Mail Archives: geda-user/2015/03/06/10:57:18
On Mar 6, 2015, at 3:58 AM, gene glick <geneglick AT optonline DOT net> wrote:
> we use this trick at work as well - converts a 12-bit ADC to 16-bit precision. I think the relationship requires you need 2^n samples for each bit of additional resolution.
If the digitization errors are random and uncorrelated from sample to sample, you need 2^(2n) samples for n additional bits (from the Central Limit Theorem). We’ll use this classic technique to get ~20 effective bits of video digitization for TESS (NASA’s next exoplanet finder), using data from analog signal chains I am designing in geda-gaf and ngspice.
If the errors aren’t random, you may be able to get better measurements by adding random noise and averaging. This is sometimes known by the horrid term “stochastic resonance”.
If you can arrange a controlled correlation between the errors in successive samples (“noise shaping”) and use a suitably weighted average, you can get a lot of bits without a lot of oversampling. This is the famous “delta-sigma” technique. Delta-sigma video ADC’s I designed using geda-gaf and ngspice will fly on the ASTRO-H x-ray observatory, soon to be launched.
For focal plane temperature control at -90±0.1C on the Suzaku x-ray observatory, I used a Pt100 RTD with a commercial 20 bit delta-sigma ADC. In an FPGA, a PID controller drove a delta-sigma serial DAC (all digital). The FPGA output controlled an adjustable power converter which powered a thermoelectric heat pump. I didn’t really need 20 bits here, but it didn’t have a significant cost in hardware, and not having to worry about digital errors made the analysis simple. That was years ago: several manufacturers now make 24 bit delta-sigma ADCs I would consider suitable for temperature control.
John Doty Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd AT noqsi DOT com
- Raw text -