LeafLabs Garden » Topic: Shiftbrite SPI

LeafLabs Garden » Topic: Shiftbrite SPI http://forums.leaflabs.com/topic.php?id=1004 A place to share, learn, and grow... en-US Fri, 22 Jan 2016 00:13:15 +0000 http://bbpress.org/?v=1.0.2 <![CDATA[Search]]> q http://forums.leaflabs.com/search.php gbulmer on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6138 Tue, 30 Aug 2011 17:58:19 +0000 gbulmer 6138@http://forums.leaflabs.com/ asterix - <code>(command)&0b11</code> uses the binary, or bit-wise, AND operation '&', and 0b11 is a way of writing a binary literal value, in this case decimal 3. 0b is the binary cognate of 0x for writing hexadecimal literal numbers. The result is an unsigned value. So it all means binary AND the value of command with binary 011 i.e. do a bit-wise AND operation between the contents of the variable labelled <code>command</code> (an int) and the 32-bit unsigned integer 0000 0000 0000 0000 0000 0000 0000 0011 HTH asterix on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6131 Tue, 30 Aug 2011 07:43:52 +0000 asterix 6131@http://forums.leaflabs.com/ hmmm it is spitting an error when I compile using the example sketch posted on the shiftbrite information page I posted earlier: libraries\SPI\SPI.cpp:52: error: 'class SPIClass' has no member named 'nssPin' asterix on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6130 Tue, 30 Aug 2011 07:31:47 +0000 asterix 6130@http://forums.leaflabs.com/ Thanks mbolivar. I will try out the port and report back. I personally found maples implementation much simpler and easier to implement. Is there any advantage to arduino's SPI lib? Because really, I find it abstracted and annoying. mbolivar on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6124 Mon, 29 Aug 2011 18:46:29 +0000 mbolivar 6124@http://forums.leaflabs.com/ gbulmer, <blockquote> Very good question! I went digging through the Maple IDE source files, and found this: [...] but it would be nice to have this in the documentation. </blockquote> Good point. I've pushed this to the docs sources; it'll go out in the next release: <a href="https://github.com/leaflabs/leaflabs-docs/commit/202d0ae14f235c385c4f06b60a459164f09a308b" rel="nofollow">https://github.com/leaflabs/leaflabs-docs/commit/202d0ae14f235c385c4f06b60a459164f09a308b</a> mbolivar on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6123 Mon, 29 Aug 2011 18:38:40 +0000 mbolivar 6123@http://forums.leaflabs.com/ Hi asterix, I recently made a port for the Arduino SPI lib. If you've got time, I'd love it if you gave it a try and let me know if it works with your example: <a href="http://static.leaflabs.com/mbolivar/SPI.zip" rel="nofollow">http://static.leaflabs.com/mbolivar/SPI.zip</a> To try it out, drop it in the "libraries" folder in your sketchbook (create this folder if it doesn't already exist). Then restart the IDE; you should be able to use it from the Sketch > Import Library menu. asterix on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6115 Mon, 29 Aug 2011 18:11:28 +0000 asterix 6115@http://forums.leaflabs.com/ Gbulmer thank you for such a thorough and informative reply post :) Sending the resulting byte via SerialUSB is fixed (just a bug in my code). I managed to isolate a big part of the problem which was me connecting the shiftbrite V+ and ground rail directly to a 9v source. I patch it into a neighbouring arduinos VCC / Ground and Im getting somewhere (slap). For learning sake, can you explain briefly what happens in your example code when you prefix &0b11 to command :(command)&0b11 ?? I'll post up my code shortly once I've had a play. I'll actually be sending video wirelessly using Xbee's to maple > shiftBrites. The arduino clock cycle of 16mhz is too slow for high data transfers which is why I'm playing with the maple. gbulmer on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6105 Sun, 28 Aug 2011 16:55:02 +0000 gbulmer 6105@http://forums.leaflabs.com/ I apologise for the weird backquote and html markup nonsense, instead of code outline, but this dreadful forum software isn't capable of converting a post with