Why are IGBTs used instead of SCRs in VFDs
Simple question. Just seems SCRs have much more ability and cheaper but I commonly just see the IGBT in VFDs. What am I missing? Has to be a reason for it.
You can turn an IGBT on and off at will, like most transistors. An SCR or Triac will not shut off
until the current flowing through them drops below some threshold. I remember some type of hybrid about 20 years that used a mosfet or darlington that would shunt the current on an SCR to lower the current, but I don't think they were very successful.
I guess I thought there were some creative ways to turn them off in a VFD application. Maybe I am thinking of some old design before the advent of the IGBT. Who knows.
You can "commutate them off", by firing another SCR that "steals current" through a capacitor, That shuts the 1st SCR off. The second shuts off when the capacitor is charged.
SCRs are still used in low frequency inverters of ridiculously high power and voltage, as with multi megawatt electric utility applications, because they are rugged and high voltage capable.
But SCRs are slow, slower than even high voltage high current IGBTs, which still tend to be slowish. Slow means high losses, or else a low frequency limit usually well under 1 kHz.
For most purposes, higher frequency switching is desirable, to reduce the size of the magnetic components.
guess that is why SCRs are the common grab for all the GPUs (ground power units) we used to work on... Cute little pancake critters but GD could they handle some power.
IGBT can be turned on and off. SCRs canīt without tricks. There are special SCRs that can be turned on and off, called GTOs but thatīs another breed and far more expensive.
Another nice feature of a IGBT (insulated gate bipolar transistor) is that it needs nearly no drive current to switch on and off. You could drive it directly from a microcontroller or an optocoupler. This makes interfacing far easier.
A bad thing with IGBTs is that they donīt like inductive loads and are hesitant to switch off those but there are generation three and generation four devices now that are bred to do exactly that very well. Those can be found in automotive applications, like switches for high voltage/high current ignition coils.
I thought the slow turn off was actually a good thing and the switching actually "rolls" off to get a more sinusoidal wave form instead of square? Do I have that wrong?
I guess I could expand to get get a few questions answered on the IGBT???
1. When looking at IGBTs, there are really only 2 packages, TO style, and bolt down modules. Do the modules include multiple IGBTs on them in parallel to carry more current? Seems all the higher rated IGBTs are module style.
2. In looking at the specs for IGBTs (really even mosfets too), there is usually a spec for continuous current but rarely peak or transient current to cover inrush. I know the IGBT is well known for tolerance in this area but how does one determine peak values over time without failure testing it?
3. Looking at the TO style IGBTs, the current rating is usually just scary high when compared to the package size and leads. Obviously heat sinking is paramount. How the heck does such a small (.040" wide) lead survive all that current? I remember replacing mosfets in the past that are high current and they seem to make it but man, it woule sure seem they would need a more robust build and leads...
Older "Current Source Inverters" did indeed use SCRs, or GTOs actually, and they were hell-for-stout drives because the energy it took to destroy them was usually higher than what it took to destroy the motor. But in motor applications they had to be impedance matched to the motor. That still carries over to this day in a lot of specifications written for larger VFDs where they say the motor and VFD must be designed to operate together. It's no longer true, but engineering specifications change slowly in a lot of places.
SCRs are also much more prone to "self commutation" (turning themselves on) when used in inductive circuits, so as the switching speed goes up, the likelihood of failure goes up exponentially. CSI drives therefore did not have high speed switching rates, which while good for the motor meant a lot more losses (heat rejection) in the components and a lot more heat sink. This was especially true for GTOs, which may need upwards of 20% of the cathode current to force them off!
I once commissioned a 10HP Robicon CSI drive on a pump station about 15 years ago. The CSI drives were the preferred choice in a lot of muni applications because if the pump station flooded, the VFD could be disassembled, the power stacks hosed off and dried, the PC boards run through a dishwasher, and they would return to service! With a modern PWM drive, the entire thing is toast; there is no rescue procedure. But that 10HP VFD was the size of a small refrigerator!
If you are talking about PWM a slow transition will generate more power loss in the switch
Originally Posted by viper
device. A bad thing.
Your theory about "sinusoidal wave form instead of square". Hmm. You ought to look at a
Pulse Width Modulation for Power Converters, IEEE Press Series.
