possible tempco with KTY81

[Ian Fritz]

Rene --

Looks pretty good. I've seen this with thermistor compensation schemes
before -- you get a small, non-monotonic error over some fairly small T
range, then rapid increases outside that range. The way I see your data
is that between 10 and 40C you have a total deviation of about 0.1%,
which roughly averages out to about 30 ppm/K. It's hard to measure this
small a change! And, as you say, other sources of drift will be more
important.

I don't understand what you mean by the error being the error of the
compensation. I think a 1% deviation from the ideal R(T) translates
directly to a 1% pitch error.

  Ian 


Rene Schmitz wrote:
> 
> Hi all!
> 
> At 07:58 17.11.99 -0700, Ian Fritz wrote:
> >UNLESS R(T0)=C1*T0, which I believe is pretty close for the wirewound
> >Q81. (Terry Michaels showed some results from a Q81 spec sheet once, but
> >I don't have them any more.) Different kinds of tempco's would have to
> >be checked separately.
> 
> I think this is correct.
> 
> I've done some calculations comparing an ideal tempco resistor and
> the KTY81 + series 1350 Ohm. I've chosen the Ohms/K factor for the wirewound
> so that both devices have equal resistance at 25 deg.
> 
> T    R1(KTY+1.35k)       Relative error in percent
>                         R2=C*T
> -55     1825            1.712,13        -4,8236
> -50     1850            1.751,37        -4,2150
> -40     1902            1.829,85        -3,0832
> -30     1959            1.908,34        -2,1650
> -20     2019            1.986,82        -1,3751
> -10     2083            2.065,31        -0,7562
> 0       2152            2.143,79        -0,3509
> 10      2224            2.222,27        -0,0738
> 20      2300            2.300,76         0,0324
> 25      2340            2.340,00         0,0000
> 30      2379            2.379,24         0,0103
> 40      2458            2.457,73        -0,0117
> 50      2542            2.536,21        -0,2474
> 60      2628            2.614,69        -0,5686
> 70      2719            2.693,18        -1,1035
> 80      2812            2.771,66        -1,7238
> 90      2909            2.850,15        -2,5151
> 100     3009            2.928,63        -3,4346
> 110     3112            3.007,11        -4,4823
> 120     3217            3.085,60        -5,6155
> 125     3269            3.124,84        -6,1607
> 130     3320            3.164,08        -6,6632
> 140     3415            3.242,57        -7,3690
> 150     3495            3.321,05        -7,4338
> 
> C was chosen to be 2340 ohms/298.15 K (as for a 2.34K TC)
> The value for R1 is the value for typical R of the KTY81-1xx from the
> datasheet plus a series resistor of 1340 ohms.
> The relative error calculates to ((R2-R1)/2340)*100
> 
> Now what we can see here is that the KTY+R combination behaves pretty much
> like the wirewound within resonable limits. (-10..60 degrees, less than 1%
> error.) This shows that the KTY arrangement will pretty much work like the
> wirewound, within that range. Assumed you never leave that range it will
> behave like a wirewound resistor.
> I think if you'd take your measurements only within that range and you'd
> fit a line into the values it will also go thru the origin of your R vs. T
> graph. It is not required that the values outside the ususal range fit on
> that straight line.
> 
> I think the -10..60 degrees range is even more than we actually require.
> The 1% error
> is 1% error of the total compensation (i.e. 33ppm!). This means the drift
> will be compensated to >99% within that range. So we get >99% of the
> performance for 10% of the price of a Pt resistor!
> 
> That remaining error probably gets swamped by other factors, like opamp
> drift, capacitor drift, mismatch of temperatures (which also applies to
> Q81!)...
> 
> What do you think?
> 
> Bye
>  René
> 
> opinionsexpressedher   | uzs159 at uni-bonn.de
> einarethatofmyemploy   | http://www.uni-bonn.de/~uzs159
> erandnotofmyself....   | http://members.xoom.com/Rene_Schmitz
> 
> 
> 
>