MJL4302 & MJL4281 Models

[Mambo]

I noticed a great difference in the BF value of their models, with one = 191 and the other = 505.

From the datasheet they should be very similar;It seems another OnSemi error

[lineup]

so what should we use?
191 or 505

[AndrewT]

Others have commented that ONsemi's models are inaccurate.

Might be best to try and find better models.

have a look at this thread for ideas/info.
http://www.diyaudio.com/forums/showt...67#post1256867

[lineup]

here is link to on-semi models for
MJL4281A and MJL4302A

http://www.onsemi.com/PowerSolutions...s&rpn=MJL4281A
http://www.onsemi.com/PowerSolutions...s&rpn=MJL4302A

[lineup]

Originally posted by AndrewT
Others have commented that ONsemi's models are inaccurate.

And others have said On-Semi models are pretty good
And if I owned a such big company, I would not tolerate to have bad models online
if there were people complaining .. anywhere at www

A few more bad models, my friend, and you are FIRED from working at On.Semi

[AndrewT]

Originally posted by lineup


And others have said On-Semi models are pretty good
And if I owned a such big company, I would not tolerate to have bad modles online
if there were people complaining .. anywhere at www.

I have already complained to ONsemi (about datasheet errors) and their reply said they were currently looking at eliminating the contradictions in their Power Semi datasheets.

[lineup]

Originally posted by Mambo
I noticed a great difference in the BF value of their models, with one = 191 and the other = 505.

From the datasheet they should be very similar;It seems another OnSemi error

My question is, as I do not know what BF is:
Can BF be as different as 191 to 505 for two NPN, PNP in the same class?

MJL4281A NPN has BF=191
MJL4302A PNP has BF=505
Both models were done August 2003

[andy_c]

For an ideal BJT whose beta does not vary with current, BF is the same as beta. When beta varies with current, things get a bit messier. See Figure 9 on this page. That shows a plot of ln(Ib) and ln(Ic) vs. Vbe. Decreasing beta at low collector currents comes from the Ib characteristic deviating from the ideal exponential at low currents. Decreasing beta at high currents comes from the Ic characteristic deviating from the ideal exponential at high currents. If the high current and low current regions do not overlap at all, then beta will be constant at mid-current values and BF will correspond to the value of beta in this region. However, for a lot of devices there is quite a bit of overlap between the region where Ib is non-ideal (low currents) and where Ic is non-ideal (high currents). For such a device, BF can be much larger than the actual maximum value of beta.

The actual formulas that SPICE uses for all this are shown starting with equation (1) in the link above.

[wg_ski]

Originally posted by andy_c
and BF will correspond to the value of beta in this region. However, for a lot of devices there is quite a bit of overlap between the region where Ib is non-ideal (low currents) and where Ic is non-ideal (high currents). For such a device, BF can be much larger than the actual maximum value of beta.

For a device with a beta spike in the mid-current region like alot of the old TO-66 case power devices, one would expect this. But for a sustained-beta device one would expect the BF to come out at least in the ball park of the flat part of the beta curve. Model extraction algorithms can give stupid non-physical values to parameters if left unconstrained - and that's just operator error. Such models may be "what comes out of the optimizer", but in my particular experience with microwave nonlinear models they typically aren't as robust.

[andy_c]

Yeah, the problem is that the datasheets don't usually have the needed plots of Ib and Ic vs. Vbe using a log current scale. If those were available, one could find the regions where ln(Ic) and ln(Ib) fit the ideal behavior. By extrapolating each of these lines, BF could be found by noting that the vertical displacement between the two is ln(BF). Without those, it becomes a guessing game. For the MJL1302A, one would certainly expect BF to match the value of beta in the region where it's constant. But for the MJL3281A, beta increases with increasing current up to pretty large Ic. One needs to have telepathic powers to really know what's happening here without the needed data.