Is it possible to control where the ADC samples within a period? MATLAB Simulink C2000 F28379D

Hi,

I have a waveform running at 25 kHz. I am currently running this waveform to the ADC on board which is the LAUNCHXL-F28379D. The programming of this is done with MATLAB Simulink using the C2000 addons.

The sampling period of the ADC in this case is set to 4e-6s, or exactly 25 kHz. The idea is to get a single sample within the period. But, I would like to change the location of this single sample and pass the result to the onboard DAC. For instance, given the sample period is 4e-6s, is it possible to get a single sample at 1e-6, 2e-6, 3e-6, and so on to my choosing? I attempted to change the SoC with ePWM3A as the trigger source and timer for the ADC and visually, the waveform going into the ADC seems to be relatively identical to the waveform the DAC outputs, besides the obvious quantization effects.

Is it possible to take a sample at a specific count and hold that value? What settings can I change to achieve that?

What is my best approach to this problem? Any individuals with experience and knowledge with C2000 Matlab Simulink would be greatly preferred and appreciated! Thanks!~

[Joeny Zhen]

I believe I may be able to do this with an interrupt, but I would like to see what others think and the execution of it.

[Fangjun Jiang]

The interrupt should certainly work, although function-call generator is the same. All you need is a periodic trigger with a time offset.

Make sure your simulation step size is small enough to hit the time offset. For example, if your sample time is 4s, then your simulation step size could be as large as 4s. But if you want a time offset of 1s or 3s, then the step size needs to be 1s or smaller. If the time offset is 2s, then the step size could be 2s or smaller.

[Joeny Zhen]

I tried this attempt out. The problem with this solution is that the frequency going into the ADC, and the frequency at which this periodic trigger happens is NOT the same. This causes the location of the sample to essentially shift in time due to the time jitter between the two parts. This is exactly why I am looking for a interrupt done specifically by the c2000 hardware on the F28379D launchpad board. I made my attempts, but it did not work out which is why i'm attempting to ask an expert to how to do so. Do you have a possible solution that involves making an interrupt using the C2000 blocks instead of standard simulink blocks that occurs at what i've described at my first post? For instance, if the sampling period is 4e-6, can I send an interrupt out at 1e-6 to obtain what the value is at that sample? If so, how can I go along doing so with c2000 simulink blocks?

Answer 1

[Fangjun Jiang]

Not too many people noticed this.

In the case of a discrete sample time, the vector is [Ts, To] where Ts is the sampling period and To is the initial time offset.

Specify Sample Time

[Joeny Zhen]

Hi,

This does not work out for the ADC block. It seems to only take one value, which is Ts. I tried this vector notation and it spits out an error saying: "The sample time must be a positive scalar or -1 (for inherited)"

[Fangjun Jiang]

1. Request the Mathworks to improve it. Its sample time specification is not up to the standard.
2. Try this workaround. Put the ADC block (with -1 sample time) inside a function-call triggered subsystem block, which is triggered by a Function-call generator block with sampel time specified as [Ts, To]

[Rik]

@Joeny Zhen regarding your flag ("Not accurate for this application."):

Flags are meant to attract the attention of site admins, not for comments.

[Joeny Zhen]

@Fangjun Jiang Although this workaround did allow the code to upload and run, it does not behave as what I expected to behave. The resulting waveform is still quantized version of itself, not necessarily take a value at an instance and hold it to inf.

For instance, let's say the total count in a single period is 5000. Is it possible to get the value from sample #2000 and hold that value?

[Fangjun Jiang]

Not sure if we are still talking about the same topic. Open this example model, hook a Scope block to the output of the Rate Transition block.

ex_specify_sample_time

Specify the sample time of the Rate Transition block as [0.4] and later [0.4 0.1], observe the timing of the first sample, it is 0.4 and later 0.5.

I think the latter is what you want, right? Everything else in your model is executed at time 0, 0.4, 0.8, 1.2, ... but you want the ADC to happen at 0.1, 0.5, 0.9, 1.3,...

https://www.mathworks.com/help/releases/R2022b/simulink/ug/types-of-sample-time.html

[Joeny Zhen]

Sorry but no, we are NOT talking about the same topic. All your solutions involve an idealized simulink environment. It seems that you are not considering how the C2000 hardware works in real time. By hooking a scope block and attempting to monitor a signal in it on c2000 hardware, I lose information simply based on how the hardware is designed due to many factors such as baudrate, and so on.

I am specifically looking for a potential solution that involves using the C2000 Simulink blocks. Hopefully an expert that involves the C2000 Simulink hardware addons can answer this problem. Thanks.