pulse current survivability of small BJTs

[zilog]

Im asking this question here and not in the class d forum because it has to do with basic BJT properties.

I want to design an as weak switchmode mosfet driver as possible. I will do this using PNP transistors with as low voltage and current rating as possible to avoid adverse switching performance (t_rise/t_fall). The currents that are switched arise from the charging and discharging of the gate- and miller capacitance of a mosfet, and the current peak only lasts for ~100ns or so. If a BJT is rated for 300mA repetitive current (say 2n3906), how much can this rating be abused if the current only lasts for 100ns, and the transistor has plenty of time to rest between charging cycles? I want it to stand at least 6A peak current, is this feasible?

[Nelson Pass]

The transistor's data sheet will give you thermal response times,
but they are all longer than 100 nS. I think you will want to
integrate the current x voltage over a given cycle and see what
the average is. A chip which is rated at peaks of 6A will probably
hold up fairly well.

[powerbecker]

zilog:
"I want it to stand at least 6A peak current, is this feasible?"

There is a difficulty :
Beta drops down much earlier and Uce_sat is huge!

[sawreyrw]

The 2n3906 is a 200 mA part. It makes no sense to consider this part in an application requiring 6 amps. I strongly suggest you use something with a higher current rating.

[zilog]

So what you say is that I will not reach that high peak currents because of negative impact on Beta at high Ic? The transistor I am thinking about if for the discharge of the gate charge in a UcD amplifier.

Will BJTs with higher current rating hold their Beta better at higher Ic and thus speed up gate discharge?

[sawreyrw]

Yes, consider something like the D45VH10.

[zilog]

Originally posted by sawreyrw
Yes, consider something like the D45VH10.

Heh, that transistor is HUGE and slow :/ I guess the ZTX718 would be better then, but I really hate having to leave the idea of SMD for this project :/ And availability is always a bitch when the sellers are located in the states.

[powerbecker]

Zetex has SMD´s and is located in GB
Or You try to find something from RS or Farnell!

[Elvee]

Hi zilog,

BJTs are not components that can tolerate high current densities; you're not allowed to average the current over any arbitrary period of time, because destruction mechanisms exist that have nothing to do with thermal damages.
One of these mechanism is the junction delocalisation which results in a much lowered breakdown voltage; another is the Kirk effect, which also affects the transition frequency making them useless for your purpose.
Instead of abusing defenseless BJTs, why don't you use specialized drivers for SMPS: the more moderns now have transition times in the ns range and are designed to handle peak currents of several amperes.
LV

[peranders]

You haven't thought of using a mosfet driver?