MCU recommendation for a failed parametric search, ADC and GAIN

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

Hello all, 

Hoping the group can assist me with a search that neither the Microchip Parametric or MAPS searches have yielded ready results for
MCU w/

ADC > or = 12-bits.

with gain > or = 100x.

(could be PIC as well)

Thank you, and look forward to the forum now that I'm finally registered after lurking for awhile. 

Last Edited: Sat. Nov 14, 2020 - 10:23 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ADC > or = 12-bits.

with gain > or = 100x.

That might not be a good combo for them to produce, unless everyone (including you) intend to do a cal.  The AVR is not a high precision opamp, so at a gain of 100, even a small gain or worse, nonlinear, error will be many counts off (at 12 bits).   If it were an 8 bit ADC, it might not even show up as a count difference.   Think about it, if you have 4096 steps...that is a small amount per step.  Now you ask for those small amounts to be 100x smaller...noise levels, offset, nonlinearity, etc then play a major role. Then comes the question of what very steady reference will be used for this high resolution measurement....and you want it all for $0.97).   If you are trying to measure really small voltages, a nice quiet opamp or specific ADC is your friend.

 

Do you need the accuracy or just the resolution??  What is your signal?  How slow can you go?  Slow is your friend, since you can integrate/avg out noise.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Sat. Nov 14, 2020 - 10:37 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My signal is 270uv-9mv, and my goal is to eliminate an instrumentation amp, if possible. 

~10hz, essentially DC.

 

10-bit would get the job done, in theory, but 12-bit would give me the headroom to toss some error out.  1% accuracy system (though I have huge room to cheat here, up to 5% or so).

 

just evaluating options. 

 

thx :)

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

Note you can get some amazing opamps these days that are amazing cheap (if you are worried about  $5 opamps) ....remember a lot of opamps can have millivolts of offset, and even worse over temperature.

 

https://www.onsemi.com/pub/Colla...   ...only $0.30   in quantity.

 

 1% accuracy system (though I have huge room to cheat here, up to 5% or so). ...then you could prob get there even with a poor opamp & cal.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

avrcandies wrote:
My signal is 270uv-9mv, and my goal is to eliminate an instrumentation amp, if possible. 

 

Honestly I'd keep the external amp, internal amps in most MCUs are not usually high precision. There are exceptions like  https://www.analog.com/en/products/aducm300.html#product-overview to find more try the keywords analog front end MCU.

 

You will notice such devices are not exactly dirt cheap but they come with lots of analog goodies.

 

I will also recommend a cheap fixed gain precision in-amp  https://www.analog.com/en/products/ad8293g160.html#product-overview (yeah, maybe I'm kind of partial to Analog Devices for analog stuff but Microchip may also have offerings).

 

edit: Since we are talking instrumentation amplifiers, let me add that the classic 3 opamp topology is just so 20th century... In the modern age I often prefer current mode in-amps:

https://www.microchip.com/ParamChartSearch/Chart.aspx?branchID=11017

https://www.maximintegrated.com/en/design/technical-documents/tutorials/4/4034.html

https://www.analog.com/en/products/ad8237.html

https://www.analog.com/en/products/ad8420.html

Last Edited: Sat. Nov 14, 2020 - 01:01 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There’s plenty of 20-24bit adcs from various vendors that will meet your requirements. Note that down at the microvolt level, noise is an issue so achieving much better than 14 bits is a challenge.

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


If you are willing to cal, things can get cheaper (but really, even the top devices are much cheaper these days).  What temp range do you need?  Watch that some high accuracy/low cost opamps can have poor temperature drift.

 

This ina180 part is extremely cheap (like $0.15)  & comes in 100x gain flavor 

https://www.ti.com/lit/ds/symlin...

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

enginerding wrote:
ADC > or = 12-bits.

with gain > or = 100x.

Gain is the limiting requirement; to a point, resolution can be adequate (reduce gain so add bits)

Consider an nJFET amplifier as a LNA into the MCU; MCU ADC noise data can be difficult to find, non-existent (you measure noise), or is iffy (therefore LNA)

AVR don't have a stated PSRR; lots of bits means a low noise VCC.

