use this to set or check values in a register. The name is made up of three parts:

CLK_SYSCLKConfig

The next task is to set the prescalar for the system clock. This is being set to 1 to ensure the system runs at 16 MHz.

CLK->CKDIVR  &= (uint8_t) (~CLK_CKDIVR_HSIDIV);
CLK->CKDIVR |= (uint8_t) ((uint8_t) CLK_PRESCALAR_HSIDIV1 & (uint8_t) CLK_CKDIVR_HSIDIV);

The first line resets the prescalar to a known value whilst the second selects the divider which will be used.

CLK_ClockSwitchConfig

The final operation is to switch the system clock to the HSI and this is achieved with the following code:

CLK->SWCR |= CLK_SWCR_SWEN;
CLK->SWCR &= (uint8_t) (~CLK_SWCR_SWIEN);
CLK->SWR = (uint8_t) CLK_SOURCE_HSI;
uint16_t downCounter = CLK_TIMEOUT;
while ((((CLK->SWCR & CLK_SWCR_SWBSY) != 0 ) && (downCounter != 0)))
{
    downCounter--;
}

Direct Register Access – Method 2

This method uses the register declarations found in the header files provided by IAR. So for the STM8S103F3 we will be looking in the file .

CLK_DeInit

As before, the first thing we will do is to reset the registers to a known set of values:

CLK_ICKR = 0;
CLK_ECKR = 0;
CLK_SWR = 0xe1;
CLK_SWCR = 0;
CLK_CKDIVR = 0x10;
CLK_PCKENR1 = CLK_PCKENR1_SPI | CLK_PCKENR1_TIM2;   //  Enable the peripheral clocks we need.
CLK_PCKENR2 = 0;
CLK_CSSR = 0;
CLK_CCOR = 0;
while (CLK_CCOR_CCOEN != 0);
CLK_CCOR = 0;
CLK_HSITRIMR = 0;
CLK_SWIMCCR = 0;

The first thing you will notice is that by using this method we are not using the pointer dereferencing operator. Instead the application is accessing the registers directly. So let’s have a look at the header file and dissect the reset of the ICKR register. Searching for CLK_ICKR leads us to the following code:

typedef struct
{
  unsigned char HSIEN       : 1;
  unsigned char HSIRDY      : 1;
  unsigned char FHW         : 1;
  unsigned char LSIEN       : 1;
  unsigned char LSIRDY      : 1;
  unsigned char REGAH       : 1;
} __BITS_CLK_ICKR;
__IO_REG8_BIT(CLK_ICKR,    0x50C0, __READ_WRITE, __BITS_CLK_ICKR);

The first things we see is the definition of the structure which maps on to the format of the ICKR register. Each bit field is broken out and maps on to the sections of the register as defined in the data sheet.

The final line of code in the above snippet uses the __IO_REG8_BIT macro to map the data structure onto the address 0x50C0 and create a new name with bit level access.

The next thing to note is the while loop which checks the CCOOEN bit in the CCOR register – while (CLK_CCOR_CCOEN != 0);. As above, this uses a three part notation to form a reference, this time it is to a particular bit in a register. This is not a mask as in the previous example. This is broken down as follows:

Some more digging in the file leads to the following definition:

#define CLK_CCOR_CCOEN           CLK_CCOR_bit.CCOEN

The CLK_CCOR_bit declaration was created by the __IO_REG8_BIT macro. This is the name which has been given to the location in memory of the ICKR register.

CLK_SYSCLKConfig

The next task is to set the prescalar for the system clock. This is being set to 1 to ensure the system runs at 16 MHz. Note that a prescalar of 1 maps to the prescalar bits in the register being set to zero.

CLK_CKDIVR = 0;

CLK_ClockSwitchConfig

The final operation is to switch the system clock to the HSI and this is achieved with the following code:

CLK_SWCR_SWEN = 1;
CLK_SWR = 0xe1;                     //  Use HSI as the clock source.
while (CLK_SWCR_SWBSY != 0);        //  Pause while the clock switch is busy.

Conclusion

So there you have it, three different ways of performing the same task. The method used will be down to individual preference. Happy experimenting.

If you have been reading my recent posts you will have noticed that I have had some problems setting up the UART on the STM8S. IN fact I spent several days getting this working. This post gives a couple of code snippets which I have used to get around this problem. You may need to refer to the data sheet on the processor for more detail on how this code works.

Setting Up the UART

This code makes a fundamental, namely that you have configured the chip and your circuit to run at 16 MHz. I did this by setting the chip to use the internal oscillator as it’s clock source and using a prescalar of 1. If you have this set up then the following method will allow you access to the UART running at 115200 baud, 8 data bits, 1 stop bit and no parity.

Firstly, we need to make sure we are using the right include files:

#include "stm8s.h"
#include "stm8s_uart1.h"

Next we need to set up the port:

void SetupSerialPort()
{
    //
    //  Clear the Idle Line Detected bit in the status register by a read
    //  to the UART1_SR register followed by a Read to the UART1_DR register.
    //
    (void) UART1->SR;
    (void) UART1->DR;

    UART1->CR2 = UART1_CR2_RESET_VALUE;
    UART1->CR4 = UART1_CR4_RESET_VALUE;
    UART1->CR5 = UART1_CR5_RESET_VALUE;
    UART1->GTR = UART1_GTR_RESET_VALUE;
    UART1->PSCR = UART1_PSCR_RESET_VALUE;
    //
    //  Setup the port.
    //
    UART1->CR1 = (u8) UART1_WORDLENGTH_8D | (u8) UART1_PARITY_NO;   // Word length = 8, no parity.
    UART1->CR3 = (u8) UART1_STOPBITS_1;                             // 1 stop bit.

    UART1->BRR1 &= (uint8_t) (~UART1_BRR1_DIVM);
    UART1->BRR2 &= (uint8_t) (~UART1_BRR2_DIVM);
    UART1->BRR2 &= (uint8_t) (~UART1_BRR2_DIVF);
    //
    //  Set the clock prescaler for 11520 baud.  This assumes a 16 MHz clock speed.
    //
    UART1->BRR2 = 0x0b;
    UART1->BRR1 = 0x08;
    //
    //  Disable the Transmitter and Receiver before seting the LBCL, CPOL and CPHA bits
    //
    UART1->CR2 &= (u8) ~(UART1_CR2_TEN | UART1_CR2_REN);
    //
    //  Clear the Clock Polarity, lock Phase, Last Bit Clock pulse
    //
    UART1->CR3 &= (u8)~(UART1_CR3_CPOL | UART1_CR3_CPHA | UART1_CR3_LBCL);
    //
    //  Set the Clock Polarity, lock Phase, Last Bit Clock pulse
    //
    UART1->CR3 |= (u8)((u8) UART1_SYNCMODE_CLOCK_ENABLE &
                       (u8) (UART1_CR3_CPOL | UART1_CR3_CPHA | UART1_CR3_LBCL));
    //
    //  Set the Tx and Rx state
    //
    UART1->CR2 |= (u8) ((u8) UART1_CR2_TEN | (u8) UART1_CR2_REN);
    UART1->CR3 &= (u8) (~UART1_CR3_CKEN);
}

Having set up the UART we need a method to write data to the port. The following method will send simple text strings to the UART:

//
//  Send a message to the debug port (UART1).
//
void Printf(char *message)
{
    char *ch = message;
    while (*ch)
    {
        UART1->DR = (u8) (*ch);
        while ((UART1->SR & (u8) UART1_FLAG_TXE) == RESET);
        ch++;
    }
}

Using the UART

The use of this code can be illustrated by the following program:

void main()
{
    SetupSerialPort()
    Printf("Hello from my program");
}

A couple of days ago I was working on a project and was having some problems debugging. It was nothing to do with the debugging environment more to do with the high speed interaction between two devices. So I needed some debug output – enter UART1. I decided that I would send some debug information to UART1 and capture this in PuTTY. At least I would be able to see the sequence of events and hopefully work out where my code is going wrong.

This is where my problems started…

UART Examples

So I started looking through the Standard Peripheral Library example programs. Sure enough there is a program there in the directory UART1_Printf Sounds ideal, I should be able to throw some text data at this and have it appear in PuTTY. Great.

Quick check of the readme.txt in this directory and it looks like this program is compatible:

Hardware and Software environment

This example runs on STM8S High density and Low density devices and on STM8A High density devices.

So I am running on the STM8S103F3 which is one of the low density STM8 devices so I should be fine with this code. A little further down the file it gives the specification for the serial comms as:

Hyperterminal configuration:

• Word Length = 8 Bits
• One Stop Bit
• No parity
• BaudRate = 115200 baud
• flow control: None

Great – we are in business!

So what happened?

Creating a project was simple and I had the code compiling within a few minutes. Next job is to hook up PuTTY and send some data through down the serial connection and see it appear on the PC. This is where life got a little messy. Nothing appeared. Double checked all of the settings in both the program and Putty – they both match. Ahhh – should Tx and Rx be crossed? No that did not work.

Hmmmmmmmm….. This could be a long day.

Next step, break out the trusty logic analyser from Saleae. Connected this up and managed to show that data was in fact being output from the application but at 53 baud. This is a long way from the 115200 baud promised.

Hmmmm, so lets change the baud rate requested – 9600 baud is normally safe. No change, still 53 baud.

Starts clutching at straw, lets just make sure the system clock is running using the HSI 16 MHz signal. Still no joy, 53 baud is all I seem to be getting.

The Problem

I finally broke out the data sheet for the chip and started single stepping through the STD Peripheral library code. The example program contains the following line of code:

UART1_Init((uint32_t)115200, UART1_WORDLENGTH_8D, UART1_STOPBITS_1, UART1_PARITY_NO,
UART1_SYNCMODE_CLOCK_DISABLE, UART1_MODE_TXRX_ENABLE);

This should set up the port to operate at the rates as per the read me file. If you have a look at this method, about half way down you will see the following code:

/* Set the UART1 BaudRates in BRR1 and BRR2 registers according to UART1_BaudRate value */
BaudRate_Mantissa    = ((uint32_t) CLK_GetClockFreq() / (BaudRate << 4));
BaudRate_Mantissa100 = (((uint32_t) CLK_GetClockFreq() * 100) / (BaudRate << 4));
/* Set the fraction of UART1DIV  */
UART1->BRR2 |= (uint8_t)((uint8_t)(((BaudRate_Mantissa100 - (BaudRate_Mantissa * 100)) << 4) / 100) & (uint8_t)0x0F);
/* Set the MSB mantissa of UART1DIV  */
UART1->BRR2 |= (uint8_t)((BaudRate_Mantissa >> 4) & (uint8_t) 0xF0);
/* Set the LSB mantissa of UART1DIV  */
UART1->BRR1 |= (uint8_t) BaudRate_Mantissa;

So I am single stepping through the code and notice something odd, the Baud Rate Registers (BRR1 and BRR2) are not being set correctly. By my calculations these should be 0x68 and 0x03 respectively (for 9600 baud).

A quick modification to the source file to put these explicit changes in and….

It Works !

I now have PuTTY talking to my STM8 at 9600 baud and so my debugging life should be a little easier. On the downside, I have lost a little faith in the Standard Peripheral Library. I think that in future I will be using this more for guidance rather than as a method of programming this family of processors. Coupled with the data sheet for the chip is is going to be an invaluable tool. Going forward I will be building up my own set of methods working with the registers on the processor. This may be slower but it will bring me closer to the chip and I may occasionally come across the odd gem in the data sheet which I may not otherwise have become aware of.

Now, back to the original problem.

So having received a Raspberry Pi I suppose I should start to configure it to make it a little more to my taste but also to fit in with my local systems at home. To do this I am going to perform the following tasks:

1. Install the operating system of choice (Debian 6)
2. Resize the partitions
3. Backing up your system
4. Enable SSH
5. Changing the message of the day
6. Changing the hostname
7. Changing prompt colours
8. Using History

I have a habit of configuring the device using the root user. There are many debates on whether this is advisable or not. I do not normally work with the root account but for configuring machines I find it more convenient to login as root. I do this because you are often executing a large number of commands which require super user privileges and it is quicker to do this if you are not prefixing every command with the sudo command.

So in this tutorial I will be assuming that you have changed to the root account before running any of these commands. You can do this by executing the command:

sudo su


Remember, if you do this you can find yourself putting the system stability at risk if you mistype a command or execute a command accidentally. The most infamous example is deleting the entire file system by a mistyped command – you can do serious damage as superuser.

