# Safely driving a high power h bridge with optocouplers

I was looking into some high power H bridges (I was hoping to make an induction heater or at least a high voltage power supply). Some I came across are: this, this and this, to name a few.

I noticed this related question about driving a mosfet with an optoisolator. However, I may wind up using IGBTs or SCRs, which brings me to the next related question about how to use IGBTs as switches. Based on the answers to those questions, I should be able to use 8 optocouplers (one on either side of of the gate of each IGBT/SCR) to drive one or more of these devices. I already have plenty of optoisolators with some 5000V isolation voltage.

I was wondering if I could safely use those to drive one or more of the above devices, or possibly a general class of devices.

I could infer an answer and I suspect it to be yes, but I wanted to be very sure of how to use one of these devices before I shell out the money or tinker with the voltages they can put out. If the answer is no, what would recommend using to drive these devices?

• Bus voltage range = ??? 50V? 200V? 1000V? Apr 25 '12 at 2:11
• I suspect somewhere around 500V, but I am trying to get parts that work together more than anything. It all depends on the parts I can find that fit together. I do not have experience in this area and I am trying to learn as I go. Apr 25 '12 at 2:15
• OK. A 600V IGBT is going to work well up to about 400-425V (you need to allow for overshoot transients -- get some good high-voltage film capacitors and place them right across the half-bridge so you keep the loop area small); a 1200V IGBT is good for twice that. Apr 25 '12 at 2:20
• BY THEMSELVES optos are not a good choice for high speed H bridge use. At slow speeds where the switching tin=me is not crucial the low current capability of the opto that leads to low gate switching times can be OK. Once you start using serious PWM you may need 100's of mA of gate drive and maybe 1A+ in order to transfer gate charge rapidly and few optos will suit. Apr 25 '12 at 4:35
• You need to walk before you try to run. Based on your other questions which help give an overall picture of what you are trying to do, you have every chance of destroying a lot of expensive equipment if you try to design high power high voltage drivers. You need to obtain a feel for real world gate driving. IGBT looks like a MOSFET for drive purposes. SCR's are either ancient history or black belt master stuff. Apr 25 '12 at 4:36

Bridge parts selection. Heard a lot about packaged bridges lately. I have never used a packaged bridge. Always discrete fets and diodes. The diode augments the fet body diode. Once the voltage goes above 1200 Vpk then it's IGBTs. Discrete approach gives more control over efficiency, deratings, gate drive charge, etc.

Driving the gate with optos. Well I guess that's ok if there is something to limit the gate voltage between 0 and 15 volts. My experience is with isolated flyback supplies and electronic drive for the gates. I guess the opto drive would be ok to about 10 kHz-ish if the switching timing didn't have to be precise and the gate charge wasn't too big and the load was small and efficiency didn't matter. Check fet datasheet for "total gate charge" or something like. Also see above note about picking your own fets. A 10k pull up and 10 nC total gate charge to 10 V means 30us-ish switching time, give-or-take for gate drive nonlinearities, limiting you to sub 30kHz switching frequencies or smoking the fets in a somewhat analog mode. ie. slow fet transitions from off-on burn power in the fet.

There are lots of full-bridge fet drivers on digikey that may meet your gate drive needs.

Perhaps you would like to post separate questions for 1. bridge component selection, 2. fet drive dynamics, 3. HV isolated gate supplies, and any of the other issues that come up making HV bridge drive.

• Any good designer must learn to write a good spec before they can achieve a good design and then test against that spec. What is your spec? Apr 25 '12 at 0:21
• I am flexible at this point because this is my first HV experiment. I figure ~30 amps at about 600V should be safe, but more current would probably be better, and higher voltage ratings would probably be safer (for the chips that is.) Apr 25 '12 at 0:27
• Also, I already know how I want to determine when to switch it. I have a pair of wires with ~30V ac (frequency controllable), and I simply want the h bridge to switch depending on which way the voltage drops. Apr 25 '12 at 0:43
• Wait...are you sure you don't want a bridge rectifier?
– W5VO
Apr 25 '12 at 8:53

A few odds + ends:

• MOSFETs vs. IGBTs: If you are switching from a DC link voltage of < 200V, MOSFETs will probably win; if > 200V then an IGBT will probably win -- where "win" here = better performance for cost.

• Driving MOSFETs and IGBTs -- look into gate driver ICs (MOSFET and IGBT drivers are very similar). Avago makes some decent optocouplers.

The key thing here, as Tony Stewart has pointed out in, is a spec:

• What voltage range do you have available? (both power and control voltage)
• What current are you trying to switch?
• What frequency are you trying to switch at?
• Do you need to achieve 100% duty cycle on any of the switches? (many gate drive ICs assume a charge pump power supply which precludes the use of fully-on switches otherwise the charge pump won't get replenished)
• Temperature range?

IMHO, get yourself a decent power electronics book and start reading. (Kassakian/Schlecht/Verghese or one of Ned Mohan's books) This can be a difficult and/or hazardous field of electronics to get into if you're not careful.

edit -- they're not cheap, but I'm pretty sure you can buy H-bridge IGBT modules from Semikron that include gate drive circuits, so you just put a film capacitor on the DC link next to the module, and feed in logic level drive signals and you're good to go.

• Let me know if you find one of those Semikron modules. If <$70 I will consider buying one, if <50 I will probably buy it. Apr 25 '12 at 2:27 • Depending on what I would be spending for a home made driver, I might consider shelling out$100 but I cannot imagine building a simple driver would cost too much. Apr 25 '12 at 2:35