[time-nuts] uC ADC resolution (was: Poor man's oven)
kb8tq at n1k.org
Wed Jun 7 14:32:26 EDT 2017
There is a gotcha with the initial assumption: You want the loop to be
*quiet* at a level well below 0.1C. If it is bouncing around that much,
the second order (rate defendant) tempco of a normal crystal will
become a pretty major issue.
Simple rule of thumb - add at least two bits past whatever the target is.
More or less, if you *are* after 0.1C and that comes out to 6 bits, you need
eight solid bits to get things to work properly.
> On Jun 7, 2017, at 2:10 PM, Chris Albertson <albertson.chris at gmail.com> wrote:
> One question for the control theory experts.
> Assume me goal is to regulate temperer of an aluminum block to within 0.1C,
> how good must my ADC be? Is an effective 6-bits good enough?
> It seems to me the problem with fewer bits is only quantization noise.
> Lets assume 6-bits. This is 1 part in 64. If I scale the input to the
> ADC such that it os 1.0C from 0 to 63 counts then each cunt is 1/64 C
> which is about 6 times better then my allowed error of 0.6 C.
> My gut-feel is that this is marginal but could work ("work" is defined as
> holds temperature within the range) but I'd be happier using 8 bits. Im
> pretty sure I can get 8-bits by over sampling and filtering.
> I don't know how to analyze this but I'm guessing with n-bits each each
> sample has a 1/2 bit error so my I and D terms in the PID controller will
> accumulate lots of 1/2 bit errors. I thing I want them "a couple orders
> of magnitude" smaller then the allied temperature range.
> Of cose one could buy the best ADC on the market. But this is POOR MAN's
> project. So he asks, "What is the lowers performance/cost part that will
> allow the system to meet its specification?
> BTW, a related story. I'm on another couple lists that deal with vacuum
> tube audio. We see the same things there people correctlypointing out how
> to make something better but the question is always how much better and at
> what cost an does it matter. So a fun project was proposed. Set a budget
> of $200 to build a tube based stereo Hi Fi amplifier. Who can do the
> best. Youhade to publish the BOM with prices and suppliers. Extra points
> if you came in under budget. This eliminated all the suggestions to buy
> high end hand made transformers from Sweden.
> IT turrets out that you see MUCH more interesting designs when you lower
> the budget. Anyone can make a high performance system even enough money.
> They waste half the cost on useless stuff and the product costs double what
> it should and is over complex but is works real, really well. That's
> easy. Harder and more interesting is "Can you make something just as good
> at 1/2 the price?" Answer is usually Yes. Then you say "what much do you
> loose if I set the price to 1/4? The answer is surprisingly little if you
> get smart about sourcing parts. Turns out about $180 is the minimum
> for pretty decent quality HiFi vacuum tube.
> An interesting graph would be Oven Specification vs. Price. What is the
> minimum cost for keeping temperature to within 1.0 C, for 0.1C, 0.01 C?
> Can you do 1.0C for under $5? or 0.1C for under $10. I bet yes.
> I did an exercise a while back to see what is the minimum price and
> complexity to build a GPSDO that was good enough only to drive the lab
> bench instruments I have. I implements only 1/2 od Lars W's design and
> cut his lines of code by about 90%. Turns outhe cost is the XO and about
> $10. Compared to my Thunderbolt, performance was not nearly as good but
> the ratio of performance over parts cost might be better.
> On Tue, Jun 6, 2017 at 2:39 PM, Mark Sims <holrum at hotmail.com> wrote:
>> Another thing to watch out for on processor ADCs is their performance near
>> the supply rails... the AVR ADCs are particularly entertaining below
>> around 300 mV (with a 5V Vref).
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/
>> and follow the instructions there.
> Chris Albertson
> Redondo Beach, California
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts