• pin
  • Member
  • Offline

Topic: Number format for trendline's constants

In latest beta (4.4/512 - earlier versions not examined), when specifying user defined trendline function, initial values of constants cannot be entered in exponent notation - not 1E-6, but 0.000001. For very big/small values this is rather annoying. Then, going back to edit those numbers, they are shown converted to exponent format. It looks strange.

Re: Number format for trendline's constants

I am not able to reproduce the problem. Or maybe I just don't understand what you are doing. When I insert a tangent both 0.000001, 1E-6 and 10^-6 works in the x= field. Can you maybe clarify what you are doing?

3 Reply by pin (edited by pin 2011-11-20 00:05:10)

  • pin
  • Member
  • Offline

Re: Number format for trendline's constants

OK. I have measured some points on LED I-V curve and I try to find constants N, Is, Rs, which come from Shockley's equation:
Vf = N.Vt.(ln((If-Is)/Is) + Rs.If/N/Vt) ; Vt=0.026 for room temperature
Please, follow me:

Insert the point series:
x .......... y
0.000000944, 1.339
0.00000298, 1.404
0.00000779, 1.455
0.0000279, 1.505
0.0000922, 1.558
0.0009172, 1.657
0.000277, 1.603
0.00276, 1.710
0.00910, 1.810
0.027, 2.100

Insert User defined trendline:
$N*0.026*(ln((x-$Is)/$Is) + $Rs*x/$N/0.026)

Insert initial values:
$N : 2
$Is: 1E-22
$Rs: 5

In my case, I had to input $Is as
0.0000000000000000000001

When editing the model, value 1E-22 is shown, but if you erase the 'E', you cannot put it back, all keys except numbers and decimal point are ignored.

Re: Number format for trendline's constants

Sorry, my mistake. For some reason I thought you were talking about tangents when you clearly wrote trendline. Thank you for the clarification. I can reproduce the problem now. I will try to get it fixed soon.

Re: Number format for trendline's constants

I have now tried to fix this in the latest beta version. You can now use either 0.0000000000000000000001 or 1E-22 and Graph will remember what you wrote.

  • pin
  • Member
  • Offline

Re: Number format for trendline's constants

Many thanks! I think I will highly prefer 1E-22 format. smile