EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Miyuki on November 18, 2020, 08:42:34 am
-
Hi folks,
I have a question about bipolar transistor speed
If I have a PNP transistor like in the attached picture
Does it have its delays td and tr what are commonly about 20-30ns each
And how it affects MOSFET fall time
Can this solution offer 10ns or faster MOSFET turn off speed?
I know at the time when gate charge is at miller level that bipolar transistor is already half conducting, so actual switch speed can be reasonable
Or will I need some other solution?
I want to reach about 50V/ns @ 0.5A/ns switching speed from 5A peak current and 400V
-
Some more information about your application is required. Topology? MOSFET? Are you commutating a diode?
Why not use a standard gate driver? Can you have a negative gate drive supply? The negative turn off voltage will help discharge the gate faster.
-
BJT transistors can turn on very quickly, its the turn off that is slow.
The speed of turning off is also very dependent on the conditions, generally they take much longer to turn off when driven far into saturation.
But yes as Phoenix said getting the best possible switching time you need to use a negative gate drive voltage. If the extra complexity of doing that is worth it for the bit of extra slewrate is up to you to decide.
-
I need it for interleaved PFC
And this circuit is in the recommended layout, even as the controller have 1.5A sink capability, the layout can be an issue here
It is CRM so MOSFETs turn on softly
Turn off if only hard switch and goes trough small SMD diode, film capacitor followed by big electrolytic
Mosfet used will be something like IPA60R099P6 with Qg up to 70nC but might be replaced with a better one or SMD
-
I need it for interleaved PFC
And this circuit is in the recommended layout
Who is recommending it? Looks like a TI figure.
-
I need it for interleaved PFC
And this circuit is in the recommended layout
Who is recommending it? Looks like a TI figure.
The upper figure is from TI papers - as a general concept
Here is how they have a reference design - On Semi/Fairchild
I see why they do it, it makes the layout much simpler as that transistor is directly at the leads of power MOSFET
Plus another benefit as this bipolar transistor cannot turn off fast, it provides resistance against false turn on
And also it bypasses the sense resistor
But how fast is actual MOSFET turn off / BJT turn on?
-
I need it for interleaved PFC
And this circuit is in the recommended layout
Who is recommending it? Looks like a TI figure.
The upper figure is from TI papers - as a general concept
Here is how they have a reference design - On Semi/Fairchild
I see why they do it, it makes the layout much simpler as that transistor is directly at the leads of power MOSFET
Plus another benefit as this bipolar transistor cannot turn off fast, it provides resistance against false turn on
And also it bypasses the sense resistor
But how fast is actual MOSFET turn off / BJT turn on?
Its mostly about keeping parasitic inductance on the MOSFETs gate as low as possible. If you have a BJT right next to it then the total loop area is very small.
The speeds are probably mostly determined by the MOSFET itself since different partnumbers can have very large differences in total gate charge. So if the gate capacitance is twice as big it will take twice as long to drain it.
-
I would say do not bother with discrete muller clamps as, while deceivingly simple, they are a pain in the back to get working reliably, moreover it will probably not help with the turn ff all that much, that circuit main purpose is to actually keep the mosfet off once it has been switched avoiding parasitic turn on due to miller effect when there is no negative supply.
The easiest and almost only way to speed up commutation is to get a bigger gate driver with more current sink capability, and decreasing gate resistance (you can use a diode to only decrease turn off one while leaving the turn on unchanged), an external current gain stage is also an option (push pull or totem pole bipolar topologies).
Also depending on the mosfet itself there is only so much you can do to speed it up, since internal gate resistance, package impedance, etc will put a hard cap on your current and thus switching speed.
Now if your transistor choice is not set in stone I would switch to a Silicon carbide one, as they have much lower gate capacitances and are thus much easier to drive, also stay as far as humanely possible from through hole packages, as their parasitics are horrible, and go for a SMD device (I know cooling is more difficult, but not impossible).
- If you felt extra adventurous you could even try Gallium nitride, which can quite easily reach 5 ns, however I would advise against it if you want something easy, as these type of devices are quite finicky to use
- gate driver must be very very clean, otherwise they will blow up
- soldering is a major PITA (think BGA/QFN only much worse as they must be directly connected to large copper planes that wick away heat ( no thermal reliefs of course), and must be done right otherwise the temperature unevenness And other effects will blow up the thing
- cooling must be on point, as their on state resistance more than doubles over the temperature range