Hello Every one,
I am trying to communicate my PSOC microcontroller chip 5LP (LP035) with Raspberry Pi. I have written Pi program in C and java(in order to check the speed of communication between two languages)
PSOC(microcontroller) is acting as slave and Pi is a Master.
Between the communication I set up a mask that is responsible for giving back the required channel value to the master (Pi).
My code was working properly but now it start behaving different (I have not change any thing!) .
Firstly, It only works correct when master send a mask value either 1 or 3.
Secondly: If my program is working (communicating) properly with Pi (java programed) and if I run the same program with Pi (C programmed) and vise versa. I start getting garbage values and sometimes correct value along with garbage value.
I dont understand where the problem is laying.
I will be very thankful to you If any body can help me in it!
Looking forward to hearing from you!
Regards
Usually the bug is in front of the keyboard. (That is: your code is not appropriate.)
Especially if everything seems to work fine in Java but not in C. Most likely your C program is inferior to the Java one.
BTW: do you use some "Slave Select" signal? SPI is susceptible to phase shifts, thus the SS signal is employed to synchronize master and slave. But there is a wealth of other things that can go wrong, whether you use some hardware SPI implementation or some "software SPI": starting with CPOL and CPHA settings, violation of timing parameters, bad connections, EMI and so on...
Eventually you could monitor the signals with an oscilloscope to check whether the signals are as they should be.
Regards
Thank you Dreher for your reply.
Yes, I am using slave select signal.
The configuration is correct (Mode 0)
I also checked the signals with oscilloscope but the problem is laying on the mcro controller side (when it is connected with Pi-C programmed).
The problem is on the PSOC side? What about the settings there? And what about dead times and alike on both sides?
In a setup like yours, both communication partners are prone to implementation errors (bugs). A rather challenging task.
My former boss stated: "There are 2N possible programs - and only 1 is error-free."
(N being the number of bits of your program.) While exaggerating, there is some truth in that statement.
Yes It seems to be because the miso line is showing some garbage values. mosi line has the correct mask.
I also try to make communication between two psocs(one as a master and one as a slave). And in that case it was working all correct.
Regarding settings on the PSOC side.
Mode o (CPHA 0, CPOL 0)
Data lines: MOSI +MISO
Data bits 16
Bit rate (Slave) = Master Clock / 2 (122 kHz)
Shift direction : MSB first
"what about dead times and alike on both sides?"
how i can measure it? you mean the time in which no sampling is going on? or the time before and after communication start and end?
By the way I am very impressed with the saying said by your former boss. It's so true.
In my case N = 16. So........ :(
N = 16? No! N = 16 * !!!
Regarding data setup and hold times and alike: you cannot measure them directly. But indirectly as there is some specification relating to the SPI clock edges. And you have to do this for both receiving end points.
"bit rate (slave)" ? Is the clock rate on the SPI clock line. I would not relate it to any otzer clock - just measure the value to make sure it is within the specs.
BTW: what kind of "garbage" is MISO showing? Good signal - "only" the wrong values or disturbances, EMI or al8ke?
Yes the clock is correct.
By means of garbage value for example, if i supposed to get at 16 bit value 0xF00F.
I am getting sometimes 0xF00x, where x is a togalling/ varying number. And some times I got totally wrongs values and some times only first two bits i.e 0xFFxx where xx = different number.
BTW: Does this also means Buffer over flow?
I strongly recommend to use an oscilloscope to check the signal lines - and check vs. the datasheets. I'm not familiar with the driver specs of the RPi, but could well be that they are not the strongest. SPI is not intended to span distances beyond some cm.
If you're transmitting more than a single word, buffer overflow might occur. But then another one of the values transmitted is expected to be received. Data corruption within the value received is certainly no effect of overrun.
Hi Muhammad, there can be a few issues which cause garbage with serial communications. Generally, timing is everything, having too many statements and loops in the data collection part of your program can cause errors reading the SPI buffer as the statements take too long to process and the SPI register has moved on! (This can vary with the compliation options you set when writing the code!) Making sure the MC is has the correct oscillating frequency and adjusting the OSCAL variable if it is off can also help. Also check there is a clean switching between receiving and sending data, in many cases it can take some time for the MC or supplementary interface chip to switch between receive and transmit, and this can cause the data to be corrupted.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages