Transistors getting hot

Thread Starter

CrazyAl

Joined Aug 27, 2010
20
Hi, I am using this setup below.



But the problem is the 4 transistor driving the gate drive transformer get hot even with heatsinks.

What can I do to fix this?

Thanks.
 

thatoneguy

Joined Feb 19, 2009
6,359
I am unable to decipher your schematic.

Please download a Proper schematic drawing program, such as LTSpice, mark resistor values, Transistor numbers and make connections more clear.
 

Thread Starter

CrazyAl

Joined Aug 27, 2010
20
I am unable to decipher your schematic.

Please download a Proper schematic drawing program, such as LTSpice, mark resistor values, Transistor numbers and make connections more clear.
Sorry, the two BJT's are BD243B's and BD244B's which are NPN and PNP respectively and they have 180 ohm base resistors and are being switched by the SG3525 whose two output pins are 180° out of phase from each other. Meaning that the BJT's are push-pull through the gate drive transformers primary.

They are in totem pole configurations in order to reduce the chip heating and help with making sure the MOSFET's on the other side of the gate drive transformer switch quickly. The MOSFET's are driving a transformer in a half bridge configuration.

Hope this helps.
 

Thread Starter

CrazyAl

Joined Aug 27, 2010
20
Those BJT that you use are too slow. Use switching BJT
Thanks. Is this why they are getting hot? What in the datasheet should I be looking at that says how slow BJT's are?

Can you recommend some BJT's I should use? The main thing is to quickly switch the MOSFET gates.

Thanks.
 

PaulEE

Joined Dec 23, 2011
474
Thanks. Is this why they are getting hot? What in the datasheet should I be looking at that says how slow BJT's are?

Can you recommend some BJT's I should use? The main thing is to quickly switch the MOSFET gates.

Thanks.
If transistors get hot, that implies excessive current through them with low voltage drop between emitter/collector, or low current and high voltage; either way, power dissipation.

I'm not 100% sure, because the schematic is tough to concentrate on the way it is drawn, but I would guess that the transistors are (npn/pnp pairs) on just long enough that they're shorting from npn's collector to pnp's emitter, pulling excessive current, and getting too hot to work properly before long.

Be sure that each transistor is getting a true square wave with no DC offset. DC offset will throw the whole operation off. If the SG chip is giving you 0 to +5v, that means the npn is happy but the pnp isn't as happy as it would be with 0 to -5v...I would check those waveforms.

I haven't looked into the transition frequencies or the capacitances of the transistors, but if the frequency is near 10-20 kHz, you should be O.K.

Also, on the transistors, for the purpose of testing, you might consider a few 1 ohm emitter ballast resistors. It's a quick method of ensuring that the output impedances of the transistors, in their fully-on state, have near uniform output impedances. And, they're a form of negative feedback. If the current begins to run away through the bjts, the voltage drop on those resistors rise, which limits the overall current that eventually gets through. You can also set up a protection circuit that monitors the voltage drop on these resistors and kicks the transistor base actuation signal upon a failure condition.

Hope this helps a bit.
 

Thread Starter

CrazyAl

Joined Aug 27, 2010
20
If transistors get hot, that implies excessive current through them with low voltage drop between emitter/collector, or low current and high voltage; either way, power dissipation.

I'm not 100% sure, because the schematic is tough to concentrate on the way it is drawn, but I would guess that the transistors are (npn/pnp pairs) on just long enough that they're shorting from npn's collector to pnp's emitter, pulling excessive current, and getting too hot to work properly before long.

Be sure that each transistor is getting a true square wave with no DC offset. DC offset will throw the whole operation off. If the SG chip is giving you 0 to +5v, that means the npn is happy but the pnp isn't as happy as it would be with 0 to -5v...I would check those waveforms.

I haven't looked into the transition frequencies or the capacitances of the transistors, but if the frequency is near 10-20 kHz, you should be O.K.

Also, on the transistors, for the purpose of testing, you might consider a few 1 ohm emitter ballast resistors. It's a quick method of ensuring that the output impedances of the transistors, in their fully-on state, have near uniform output impedances. And, they're a form of negative feedback. If the current begins to run away through the bjts, the voltage drop on those resistors rise, which limits the overall current that eventually gets through. You can also set up a protection circuit that monitors the voltage drop on these resistors and kicks the transistor base actuation signal upon a failure condition.

Hope this helps a bit.
Thanks! This helps. I do not have a scope but I will try what you said, I am switching them @ around 50khz.
 

Skeebopstop

Joined Jan 9, 2009
358
I am fairly in line with PaulEE with some other recommendations.

Generally push pull stages have dead time in built. Normally you'd only see something like you are attempting in a linear application whereas here it sounds like you are trying to modulate and only let through transformer pulses for secondary side pulsing. I'm a bit confused by the ultimate desire, but at least it seems you just want to shoot through bipolar pulses. Regardless, if you are in a switching application you've got a setup for a linear type output.

Normally for a switching application you'd have 4 gate drivers in this type of full bridge configuration, so that 2 of them can be dead band compensated. In your circuit in particular, you've got power BJTs heavily biased (due to low Beta). This means you'll generally have a HUGE storage time in the transistors (Ts is the parameter to look for depending on how heavily you bias). So you are definitely getting some HEAVY shoot through as the storage time is much longer than the turn on time of said power BJTs. You have to get into some really funny ESBT type topologies if you want to overcome this problem.

Generally also with BJT type application, assuming you can overcome the charge storage issue, will benefit greatly from a finely tuned 'on pulse' of charge to help reduce how much bias current you require to get them nicely diffusing. This is hard to achieve and I think beyond the scope of this exercise.

You seem to have few options unless you do some of the following:

1. change the chip you are using to have dead band compensation
2. change to mosfets, they are designed far better for a switching type application

Maybe you are trying to do this in a linear type application at which point you're just exceeding your power budget and driving BJTs in linear regions WILL get hot. There is a reason valve amplifiers glow. BJTs would too if only they could emit light.

Skeeb
 
Top