| View previous topic :: View next topic |
| Author |
Message |
soonc
Joined: 03 Dec 2013
Posts: 215
|
| HW SPI and LCD Not working |
 Posted: Sat Jun 27, 2020 1:31 pm |
 |
|
Using a LCD with ST7565R Controller and bit bang the SPI everything works fine.
I'd prefer to use hardware SPI but can't make it work.
The display is a write only device so DI from the device is not needed.
The help file mentions:
"pins must be specified with DI, DO, CLK or SPIx"
What am I doing wrong ? Thanks in advance for any help.
| Code: |
#include <18F47K42.h>
// compiler V5.093
#FUSES NOWDT
#FUSES NOPRLOCK1WAY
#use delay(clock=48000000,crystal=12000000)
#pin_select SCK1=PIN_B4
#pin_select SDO1=PIN_B5
#use spi (MASTER, SPI1, MODE=0, BITS=8, IDLE=0, FORCE_HW, STREAM=LCD)
/////////////////////////
// does not work
void SPIout(int8 out)
{
spi_xfer(LCD, out ); // also tried spi_write() also did not work.
}
////////////////////////
// Bit Bang works
void SPIout(char out)
{
char mask;
mask = 0x80;
do{
output_low(PIN_B4);
if(out & mask)
{
output_high(PIN_B5);
}
else
{
output_low(PIN_B5);
}
output_high(PIN_B4);
mask >>= 1; // next bit
}while(mask > 0);
}
|
|
|
 |
jeremiah
Joined: 20 Jul 2010
Posts: 1354
|
|
| Posted: Sat Jun 27, 2020 2:30 pm |
|
|
Try a couple of things:
1. Set the input signal to the same pin as the output signal
2. Set the baud in the #use spi() directive to a slower speed than max. Maybe you are going too fast. |
|
|
soonc
Joined: 03 Dec 2013
Posts: 215
|
| Thanks |
| Posted: Sat Jun 27, 2020 4:29 pm |
|
|
| jeremiah wrote: | Try a couple of things:
1. Set the input signal to the same pin as the output signal
2. Set the baud in the #use spi() directive to a slower speed than max. Maybe you are going too fast. |
1. Help file says: "SPI cannot use the same pins for DI and DO."
2. Display is rated for 20MHz. Should be fast enough. I'll try reducing Baud and post results tomorrow. |
|
|
jeremiah
Joined: 20 Jul 2010
Posts: 1354
|
| Re: Thanks |
| Posted: Sat Jun 27, 2020 4:40 pm |
|
|
| soonc wrote: | | jeremiah wrote: | Try a couple of things:
1. Set the input signal to the same pin as the output signal
2. Set the baud in the #use spi() directive to a slower speed than max. Maybe you are going too fast. |
1. Help file says: "SPI cannot use the same pins for DI and DO."
2. Display is rated for 20MHz. Should be fast enough. I'll try reducing Baud and post results tomorrow. |
1. That's probably if you are specifying them in the #use spi(), which you should not be for a hardware SPI. Just try setting the pin_selects for both to the same pin. If it doesn't work, then at least you tried. Obviously, if the LCD did output data, you couldn't make them the same (you would have dueling drivers for the I/O), but in this case you don't have a physical I/O line from the LCD going to the DI line.
2. That's fine, but consider that software SPI is way slower than hardware SPI. So in situations where I find software SPI works and hardware SPI doesn't, it is often times the speed. Again, just give it a whirl (datasheets can have errors, parts can have errata, parts can be counterfit or mislabeled...seen all of that). If it doesn't help, then at least you tried.
All else fails, send an email to CCS support and ask them what the recommended setup is for this situation. Maybe they have one in mind that isn't obvious from the options in the help. |
|
|
soonc
Joined: 03 Dec 2013
Posts: 215
|
| Re: Thanks |
| Posted: Sat Jun 27, 2020 7:23 pm |
|
|
| jeremiah wrote: | | soonc wrote: | | jeremiah wrote: | Try a couple of things:
1. Set the input signal to the same pin as the output signal
2. Set the baud in the #use spi() directive to a slower speed than max. Maybe you are going too fast. |
1. Help file says: "SPI cannot use the same pins for DI and DO."
2. Display is rated for 20MHz. Should be fast enough. I'll try reducing Baud and post results tomorrow. |
1. That's probably if you are specifying them in the #use spi(), which you should not be for a hardware SPI. Just try setting the pin_selects for both to the same pin. If it doesn't work, then at least you tried. Obviously, if the LCD did output data, you couldn't make them the same (you would have dueling drivers for the I/O), but in this case you don't have a physical I/O line from the LCD going to the DI line.
2. That's fine, but consider that software SPI is way slower than hardware SPI. So in situations where I find software SPI works and hardware SPI doesn't, it is often times the speed. Again, just give it a whirl (datasheets can have errors, parts can have errata, parts can be counterfit or mislabeled...seen all of that). If it doesn't help, then at least you tried.
All else fails, send an email to CCS support and ask them what the recommended setup is for this situation. Maybe they have one in mind that isn't obvious from the options in the help. |
Thanks. You nailed it...
It did compile with same pin for DO and DI.
Next I reduced Baud to 400,000 and it worked.
The I increased the Baud number until it stopped at 8MBaud.
I got a little more curious and put the scope on the display CE line.
This code simply transfers 1024 bytes to the display each time it writes anything.
BitBang took 40mS to transfer 1024 bytes.
I ended up using 3.5MBaud and it now does the same transfer in 17mS.
Although everything compiled with higher Baud numbers 17mS was the fastest transfer time.
So I'll back off until the transfer is close to 20mS which is a nice 50 frames per second.
Although that sounds wonderful I just noticed the LCD does not keep up with the animation so I'll most likely have to slow it some more.
Anyway great help thanks I appreciate it. |
|
|
Ttelmah
Joined: 11 Mar 2010
Posts: 19539
|
|
| Posted: Sun Jun 28, 2020 1:47 am |
|
|
I don't see you turning off slew rate control.
The pins on this PIC have hardware slew rate control on their outputs.
This is why there is a warning in the data sheet for the fastest SPI rates:
| Quote: |
the slew rate control must be disabled on the clock and data pins
|
This may be why your faster attempts were failing....
set_slow_slew_b(FALSE);
Finding that the LCD can't actually keep up, is common....
However normally the key is to send the data nice and quickly, and then
use the extra time to do other things.  |
|
|
soonc
Joined: 03 Dec 2013
Posts: 215
|
| Nice Thanks |
| Posted: Sun Jun 28, 2020 3:48 pm |
|
|
| Ttelmah wrote: | I don't see you turning off slew rate control.
The pins on this PIC have hardware slew rate control on their outputs.
This is why there is a warning in the data sheet for the fastest SPI rates:
| Quote: |
the slew rate control must be disabled on the clock and data pins
|
This may be why your faster attempts were failing....
set_slow_slew_b(FALSE);
Finding that the LCD can't actually keep up, is common....
However normally the key is to send the data nice and quickly, and then
use the extra time to do other things. |
I had no idea such a thing was available...!
I found anything over 9MBaud did not improve the speed.
That got the transfer speed down to 13.94mS for 1024 Bytes.
Thanks great help .... |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
