I'd like to make DMA transfer 2 data from SAR ADC to Memory as switching AmuxHW, but the output of DAC was mixed signal of 2 data (reference to attached picture:Blue= SIG1, Yellow=SIG2, Green=output of DAC).
Digital delay circuit was referenced to CE95299 (http://www.cypress.com/documentation/code-examples/ce95299-delta-sigma-adc-using-16-multiplexed-sing...).
This is not enough: Can you please post your complete project so that we all can have a look at all of your settings. To do so, use
Creator->File->Create Workspace Bundle (minimal)
and attach the resulting file.
You ought to use the eos signal to switch your MUX to the next channel. Look in ADC datasheet for "eos" to get an explanation.
You could save some resources (edge detector logic) when using the "pulse" option of the control register.
You do a GetResult() but you do not know which channel was converted nor do you wait for a conversion ready.
Thank you for your kindness, Bob.
I intended to distinguish 2 data by setting 2 TD chains of one DMA. How can I intake multiple data as fast as possible by a ADC? I thought continuous HW trigger with using ADC and AMuxHw is best, but I couldn't understand how to distinguish multiple data in CPU.
ADC SAR Seq is a bit slow.... No, it is the very same. Data is stored internally into an array. Your main problem will be: What are you going to do with the data, sending over any communication interface will take too much time, even when you use SPI. Retrieving the data directly (no DMA) would just work, but what are you going to do with it.
I'll do PID control, so it's necessary to calculate from these data in CPU. I have to decrease time it takes 1 For loop to increase sampling speed, and it found out data intake time of ADC was dominant.
What kind of application makes it necessary to control a PID in µs range? I thought even for flight control ms range would be quite sufficient.
You need to estimate the amount of instructions you can execute within 30µs and how many instructions you need to process the incoming data with your algorithms (Madgwick?). I can assure you it is looking a bit tight 😉