Throughout this article we will be editing text files. The editor being used is the nano text editor. It’s small, simple and already installed on the system.

These steps will also require a number of reboots. I have found that entering the following command:

ping -t 192.168.10.5

into a command prompt on my Windows machine will give me a good indication when my Raspberry Pi is ready for me to SSH back onto the device.

Installing Debian

First job is to download the operating system image from the Raspberry Pi web site. Extracting this to your hard drive leaves you with a 1.8 GByte disc image file.

This file can be written to a SD card using the Win32DiskImager (the link can also be found in the downloads section of the Raspberry Pi web site). Start this application, select the image file you have just extracted, select the location of the SD card and hit the Write button.

Once completed the SD card should be ready to use. Insert the SD card into the Raspberry Pi, connect a keyboard, mouse (optional) and monitor and you are ready to go.

Resizing the partitions

The default installation allows for the use of SD cards starting at 2GB and going upwards. This means that by default you have a single partition available to you and possibly a large amount of unused space. To use this space you really have two options, create alternative partitions and mount these or resize the existing partition to the full extent of the SD card. I personally went for the later option as I have always preferred not to have to worry about partition sizes and where I can put files.

Rather than describe this fully here I will simply point you to a rather good video on the subject on the unofficial RaspberryPiBeginners channel on YouTube.

Backing Up Your System

An important part of maintaining these systems is backing up on a regular basis. I have an 8 GByte SD card for the Raspberry Pi and a laptop with over 200 GBytes of free space. I have decided that for me I can afford a few GBytes on the laptop for an occasional full system image with some incremental backups of user files.

For a full backup of the SD card we need to use the same tool as we used to write the original image file. This time instead of writing the image we read the SD card and generate an image file. So insert the SD card into the computer’s SD card reader and select the drive containing the SD card. Enter a name for the image file and hit the Read button. At the end of the process you will have a file which can be written back to the SD card should any changes make the system un usable.

Enable SSH

I find it a little intrusive having to connect the Raspberry Pi to my monitor and USB hub as it disrupts the PC setup. To help avoid this I have setup the system to use SSH, I can then connect to the Raspberry Pi using PuTTY from the PC.

Enabling this is as simple as renaming a file and rebooting the device.

So logon to the Raspberry Pi and execute the following commands:


cd /boot
mv boot_enable_ssh.rc boot.rc

Next we need to make a note of the IP address of the device using the following command:

ifconfig

Make a note of the IP address of the device (look for the IP address in the eth0 section of the output from this command).

Now reboot the device with the following command:

reboot


The final thing to do is check that everything works. So start PuTTY (or your favourite SSH client) and connect to the machine using this client. You should see the login prompt in the client.

Changing the Message of the Day

This is a trivial task and I do this on each of my Linux machines so that I can see which machine I am connected to as soon as I login. The default message for the Debian distribution I have is as follows:

Linux raspberrypi 3.1.9+ #90 Wed Apr 18 18:23:05 BST 2012 armv6l

Debian GNU/Linux comes with ABSOLUTELY NO WARRANTY, to the extent
permitted by applicable law.

This file is created on system boot by a shell script. You can change the end of the file by editing the text in the file /etc/motd.tail. So simply log on to the machine and edit this file adding the text you wish to see when you login. Remember, you will need to be superuser to do this.

sudo su
cd /etc
nano motd.tail

Now restart the system and logon once more. This time you will see your modified message appearing instead of the default message.

Changing the Host Name

Changing the host name is simple and requires that two files are edited.

• /etc/hostname
• /etc/hosts

The hostname file should be edited to contain the host name and nothing else. In my case I am going to call this device RPiDev. So edit the hostname file using an editor such as nano, remove the existing entry and add your selected name and save the file.

