[time-nuts] Control Systems (was: uC ADC resolution)

Bob kb8tq kb8tq at n1k.org
Sun Jun 11 16:13:21 EDT 2017


> On Jun 11, 2017, at 2:21 PM, Chris Albertson <albertson.chris at gmail.com> wrote:
> On Sun, Jun 11, 2017 at 9:28 AM, Attila Kinali <attila at kinali.ch> wrote:
>> Hi,
>> You are asking a lot of question regarding control systems.
>> But, there are no easy answers there. Especially if you want
>> to build it cheap. The cheaper you want to be the more you
>> need to know and understand the problem.
> This has been my opinion for a LONG time.  It is easy to come up with good
> solutions if to just throw money at the problem.    So you see here people
> proposing just going top of the line all across but an engineer earns his
> money
> by comping up with cost effective solutions that meet all the stated
> requirements.
> This is my interest in mechanics too.  Can a $200 3D printed plastic robot
> arm
> with poor absolute repeatability place an M6 screw into an M6 nut?
> Certainly
> not if it runs open loop.  But what if you add visual feedback?
> Yes everything requires more expertise if you reduce the budget
> So you understand my questions are all like this:  "If you back down from
> top of the line solution how does that effect real world performance?"
> No one answers.

Would it help if each and every time we answered “that depends”. This is always
the real answer to any real world engineering problem. Quantifying and qualifying 
*all*  the dependencies is what people get paid to do for a living. There is *always* 
both theoretical and experimental work involved. There is never a simple one line 
answer. If you want the more detailed answer, start digging into it. Try this and that.
Measure what you see. Report your results and we’ll help you analyze them. 

Often this spirals instead into a game of liars poker. I can do it for $100, he can do 
it for $10, somebody else can do it for $1, the next guy is at $0.10 and we are at
$0.01 in no time. There is no data, not qualification of anything, just a bunch of random
cost numbers. That’s not how things work on any real design I’ve ever seen. Raw cost per unit, 
time / cost to implement, margin requirements,  volume of production, and performance 
are all tightly related  to each other. Until you nail all that down, talking about a price 
per unit does not make any sense.  

As an example, we are talking here about OCXO’s. Can you build a ten cent OCXO? Sure 
you can. It only takes a well stocked junk box of “free” parts. Can you spend a few years 
tweaking the one uint for performance, indeed yes again. Is it still a ten cent OCXO after you 
spend a year of your time tweaking it?  Would you sell someone a year of your time for ten cents? 
Would you build, test, and sell somebody a thousand of them for ten cents each?  


>> I suggest you reading an introductory text into control systems
>> like e.g. "Feedback Control of Dynamic Systems", by Franklin, Powell,
>> Emami-Naeini.
> -- 
> 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 mailing list