2
\$\begingroup\$

I tend to design my circuits around H-bridges, but the problem with making them out of standard mosfets is that I always have to compare the rise times, fall times, threshold voltage, etc of my n-channel transistors and my p-channel transistor. I figured this circuit would be best as an IC, but the h bridges I found all seemed for relatively low current and voltages. I figured if this type of device was mainstream, it might exist under a different name.

Are there any standard ICs that consist of well matched p-type transistors and n-type transistors?

I am looking to switch ~3kW at frequencies on the order of ~1kHz-1Mhz (depending on what is feasible.) The current and voltages being used will depend on what is feasible. This may require two separate devices depending on what I choose to do.

My high voltage applications would probably not exceed 1kV (to start), and my high current applications would probably not exceed 50Amps (to start.

\$\endgroup\$
  • \$\begingroup\$ "high" and "low" are not valid specifications. Numbers, please! \$\endgroup\$ – Kevin Vermeer Feb 29 '12 at 16:24
  • \$\begingroup\$ @KevinVermeer, my fault--I forgot to put that in. \$\endgroup\$ – Alex Eftimiades Feb 29 '12 at 16:57
  • 3
    \$\begingroup\$ Think about it. No IC is going to be big enough to handle that kind of power. Even at 99% efficiency that would be too much heat for any IC to deal with. \$\endgroup\$ – Olin Lathrop Feb 29 '12 at 18:43
  • \$\begingroup\$ I knew seemed like a lot, but I really did not know that it was impossible. Well, thanks for not downvoting it (knock on wood)! So then I suppose contractors are the only solution? \$\endgroup\$ – Alex Eftimiades Feb 29 '12 at 19:31
  • \$\begingroup\$ I was reading some related questions, and came across this electronics.stackexchange.com/questions/15107/… The ic, IRF7739, was said to work at 40V 375A. That is 15kW--way above what I was looking to do. So it seems like there are ICs that can handle that kind of power--easily. I just wanted an H-bridge. \$\endgroup\$ – Alex Eftimiades Feb 29 '12 at 19:36
8
\$\begingroup\$

There seems to be some confusion as to the power requirements. This device does not need to dissipate the energy produced by a 1kV drop at 50A; that power is dissipated by the load. This device will operates in two states: 1kV blocking at milliamps (or less) of leakage current, and 50A of current at milliohms (or less) of resistance. That results in reasonable power dissipation levels. It doesn't have to dissipate 50kW of energy.

At the voltage and current levels you're working at, you should investigate using IGBTs, or Insulated Gate Bipolar Transistors. A MOSFET has a necessarily small switching region where the electric field (the FE in MOSFET) can work; bipolar transistors have the advantage that the switching region can be a large plane or plate of silicon. This allows IGBTs to exceed the capabilities of MOSFETs in extremely high-voltage and high-current situations. This IRF appnote describes some of the decisions to be made between IGBTs and MOSFETs, summarized concisely in this graph:

enter image description here

At 1kV, you're at the upper edge of MOSFET capabilities, and should probably use IGBTs. You mention that you want to go to 1 MHz "depending on what is feasible" - Most anything can be feasible given enough budget. I suggest that you try to make an IGBT work for you.

With respect to H-bridge ICs and arrays, sure, there are lots of these. However, they're spec'ed out to typical use cases. Digikey claims to have over 3,000 FET arrays. Unfortunately, the highest Vdss capacity is only 300 V, much less than your 1kV requirement.

IGBTs are more available in arrays too. Digikey shows a few IXYS parts, and their web page does in fact show quite a few parts which meet your spec: Take a look at their website and follow the links to IGBT Modules -> Full Bridge IGBT Modules (their website isn't very good at deep linking). Here's an example datasheet. Notice that this isn't a typical surface-mount PCB item; it's a 120 mm by 60 mm chassis-mount module. These things can put out some serious heat, and can switch some serious power.

You're in exclusive territory here; this isn't something you'll find at your local Radio Shack (not that you'll find much there anyways...). Finding parts with similar rise and fall times is the least of your worries!

\$\endgroup\$
  • \$\begingroup\$ Thank you for the information. I knew this was possible (otherwise there would be a lot less interesting DIY projects), I just did not know where to look. \$\endgroup\$ – Alex Eftimiades Feb 29 '12 at 22:21
  • \$\begingroup\$ Stumbled upon the question and was coming in to make sure someone has said IGBTs, thanks kevin :) We drive much large loads then this where I work and IGBTs are very dependable. \$\endgroup\$ – Kortuk Sep 27 '12 at 21:57
-1
\$\begingroup\$

3Kwatts! Consider how hot a 300 watt light bulb will get and multiply by ten. 3000 watts is equivalent of the heating element in your electric oven.

But admittedly much of that heat is dumped in the motor and not the h-bridge. Also, is that 3kwatts at motor stall or motor no load? For any h-bridge to survive, it has to be engineered to handle the current load of a stalled motor that may be about 5 times higher than the no load current. Of course, fuses do help as protection, but start up is a stalled condition and you have to provide enough current under load to make the motor move. Fuses have to allow the motor to start and may not eliminate all the stress

Then look at IC packages and you will generally notice that they dissipate about 1 watt. A TO-220 package with heat sink might handle about 50 watts.

In order for an IC to work and work well, the circuit would have to be about .9997 or better in efficiency. It is never going to happen.

With a BJT, the best you are going to do in saturation mode is about 0.3v of voltage drop and that is where your heat in the transistor is created. With Darlingtons, the best you are going to do is about 1.0v of voltage drop - so they are even worse about heat. If you know your operating voltage (lets say 36VDC) you can figure your current. 3000watts/36volts = 83 amps.

There are transistors that will handle 36 volts and 83 amps, but how much heat with the .3v drop produce? 0.3v x 83 = 24.9 watts of heat.

MOSfets come with a different number. The ON resistance and the best that you can do is about .05ohms of on resistance. Also, MOSfets can be wired in parallel, so two 50 watt packages can be used if you need to handle 100 watts.

MOSfets run the coolest, but all create some heat in managing as switches and all have limits to how much heat their packaging will tolerate before failure. Silicon generally fails at about 150 degrees centigrade.

If you want power you just cannot get away from heat and it is the heat that is destructive.

But I doubt that MOSfets are going to handle your frequency requirements.

\$\endgroup\$
  • \$\begingroup\$ I think Kevin really nailed this question, the issue is not your IC dissipating 3kW, your IC just has to switch that much power. IGBTs are very good at this and I have proof of it, I will admit though when ours fail it is quite spectacular, I consider that a feature though. \$\endgroup\$ – Kortuk Sep 27 '12 at 21:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.