Help! Best Microcontroller which has 20+ channels, 50,000+ readings, 24bit+

Using Arduino Microcontrollers
1

Hi,

I have been building a headset and have been trying a few different microcontrollers that fit my specifications but I can't find any and I was wondering if anyone knew of any. I am aware that Arduino is basic and not made for my specifications so if you have any other boards that fit the criteria that would also be perfect.

Specifications:

  • Needs 20+ channels (10 for photodiode input (Something like analog read) and 10 for LEDs)
  • Needs to be able to take 50,000+ readings per second (From a photodiode using something like analog read to see how much light is being captured really quickly)
  • Needs to have accuracy on Analog read (I think it would be 24+ bit (Needs to have an accuracy of 0.0000001 or more for analog read working at 50,000+ readings per second))

I am aware that these are high specifications I just need to take a lot of readings accurately and Arduinos are not that good for this but if you know any other boards which can do this which are not too hard to program this would be very helpful.

Thanks,
J

2

What have you tried so far ? what results did you get ?

3

Needs to have an accuracy of 0.0000001 or more for analog read working at 50,000+ readings per second

Would that be 0.1 microVolts?

What is the closest example that you have been able to find?

4

ard_newbie:
What have you tried so far ? what results did you get ?

Currently, I have been using an Arduino shield
This one is 24 bit so had a good range but was very slow and could only take about 7-14 readings a second but had a good range for accuracy. ARD-LTC2499: 16-Channel 24-Bit ADC Data Acquisition Shield for Arduino (3.3V IO Version) Then I started to use the ARD-LTC1867: 8-Channel 16-Bit 200ksps ADC Data Acquisition Shield for Arduino which is much faster but doesn't have good enough accuracy. I was getting a few thousand readings a second but the range was only a little better than a standard Arduino for accuracy.

Any ideas?

5

jremington:
Would that be 0.1 microVolts?

What is the closest example that you have been able to find?

I'm not sure, I am not very skilled in electrical engineering. Using 24 bit I got around 0.00001 accuracies. I'm not sure if I am saying it correctly but that is how accurately it could monitor the photodiode.

6
  1. You likely don't need 24 bit precision.
  2. It is not easy to get 24 bit precision - you will measure noise unless you are a very skilled EE.
  3. You can get either high precision OR fast sampling speed. Both will be very expensive AFAIK.
  4. Even if you are skilled and wealthy enough I doubt the photodiodes have such precision.

It sounds like an "XY problem". Tell us what you want to do and someone may show you a way how to do it.

7

Smajdalf:

  1. You likely don't need 24 bit precision.
  2. It is not easy to get 24 bit precision - you will measure noise unless you are a very skilled EE.
  3. You can get either high precision OR fast sampling speed. Both will be very expensive AFAIK.
  4. Even if you are skilled and wealthy enough I doubt the photodiodes have such precision.

It sounds like an "XY problem". Tell us what you want to do and someone may show you a way how to do it.

Hi,

Thanks, we are building a brain-computer interface, just a simple demo we just need accuracy as good as possible. The sampling rate can be lower to get precision. Maybe 5000+ but the accuracy is essential. The photodiodes are very high precision as well.

Thanks

8

What photodiodes are you using? Please post a schematic diagram of the transimpedance amplifier that you have selected.

9

Needing a 24 bits A2D means you expect a range of values with 16e6 to 1 ratios - ie you need to measure 1V and also 0.1uV (and within 20us of each other!) There are other (much easier and cheaper) solutions for measuring very tiny voltages within a relatively small range (measuring 0.1uV to 25uV with 0.1uV accuracy only needs an 8bit ADC (and a somewhat carefully designed analog amplifier.)

Doing anything meaningful with a million samples/second (20 chan * 50k samples/s) is going to be challenging entirely separately from the ADC issues...)

10

jamesdacombe:
... just a simple demo we just need accuracy as good as possible. The sampling rate can be lower to get precision. Maybe 5000+ but the accuracy is essential. The photodiodes are very high precision as well...

Design your machine first, and know the range of precision required to support your output. Then specify your needs to make hardware selections. If you ask what's available, then design, you'll always be a bit behind.