There is, it is the SOA diagram. You gain safety margin by using running a 72A@25C device at say 15A so when the machine is working hard on a hot day and the heatsink is at 80C there is still some reserve capacity. The peak values are determined by spice modelling.
Originally Posted by viper
That scary high current is at 25C heatsink temperature, nobody would run a device in production at those currents. The leads carry the current while getting very hot but their short length means that they can conduct heat to the circuit board and device package.
3. Looking at the TO style IGBTs, the current rating is usually just scary high when compared to the package size and leads. Obviously heat sinking is paramount. How the heck does such a small (.040" wide) lead survive all that current? I remember replacing mosfets in the past that are high current and they seem to make it but man, it would sure seem they would need a more robust build and leads...
IGBTs are fine with inductive loads, the commutating diodes take the "kick" to the DC bus.
The "low drive power" is not entirely true..... Average power is lowish, way lower than a regular bipolar transistor. But you NEED a high current to get fast switching and ow losses. The gate is a big capacitor, and a lot of current has to be sourced or sunk to turn the device on or off.
if you have all day because the switch speed is OK being slow (turn on or turn off and leave it for a while), then the low power sources can handle it.
Speed of switching makes EMI (interference hash) and speed can be traded off against EMI by adjusting the gate resistors..... if you can stand some extra losses.
yes, derating...... I don't know why they bother with some ratings.... for initially sorting out devices, all I am really interested in is the ratings at 125C or 150C.... they are usually the limiting case, and worst case for switching losses etc. After the sorting, the other stuff is of interest too.
As for surge current ratings.... the toughest devices are usually diodes and SCRs... then Mosfets, then IGBTS, and finally bipolar transistors and Triacs.
IGBTs are bipolar transistors with mosfet gates on them.... they still have the issues that the faster mosfets don't, limits on SOA (safe operating area), slower switching, "tail current" issues. They are still a "minority carrier device" while a mosfet is a "majority carrier device", so they have what amounts to recombination times, etc. And they are usually much higher in effective "on resistance", or "saturation voltage".
As for the choice of IGBT's over say mosfet's
The IGBT's forward voltage is almost independent of reverse blocking voltage for any given bandgap but a FET's on resistance and consequently forward voltage scales with voltage as does the die area for a given on resistance. Same for the gate capacitance. This means that IGBT's are smaller with lower losses and less drive requirements at higher voltages than mosfets. IGBT's are slower to turn off.
An IGBT is a bipolar transistor with an integral mosfet base driver, they have replaced the older darlington and triplington transistors. They are majority carrier devices like SCR's and they are driven into saturation like SCR's making both of them slow to turn off.
That would be "MINORITY CARRIER" device................
Originally Posted by HelicalCut
Dishwashers and flooded inverters...
Hee hee... JRAEF's comments brought a fond memory to me... while teaching GTO-based traction inverters at a major transit authority, flooded car recovery procedures was one of the training segments.
My students were aghast, and even somewhat offended when I dragged a dishwasher and garden-hose into the shop for that training day.
I still like using the dishwasher and tomato sauce... for cleaning up old tube-type radio transmitters and receivers... works great, and gets those variable bread-slicer caps CLEAN. Makes the wife mad, though... so does warming up pistons in the oven... she says it makes the 'food taste funny'... I never noticed... and the chicken-grease helps those wrist pins slide in nice. Ah, the aroma of a rebuilt engine on startup always makes me hungry... and my buddies say "you runnin' that on Biodiesel?"
Yeah, the dishwasher! When I tell colleagues about they always look at me as if Iīm crazy. But thereīs no better way to degrease parts.
Oop's My mistake.
Originally Posted by JST
"...An IGBT is a bipolar transistor with an integral mosfet base driver,..."
OK, I never knew that. Thank you.
An aside, seeing as it's topical: IF you ever wanted to heat a motorcycle
gearbox which has had 90wt oil in it, to get it hot enough to disassemble,
by plunking it into your kitchen oven, there's a trick. It's called a
Reynolds Roast-in Bag. Basically a high temperature plastic bag - and I
think they are polyimid - which contains the majority of the stink and
tends to keep marital discord to a bare minimum....
That one goes down in the book right now. Great idea, should work for cylinder heads and valve guides, too.
It's called a Reynolds Roast-in Bag
Must be referring to Guzzies... Most Harley gearboxes I had... emptied themselves naturally... just sitting on the sidestand ;-)