The next file is a little more complex as this contains routing information for the machine. By default, this file looks like this:

::1 raspberrypi localhost6.localdomain6 localhost6
127.0.1.1 raspberrypi

127.0.0.1 localhost
::1 localhost ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters

We need to edit this file and replace all instances of the text raspberrypi with your chosen name. My hosts file now looks like this:

127.0.0.1 localhost
::1 localhost ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters


Changing the Prompt Colour

Working with the system in console mode can be a little dull as you only get to see text in black and white. You can add a little colour to the system by changing the prompt. I did this by editing the /etc/bash.bashrc file. You will need to be the root user to edit this file. So edit this file and about half way down you will find the following text:

PS1='${debian_chroot:+($debian_chroot)}\u@\h:\w\$ '

Try changing this to the following:

PS1='${debian_chroot:+($debian_chroot)}\! \[\033[01;32m\][\u@\h] \[\033[01;34m\]\w \$ \[\033[00m\]'

Now logout and then back in to the device. You should find your prompt has changed with the user and system name appearing in light green and the current directory appearing in a light blue.

The colour coding is performed by the text in the square brackets, particularly the [01;34m\] commands. The following colour codes can be used:

Using History

In the section on changing the prompt you will see a little piece of magic when setting the PS1 variable. About half way along you will see the text:

\!

This piece of text is a special flag which indicates that the shell should insert the current history number for the command into the prompt. So on my system I use the above value for PS1 and I see the following prompt:

146 [root@RPiDev] /etc #

The number at the start of the line can be used to re-execute a particular command again. This is done by prefixing the command number with the ! character. So let’s say that command 130 was cd /etc/networks and I want to get back to this directory. I can either type the command in again or as I know this was command number 130 I can type in !130.

You can display the full history by executing the command history from the command prompt.

I have been waiting for this event for more than a year now. The day when I would finally get my hands on a Raspberry Pi. Today it finally happened. It has been a roller-coaster ride with the initial release expected in December 2011 and launch delayed, finally announced and then over subscribed, it has been an eventful 12 months.

Today all of the waiting came to an end as the board finally arrived.

After waiting for 12 months, waiting for the end of a working day to play with the device was probably the most difficult. But at the end of the day it was time to hook up the device and power it up.

And connected to a monitor:

So now I know the device is working it is time to find time to play but that will have to wait as other projects are demanding my attention. In some way this is a shame as I really believe in what this project stands for but the delays and hype have dulled the edge.

Back in the bag for another day.

The 8 x 8 x 8 LED cube has been on it’s travels again. This time it has made an appearance at the Bay Area Maker Faire:

Thanks to fellow Netduino community member Chris Hammond for permission to use this photo.

In early May I started working on a module to display a 4 digit number on a 7 segment LED display. I developed this as far as a working proof of concept developing the hardware and software necessary to prove that the module was feasible. At this point I discovered that development was ongoing by two other teams and decided to stop development at this point taking what I had learnt onboard and move on to other projects.

This article presents the work so far as this is a working project and demonstrates some of the principles used in the STM8S articles written so far this month. Namely:

• GPIO output
• Timers

As already stated, this project is working but incomplete and there is no intention to take the hardware further than the proof of concept as presented here.

Hardware Design

The system will use four 7-segment LED modules in place of a single 4 digit, 7 segment module. The hardware design requires the following circuit components:

• Microprocessor controller
• Power supply to the 7-segment display modules
• Decoding for the 7-segments

The software will use multiplexing to allow one chip to be used to control the four 7-segment modules. The basic process is as follows:

1. Turn power to all segments off
2. Send the data for a digit to the 74LS47 chip
3. Turn on the power for the digit
4. Wait a short time
5. Move on to the next digit and go back to start

This will give the illusion that all of the LEDs are powered all of the time.

Microcontroller

The microcontroller chosen was the STM8S003 as this is supported by the Netduino team for designers who wish to create their own modules. This chip has several free development environments available and the IAR environment has been used in this blog to document several small projects using this family of chips.

The project as it stands requires the use of 9 pins from the STM8S. Four pins will be used for power control, one line per digit. A further four pins will be used to tell the 74LS47 which of the segments are to be lit. The final pin will be used to indicate if the digit zero is to be shown for a zero value. More on this in the decoding section below. The use of this number of pins makes this unsuitable as it stands for use as a Netduino GO! module. Further refinement is required to take this forward as a module.

7-segment Power Control

The LED power is controlled by using a PNP transistor (the 7-segments share a common anode). In this circuit the power is run through a single current limiting resistor. This will mean that numbers with few digits (for example, 1) will be displayed more brightly then those with a large number of segments (say 8). The final circuit would need to refined to have a current limiting resistor per segment or using a constant current driver. This component of the circuit looks like this:

The resistor R5 is connected to the output of the microcontroller and the output of R9 going through to the power pin on the 7-segment display.

