hello,

I'm using this circuit to control a dc motor speed using a PWM from a microcontroller.




details:

using n-channel enhancement mosfet , 400V , 30A ratings.

1 KHZ pwm signal connected to the mosfet gate.

30A flywheel diode

dc motor ratings : 160V , 20A .


I first tried this with VGS value of 5volts (connected the pwm directly from micro to mosfet gate) , but the mosfet got burned (drain and source were short circuited) .

the mosfet datasheet mentions that the VGS value could be up to 30V , so I used an op-amp to amplify the pwm signal to a max of 15v and connected that to mosfet gate , the motor worked fine for about quarter the scale , then the mosfet got burned again and I lost control of motor speed (rotated full speed regardless of VGS)

also the motor is not loaded , so current can't have reached more than 30A.

can any one help me with that?

 

Are you trying to get an electro-mechanical relay to pass 1kHz PWM? Surely that's not on: those things are much too slow.

 

The MOSFET (VN0106) you are referring have different parameters as per the datasheet.

Drain-to-source breakdown voltage is 60V not 400V
On-state drain current (min) 2A not 30A
Rds(on) is 3 ohms.

Datasheet : http://www.redrok.com/MOSFET_VN0106_60V_2.5A_5O_Vth2.4_TO-92.pdf

Or may be you are referring some other mosfet...

And that circuit is using a relay and I think in place of that relay you are using a DC motor.If yes then you cant do it like that anyway provide us a proper circuit diagram of your circuit so that we could help.

Good Luck

 

I'm sorry for the ambiguity.

this scheme describrs only the connection , I'm using the(k1519) mosfet from hitachi.

and instead of the relay coil in the scheme , I'm connecting the motor.

why can't it work?

 

You've given us a drawing that has maybe one part that is really the way you're doing it. Too much wrong information to hold in my head while trying to figure out what went wrong. Please provide a drawing of the real circuit with the real part values and numbers.

 

sorry about that.

this is the circuit I tried first and there was no result , once I increased the potentiometer a little bit , the motor rotated full speed as if I was connecting the voltage to its leads directly , and I found that the drain and source of mosfet were short circuited even when VGS =0v (mosfet was burned)




then I thought that it might be that VGS is not large enough to open the mosfet channel (in the datasheet saturation starts at VGS=5v) , so I connected the op-amp to amplify the voltage ( multiply by 3 ) for the PWM signal to get to a peak of 15V instead of the 5V coming from the microcontrollter , the motor operated very well , I was aple to control the speed and then reduce PWM and make it stop again , but when I increased the PWM over a certain value , the motor started rotating at its full speed and I found mosfet drain and source to be short circuited whatever VGS is again , and I stopped being able to control motor speed.




thanks for your attention.

 

Here's a bit of help. I have to go eat supper now. BRB

 

these are the datasheets , I already have them :] .

 

Here are my thoughts:

1. Gate resistor of 1k is TOO LARGE. This should be on the order of a few ohms to 10s of ohms. Try using a 10 ohm resistor to start. With a 1k resistor, it is taking FOREVER to turn on the MOSFET, which means it is burning a ton of power when you try to switch it and increasing the temperature of the device.

2. The gate threshold of the MOSFET is 2-3V. RDSon is spec'd at 10 VGS. You should therefore use at least 10V for the gate so that the MOSFET is in full saturation.

But I think the cause is a weak current output via your PIC + a huge gate resistor. The gate of a MOSFET is like a capacitor that needs to be charged, and its probably taking way too long in your case and the MOSFET is in the linear region too long, burning power and increasing temperature.

Another option is to use a Gate Driver in-between the PIC and MOSFET, which will provide more current to drive the MOSFET (you will need a 12-15V source though). Something like the IRS2117PBF for example:
http://www.irf.com/product-info/datasheets/data/irs2117pbf.pdf

Good luck.

 

and do I have to use a mosfet driver , I think at this low freq. (1KHZ) there would be enough time to charge the gate without the need of a mosfet driver.

 

and do I have to use a mosfet driver , I think at this low freq. (1KHZ) there would be enough time to charge the gate without the need of a mosfet driver.

It is up to you. If you have available a 10-15V supply, the gate driver is better than using an amplifier circuit. The gate driver can be controlled by a 5V signal and will switch the FET up to your 10-15V supply with higher current output.

I would first change your gate resistor value to 10 ohm with your amplified voltage output of 15V and see what happens. I think your PIC is capable of sourcing very little current to drive the FET, so maybe you can even use a 1-2 ohm resistor for the gate.

