Micro with 20-bit A/D?

Go To Last Post
22 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

We just had a visit from a manufacturer's rep for a device with claimed 20-bit A/D performance. He had a board in-hand, and I could see no A/D chip. I couldn't get a good enough look at the markings on the main chip to identify it.

I wasn't aware of micros that incorporated 20-bit/24-bit A/D (probably sigma-delta?). If y'all know of any, pop out a few examples. Thanks.

(Of course, when we look at something cool we immediately think "well, WE could do that" with a bit of reverse-engineering...)

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What manufacturer?

It ought to be possible to integrate a delta-sigma ADC into a processor. They historically have had problems with muxing but, hey, 20 bits is 20 bits!

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At least Silabs, AD and TI do offer even 24-bit ADC on their MCU chips for a long time.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The ADuC7060/ADuC7061 are fully integrated, 8 kSPS, 24-bit data acquisition systems incorporating high performance multi-channel sigma-delta (Σ-Δ) analog-to-digital converters (ADCs), 16-bit/ 32-bit ARM7TDMI® MCU, and Flash/EE memory on a single chip.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

..as already has been shown in the links above.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The new cypress PSOC3/5 can hit 20 bits on their top of the line models. My impression was the hardware was built for 16 bit sampling, anything after that was oversampling post processing as the sampling rate dropped in line with what you would expect using a 16 bit ADC and oversampling for the last 4 bits.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The noise performance is nowhere near as good on the ADuC834 vs an external 24bit adc. I get a useable 14bits on the slowest sampling time with the '834.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To get near 20 bits, doesn't it take some unusual PCB design methods, low noise amps, and so on?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Not necessarily 'unusual', but attention to detail helps as the lower bits can mean nanovolts so every connection is a potential thermocouple, leakage currents can become significant etc. In my app there is little in the signal path apart from some filtering and protection. Using an external 24bit adc, I would probably achieve 16 bits or more.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks, all. I guess a good keyword is "mixed signal". Curiously the TI offerings are on an x51 core and not one of there proprietary cores.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

decoupling and filtering of every singel signal is becomming very interesting when you have such high number of bits.
I really wonder if when having 24bits adc you can actually say I have a 24 bit value, or that you have to strip a number if bits because the small amount of noise you always have will give garbage on those bits.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

theusch wrote:
Thanks, all. I guess a good keyword is "mixed signal". Curiously the TI offerings are on an x51 core and not one of there proprietary cores.

The TI offerings are nothing but a 12xx family designed by Maxim and sold to TI.

BTW, I easily get stable 19 bits @10 sps from a Silabs C8051F350 chip. Requires just a proper hands.sys driver :D

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Where do we find this "hands.sys" driver?
On a penis.exe?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

KArtman .. on Sariche's orbital ..or was it Wanke(l)r engine

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

decoupling and filtering of every singel signal is becomming very interesting when you have such high number of bits.
I really wonder if when having 24bits adc you can actually say I have a 24 bit value, or that you have to strip a number if bits because the small amount of noise you always have will give garbage on those bits.


I think in this app that the intent isn't to get the full 24 (or whatever) bits, but rather to get something a bit (or a few bits--pun intended) more than the 10 or 12 commonly found on a microcontroller.

If we actually purchase one of these OEM boards, I'll try to remember to report back on the signal conditioning used.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Bear in mind that all the delta-sigma ADCs initially are the 1-bit ADCs. Higher resolution there is achieved by oversampling, averaging, filtering and decimating this primary 1-bit data stream.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Would be good to see more 12/16bit adcs on mcus. I guess the diminishing returns past that point doesn't make it all that viable for general purpose applications. Having at least 12bit would be terrific, and the fact that the xmega has gone in this direction is awesome even though I believe there have been a lot of issues with it.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Some of the Freescale and TI Cortex M3 based MCUs have the on-chip 16-bit ADC.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Cameron707 wrote:
Would be good to see more 12/16bit adcs on mcus. I guess the diminishing returns past that point doesn't make it all that viable for general purpose applications. Having at least 12bit would be terrific, and the fact that the xmega has gone in this direction is awesome even though I believe there have been a lot of issues with it.

All of the STM32 line of ARM Cortex M3 micros come with 12-bit :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The key factor with 24 bitters is that they are linear out to 24 bits, meaning you can get an accurate 24 bit result by averaging out the system noise, especially if you use an FFT and if you are dealing with AC inputs (vibration monitors for instance), in this case it may take 4 seconds for each update.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The same oversampling/averaging/decimating technique can be very successfully applied to the standard 8/10/12-bit MCU SAR ADC.

Warning: Grumpy Old Chuff. Reading this post may severely damage your mental health.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Reference temperature drift is a big consideration at 24-bits. Linear Tech has several apnotes on circuit layout at 20-bits with a lot of discussion regarding copper solder junctions and air currents, etc.