EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: XaviPacheco on June 06, 2018, 02:14:53 pm

Title: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 06, 2018, 02:14:53 pm
Hi,

I have PMDC motor which nameplate says:
VOLTS: 125
R.P.M: 5000
AMPS: 27
H.P.: 4.0
DUTY: CONT

I want to use that motor in a treadmill.  When high load, can that current be much higher than the one in the nameplate?

I want to use PWM @25kHz from a MCU which voltage is 3.3V.  to control this motor. I've seen some MOSFETs and IGBT options. I will get away with an IGBT. I like these ones: AUIRGPS4070D0  and IRGPS46160DPbF. But still not sure. I haven't selected a proper gate driver.  I will appreciate your suggestions and comments if you suggest another transistor and gate driver. I'm relatively a new designer looking for support. Thank you. 
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: Siwastaja on June 06, 2018, 02:49:11 pm
Nameplate current refers to continuous load at rated RPM.

Maximum current, aka. stall current, is when the RPM is zero, due to no back-EMF voltage, which is generated by the rotating magnetic field.

You can think this as a power source you are going to design (ideally, should really be a current source, but due to lack of competence in PWM controller design, voltage sources are often created instead), which drives a voltage source (the motor!) through a series resistor (wires, motor windings, brushes). The motor "voltage source" voltage depends on its RPM only. The DC current is then limited by the series resistance only!

Big and efficient motors try to optimize the series resistance to zero. Hence, they can take almost unlimited current; similar to LEDs, you drive them with current source.

In some cases, where you can guarantee that the motor will spin up just fine and then provide power for a controlled mechanical load, and you have a strong power source (such as a huge battery), you do connect the motor "directly". It jerks, takes an enormous current spike, and runs. Fuse can be used to protect it. But this use case is not very useful if you want to control pretty much anything.

So, we want a speed and/or torque controller. And for that, we need a current mode controller.

Since we don't have all the information, we need to assume, but this is going to be an illustrative example:
Let's assume that 125V, 27A, and 2980W (4HP) ratings all apply at the same time, and the 4HP is mechanical power output. This means the motor is 2980W/(27A*125V) = 88% efficient - which sounds about right for a big motor with continuous duty. If we further assume that 60% of these losses are resistive (rest being magnetization hysteresis aka iron losses and friction losses), this means I^2*R losses of 237W. From this, we'll calculate the resistance of R = P/I^2 = 0.33 ohms. Sounds about right magnitude.

From this, let's apply the 125V voltage to the motor which is not running, the rpm is at 0. The current will be:
I = U/R = 125V/0.33 ohms = 379A (!)

This is the stalled rotor current: if you force the rotor to stall, you get this until the motor windings burn out (a few seconds, maybe tens of seconds). Of course, only portion of this current is generating actual torque: the torque capability is maxed out (with iron completely saturated) way earlier, typically at around 2-4 times the nominal current rating. Which is why you need active current control as well.

Now, luckily, there is some inductance in the motor. Quite a lot of it. It takes milliseconds to reach this current. So, doing PWM on the "block diagram level" is fairly easy. With a big motor, the PWM could run at 2-3 kHz, and even an Arduino analog input could be used to produce the current measurement feedback!

Power stage PCB design and layout is the most difficult part.

But, you definitely need a current feedback. High bandwidth, isolated hall effect module from LEM or similar is often easiest. Or go with ground referenced shunt and an amplifier. But without current limit, you can never know what you are doing. Of course, you could write some code that gradually ramps up the PWM duty to smoothly rise the motor RPM and experiment so that you never exceed sensible currents. But any mechanical issue or any corner case would throw it of and require extra logic. So that's why you simply use current mode control. As an additional bonus, you can use the current setpoint as fairly good torque control directly. In a treadmill, you would most likely want to limit both torque and speed. The former limits sudden jerks very well. Usually, a torque control with maximum speed limit is very useful - similarly to LED power supplies that provide adjustable constant current with an open circuit voltage safety limit.