Offered up is XMEGA E and the forthcoming (imminent?) tinyAVR 2-series.

 

ATxmega32E5 - 8-bit AVR Microcontrollers (hardware oversample-and-decimate)

Assignment 3: 17-Bit Resolution | Getting Started with tinyAVR® 2 Family ADC Hands-On

 

"Dare to be naïve." - Buckminster Fuller

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

enginerding wrote:
... and my goal is to eliminate an instrumentation amp, if possible. 
AVR DB has enough op amps.

enginerding wrote:
... to cheat ...
... for margin ... wink ... that's reassuring.

 


TB3286 - Getting Started with Analog Signal Conditioning (OPAMP)

[page 39]

11. Instrumentation Amplifier

 

"Dare to be naïve." - Buckminster Fuller

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

enginerding wrote:
(could be PIC as well)
Advantage PIC as very recent PIC have DMA whereas DMA is in most XMEGA AVR.

What are the AVR28DA128 AVR32DA128 AVR48DA128 AVR64DA128 ?? | Page 4 | AVR Freaks

 

"Dare to be naïve." - Buckminster Fuller

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

gchapman wrote:

enginerding wrote:

... and my goal is to eliminate an instrumentation amp, if possible. 

 

AVR DB has enough op amps.

 

Enough to make an in-amp in classic 3 opamp topology, but not with the gain desired by the OP.

 

Though the gain can possibly be achieved if you create an in-amp in 2 opamp topology as described here

https://www.ti.com/lit/an/sboa281/sboa281.pdf

https://www.analog.com/media/en/training-seminars/tutorials/MT-062.pdf

and then use the 3rd opamp for further amplification.

 

edit: as described, but with RG = infinite (open circuit), i.e. like this device https://www.analog.com/media/en/technical-documentation/data-sheets/AD8236.pdf

I'm waiting for AVR-DB samples to arrive from Microchip. It will be an interesting experiment to implement this.

Last Edited: Sat. Nov 14, 2020 - 09:11 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Forget that....as soon as I saw micropower, I became suspicious...good in one spec, often means sad in another

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

I'm not recommending that device per se, but that it's topology can be generated using the internal opamps and resistor ladders of the AVR-DB with just 2 opamps, leaving one for a further amplification stage.

 

edit: and yes, I fully expect it will suck in practice compared to a good instrumentation amplifier, but I'm curious to test.

Last Edited: Sat. Nov 14, 2020 - 09:31 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

edit: and yes, I fully expect it will suck in practice compared to a good instrumentation amplifier, but I'm curious to test.

It would/should be decent (better than a single opamp)... The hard part is getting matching/tracking resistors,,,eventually the CMRR goes to crap, but this start out mucho better.

Back in the day this was not too expensive while the all-in-one instrumentation amp would set you back $$$$.  Amazing how far precision parts have come.

 

A few years ago worked with a company ...asked them why they always used 5% resistors...said 1% would be way too expensive...take a look, now they can be almost cheaper than 5%, due to volume.  Anyone use 10% parts??!!?

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

Thank you all for taking the time on this thread.

I have a few op-amps on hand, don't you think :)

The LTC1100, MAX4209, and LMP8358 (TI) are the in-amps I selected and design verified for this proj.  
They've sensed down around ground, rail-to-rail in/out on a single-supply, which is no easy task.  Most in-amps will outperform their specs with a high impedance load on the output.
 

I picked up an xMega-A3BU eval board for my first foray in AVR (minus the obligatory arduino time).
The ATxMega256-A3BU docs assert a 64x differential gain available.  So I though I'd see if I there was more gain + 12-bits ADC in the ecosystem.

I'll reply all in the same post, to keep the thread tidy :)

@avrcandies, @eltangas, @kartman, @gchapman, thanks for the input and feedback.  Good suggestions all around. thank you.

Also, thanks for the op-amp suggestions.  Can never have too many.  Have another ESD box to fill. 

