Re: For all the group

  Jim Thompson wrote:

I'm a good guy! Honest! :-)

Obligatory electronics related question: What are the
limits on ADC resolution? How close are we?

Jeroen Belleman

 On Tue, 18 Mar 2008 17:29:03 +0100, Jeroen Belleman


I already added you to the "good guys" column ;-)


I'm sure it's noise related, and thus depends on the configuration...
resistors; type: charge redistribution, current-steering, etc.; device
types: BJT or MOS....


                                        ...Jim Thompson
--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
            
         America: Land of the Free, Because of the Brave

 Jim Thompson wrote:

If you want to be a tight-ass (or smart-ass) about terminology, there's
no limit to _resolution_ -- how many useless bits do you want, though?

The limit to the useful number of bits in current-day converters seems
to be noise dependent.

Monolithic 16-bit converters that sample at 100ksps have about 1-2
counts of RMS noise -- this means that your output number will routinely
vary by four to eight counts from nominal.  This can be leveraged,
however; if you have the processing power you can sample up to the speed
limits of the converter then average down to your desired 'actual'
sample rate you get a nice anti-aliasing filter for closed-loop control,
and considerably more precision (if not accuracy) than 16 bits.

Ditto for 14-bit monolithic converters sampling at 40MHz, with perhaps
more LSB's of RMS noise.

I suspect that you could do better with a hybrid converter, but I'm not
Jim, so I don't know how much better, or how far you'd have to move
heaven and earth to keep the critter in production if you did.

If your process is slow you can get 24-bit sigma-delta A/D converters
that sample out at 60Hz.  I suspect the same (or perhaps only slightly
better) noise spectral density as the above converters, but I've never
done the math (it'd be interesting to do so).

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

 wrote:


Yes, I mixed resolution and accuracy :-(

And I agree with "how many useless bits do you want?"  But I'm sure
the Monster Cable crowd can "hear" the difference ;-)


Once upon a time, there was a plane called the L1011 (IIRC).  They
thought they needed a 16-bit DAC for the seat audio system.  I
demonstrated an 8-bit resolution plus 8-bit "gain-rider" that sounded
great and was CHEAP, but they opted to spend a lot of money.

Marketing people have no brains ;-)


                                        ...Jim Thompson
--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
            
         America: Land of the Free, Because of the Brave

 Jim Thompson wrote:

The 1011 is a very nice airplane.



I married one with brain :-)



I wish they'd invest the savings in better headphones. Just came back
across the ocean on American Airlines and their new "stick-in-the-ear"
thingies are horrible. THD was about 90% on both long-haul flights, you
had to crank it to max volume to hear anything. Terrible.

--
Regards, Joerg

http://www.analogconsultants.com/