2
\$\begingroup\$

I am simulating circuits on SPICE (LTSpice and PSpice), I have assigned tolerances to both resistors, capacitors and inductors.

But now I would like to try to understand how much a transistor can deviate from its parameters.

Assuming to use the pair of transistors 2N5551 and 2N5401, I would like to understand what is the percentage of possible variation of all the SPICE parameters, not just the beta. In this way I would be able to reduce errors and predict the behavior of circuits in montecarlo simulations.

\$\endgroup\$
4
  • 4
    \$\begingroup\$ That is an extremely difficult question to answer. Look at the data sheet, it will give you most of the information, they guarantee the parts to meet that specification. Look at different manufacturers parts, the data sheets may be different. Will it exceed it, for sure but there is no guarantee by how much. They make transistors to make money and they need a little guardband to cover their processes. They guarantee the part to meet the specification not exceed it. \$\endgroup\$
    – Gil
    Aug 31, 2021 at 21:28
  • 1
    \$\begingroup\$ Brando, are you considering varying IS, BF, and TEMP? I would be wanting to do those, at least, for Monte Carlo. \$\endgroup\$
    – jonk
    Aug 31, 2021 at 22:23
  • \$\begingroup\$ You are not likely to ever get data for standard deviation on Is, BF,NS or NE. These process controlled by the each OEM to meet the 100% test validated product shipped. You need to calculate your design margin to worst case instead and potential yield loss with thermal stress. \$\endgroup\$ Sep 1, 2021 at 2:10
  • \$\begingroup\$ Brando, the IS controls the VBE and this will vary by a factor of 3 or so, either direction, from nominal (almost a decade span.) BF will vary more like about +/- 50% around nominal. And you most certainly will want to bracket your operating temperatures. You can ask Spice to specifically vary the temp of individual devices using DTEMP or TEMP. And all the rest isn't difficult. Just tedious to set up. Once done, though, you can just have at it. There is also an AREA, AREAB, and AREAC parameter set to vary, and it would be worthwhile to look into those, as well. \$\endgroup\$
    – jonk
    Sep 1, 2021 at 3:11

3 Answers 3

6
\$\begingroup\$

The question applied to a 2Nxxxx (JEDEC system) part is especially difficult; those parts are made to a specification which is loose enough that multiple manufacturers can fill the needs with very different devices, as long as the limits are honored.

The detailed data on old National Semiconductor 2N5551, for example, starts here

but that specifies 'process 16' from this manufacturer, and further info on process 16 is in a second data sheethere.

This manufacturer, and ONLY this manufacturer, will commit to the data on the second data sheet, for their "2N5551" devices.

Even disregarding the possibility of finding a counterfeit device from an unscrupulous source, the actual variations in a practical situation will be unpredictable (very small for adjacent dice from a single wafer, for example). If some variance is critically important, most manufacturers will select to suit a customer's needs, at a premium (to pay for the 'extra' testing).

\$\endgroup\$
1
  • \$\begingroup\$ For many 2N2222 variants you are lucky to get a 'typical' rating on the data sheets, with no min/max values offered. If designing a transistor circuit that depends on specific device parameters, well, that is often a bad choice. \$\endgroup\$
    – Jon Custer
    Sep 1, 2021 at 13:48
5
\$\begingroup\$

If you actually plan to build anything you will need an orderable part number from a specific manufacturer. In that case the manufacturer datasheet will usually list min/max/typical parameters for the parts. They don't necessarily list the same parameters used in a SPICE model, but its possible to make some inferences.

Here are some datasheets for some typical examples of these parts.

2N5551RLRPG:
https://rocelec.widen.net/view/pdf/ayujbdzwfc/ONSMS13513-1.pdf?t.download=true&u=5oefqw

2N5401:
https://www.nteinc.com/specs/original/2N5400_01.pdf

Having said that, the manufacturer only guarantees what is in the datasheet. If you make your design rely on anything not specified in the datasheet then you could find yourself with problems later on.

\$\endgroup\$
4
\$\begingroup\$

Assuming you simulate over temperature, then IS will be varied appropriately by the simulator.

The most variable parameter is likely device beta. You can put in values from the data sheet, or perhaps guesstimate at -50, +100 % from typ.

In a good design, you won't depend on precise values of these parameters, just 'sufficient'.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.