I'm gonna bench the xMega-A3BU, tonight or tomorrow, and see what I get out of it at 64x gain stage, just out of curiosity.  

But doesn't seem like their is any MCU in the AVR ecosystem that has differential gain of 100x or more, and 12-bit or more on the ADC.

 

 

Attachment(s): 

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

Note the ADC has "gain" in terms of the Vref...if you use 1V ref vs 3V ref, your numbers will be 3x larger.   Theoretically you could use a Vref of 50mV and have a huge gain. However, that won't (supposedly) work, the ADC will fall apart with such a low Vref....too bad!!

 

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

You're welcome.

enginerding wrote:
Have another ESD box to fill.
Those are handy.

enginerding wrote:
I'm gonna bench the xMega-A3BU, tonight or tomorrow, and see what I get out of it at 64x gain stage, just out of curiosity.
Consider XMEGA E as there are reasons for that ADC (high source impedance, oversample-and-decimate); XMEGA A ADC is an order-of-magnitude faster (pipelined)

enginerding wrote:
But doesn't seem like their is any MCU in the AVR ecosystem that has differential gain of 100x or more, and 12-bit or more on the ADC.
High gain is difficult with noisy MCU.

 


Super Kits Enclosure for SMT Resistors, Capacitors and Inductors : Analogtechnologies.com

 

ATXMega32E5 breakout board from Geppetto Electronics on Tindie

OSH Park ~ nsayer - XMega_E5 Breakout v1.0.2

 

AVR MCUs | Microchip Technology

[right column]

View our Quick Reference Guide to get a quick overview of Microchip's 8-bit AVR portfolio, peripheral integration, and migration options.

Mar'20 so doesn't have AVR DB.

 

edit :

MT-DB-X5 User Guide

 

"Dare to be naïve." - Buckminster Fuller

Last Edited: Sun. Nov 15, 2020 - 04:15 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I can always bring VREF down and rely on more bits, but coming from the analog perspective, my instinct is to raise small signals as much as possible to get them into the "goldie lox" zone where they're more resistant to small-signal problems (interference, thermal EMFs, etc).  

 

@gchapman, i'll take a look at xmega e.  In a "doh" moment, I assumed the A3BU dev board had display support on the MCU since it had an LCD on the dev board.  Nopers. The LCD is driven separately. 

Looks like the xmega B dev boards w/ LCD are hard to find.  Might have to just solder something together with an atxmega128b1 chip.

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

I can always bring VREF down and rely on more bits,

LIke I mentioned it's a wonderful thought, but you are only allowed to go so low (like maybe 1V full scale) ...too bad you can't go 100mV full scale.

 

https://www.edn.com/measure-twic...

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

enginerding wrote:
Looks like the xmega B dev boards w/ LCD are hard to find. 
XMEGA B likely won't be updated; if superseded then a design closer to AVR DA (AVR DB, ...) is more likely.

Segmented LCD PIC24 have been updated.

New PIC24F MCUs Feature Low-power Animated Display Driver for Battery-powered Devices | Microchip Technology

New PIC24F GU/GL Family of Low Power MCUs - YouTube (5m29s)

 

"Dare to be naïve." - Buckminster Fuller

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

AVR DB op amps are very noisy.

Given the AVR DB op amp SPICE model, LTspice 17.0.18.0 didn't converge and produced

ERROR: Node U2:98 is floating and connected to current source G:U2:70

(U2 is that model)

 

AVR128DB28 - 8-bit Microcontrollers

[Documents tab, bottom]

SPICE Model for AVR_DB

 

"Dare to be naïve." - Buckminster Fuller

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

enginerding wrote:
Might have to just solder something together with an atxmega128b1 chip.
Possibly quicker to attach an LCD to a STK600 such that CPU, memory, power, programming, and the LCD portion of I/O will be done; then, any soldering is for specific I/O.

 

STK600-LCDX

via

ATxmega128B1 - 8-bit AVR Microcontrollers

[Development Environment tab, Demo & Evaluation Boards button]

"Dare to be naïve." - Buckminster Fuller