I am generating a 100 kHz interleaved signal (image attached below) with 80 % duty cycle using the following code for Arduino UNO. However, when I measure the signal frequency with oscilloscope, I get 48 kHz only. If I try to reduce the delayMicroseconds() the maximum achievable frequency is 61 kHz.
Why this is happening? Either with UNO I cannot get more than 61 kHz or the oscilloscope measurements are incorrect... I could figure out the reason. Please help me with this issue.
"This function works very accurately in the range 3 microseconds and up. We cannot assure that delayMicroseconds will perform precisely for smaller delay-times."
I don't know anything about the Arduino, but this is a very scary statement to me. This is a digital system and everything SHOULD be deterministic, right?
What is the overhead of the digitalWrite statement? Is it a single clock-cycle? 100 clock cycles?
Yes, have a search on the Web for some way to configure compilation to generate output listing of disassembled code, so that you can precisely determine how long each microprocessor instruction take from the core.
I've not done this myself on the Uno but doing a little Googling I found https://www.arduino.cc/en/Tutorial/SecretsOfArduinoPWM that mentions the MCU in the Uno has 3 hardware PWMs and shows how to manipulate the registers directly. Using the hardware is generally a better approach than bit-banging when timing is important.
Susan
Don't forget that everything is done via library code written in C++ on Arduinos. That and the library code tries to handle all of the variants of processor that can make up an Arduino.
Susan
With some examples I have managed to get one of the required signals with the following code using Time 1. How I can get the other phase-shifted signal?
One thing is modulation by pulse width varying in time, where the signal period is constant, but another thing is phase modulation where the pulse width can be constant. Should we assume that you mean by "phase-shifted", another PWM signal? Keep in mind that many uC's PWM peripherals have the same source for the period, therefore you likely would not be able to generate another PWM output shifted in time. Anyway, how about doing some calculation on you own? There are numerous tools available on the web that could help you use another timer to provide such a delay; this is a straightforward task.
Using assembly instructions and bit bang, it is easy to achieve your waveforms. A single cycle of 100Khz translates to 160 processor cycles. More than enough.
- - - Updated - - -
Using even fast PWM you will not get more than 62.5Khz frequency
What does all this mean ? Kindly explain. without explanation it does not mean anything at all. It's just a bunch of words written together which somehow might sound very smart.... but it doesn't make any sense. So please clarify.
:thinker: not more explanation than in post #10 ...
did you click on the link ?
even explanation are in french, C code or ASM code is universal ..
just one loop to toogle a bit (output of MCU) ON then OFF .. in asm ..
Frequency depend of the FOSC of the used MCU ..and number of MCU cycles to do an operation.
NOP instruction is used to adjust the period, so the frequency .
in this case a PIC18Fxxxx is used , not an Arduino.
...
just one loop to toogle a bit (output of MCU) ON then OFF .. in asm ..
Frequency depend of the FOSC of the used MCU ..and number of MCU cycles to do an operation.
NOP instruction is used to adjust the period, so the frequency ....
Yes, the name given to such a method is 'bit banging'. Arduino boards are typically run from 16Mhz crystal/ resonators, hence 1 processor cycle is 62.5nS. Most asm opcodes are single cycle (like NOP), while some are 2-cycle (like SBI, CBI), and some 3-cycle (like JMP).
Of course the downside is that cpu gets totally tied.
- - - Updated - - -
Precision of frequency depends solely on accuracy of crystal, and in case of 100Khz the duty cycle can be varied with precision of 1/160.