code tags back and forth. Frankly I got sick of trying to fight it. I've mentioned some of the many failings of the forum software before, but I would much prefer LeafLabs to get Oak shipped rather than fix it. Also, please note, I haven't compiled and tested any of my posted code. gbulmer on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6104 Sun, 28 Aug 2011 16:48:54 +0000 gbulmer 6104@http://forums.leaflabs.com/ <blockquote>`// reworked to 7 as the first bite is insignificant, and the second should be 0 so that command mode is 0; spi.write((cmode << 7 | b >>4)); spi.write((b << 4 | r >>6)); spi.write((r << 2 | g >> 8)); spi.write(g); ` </blockquote> You probably don't want to do this. As I read the example at <a href="http://docs.macetech.com/doku.php/shiftbrite" rel="nofollow">http://docs.macetech.com/doku.php/shiftbrite</a> the top two bits are command, so I would try to be consistent and allow it to be two bits. If cmode it is 0, then set it to 0. I wasted time trying to figure out whether you had made a mistake. Hence: <pre><code>int cmode = 0; spi.write((cmode << 6) | (b >>4)); spi.write((b << 4 | r >>6)); spi.write((r << 2 | g >> 8)); spi.write(g);</code></pre> Also, for clarity, because int's are 32 bits, and to be explicit, you might consider trimming the bits to be the ones required: <code>spi.write( (cmode << 6)&0b11000000 | (b >>4)&0b00111111);</code> I might write some macros or inline functions to do this stuff: <pre><code>#define MAKE_SHIFTBRIGHT_DATA(command, red, green, blue) (((command)&0b11)<<30 \ | ((blue)&0x3FF)<<20 | ((red)&0x3FF)<<10 | ((green)&0x3FF) #define SHIFTBRIGHT_BYTE0(val) ((unsigned char)((val)>>24)&0xFF) #define SHIFTBRIGHT_BYTE1(val) ((unsigned char)((val)>>16)&0xFF) #define SHIFTBRIGHT_BYTE2(val) ((unsigned char)((val)>>8)&0xFF) #define SHIFTBRIGHT_BYTE3(val) ((unsigned char)((val)>>0)&0xFF)</code></pre> and do: <pre><code>uint32 val = MAKE_SHIFTBRIGHT_DATA(cmode, r, g, b); spi.write(SHIFTBRIGHT_BYTE0(val)); spi.write(SHIFTBRIGHT_BYTE1(val)); spi.write(SHIFTBRIGHT_BYTE2(val)); spi.write(SHIFTBRIGHT_BYTE3(val));</code></pre> This has the small advantage that it is easy to print out the values over USB that are also being written via SPI. 1. when you writre "shiftleft or shiftright functions", so I assume you mean the left and right shift operators, '<<' and '>>'. The shift operators yield the same type as their left operand, so if the operand is a 16 bit signed int, the result will be a 16 bit signed int. An 'int' is 32 bits on an ARM with gcc, not 16 bits. The type the shift operators yield can not override the result type, which is truncated (high bits are thrown away) So: <code>1023 << 4;</code> is binary 000000000000000000011111111110000 (I hope I counted the 32 bits correctly there :-) when this is assigned to an 'int8' yields binary 11110000, or answer a) <blockquote>I've tried SerialUSB sending the result but I get a blank line. </blockquote> With these sort of things 'the devil is in the detail' so it would be helpful if you gave the code that is being used. I assume the code is trying to print the int8 variable <code>rshifted</code>, which has a decimal value of 240. There is no ASCII value, so the terminal program may have no visible glyph to print for that code, and uses space in lieu of one. instead, cast the value to be an <code>(unsigned int)</code> and see what it prints. 2. Please provide the code for the whole program. What happens in any particular case depends on how the peripheral is setup. Assuming everything is simple and vanilla, then SPI write will send 8 bits. It is irrelevant what is contained in the 8 bits, so when sending 1, it will send eight bits, seven 0's and one 1. <blockquote>3. In SPI's begin(frequency, bitOrder, mode); what exactly is the mode? </blockquote> Very good question! I went digging through the Maple IDE source files, and found this: <blockquote>`/** * @brief SPI mode configuration. * * Determines a combination of clock polarity (CPOL), which determines * idle state of the clock line, and clock phase (CPHA), which * determines