Teensy 16bit ADC

Status
Not open for further replies.
Amazon and Ebay have many that can be made to work as a separate board wired to a teensy. But has anyone actually measured 16 bits of SNR from such a module?
 
What is the highest res for a teensy? Anyone know? I am trying to read a pressure sensor and looking for higher than 12 bit resolution and wondering if I can get it without adding an external ADC.
 
Is there a teensy board that incorporates a 16Bit ADC?

Thanks

What kind of ADC are you meaning? And what bandwidth? For audio there's the Audio Shield, but that's no
use for data acquisition for instance as audio ADCs aren't calibrated, may not even let DC through, and can
only be run at particular sample rates whether you want that or not.

Outside of audio there are many ADCs with various max sample rates and performance levels, the most
recent one I've used (though not with a Teensy) is the ADS8885, an 18 bit SAR ADC upto 400kSPS with
100dB SNR.
 
Low bandwidth reading a DC signal typically 4-20ma converted via a shunt and a opamp circuit to whatever input is required by the ADC. I was hoping there was a higher ADC input option on a teensy, more than 12 bits, maybe 16 or an over sampling option?
 
Low bandwidth reading a DC signal typically 4-20ma converted via a shunt and a opamp circuit to whatever input is required by the ADC. I was hoping there was a higher ADC input option on a teensy, more than 12 bits, maybe 16 or an over sampling option?

Trying for 16 bits on a noisy processor chip would never work anyway, you need quiet analog surroundings for
spur-free 16 bit precision, that's 50µV resolution for a 3.3V chip... And talking of precision do you need accuracy as
well, because that's going to need a quality stable voltage reference and probably require differential mode measurements,
and a high precision opamp, perhaps a CAZ or similar.
 
Mark, thanks for your insight on this subject. I have been developing with 16 bit ADCs with other microcontrollers and mitigated noise using various methods with much success. But back to the orginal question, are there any teensy variants with 16 bit (or greater than 12) inputs? (BTW, I don't have to use all 16 bits, one method I use is by shifting off the noisy bits)
 
You can easely average out a couple of thousand onbaord teensy ADC readings to get the resolution and noise kevel you require.
 
Mark, thanks for your insight on this subject. I have been developing with 16 bit ADCs with other microcontrollers and mitigated noise using various methods with much success. But back to the orginal question, are there any teensy variants with 16 bit (or greater than 12) inputs? (BTW, I don't have to use all 16 bits, one method I use is by shifting off the noisy bits)

As I said that sort of accuracy wouldn't be attainable on the same chip as a digital processor due to the high levels of
switching interference generated by digital circuitry. It wouldn't make much sense to put anything more than 12 bits
on a processor, and you typically won't see 12 effective bits anyway in that environment.

If you have noisy bits you should average, not discard. Averaging is a major technique for reducing (random) noise, and
can increase the effective number of bits in some circumstances. Averaging can only reduce random noise though,
interference is a real signal!
 
> can increase the effective number of bits in some circumstances.

Exactly - I once tried it with a teensy and averaging only helped somewhat - it stopped providing improvement after a point far short of 16 bits. Possibly a lock-in amplifier technique would have helped to reduce low frequency interference/noise. I also found that in some cases, median outperforms mean.
 
Unless you happen to synchronize perfectly, 1Hz noise doesn't average out if you average over anything close to 1 second.
 
It gets even worse when you try to write the data to an SD Card. Parts of the write cycle can involve short current spikes of 100mA or more. That puts a lot of stress on the Teensy's 3.3V regulator. The SD card itself has an internal processor generating a lot of high-frequency noise. I've worked on systems where the SD writes generated enough RF noise that you had to spend a lot of time and effort shielding the ADC inputs.

The problem is even tougher in that the SD current spikes can occur well after the Teensy thinks it has finished writing to the SD card. Thus, the "Stop collecting while we write to SD" approach fails.
 
if you average 256 times, you gain 4 bits. 16 bits dynamic range from a 12bit ADC really is no problem at all at low sample rates.
 
> if you average 256 times, you gain 4 bits

Only if the noise is high freq, white noise - and it often isn't either one. Go ahead, try it.
 
you sum all 256 measurement values, and divide by 256.
conviniently, dividing by 256 is the same as bitshifting by 8 which is mucj faster then actual division.
 
> if you average 256 times, you gain 4 bits

Only if the noise is high freq, white noise - and it often isn't either one. Go ahead, try it.

Where did you get that from? Not only are you contradicting your self (HF noise is by definition not white). But it is nonsense too.

I do it every day.
 
> Can you give me an example of how this is done?

Second that - let's see a program and output yielding 16 bit results from a teensy with 256x oversampling.
 
This has nothing to do with teensy. It is very basic signal. processing material. I am not going to do your homework and I am certainly not to put any efforts in teaching you.
 
Status
Not open for further replies.
Back
Top