Hello,
I am trying to communicate with part WS2812B (spec attached) over a single wire. Per spec, it requires the signal to be held high for 400 nanoseconds +- 150 nanoseconds. Is this feasible using Terminal A9, NHET1[27] pin?
Currently, I am able to get 1900 nansecond resolution, but not much smaller.
Thanks.
Ubaid
Attaching HET1 Timing Configuration for reference. As well as some sample code on how we evaluate how much time has elapsed:
void BSP_OS_TimeDlyNs (CPU_INT32U dlyNs)
{
_pmuResetCounters_();
_pmuStartCounters_(pmuCOUNTER0);
while (_pmuGetCycleCount_() < dlyNs) ;
}
Ubaid
Ubaid,
This example het IDE project might get you started.
Run for some cycles like 50 loops then enable the program by writing a '1' to the data field of the 1st instruction.
(you can just 'poke' this value into memory in the IDE) ..
That enables the shifting. Shift register is 'SHBIT' and it's loaded with 0xAA to start. Shifts LSB first, you can change this though if you want by adjusting the shift type, direction, count.. Just study the syntax of the ADD instruction and it's shift options.
I had to set the prescale to 5 to make this work - which leaves 32 cycles for theHET program.
There might be some trickery to stretch to 64 cycles but this program is pretty short, 10 cycles to execute during the longest loop so you could add more functionality before you hit a limit.
Some comments to read once you get into the code and understand it:
a) The Macro 'CONST' doesn't work (syntax error) if you pass a value that's a float to it.
This I think is a limitation of the assembler or at least a limitation on my ability with it.
Since I set the HET clock to 100MHz or 10ns, 800/10 comes out as an integer and the macro works ok.
If you set to 9.9ns for 110MHz then you'll need to do the computation outside the assembler and put an integer as the 2nd parameter of the macro. I just wanted you to understand what's going on here:
CONST TIME_ONE_H, ((800/HR_CLK_PER)*4)
So I left as a formula. I couldn't figure out a way to get the power or exp function in the assembler, so the '4' is hard coded and related to the LRP prescaler of '5'. It would be 2^(7-LPRES) in the forumula - where ^ is 'power' not XOR.
I didn't bother testing this but you should be albe to try writing new data to SHBIT and then writing length-7 to NXTBIT (data fields) when the transmission finishes to start off a new transmission.
Anyway this is only meant to get you started.
Picture of 0xAA transmitting below. In Synapticad, you can measure the time by left clicking on 1 edge then moving the cursor to another edge. The upper right window shows the absolute time in black and the delta time in blue - so the delta time is the edge-to-edge time. You can measure at 800ns, 450ns, etc... of course in the screen capture I took I botched this and the delta shows -6.8 us!
-Anthony4064.starter_microwire.zip
Hi Ubaid,
In HalCoGen, you would select the custom program with the box below
Also use HalCoGen to set the pin direction - you have an output pin so you need to make sure it's port is setup as an output.
I used pin '0' but you can easily change this in the assembly .equ statement. EDIT: And use HalCoGen to set the LR prescaler
value to 5 (/32) as well.
In your program, you need to include the header file simp_pwm.h in addition to het.h. You can use the structures defined in this header to write to specific addresses in the custom program.
You only need to write to 3 locations -
pHET_START_0 - write '1' to the data field to enable.
pHET_SHBIT_0 - this is your 'transmit buffer' of sorts
pHET_NXTBIT_0 - this would be your bit count - 1. write value of 7 for 8 bits ..
After you include simp_pwm.h you should be able to create pointers to the particular instructions like this:
hetINSTRUCTION_t *pSTART = &(hetRAM1->Instruction[pHET_START_0]);
(make sure to pick the correct hetRAM - has to match the HET that you load the program into and you have 2 HETs..)
pSTART->Data = 0U;
to write to the 32-bit data field of the start instruction.
But please look at the program in the HET IDE first, and make sure you know how it is works so you'll understand it when you get to silicon. It's harder to debug HET issues on silicon than the simulator.
-Anthony
