ADC speed?

[ESP_Sprite]

There's no filter we put intentionally in the signal path. The bandwidth of the ADC just is more-or-less 6KHz if you want a linear frequency response.

[Archibald]

Within the Arduino IDE and using the micros() function to time ADC sampling, I am measuring the sampling time to be close to 9.5μs (including storing measurements in an array).

Do we know if ESP32s contain sample-and-hold capacitors like the Atmel chips? If not, one wonders how well the successive approximation works with high-frequency high-amplitude signals. I have not found any mention of sample-and-hold in the Technical Reference Manual.

[bobolink]

There's no filter we put intentionally in the signal path. The bandwidth of the ADC just is more-or-less 6KHz if you want a linear frequency response.

Oh, OK. So it could be things like layout within the chip or something that limits to 6KHz.

I add an external anti-alias analog filter to limit the bandwidth to speech (4KHz) so I should be OK. I saw all these 44.1 KHz projects and wondered what was going on. Maybe they equalize.

Thanks

[eric.osowski]

Hello, how are you?
You said that are have a vibration analysis project. I'm reseaching about the same project, for my final papper of Electrical Engineering.
I thought that ASP 1115 could do the job, but I didn't see that its SPS is very low.
Now,I have the same issue than you.
So the question is, did you find a solution for the rate of samples? I beliave that have to be biger than 30kSPS, what do you think?
could you help me?
if yes, my e-mail is eric.osowski@hotmail.com
if not, thank you anyway =)
Thanks.

[Strontium]

I am working on a development module that will have (among other things) 7 ADC inputs capable of 12 bit @ 1Msps. I am just finishing the layout now and will post more information elsewhere, however if you are interested in a small ESP32 based module with this feature drop me a PM and i will keep you updated.

[mbalci]

There's no filter we put intentionally in the signal path. The bandwidth of the ADC just is more-or-less 6KHz if you want a linear frequency response.

So what is the conclusion ? what is the maximum sampling rate of adc in esp32? I could not see such information in datasheet , very interesting why it did not included in it.

[BaartCM]

I found out by experimenting that I can read the adc’s at a much higher rate than 6k but that the data is garbage. If I sample vibration readings from a motor at say 4K sps, I get a peak at the correct frequency when I do an fft but if I sample at a higher rate than 6k, the peak is not at the correct frequency and the fft content is rubbish.

[stephensworld]

I also need a higher sampling rate. Is there any way to get good results with higher sampling rate than 6Khz? External adc is no solution for me.

Thanks

Stephan

[BaartCM]

I also need a higher sampling rate. Is there any way to get good results with higher sampling rate than 6Khz? External adc is no solution for me

I’m afraid it would seem not. Your only option with the ESP32 is an external ADC. The STM32duino (BluePill) can sample much faster but has no WiFi or Bluetooth and much less memory and runs a lot slower.

Can I ask what is the problem with using an external ADC? An mcp3208 is easy enough to connect to and has libraries available and the cost is low.

Steve.

[mbalci]

How to set adc sampling rate when we get data via I2S ? As we can see I2S uses WS signal to sample adc , so what is the relation with
i2s_config.sample_rate = 120KHz; and ws signal ? is I2S sampling rate frequency of WS signal or frequency of i2s clock signal ?