Gate Driver ICs vs Totem Pole using Transistors

[chinuhark]

I have been wondering about this for a while now, and couldn't find any proper comparison between the two.
My question is that apart from the saving in space which allows for a compact design, what are the benefits of using off the shelf Gate Driver ICs ( like say IR2110,IR2186,IR2136) as opposed to making your own gate driver using a pair of npn,pnp transistors in totem pole configuration. (Especially considering the fact that I will be using isolated supplies as it is and not be going for bootstrapping, and will have dead time in software itself)
What is drawing me towards the totem pole route is that a) It costs about 10 times less b) Current capacity is virtually no issue at all
Thus I can get an extremely powerful high current gate drive circuit at much less cost. Also it is extremely convenient to place them right next to the gate of the IGBT/MOSFET, so no stray inductance issues(Just 2 small transistors and a few smd resistors fit nicely next to the gate).
Also most importantly, does anyone have any idea about what commercial VFD manufacturers use, i.e. Half Bridge/3Ph driver ICs or independent totem pole for each switch. A Youtube video on VFD breakdown I saw seemed to have independant totem poles for each but I'm not sure.

 

[dick_freebird]

The big value is the on-chip phase control, which you
would be hard pressed to keep tight with a discrete lashup.
The faster you want to switch, and the closer you want to
control the turnoff-turnon tango, the more you like integrated
bridge drivers. Fault detection and protection can be another
bonus.

I think you'll also find that fast transistors are not high current
and high current transistors are not fast, when you get down
to actually designing and debugging the circuit, and that
"fast" demands clamping those switching BJTs in ways that
bloat your "simple" design, and can be very temperature
dependent (any saturation at all will blow your phasing
right out of the water and maybe burn you some power
MOSFETs).

 

[schmitt trigger]

I'm with Dick on this one.
On high power circuitry, protection and fault tolerance are the number one concerns. And those functions have to be implemented in hardware, not software.

You can always roll your own with discrete devices. But the devil is in the details, and by the time you have ironed out everything, the resulting circuit may be as costly as an integrated solution.

But this is only an opinion, without having any details or knowledge about the cost/ size/ performance requirements of your project.

 

[juicedry]

In terms of the NPN-PNP emitter follower configuration their is a good article on this by Robert Kollman on eetimes. Click this link for the Power Tip 42 (Part 1): Discrete devices—a good alternative to integrated MOSFET drivers.

Some key points about using NPN-PNP totem pole are:

1. The output voltage tracks the input voltage therefore if you only have 0V to 5V input to the totem pole, it will only give you 0V to 5V on output (Not quite 5V as it is a little less). Even if the Vcc rail voltage to the NPN on the top side is at a higher voltage, let's say 10V.

2. The gain in the NPN and PNP transistors is important, NOT necessarily the absolute maximum peak current that the transistors can handle but the actual high gain hold-up. This means that input circuitry to the totem pole only needs to provide a small amount of current for the output to drive to a high current. As the input voltage to the totem pole is being switched, and the raising/falling voltage edge approaches the input voltage rail, then the input current significantly drops off. For low gain transistor at high currents (<100), like the MMBT3904 3906 this means that the output current droops and it takes longer to charge or discharge the IGBT/MOSFET gate (capacitive load).

Check out devices like the ZXGD30xx series from Zetex (Diodes Inc), especially the newer ZXGD3005 and ZXGD3006 as these appear to have very high gain 1mA input gives 4Amps output. They are spec'd up to 10Amps.