I have come across a possible issue with a few of Bob Cordell's transistor models for LTspice.
Is there an explanation for the apparently incorrect variation with temperature of Vbe with the Cordell models for 2SC3503 and 2SA1381; 2SC3601 and 2SA1407? Is there some way to tweak the models so they more accurately model TC(Vbe)?
As a rule of thumb, TC(Vbe) is around -2.1mV/degC (NPN). Here are typical plots of Vbe against temperature for NPN and PNP devices at collector currents Ic ranging from 100uA to 100mA.
These are straight lines which open out slightly to the right; if one were to extrapolate the lines to the left, they should, in theory at least, all converge on the same point (~1.2V) at -273 deg C. All the LTspice stock models I looked at seem to be reasonably accurate in the way they model TC(Vbe).
In contrast, here are the same plots using Bob's models for 2SC3503 and 2SA1381.
The 2SC3505 is only showing -0.50 mV/deg C. The 2SA1381 is even more confused with the gradient actually changing direction for higher collector currents; TC(Vbe) varies between 0.44mV/degC at Ic=100uA and -0.14mV/degC at Ic=100mA (note change of sign). Here is a detailed plot of TC(Vbe) for these two transistors at Ic=16mA.
The plots for 2SC3601 and 2SA1407 show a similar pattern.
Here are the LTspice files used to obtain these plots - unzip the folder before running.
View attachment TC_Vbe.zip
If you are interested in seeing more results, download TC_Vbe.xlsm from TC_Vbe.xlsm - Google Drive - there are results from 16 transistor models, though they don't really add anything new to the above. There is a worksheet with calculation of TC(Vbe) for temperatures 0K to 373K at Ic=1mA. The spreadsheet loads into LibreOffice Calc in addition to MS Excel.
Comments/explanations/corrections appreciated.
Stephen
Is there an explanation for the apparently incorrect variation with temperature of Vbe with the Cordell models for 2SC3503 and 2SA1381; 2SC3601 and 2SA1407? Is there some way to tweak the models so they more accurately model TC(Vbe)?
As a rule of thumb, TC(Vbe) is around -2.1mV/degC (NPN). Here are typical plots of Vbe against temperature for NPN and PNP devices at collector currents Ic ranging from 100uA to 100mA.
These are straight lines which open out slightly to the right; if one were to extrapolate the lines to the left, they should, in theory at least, all converge on the same point (~1.2V) at -273 deg C. All the LTspice stock models I looked at seem to be reasonably accurate in the way they model TC(Vbe).
In contrast, here are the same plots using Bob's models for 2SC3503 and 2SA1381.
The 2SC3505 is only showing -0.50 mV/deg C. The 2SA1381 is even more confused with the gradient actually changing direction for higher collector currents; TC(Vbe) varies between 0.44mV/degC at Ic=100uA and -0.14mV/degC at Ic=100mA (note change of sign). Here is a detailed plot of TC(Vbe) for these two transistors at Ic=16mA.
The plots for 2SC3601 and 2SA1407 show a similar pattern.
Here are the LTspice files used to obtain these plots - unzip the folder before running.
View attachment TC_Vbe.zip
If you are interested in seeing more results, download TC_Vbe.xlsm from TC_Vbe.xlsm - Google Drive - there are results from 16 transistor models, though they don't really add anything new to the above. There is a worksheet with calculation of TC(Vbe) for temperatures 0K to 373K at Ic=1mA. The spreadsheet loads into LibreOffice Calc in addition to MS Excel.
Comments/explanations/corrections appreciated.
Stephen
Last edited:
Also watch out, some of Bob's models have XTI=1 which seems very unlikely to be correct.
The default is 3 and is tied to basic semiconductor physics so I don't see how it can vary much.
I noticed this a while back, as well as the EG problem that Toni also spotted.
Maybe one is a (bad) attempt to fix the other, or both caused by a curve fit failure.
I removed them from my own models but never raised it with Bob.
Best wishes
David
Interesting, I'll keep an eye open for this.
At least with the Cordell models I was mainly interested in, 2SC3503 and 2SA1381, XTI is set to the default value of 3.
Thanks
Stephen
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.