Impedance matching for 8 ohm speaker

[crutschow]

it does help to see before building a circuit what the effects of various tweaks and changes will do, or whether a circuit works at all....

Also makes it easier to do a worst-case analysis or see the effect of component tolerances (especially if you can do a Monte Carlo simulation).
And it's handy to be able to look at all the voltages and currents in a circuit with no effect on the circuit (which is generally not possible with the real circuit).

 

[audioguru]

A simulation uses the "typical" spec's for an active device like a transistor's hFE. But since the range of hFE is from minimum to maximum then if you design a circuit that works in the sim then it probably will not work if your device has near minimum or near maximum hFE.

I always design for "worst case" using minimum then maximum hFE numbers to make certain that every one I build works perfectly, not just the ones with "typical" transistors.

Of course you can buy hundreds of transistors from different vendors and test them all, then pick only the ones with "typical" spec's if you find any.
But I don't doo dat.

 

[crutschow]

It's a little bit of a pain, but you can go into the transistor model and change its gain to the minimum, for a worst-case simulation, as well as doing that for any other device models.

 

[unclejed613]

you can simulate the "murphy's law condition" where all of the part tolerances add or subtract cumulatively.... i forget the actual wording on that one, but it's something like: "all of the part tolerances will combine so all of the errors are in the same direction"....

 

[crutschow]

all of the part tolerances will combine so all of the errors are in the same direction"

Thus for a resistive divider, one resistor would be at it's maximum resistance tolerance, and the other would be at its minimum.

 

[kubeek]

That´s is what is here called as worst case calculation, but I have yet to see a tool that would easily include that into a simulation. A straight DC would suffice for 99% of the cases.

 

[audioguru]

A tolerance of 5% is common and is very low, and a circuit will probably work. But the typical hFE of a 2N3904 transistor is 200 and its minimum is only 100, a big difference.

What about the maximum allowed voltage, current and heating of a transistor? I have simulated an audio power amplifier with tiny very overloaded output transistors and it work fine in the simulation but it blows up in reality.

 

[crutschow]

What about the maximum allowed voltage, current and heating of a transistor?

Glad you asked.
Since Spice doesn't check that automatically, the normal design procedure is to determine the voltage, currant, and power, of any possibly stressed devices in the circuit simulation for over-spec values, which you apparently didn't do.
LTspice, for example, besides plotting voltage and current, will also plot the instantaneous power, as well as calculate the average RMS power, for any device, including IC's.
That can also tell you how big a heat-sink you may need for power devices, such as the output transistors in an audio amp.