which clock edge triggers data capture. */ typedef enum spi_mode { SPI_MODE_0, /**< Clock line idles low (0), data capture on first clock transition. */ SPI_MODE_1, /**< Clock line idles low (0), data capture on second clock transition */ SPI_MODE_2, /**< Clock line idles high (1), data capture on first clock transition. */ SPI_MODE_3 /**< Clock line idles high (1), data capture on second clock transition. */ } spi_mode;` </blockquote> but it would be nice to have this in the documentation. The linked page you gave doesn't give explicit information to be sure which mode is correct, but the datasheet for the Allegro A6281 will likely say what it expects. I tried to work it out by reading the ATmega to understand what mode the setup puts its SPI peripheral into. The linked code has: <code>SPCR = (1<<SPE)|(1<<MSTR)|(0<<SPR1)|(0<<SPR0);</code> SPE bit enables the SPI peripheral (not relevant) MSTR bit sets it in master mode (I assume your set up doesn't change this) SPR1 and SPR2 set the clock rate (clock / 4, or 4MHz). I assume your setup code does the same frequency So CPOL = 0 which means the clock signal is 'idle low' and CPCHA = 0, which means data is setup on the trailing edge of the clock, and sampled on the leading edge. This looks like SPI_MODE_0, but you could try 0 and 1 (or even all 4), and see if there are any differences in results. This is covered on page 679 of the RM0008 STM32F manual. (full disclosure: I am not a member of LeafLabs staff.) asterix on "Shiftbrite SPI" http://forums.leaflabs.com/topic.php?id=1004#post-6100 Sun, 28 Aug 2011 01:36:43 +0000 asterix 6100@http://forums.leaflabs.com/ I need help sending 3 10bit integers and one 2bit command mode using SPI. I'm trying to send 10bit color values (0-1024) to an array of shiftbrite led modules. There is an arduino sketch that uses the arduino SPI class to send data: <a href="http://docs.macetech.com/doku.php/shiftbrite" rel="nofollow">http://docs.macetech.com/doku.php/shiftbrite</a> I've unsuccessfully tried reworking a similar example of the sketch using maples hardwareSPI class, something like this: //////////////////////////////////////////////// int r = 900; int g = 100; int b = 100; int cmode = 1; for (int h = 0;h<11;h++) { // reworked to 7 as the first bite is insignificant, and the second should be 0 so that command mode is 0; spi.write((cmode << 7 | b >>4)); spi.write((b << 4 | r >>6)); spi.write((r << 2 | g >> 8)); spi.write(g); } delayMicroseconds(1); digitalWrite(latchpin,HIGH); // latch data into registers delayMicroseconds(1); digitalWrite(latchpin,LOW); delay(1000); /////////////////////////////////////////////////// I've also tried using a software SPI solution using shiftOut(): /////////////////////////////////////////////////// int r = 900; int g = 100; int b = 100; int cmode = 1 for (int h = 0;h<11;h++) { shiftOut(dataPin, clockPin, MSBFIRST, c << 6 | b>>4); shiftOut(dataPin, clockPin, MSBFIRST, b << 4 | r >> 6); shiftOut(dataPin, clockPin, MSBFIRST, r << 2 | g >> 8); shiftOut(dataPin, clockPin, MSBFIRST, g); } //////////////////////////////////////////////////// Data IS being sent out to the shiftbrights through the SPI MOSI as modding the values gives me varied results. But I'm not seeing what I should (colors are different on each module, some are blank, sometimes colors change). 1. I have a hunch that the shiftleft and shiftright functions aren't outputting an 8bit integer (uint8)that can be sent as a clean byte; as a standard integer is uint16 (2bytes). Example: int r = 1023 // binary: 00000011 11111111); int8 rshifted = (r << 4); binary 00111111 11110000, or uint16 16368); when I convert rshifted to a byte, will binary result be: a) 11110000 (the last 8 bits) b) 00111111 (the first 8 bits) ; c) some other value like 255 or null? I've tried SerialUSB sending the result but I get a blank line. 2. What happens when I send less than 8 bits through the SPI using MSB first? If I send spi.write(1) (binary 00000001); does it send the 0's or just the MSB (1); 3. In SPI's begin(frequency, bitOrder, mode); what exactly is the mode?