Re IGBT: for lowest losses, consider MOSFETs. IGBTs are great at supplies well over the 200-300V range, but they have a diode-like "constant" (actually logarithmic) voltage drop, so you are always losing around 2-3V on the IGBT. At 350-400V DC bus and 600V rated IGBTs, this is not a big percentage, but for much lower, you'd benefit from the MOSFETs resistive Rds(on) behavior.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 06, 2018, 02:59:02 pm
I plan to use the LTS 25-NP current sensor from LEM for current feedback. I will also measure speed using reed switch or probably encoder. (Many treadmills have a reed switch). I will consider a MOSFET then. This one seems good, IXFH80N25X3, what do you think? By the way, I really appreciate your effort in writing that long and precise answer.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: jbb on June 06, 2018, 07:29:03 pm
Just before we get into device selection, what’s your power source? 120 V AC rectified into a DC link capacitor? 240V AC?

Also, what experience do you have with mains power and switch mode power supplies?

Finally, if this a one-off project? If so, we may recommend somewhat more expensive parts that are less likely to go bang or try to kill you. (High volume low cost stuff takes a surprising amount of engineering time and one tends to blow up many devices along the way.)

Siwastaja: what do you think of an old-school SCR drive? Too slow? Too crude? Power factor too embarassing?
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 07, 2018, 06:41:07 pm
My power source is 120VAC rectified into a DC link capacitor. I know the basics of mains power and switch mode power supplies.
Yes, that's a one-off project.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 08, 2018, 01:56:08 am
I will get away with a MOSFET. I will use low side switching. PWM signal will be from an STM32 MCU (3.3V) @25 kHz. I've chosen the IRF200P222 , datasheet: https://goo.gl/nt9Nou and this gate driver supplied at 12V https://goo.gl/SQ7sRH

What do you think of that selection? Also, I need to calculate a gate resistor I think. And a RCD snubber network probably.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: JS on June 08, 2018, 04:13:10 am
...
Siwastaja: what do you think of an old-school SCR drive? Too slow? Too crude? Power factor too embarassing?
It wasn't asked to me but this is the internet and I answer anyway.

SCR would only work at mains freq, 60Hz if you are lucky, I'm not. So you have a 120Hz pulses, over an order of magnitude away from the mentioned few kHz. So, to do current control you need to add quite a significant inductance so current stays smooth between the pulses, trying to add that inductance with low resistance is hard, if you go with the same numbers as the motor you will have more losses on the inductor than delivered to the motor.

I guess they are ok for controling the voltage to the DC link if you need to, not so much for regulating the current directly to the load.
@Siwastaja would correct me if I'm wrong.

I will get away with a MOSFET. I will use low side switching. PWM signal will be from an STM32 MCU (3.3V) @25 kHz. I've chosen the IRF200P222 , datasheet: https://goo.gl/nt9Nou and this gate driver supplied at 12V https://goo.gl/SQ7sRH

What do you think of that selection? Also, I need to calculate a gate resistor I think. And a RCD snubber network probably.
  You shouldn't go too high with your PWM as losses on the power devices go up with frequency, you need faster transistors and even then you have higher losses than needed. It's just a matter of using just a high enough frequency so current is smooth enough with the inductance they gave you. Of course if you can reach 25kHz pwm at the desired precision you can get a lower without trouble. I don't know what µC you are using, the usual blue pill runs fine at 72MHz and the counter runs at those speeds, if you just need 2 bit resolution you could have 18MHz, pins are supposed to toggle at those speeds, good luck with the power transistor. You don't even need X bit resolution, you can configure for 1537 steps, so you can control the PWM freq quite precisely and optimize for your resolution, not so much using just prescaler.

JS
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: CopperCone on June 08, 2018, 04:03:41 pm
Do they have scr igbt hybrid circuits in use for whatever reason? Like scr for voltage control and scr for current control?
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 10, 2018, 01:11:57 pm
I'm trying to calculate the gate resistors for the MOSFET that I selected which has a gate resistance of 1.3 ohms. Well, I attach its electrical characteristics (Parameters.PNG). Also, I add the Vgs vs Total charge curve. I would like to find the suitable configuration for this application (See Gate1.PNG an Gate2.PNG). Some of the gate driver characteristics are also attached [GateDriverParameters.PNG].

The gate driver power supply is 12V.
The sink/source capability of the driver is 5A.

So taking into account the worst scenario where the charging current is equal to the peak current of the driver (5A)

RGon = 12/5 - (0.55 + 1.3) = 0.55 ohms
where 0.55 ohms is the Rds(on) or source resistor of the driver and 1.3 ohms is the mosfet gate resistance.

For the discharge current
ROff = 12/5 -(0.7 + 1.3) = 0.4 ohms
where 0.7 ohms is the sink resistance of the driver.

It's supposed that the charging peak current has to be less than driver peak current (5A). So, how do I know the real charging peak current? Once I know it I think I can calculate the power rating of the gate resistor. But I need to know also the on switching time, which is the time that this current wll be flowing.

Is this approach right?
In the Gate2.PNG picture, do I need the diode and the discharging resistor?

I'm really stuck at calculating the right gate driver configuration. I will appreciate your help so much.

Datasheet of the Mosfet: https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d4625b3ca4ec015b3e42ba4a0744 (https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d4625b3ca4ec015b3e42ba4a0744)

Gate driver: https://www.mouser.com/datasheet/2/196/Infineon-2EDN752x-2EDN852x-DS--DS-v02_05-EN-1225984.pdf (https://www.mouser.com/datasheet/2/196/Infineon-2EDN752x-2EDN852x-DS--DS-v02_05-EN-1225984.pdf)

Useful document about gate resistors: https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d462518ffd8501523ee694b74f18 (https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d462518ffd8501523ee694b74f18)
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: jbb on June 10, 2018, 08:14:41 pm
I will get away with a MOSFET. I will use low side switching. PWM signal will be from an STM32 MCU (3.3V) @25 kHz. I've chosen the IRF200P222 , datasheet: https://goo.gl/nt9Nou and this gate driver supplied at 12V https://goo.gl/SQ7sRH

What do you think of that selection? Also, I need to calculate a gate resistor I think. And a RCD snubber network probably.

On the MOSFET: iffy; I think the voltage rating is too low.

Example calculation for switch V rating:

So I suggest a 250 V (or higher) rated part.

The MOSFET you suggested does have quite a good body diode, though.  The descriptions of "Half-bridge and full-bridge topologies" and "Enhanced body diode dv/dt and di/dt Capability" were good.  Do they make a similar 250V part?
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 10, 2018, 08:24:35 pm
They have this one with reduction in the drain current:
https://www.infineon.com/dgdl/Infineon-IRF250P224-DS-v01_00-EN.pdf?fileId=5546d4625a888733015a8bb640137c6e (https://www.infineon.com/dgdl/Infineon-IRF250P224-DS-v01_00-EN.pdf?fileId=5546d4625a888733015a8bb640137c6e)
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: jbb on June 14, 2018, 08:32:05 pm
That might be a good option.  But I realise that I forgot to ask something: are you going to use a single MOSFET and freewheel diode (motor only goes one direction), or a full H Bridge (motor can break and reverse)?

The datasheet mentions 10V gate drive with Rg = 2.7 Ohm.  (And an internal Rg of 1.3 Ohm).  This suggests a 4 Ohm total, which gives 2.5A peak @ 10V.  A very useful trick is to use a higher value on resistor Ron and a lower value off resistor Roff (some gate drivers have 2 pins, otherwise you can use a Schottky diode so Roff only discharges the gate). This allows you to soften the turn on (less current spikes and ringing) without needlessly increasing the turn off losses.  Also, if you go to H Bridge, having Roff < Ron makes shoot-through less likely.

For you safety during development, I would recommend that you isolate your control board from your power stage.