Decoding the 7-segments

The decoding of the output from the microcontroller is performed by the 74LS47 chip. This is a dedicated decoder circuit for common anode 7-segment LED modules. The system takes a binary coded decimal number and then sinks the current from the appropriate segments of the display. The result is that the module will display the appropriate digit.

We will also take advantage of one final line, the RBI line. By turning this line on and off we can determine if the digit will display the digit 0 when the input to the decoder is zero. So why would we change this line? We we could fix the line so that a digit will always be displayed. So when the microcontroller wants to show 0 it will actually show either a blank display or 0000. The blank display is obviously not desirable as the user will not be able to determine if the display is showing 0 or if it is turned off. Showing four zeroes will give the user the correct information. A nicer solution is for the display to show a single zero when the microcontroller outputs 0.

The resulting decoder wiring looks something like the following:

Where pins A, B, C and D will output the binary version of the digit to be displayed.

Full Circuit

This PDF File contains the full circuit diagram for the proof of concept module. The fully assembled circuit looks like this:

Software

In order to test the circuit we will write a small program which will start at 0 and count to 9999 before returning to 0 and starting counting again.

The first thing we need to do is to initialise the system. The following code resets the system clock source, configures the GPIO lines and then starts the timer which contain the code which will perform the multiplexing.

//
//  Initialise the clock
//
CLK_DeInit();
CLK_SYSCLKConfig(CLK_PRESCALER_CPUDIV1);                // CPU Prescaler = 1.
CLK_SYSCLKConfig(CLK_PRESCALER_HSIDIV1);                // Prescaler = 1, 16 MHz.
CLK_ClockSwitchConfig(CLK_SWITCHMODE_AUTO,              // Automatically switch
					  CLK_SOURCE_HSE,                   // Switch to internal timer.
					  DISABLE,                          // Disable the clock switch interrupt.
					  CLK_CURRENTCLOCKSTATE_DISABLE);   // Disable the previous clock.
//
//  Initialise GPIOs
//
GPIO_DeInit(GPIOC);
GPIO_Init(DIGIT0_BI_PORT, DIGIT0_BI_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT1_BI_PORT, DIGIT1_BI_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT2_BI_PORT, DIGIT2_BI_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT3_BI_PORT, DIGIT3_BI_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT_A0_PORT, DIGIT_A0_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT_A1_PORT, DIGIT_A1_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT_A2_PORT, DIGIT_A2_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(DIGIT_A3_PORT, DIGIT_A3_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
GPIO_Init(RBI_PORT, RBI_PIN, GPIO_MODE_OUT_PP_LOW_FAST);
//
//  Setup timer 2 to interrupt every 2048 clock pulses.
//
TIM2_DeInit();
TIM2_TimeBaseInit(TIM2_PRESCALER_2048,      // Interrupt every 2048 clock pulses.
				  1);                       // Period is one.
TIM2_ITConfig(TIM2_IT_UPDATE, ENABLE);      // Enable the overflow interrupt.
TIM2_Cmd(ENABLE);

The next thing we need to do is to look at displaying a digit. To do this we will turn off all of the digits and then convert the digit into binary outputting the value on the four pins connected to the 74LS47 decoder.

GPIO_WriteHigh(DIGIT0_BI_PORT, DIGIT0_BI_PIN);
GPIO_WriteHigh(DIGIT1_BI_PORT, DIGIT1_BI_PIN);
GPIO_WriteHigh(DIGIT2_BI_PORT, DIGIT2_BI_PIN);
GPIO_WriteHigh(DIGIT3_BI_PORT, DIGIT3_BI_PIN);
if (digit & 0x01)
{
	GPIO_WriteHigh(DIGIT_A0_PORT, DIGIT_A0_PIN);
}
else
{
	GPIO_WriteLow(DIGIT_A0_PORT, DIGIT_A0_PIN);
}
if (digit & 0x02)
{
	GPIO_WriteHigh(DIGIT_A1_PORT, DIGIT_A1_PIN);
}
else
{
	GPIO_WriteLow(DIGIT_A1_PORT, DIGIT_A1_PIN);
}
if (digit & 0x04)
{
	GPIO_WriteHigh(DIGIT_A2_PORT, DIGIT_A2_PIN);
}
else
{
	GPIO_WriteLow(DIGIT_A2_PORT, DIGIT_A2_PIN);
}
if (digit & 0x08)
{
	GPIO_WriteHigh(DIGIT_A3_PORT, DIGIT_A3_PIN);
}
else
{
	GPIO_WriteLow(DIGIT_A3_PORT, DIGIT_A3_PIN);
}
GPIO_WriteLow(port, pin);

The final statement will turn on the PNP transistor connected to the selected digit by connecting the base of the transistor to ground.

The final piece of code we need (before the main program loop) is the interrupt handler for the timer. This will turn off all of the power to the LED modules and then work out which digit to show and write the data to the 74LS47 decoder. This interrupt service routine can be found in the file stm8s_it.c. See previous STM8S post on interrupts for further information. Look for the following code in this file:

INTERRUPT_HANDLER(TIM2_UPD_OVF_BRK_IRQHandler, 13)
{
}

And replace with the following code:

INTERRUPT_HANDLER(TIM2_UPD_OVF_BRK_IRQHandler, 13)
{
    if (_currentDigit == 0)
    {
        DisplayDigit(_displayValue[0], DIGIT0_BI_PORT, DIGIT0_BI_PIN);
        GPIO_WriteHigh(RBI_PORT, RBI_PIN);
        _currentDigit++;
    }
    else
    {
        if (_currentDigit == 1)
        {
            DisplayDigit(_displayValue[1], DIGIT1_BI_PORT, DIGIT1_BI_PIN);
            if ((_displayValue[2] != 0) || (_displayValue[3] != 0))
            {
                GPIO_WriteHigh(RBI_PORT, RBI_PIN);
            }
            else
            {
                GPIO_WriteLow(RBI_PORT, RBI_PIN);
            }
            _currentDigit++;
        }
        else
        {
            if (_currentDigit == 2)
            {
                DisplayDigit(_displayValue[2], DIGIT2_BI_PORT, DIGIT2_BI_PIN);
            if (_displayValue[3] != 0)
            {
                GPIO_WriteHigh(RBI_PORT, RBI_PIN);
            }
            else
            {
                GPIO_WriteLow(RBI_PORT, RBI_PIN);
            }
                _currentDigit++;
            }
            else
            {
                DisplayDigit(_displayValue[3], DIGIT3_BI_PORT, DIGIT3_BI_PIN);
                GPIO_WriteLow(RBI_PORT, RBI_PIN);
                _currentDigit = 0;
            }
        }
    }
    TIM2->SR1 = (uint8_t) (~(uint8_t) TIM2_IT_UPDATE);  // Clear the interrupt.
}

One interesting thing to look at is the setting of the RBI_PORT. Note that this is set to high for the first digit (i.e. rightmost digit) and low for all of the other digits. This is the piece of magic which ensure that we always display a digit for zero in the rightmost digit but only display a value for no zero digits on the remaining three digits.

If this were to be taken forward then the code above would need to be optimised to ensure that the ISR completed in the shortest time possible.

Now we have all of the elements in place we can start to write our main program loop. The code will loop from 0 to 9999 displaying each digit and the restarting the count. The individual digits will be put into an array of four integers, one for each LED in the display. This gives us the following code:

enableInterrupts();                         // Make sure interrupts are enabled.
int value = 0;
while (1)
{
	_displayValue[0] = value % 10;
	_displayValue[1] = (value / 10) % 10;
	_displayValue[2] = (value / 100) % 10;
	_displayValue[3] = (value / 1000) % 10;
	if (value == 9999)
	{
		value = 0;
	}
	else
	{
		value++;
	}
	//
	//  Now delay otherwise we'll be counting too fast for anything
	//  to appear on the display.
	//
	for (int outerDelay = 0; outerDelay < 5; outerDelay++)
	{
		for (int delay = 0; delay < 0xffff; delay++)
		{
			nop();
		}
	}
}

The full source code for this project can be downloaded here.

Conclusion

This was an interesting project to put together and took about a day to work out and get everything working. To take this further would take a lot more work, many times the amount already put into the project. As two other groups are working on similar ideas I have decided to archive this for now and move on to other projects.