H bridge

[connect21]

I am trying to work with an "H" bridge,(I think that's how it's called) and I realise that I need more information. I am using four identical IGBT to make an "H" bride to control the primary of a transformer I am planning to use to make a TIG welding machine. The first problem I ran into is the same problem I had before with bipolar transistors, I have the two top IGBT, one left and one right, with their collector connected to the positive, their emitter connected to the collector of the two lower IGBT and the lower IGBT have their emitter connected to the negative side. If I use the same value resistor for the four gate (or base), the two top ones ( connected to the positive) don't want to saturate, I have tried using lower value resistors (more current) to try to saturate the two top IGBT but limited success. I am afraid to use resistor too small and fry the transistor. What am I doing wrong?<p> Thanks

[philba]

I have no experience with IGBTs so I can't help you with that. <p>In your application, (TIG welder), you should be seeing back emf from the transformer? do you have protection diodes? I don't know how this effects the IGBT. If you post schematic, you'll probably get more responses. (you can make a gif of it, put it on a website and use the image feature.)<p>On a different track - Are you planning to use PWM to control the hbridge? If so, I think that IGBTs might not be the best approach. From what I've seen most hbridges that carry significant current use MOSFETs. They are pretty easy to put into a full on state, especially the logic ones.<p>Phil

[Chris Smith]

From memory,.... IGBTs don’t use any current on their gates either. They use voltage present like a FET?<p> If this is correct, you need to terminate the feed to the gate to ground or positive depending on the gate polarity and voltage needed, through a voltage divider, and select a center tap voltage that will allow saturation because of the proper voltage. <p>What happens is the gate can become saturated or not saturated, and then the opposite wont happen such as turn off after saturation. <p>Often a diode to positive or ground is used to bleed off the “on value” once the on signal is removed.<p> This is caused by a build up of capacitance at the gate, and that means the gate cant go in the opposite direction, or go fast enough to keep up. <p>The gate voltage is best utilized when it is forced to its turn on value, then slammed, swung, or forced in the opposite ditrection or value. <p>In some cases they wont operate unless a negative value is added in a flip flop fashion to shut off the device after it is turned on. <p>What device #s are you using? Diagram also would be nice. <p>Also they make IGBT drivers because of these difficulties in design. <p>You can find them at the lighting companies that produce fluro light parts from IGBTs and the makers of the IGBts.

[connect21]

IGBTs are something new to me too. Although they have been around for over 20 years, there doesn't seem to be a lot of people that are familiar with them. IGBT goes for Insulated Gate Bipolar Transistor, they combine the best of both world, they can drive heavy current like bipolar transistor and are as easy to drive as a FET or MOSFET and they are relatively cheap compare to FET. Since I am not very familiar with FET either, I am not sure how to drive them. The IGBTs I am using, for now, are IRG4PC50KD by International Rectifier. I don't know how to post a diagram but if somebody can explain I could post it. I can convert my diagram in .pdf format. If not, I can E-mail it to whoever might be interested to look at it.
Here are my questions
1-FET, MOSFET and IGBTs are supposed to be voltage driven, but does that mean that you don't need to use a current limiting resistor to drive the gate?
2-To make my tests, I am feeding the power circuit at 24Vdc but later on, the plan is to feed it with 120Vdc. Right now I am applying 12V through a 100 ohms resistor to the gate but IGBTs Q1 and Q3 don't saturate and the voltage drop between the collector and the emitter is 12V, while Q2 and Q4 seem to saturate and the voltage between their emitter and collector is .8V Would it be safe to remove the current limiting resistor of 100 ohms?
3-I was told that a pull down resistor (instead of a diode) would be suitable to drive the gate to 0V once it doesn't need to be driven. I have also read that it would be a good thing to put a zener diode to make sure that the voltage driving the gate doesn't exede the maximum voltage allowed.
4-The IGBT that I am using have a diode between their emitter and their collector but I was reading that might need a "snubber" circuit because the diode might not be enough.<p> Thanks<p>[ February 02, 2005: Message edited by: connect21 ]<p>[ February 03, 2005: Message edited by: connect21 ]</p>

[upsmaster]

hi<p> some clues for you, we use igbt's in H bridges for dc to ac inverters. our inversion circuits are 48,120,240,360 volts dc input with single phase outputs at 120,120/240,208 and three phase 208,480 volts ac. ranges from 10 to 300 kva. manufactures spec books/data sheets will show a series resister with the gate and what value they spec. most of the time they show the diagram and a graph so you can see the effects of changing the series value. In our use this resistor will control the switching speed of the device, we also use isolated power supplies to drive each set of gates. We drive with pwm and gate voltages are 30 volts p/p with 15 volt hold off when not being gated, yes we use 18 volt zeners 2 across each gate. the drive circuit must apply a recovery voltage to shut the device off, a chip will ease this design in lew of discrete parts. hope this gives you some direction.
joe

[connect21]

Hi Joe<p>I have a few questions for you;
- What do you mean by "15 Volts hold off when not being gated".
- Are there chips specialy made to drive igbt's? Right now, I am driving mine through some cmos nand gates and I have a power supply just to drive the gates of the igbt's
- Do you use or need a protection circuitry for the collector and emitter side of the igbt's?
- What kind of igbt's are you using for these disigns you are talking about?
- In your cicuit that works at, lets say at 240Vdc on the input side, is it regulate and if so, how is it done?<p>Thanks<p>[ February 03, 2005: Message edited by: connect21 ]</p>

[Chris Smith]

A quick one for now...International Rectifier has many web pages on the design and principal of the IGBTS and drivers. Last year I was doing research on them and I remember the name IR.<p> The Resistor is not a current dropping resistor, its usually a voltage divider. <p>Like I mentioned they usually require a voltage present to turn on, and a negative value to slam them off. <p>IR sells “drivers” for this purpose because often a negative voltage is required to turn them off. <p>Drop me a Private message and Ill email you direct on the subject.

