Selection criteria for freewheeling diodes for IGBTs in DRSSTC bridge / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Selection criteria for freewheeling diodes for IGBTs in DRSSTC bridge

Move Thread

LAN_403

Herr Zapp

Thu Sept 03 2009, 07:27PM

Registered Member #480 Joined: Thu Jul 06 2006, 07:08PM
Location: North America
Posts: 644

Forum members -

From time to time interesting mid-power-level IGBTs show up on the surplus market that DO NOT incorporate an internal freewheeling diode. Can anyone provide some criteria for selecting appropriate external freewheeling diodes?

Specifically, reverse recovery time, peak voltage, peak current ratings, and any other factors that might be important for use in a DRSSTC full-bridge circuit.

For example, the Fairchild HGTG20N60A4 IGBT can sometimes be found for <$2.00. It is rated at 600V, 70A continuous, 280A pulse, with a 55ns fall time, but does not incorporate an internal freewheeling diode.

All inputs are appreciated.

Regards,
Herr Zapp

GeordieBoy

Thu Sept 03 2009, 09:39PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Choose the free-wheel diode specs to match those of the active switches. i.e. The diode specs have to complement the switch specs. Diode Peak Inverse Voltage should equal or exceed the IGBT device's maximum forward blocking voltage. Diode peak current handling should match or exceed IGBT peak current handling. (Even with accurately synchronised zero-current switching the free-wheel diodes have to handle the full primary tank current at the end of RF bursts, or if the over-current detection trips before the end of the burst.)

Average current handling of the diodes can be less than that of the IGBT since hopefully current will circulate in the free-wheel diodes for a smaller fraction of each switching cycle than it spends flowing through the IGBTs.

The thing that really heats up free-wheel diodes in bridge appilcations is driving a reactive load just slightly off resonance. This causes a moderately large current to flow in the free-wheel diodes for a significant fraction of each switching period.

One of the most important specifications of free-wheel diodes is the reverse recovery time when used in bridge applications. The reverse recovery time at the rated current should be comparable with the rate at which you turn on the IGBTs. In other words don't try to turn on the IGBT in 50ns, if the Trr of the free-wheel diodes is 200ns. Otherwise the diode will take a long time to come out of conduction. This causes a huge current to surge through the IGBT that is turning on and through the reverse biased diode that is trying it's best to turn off in time! This is usually accompanied by horrendous ringing, voltage overshoots, hot devices, loud bangs, flames and tears

-Richie,

Herr Zapp

Fri Sept 04 2009, 12:36AM

Registered Member #480 Joined: Thu Jul 06 2006, 07:08PM
Location: North America
Posts: 644

Richie -

Thank you, that is exactly the type of information I was hoping for.

Aside from the actual reverse recovery time, is the "softness" of the reverse recovery an important characteristic? Or is this a moot point, as ALL candidate devices with the required combination of Vrrm, Ifrm, and Trr will have appropriate "soft-recovery" characteristics?

Or are the diode's "soft recovery" characteristics of no real importance in a DRRSTC bridge (ignoring potential EMC issues).

Regards,
Herr Zapp

Steve Conner

Fri Sept 04 2009, 10:13AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

The faster and softer the recovery, the better. The recovery transients basically up-convert energy to higher frequencies. Step recovery diodes are used for this purpose: I've seen radio hams build devices that triple the RF output from a 432MHz transmitter to 1296, with a big parallel bank of 1N4148s. But you want to minimize this conversion, as the higher frequencies are harder to bypass, and easier to couple into unwanted places.

I once had a situation where the diode recovery transients produced so much EMI that it coupled into the control circuitry and made it malfunction. This was with the co-packaged diodes in CM600 bricks. Richie may remember this, as he was at the controls of the OLTC2 while I voided my life insurance poking the coil with an oscilloscope

This was an extreme situation where the diodes were required to conduct the negative half-cycles of primary current, about 2000A peak. In a well-tuned DRSSTC, they'll hardly conduct at all, so there will be far less recovery energy to worry about, at least until the "backwash" at the end of the burst. By the same token, you may be able to get away with smaller diodes, as the average current in them will be much less than what the IGBTs have to cope with. Although as Richie states, the peak current could potentially be the same.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob