Odp: [sdiy] really basic question .. tolerance

[Tim Ressel]

Okay, I understand now. I was a bit confused. You guys
are calculating the actual resulting TC of resistors
in various combinations. I was thinking in terms of
the effects of the TCs in combination. 

--Tim

--- Neil Johnson <nej22 at hermes.cam.ac.uk> wrote:
> Tim,
> 
> > Tc = (Tc1*R2 + Tc2*R1 + (R1+R2)*Tc1*Tc2*temp)/(R1
> + R1*Tc1*temp + R2 +
> > R2*Tc2*temp)
> 
> Agreed.  I've been working through it as well, and
> arrived at the same
> conclusion, if in a slightly different format:
> 
> Tc = R1.Tc2.(1 + Tc1.t) + R2.Tc1.(1 + Tc2.t)
>      ---------------------------------------
>          R1.(1 + Tc1.t) + R2.(1 + Tc2.t)
> 
> where t = temperature
> 
> Real nasty that the temperature coefficient depends
> on the temperature
> itself :-(  Folding this into the equation to get
> the actual resistance:
> 
> 	Rp(t) = Rp(0).(1 + Tcp.t)
> 
> shows that we have a nasty quadratic in t...yeuch!!!
> 
>  Rp(t) = actual parallel resistance at temperature
> 't'
>  Rp(0) = ideal parallel resistance at 0 deg.C
>  Tcp   = temp. coeff of parallel network, from above
>  t     = temperature
> 
> I won't expand it out, but I think its obvious we're
> going to get terms in
> t^2.
> 
> Nasty.
> 
> I think I'll stick with series combinations!!
> 
> BTW, interesting discussion on resistor tolerances
> earlier.  Did we come
> to the (correct) conclusion that tolerance is not
> some statistical 95% or
> whatever sampling, but the allowed range that a
> resistor can be advertised
> as?
> 
> When they come out of the oven, the resistors are
> not marked---they pass
> through a fast sorting machine that measures each
> one and puts them into
> appropriate bins for subsequent painting and
> bandoliering.  If you look at
> the standard resistor ranges (E12, E24, etc) you can
> see that they nicely
> fit together, with minimal range overlap (accounting
> for easy values).
> 
> E.g. the E6 range (+/- 20%):
> 
> 	Value	Range
> 
> 		80
> 	100 ----|
> 		120
> 
> 		120
> 	150 ----|
> 		180
> 
> 		176
> 	220 ----|
> 		264
> 
> 		264
> 	330 ----|
> 		396
> 
> 		376
> 	470 ----|
> 		564
> 
> 		564
> 	680 ----|
> 		816
> 
> As you can see, the ranges do overlap in some places
> (e.g. 330 overlaps a
> bit with 470) but in most cases they don't.  Same
> for the other series:
> 	E12 (10%)
> 	E24 (5%)
> 	E48 (2%)
> 	E96 (1%)
> 
> Which is why it is not that expensive to have custom
> valued resistors
> made---all they do at the factory is program one of
> the sorting machines
> to select the value you've asked for at the
> tolerance you've asked for
> (of course, they charge you for the privilege :-)
> 
> Cheers,
> Neil
> 
> --
> Neil Johnson :: Computer Laboratory :: University of
> Cambridge ::
> http://www.njohnson.co.uk         
> http://www.cl.cam.ac.uk/~nej22
> ----  IEE Cambridge Branch:
> http://www.iee-cambridge.org.uk  ----
> 


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com