Current sense: LEM sensor - isolated
Gate drive: Silabs Si8233BD-D-IS might be good (with little external 10 or 12V DC converter), or the Silabs SI8281BD (has DCDC controller built in).
DC link voltage sense: could use a high voltage differential amplifier or some kind of isolated amplifier (using DC supply from gate driver).
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 14, 2018, 09:24:44 pm
Yes, I will use a single MOSFET and freewheel diode (motor only goes one direction). I'm thinking of designing an RCD snubber network as well. For the gate driver, I thought of using the FOD3182, which is an optocoupler gate driver. I'm gonna work with relatively low frequencies 1 to 25 kHz. For current sensor, I would use this one which provides galvanic isolation: https://www.mouser.com/ProductDetail/932-MIKROE-1578 (https://www.mouser.com/ProductDetail/932-MIKROE-1578)
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: jbb on June 15, 2018, 12:56:51 am
As you’ll be making a PCB for the power stage anyway, so why not put the TLI4970 straight into it? You don’t want lots of wires in a high power converter: they pick up noise!

I also thoroughly recommend a LEM CKSR 50-NP. Very nice part, and it doesn’t mind switching transients.. But you need an op amp and ADC to go with it.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 15, 2018, 01:01:34 am
The thing is that the package of the TLI4970 is kind of unfamiliar for me. But it actually is a good idea to put it straight to the PCB. I will try it.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 15, 2018, 01:57:44 am
Actually, I just found the footprint file and was able to load it to my PCB tool!
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: Siwastaja on June 15, 2018, 05:26:45 am
Learn to make the footprints yourself and get this process into your comfort zone.

It's very common that you simply cannot find the footprint file, but it's unbelievably even more common that footprints that you find online - yes, even if supplied by the component manufacturer - or even worse, that come with the PCB package - are just wrong. If you are even remotely careful, the risk of human error on your side is far lower, for reasons I don't fully understand.

(I guess it's the summer workers / student workers mass-producing footprints for devices they don't understand or use, in a rush, using a tool they don't use, converting the resulting files to another format they don't use, using a conversion tool made by student workers in a rush...)
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: jbb on June 16, 2018, 01:46:17 am
... it's unbelievably even more common that footprints that you find online ... are just wrong.

It's so true.  Happens with many PCB packages and libraries.  What you can do is check the library footprint, copy it into your library, and then use it.  Be aware that generic library components may be 'updated' from time to time without your noticing, which can be a big pain.

At work, we use 100% our own library.  While there are no doubt a few mistakes lurking in it, we eventually find and fix them.  We use version control and issue tracking to document our activities.

Also, when you're making your own footprints, be very careful to check if the datasheet uses a top view (standard) or a bottom view (####!).  I've had some upside-down screw ups...
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 17, 2018, 04:31:39 am
Well, as for the online footprint, it might be true that they could be wrong. I usually make some of them. For the TLI4970, I found this footprint online and it seems to work: https://componentsearchengine.com/TLI4970-D025T4/Infineon
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 19, 2018, 04:23:50 am
I have an additional inquiry: is a snubber network strictly necessary in this application? I'm having hard time trying to find a proper guideline to design a RCD snubber.
Title: Re: Selecting the right IGBT and Gate driver for this application
Post by: XaviPacheco on June 19, 2018, 05:37:39 am
Extracted from Switching Power Supply Design, Abraham I. Pressman:

With MOSFETs, turnoff switching losses are considerably less than those with bipolar transistors. Current fall time with a MOSFET is so
rapid that current in it will have fallen almost to zero by the time the voltage across it has risen significantly. Thus, although turnoff snubbers are used with MOSFETs, their prime function is not to reduce overlap dissipation, which is already low. Rather, the function of the MOSFET turnoff snubber is to reduce the amplitude of the leakage inductance voltage spike. Since leakage inductance voltage spikes are proportional to dI/dt in the transistor, a MOSFET with much faster current turnoff time than a bipolar will have a larger voltage leakage spike.


I would like to add that some new MOSFETs claim to have improvements on their dv/dt and di/dt capability.
As stated previously, I'm going to use this one: IRF250P224.

So, I'm really considering if I need a snubber, but still got wires crossed.