IGBTs

[NiHaoMike]

Looking at the Fairchild web site, there seems to be many different kinds of IGBTs. At the least, there's PDP, NPT, PT, Field Stop, and Trench IGBTs. The PDP IGBTs seem to be lower voltage but also lower cost.

Which kind would be best for a 120V inverter with a maximum DC bus voltage of 210V and a PWM frequency of 15kHz or so?

[mkissin]

That's quite a low bus voltage, and quite a fast switching frequency. Unless you have a good reason to select an IGBT, I would simply use a MOSFET.

Generally speaking (and this is a fairly gross generalisation with all the new devices appearing) IGBTs are slow switching devices that are rated for high voltages and currents. They tend to have high switching losses and low conduction losses, relative to MOSFETS. The low conduction losses are due to them having a saturation voltage, rather than a drain-source resiatance so your loss is VI, rather than I^2R.

[NiHaoMike]

One of the reasons why I'm looking at IGBTs is because IGBTs generally have faster inverse (freewheeling) diodes than MOSFETs. With MOSFETs, it would be necessary to use external inverse diodes and design the circuit to prevent the slow internal diodes from operating. I'm also looking at making the design easily modifiable for 240V by simply using higher voltage components and changing the value of some voltage sense resistors.

IGBTs shouldn't have any problems at 15kHz or so. I can further reduce the PWM frequency if that would make the circuit more efficient.

[AdShea]

IGBT's don't have any intrinsic diode.  Most IGBT packages have a co-packaged fast recovery diode (FRED) (you should check to be sure that there is a packaged diode if you're depending on one being there).  You can get co-pack MOSFETS as well, or use an external diode.

[Time]

If you were switching some large amount of inductance dont you need still need a free wheeling diode across the load in addition to one for the switch?

[Zero999]

If you were switching some large amount of inductance dont you need still need a free wheeling diode across the load in addition to one for the switch?

It depends on the circuit.

An inverter consists of an H-bridge so diodes are required across the switches to shunt the back EMF into the power supply.

[mkissin]

IGBTs shouldn't have any problems at 15kHz or so. I can further reduce the PWM frequency if that would make the circuit more efficient.

I didn't say they'd have a problem, simply that a MOSFET may be more efficient. There are always exceptions of course, and they may depend on the price you're willing to pay for device.

As for your original question...for a basic application the specific device probably makes no difference. You'll just need to make sure that the device ratings are acceptable for your application, and then make a quick estimate of the device losses from your simulation data.

Also, as was stated earlier, make very sure that the exact IGBT device you buy actually has a diode in the package. Some datasheets cover two devices, one with a diode and one without, and the difference in part number is just a suffix.

[NiHaoMike]

So far, I'm planning to use Fairchild HGTG7N60A4D_Q IGBTs. ( http://www.mouser.com/ProductDetail/Fairchild-Semiconductor/HGTG7N60A4D_Q/?qs=sGAEpiMZZMtvcUztdGSumOTLm6TEfBtEPq2kPQy%2f0sE%3d ) At $1.45 each in small quantities, I could not find a cheaper IGBT or MOSFET with similar ratings, plus it has the inverse diodes built in. The 600V rating should give a lot of margin for voltage spikes and the 34A rating is a lot more than the 5A I need. (I'm hoping for 10-15A.)

[mkissin]

I can't find that specific part on the Fairchild website, but the HGTG family seems to be pretty good in general. The part will likely be fine.

[Zero999]

LMGTFY

[alm]

LMGTFY

Surely the topic starter should go through the trouble of tracking down a datasheet, not someone volunteering their own free time to help someone? What's the point of annoying the latter?

[NiHaoMike]

I have already looked at the datasheet to verify that it does have the inverse diode.

[Zero999]

I don't see why anyone should be annoyed at me. Google should always be the first place one checks.

[alm]

LMGTFY links don't serve to inform people, linking to a datasheet with the note 'found it with a simple Google search' would accomplish that much better, but to be a smart-ass about it. Look noob, this is how you do it! This may be OK if someone is asking other people for help, since you expect them to put some effort into it. But I don't see the point in this case.

Even a Google search doesn't show any datasheets on the Fairchild website, which suggests that they may be discontinued. I would avoid using discontinued components in new designs.

[Zero999]

Just Google the Fairchild site.
http://www.fairchildsemi.com/ds/HG%2FHGTP7N60A4.pdf

[alm]

That's for a different part, that one is without internal diode.

[mkissin]

OK, so that was really annoying Hero999. It was condescending and unhelpful.

What I was trying to get across when I said that I couldn't find the datasheet, was that a search of the Fairchild web site didn't turn up any hits. Neither did using their product selector. In fact, that part that the OP pointed out no longer even follows the Fairchild IGBT naming scheme on a general level.

Yes, there may be a datasheet that comes up in google. It may even come from the Fairchild website. I can probably find a datasheet for an 8086 on the Intel website if I tried hard enough. Using google as your datasheet finder will often fool you into using discontinued parts in a design. For some parts, like a TO-247 package IGBT where the pinout is standard, this may not matter. For others, it may be critical. Always check with the manufacturer.