How isolated do IGBT optical drives have to be?

[TheBrick]

I've a project that involves IBGT H bridge and in researching driving them I watched this video ( ) on driving IGBTs it suggests that the driver power supplies need to be isolated @ 9:22. My question is how isolated is isolated?

If driven from mains (as would be normal for a IBGT application) does that separate transformers for each driver, then separate, rectifiers and separate voltage regulation regulation or just the separate voltage regulation or something else?

[krish2487]

It means, that if you need to drive a IGBT H bridge, you would need 4 individual, isolated power supplies with their own rectifiers, filters and regulators. Semikron has a bunch of really good app notes that deal with the finer details of IGBT drive and its associated nuances.

PS : A suggestion, when you are working on the power supplies, I d suggest you work with a bipolar supply, generally a +15/-8 or +15/-12V supply. Will help avoid of potential pitfalls.

[T3sl4co1l]

Depends, of course.  What are you switching?

Typical examples:

- 120/240V power supply, DC output, up to 1-2kW.  The inverter is full bridge, with the switching controller and gate drivers on the line side.  The only isolation required is an optoisolator to provide feedback and regulation.  In this case, the gate drives could easily be monolithic high/low side bootstrap driver chips (IR2110, etc.).

- Same thing, but with switching controller on the output (isolated or grounded side): auxiliary isolated power is required to start the controller, and gate drive must be isolated, for example, using a gate drive transformer.  The isolation required of these two isolation barriers depends on the output ground: if the output is grounded to safety ground, 2kV Functional isolation is sufficient.  If the output is not grounded, and will not be connected to SELV (safe extremely low voltage -- safe to touch), Basic isolation is sufficient.  (It can be connected to SELV through another Functional isolation barrier of the same voltage rating -- two layers counts as Reinforced.)  If it connects to SELV directly, the isolation must be Reinforced.

- High power circuitry, like a 480VAC 3ph, 20kW motor drive.  This requires 5kV isolation, but the barrier keywords work the same way.  Now, this type of equipment requires much more drive than a gate drive transformer is capable of, and probably requires DC coupling to produce the correct waveforms.  A GDT would be impractical, so fully isolated gate drive modules are used.  This requires a Functional barrier in the power supply for each isolated driver, and a Reinforced barrier coupling signals from SELV (front panel, or PLC controls, or..) to the drivers.  The latter may use standard optoisolators (cheap, relatively slow), coupling transformers (requires modulation), monolithic magnetic couplers (logic level, moderately priced, fast, and very dV/dt tolerant), or even fiberoptic cable for very high power equipment (usually in the megawatts).

Disclaimer: don't take my word for it, read the standards!  I'm sure I've confused Basic and Functional, or what the algebra of cascaded isolation is, and required voltages, and so on.  This is just to get some flavor of what might be done and why.

Tim

[TheBrick]

@Krish That make sense as to "how far back" the driver power supplies need to be isolated. I've read about duel supplies to help pull down the gate on IGBT quicker. I found these notes from semikron. http://www.semikron.co.uk/skcompub/en/AN-7002_Connection_of_Gate_Drivers_to_IGBT_and_Controller_rev00.pdf

@Tim The very vague plan that may never get anywhere and may just become an academic design exercise is for a AC Tig welder, so constant current. So a full H bridge operating at something of the order 400 hz -> step down transformer -> something of the order 25 V -> rectified -> second full h bridge for generating rectangular AC wave form with different AC balance, pulse width e.t.c. -> current sense to feed back to first H bridge for constant current control -> torch and workpiece. I get the feeling you are on about isolation for the power and safety but I am just initial on about control of the IGBT driver chip and high dv/dt causing problems between from a control point of view. However I'm interested in what you have to say from a safety isolation point of view. Can you give me the reference for the standards you are quoting so I can do some research please. I have some old knowledge of the UK wring regulations and basic and functional isolation ring a bell from that area of application. I'm still very new to electronics as its not my day job or background.

[T3sl4co1l]

@Tim The very vague plan that may never get anywhere and may just become an academic design exercise is for a AC Tig welder, so constant current. So a full H bridge operating at something of the order 400 hz -> step down transformer -> something of the order 25 V -> rectified -> second full h bridge for generating rectangular AC wave form with different AC balance, pulse width e.t.c. -> current sense to feed back to first H bridge for constant current control -> torch and workpiece.

400Hz is ridiculously low for a switching circuit; the cost you'll spend on the custom transformer plus circuitry is probably not much different from an off-the-self buzzbox (CC transformer at 50/60Hz).

Switchers in the 50-100kHz range are the most common, being a compromise between compact size, relative ease of circuit design, and efficient switching (IGBTs are preferable for the supply voltage, but aren't very suitable above 100kHz).

I get the feeling you are on about isolation for the power and safety but I am just initial on about control of the IGBT driver chip and high dv/dt causing problems between from a control point of view.

Then you need only basic (functional) isolation, and enough CMRR (common mode rejection ratio) to tolerate the dV/dt.

Ultimately, you will need basic isolation across the transformer, if the output is grounded to safety ground.  If the output is floating, You will reinforced isolation, because a human will be in contact with the secondary side (the work being welded on... even though the O/C voltage of a TIG torch is hardly SELV, let alone with HF start, if present).

Reinforced isolation is often a good idea electrically, because the devices have lower isolation capacitance and better dV/dt tolerance.  Besides the obvious advantage of greater reliability.  Compare just any dumb DC-DC converter module to a reinforced device, like a RECOM 8kV medical grade part: the price of the latter is about double, but the former might not even survive under industrial use (like gate drivers connected to a 700VDC bus).

However I'm interested in what you have to say from a safety isolation point of view. Can you give me the reference for the standards you are quoting so I can do some research please. I have some old knowledge of the UK wring regulations and basic and functional isolation ring a bell from that area of application. I'm still very new to electronics as its not my day job or background.

These look good:
http://www.ni.com/white-paper/2827/en/
http://www.ti.com/lit/ml/slup227/slup227.pdf
http://www.avagotech.com/docs/AV02-2041EN

Tim

[TheBrick]

"Switchers in the 50-100kHz range are the most common, being a compromise between compact size, relative ease of circuit design, and efficient switching (IGBTs are preferable for the supply voltage, but aren't very suitable above 100kHz)."

That's interesting, an article I was reading on modern inverter tig weld operation suggested the frequency used for the transformer was 400 hz to allow the size of the transformer to be reduced.

[TheBrick]

No I was wrong I misremembered. The line was "20 kHz, ie the frequency is 400 times higher than a traditional 50 Hz transformer" my mistake.