[L. Daniel Rosa]

It sounds like IGBTs give you the worst of both worlds- turnoff drive requirements and saturation characteristics. Just the same, if I had some easily (and cheaply) available I'd tinker with them.<p>Your upper transistors not saturating is a result of using a totem pole output without a higher voltage supply for the gate drivers. To get saturation in this case you'll need a dedicated supply that is always higher than the main supply voltage to the transformer/H-bridge.<p>Another way involves a complementary output H-bridge- PNP BJTs or IGBTs, or P-channel MOSFETS for the upper two transistors with the appropriate gate/base drive (negative WRT gate/base).<p>Do not omit the base/gate resistor- you will regret it.

[connect21]

Mr. Rosa<p> Since the gate represents a capacitive load on it's supply, the value of it's resistor will affect the speed at which the gate will turn on.
The way I see it, it is risky to have a separate power supply with a higher voltage with the lower transistor that may have it's voltage, between the collector and emitter, varying it will change the value of the reference voltage, the emitter's voltage, for the top transistor and therefore the voltage at the gate of the top transistor will fluctuate. I would need a stable gate voltage in reference to it's emitter. They don't seem to make a lot of complementary IGBT's (in the PNP polarity) so I will have to figure a way to make the the drive circuit for the top igbt's. <p>Thanks

[upsmaster]

hi
We keep a steady -15 volts on the gate when inversion is halted..also when we start a inverter we turn on a rectifier first, this gives us our electronics through a dc/dc converter supply which puts the -15 volts on the gate this keeps them off. The rectifier also charges up all of the dc caps in the inverter circuit, sometimes 10s of thousands of mics, so the rectifier soft starts to get to this point. Then we give the electronics a signal to start and this takes this -15 volts and starts a square wave stream from the +15/-15 power supply, 30 volts p/p. Our bridge circuits use all n type igbt's. Some of the igbt's are single pack mostly 150 amps and above 200,300,400 amp types with 600 to 1200 volt ratings. Some are dual packs under 150 amps 100,50,75,125. My experience is when the gate is not clamped when you apply the dc volts to base collector you damage the gate, in our circuit we keep a 10k resistor across the gat at all times. Yes their are driver chips,, we don't use them in all of our circuits just 2 right now. The dc side is a regulate power supply which is electronicly current limited. To protect the igbt's we sample the collector emitter voltage because it is almost linear to the current flow through the junction, this voltage is compared to a set reference voltage that repersents a certain current in amps which triggers a shut off circuit when exceeded.<p> hope this helps anything i can help with let me know
joe

[upsmaster]

Hi
oh yeah you must have a dead band between turning off one set of igbts and turning on the other in a h bridge, this is so the igbt is of our circuit has a 2.5 mircosecond dead band.<p> joe

[connect21]

Hi Joe <p>Of course, I have a few questions;<p>1- Does the igbt's need to be fed with -15 Volts to be turned off? 0 Volt is not enough? Does it need to swing from -15 Volts to +15 Volts? 0 Volt to 15 Volts wouldn't work?<p>2- I don't understand what you mean by turning on a rectifier and are these caps connected on the output of the igbt's?<p>3- I don't know what a clamp is. Also you say "when the gate is not clamped when you apply the dc volts to base collector", what do you mean by that.<p>4- Who makes such high powered igbt's?<p>5- At what frequencies does your circuits operate at?<p>6- My #1 problem is how to drive the two top igbt's, I have read sevral application notes, on Intrenational Rectifiers website, but I haven't found any simple solution. Any suggestion where I might look?<p>Thanks again
Marc

[josmith]

The problem with using bipolar transtors on an "h" bridge is that the top transistors are operating in a different mode than the bottom ones. The bottom transistors are in current amplier mode and need only .7 volts to start conducting. The amount of collector to emitter current available depends on the current gain of the transtor. The top transistors are voltage followers so your emitter voltage will be the base voltage -.7.
I never worked with IGBT but if they work the same as regular bipolars the problem you are having may be the same.

[connect21]

I had the same problem with bipolar transistor and I asume it was for the same reasons.

[upsmaster]

Hi

4) Toshbia<p> 5) 8,000 Hz to about 30,000 Hz<p> 6) Toshbia...their is no easy way you must
meet the device specs.<p> 3) Clamp a device across the gate of the
igbt to keep the gate in an off state
when no gate signal is present.<p> 2) Rectifier/DC power source.. the caps are
across the dc bus.<p> 1) igbts are east to turn on, we have some
circuits with igbts in them and as a
bench test we apply a dc voltage to the
gate to see if it will turn on. Some-
times the device is already on and
removeing the gate signal will not turn
the device off, then we short the gate
with a jumper wire and then do the
test again then it works. So yes a
negative voltage makes sure that the
devices are off before you apply
real power to be switched... you don't
want to short your dc supply out
through igbts that are on when they
should be off.