~AnnexGR

 

Here are my thoughts:

1. Gate resistor of 1k is TOO LARGE. This should be on the order of a few ohms to 10s of ohms. Try using a 10 ohm resistor to start. With a 1k resistor, it is taking FOREVER to turn on the MOSFET, which means it is burning a ton of power when you try to switch it and increasing the temperature of the device.

thanks for your reply.

the reason I used this gate resistance is because of what I read in the "ART OF ELECTRONICS" , it sais "the series gate resistor , though not essential , is a good idea , because mosfet gate-drain capacitance can couple the load's inductive transient back to the delicate CMOS logic" , honestly I don't understand this.

another thing , there is actually no gate current (only 10uA leakage current in this mosfet) , so this resistance causes no voltage drop , therefore , it shouldn't affect charging speed.

and I also checked the gate charge required for the mosfet to switch and its max value was 240nc , the pic output can be up to 20mA , Q = I * t , it would take 12 microsecs to charge the gate , I'm using 1KHZ signal , so the cycle time is 1msec .

am I right?

 

I also think the gate resistor is too big.

Are you able to tell if the MOSFET heated up before it burnt? If it does it's more likely that it died because of overcurrent, if not then of overvoltage.

Did you mount it on a heatsink?

Look in the datasheet under main characteristics the diagram for maximum safe operation area, doesn't it look like you are at the limit for 1kHz (1ms) for 160V 15A?

Is the motor loaded?

10V gate voltage is absolutely sufficient.

 

there is actually no gate current (only 10uA leakage current in this mosfet) , so this resistance causes no voltage drop , therefore , it shouldn't affect charging speed.

am I right?

No, everytime you charge and discharge the gate the current is much higher, only limited by the gate resistor and the drivers current capability.
The gate acts as a capacitor.

So from your driver you have a resistor in series with a capacitor which is constantly being charged and discharged according to your PWM frequency.

from application note slup169
"For completeness, the external series gate resistor and the MOSFET driver’s output impedance must be mentioned as determining factors in high performance gate drive designs as they have a profound effect on switching speeds and consequently on switching losses. "

 

No, everytime you charge and discharge the gate the current is much higher, only limited by the gate resistor and the drivers current capability.
The gate acts as a capacitor.

So from your driver you have a resistor in series with a capacitor which is constantly being charged and discharged according to your PWM frequency.

that's right , I got it .

 

thanks for your reply.

the reason I used this gate resistance is because of what I read in the "ART OF ELECTRONICS" , it sais "the series gate resistor , though not essential , is a good idea , because mosfet gate-drain capacitance can couple the load's inductive transient back to the delicate CMOS logic" , honestly I don't understand this.

another thing , there is actually no gate current (only 10uA leakage current in this mosfet) , so this resistance causes no voltage drop , therefore , it shouldn't affect charging speed.

and I also checked the gate charge required for the mosfet to switch and its max value was 240nc , the pic output can be up to 20mA , Q = I * t , it would take 12 microsecs to charge the gate , I'm using 1KHZ signal , so the cycle time is 1msec .

am I right?

The gate resistor is to limit the turn-on slope of the MOSFET when charging the gate. A larger value resistor will reduce EMI because it switches softer, a small resistor reduces power losses but causes hard switching. A gate resistor should always be used for a circuit with a Power MOSFET, and the size depends on the application, but generally should never be more than a few hundred ohms at max.

Look at page 6 of IR's design tip for estimate switching times (tsw):
http://www.irf.com/technical-info/designtp/dt04-4.pdf

A significant mistake you are making in your switching time assumption is that the value I in the equation Q = I * t is the AVERAGE current, not the current coming from your PWM output.

The average current depends on the resistance of your driver and the gate resistor.

The resistance of your driver can be estimated as R = 5V / 20mA = 250 ohms.

The gate resistor you were using was 1kohm. So the effective resistance is 1250 ohms.

Your average current is Iavg = 5V / 1250 = 4mA. Your switching time is estimated as tsw = Qg / Iavg = 240nC / 4mA = 60 us.

60us is a VERY long time for a MOSFET to turn on (it should turn on in hundreds of ns or less, but this also depends on the application fsw)

If you reduce this resistance to 1 ohm, you get near your 12us tsw time. This is still too long IMHO, which is why I recommended a gate driver (the IRS2117PBF sources 290mA).

~AnnexGR

 

there is also a decoupling capacitor I forgot to put it in the schematic.

Are you able to tell if the MOSFET heated up before it burnt? If it does it's more likely that it died because of overcurrent, if not then of overvoltage.

Did you mount it on a heatsink?

yes it did heat up , but I don't know if it heated before it burned or after it burned (when the mosfet became a short ct.)

and there was a heat sink mounted .

also the capacitor between the voltage terminals heated up (don't also know before or after breaking)

but the motor is totally unloaded , it's unprobable current burned it .

Look in the datasheet under main characteristics the diagram for maximum safe operation area, doesn't it look like you are at the limit for 1kHz (1ms) for 160V 15A?
10V gate voltage is absolutely sufficient.

I didn't notice that before , I could increase freq. to 10KHZ and see what happens , thanks very much for that

 

Forgive me if somebody already said part of this, but...

The required switching time is not about "compared to the overall frequency". It's about how long the mosfet is between off and on because that is when it is heating up by having significant voltage across it while current flows through it.

Some of the math goes like this:

Vgate = Vin times e to the negative time/RC
Vg = dV e^-t/RC
so
R = -time over (C Log natural Vgate/Vdrive)
substituting:
time as 1 microsecond
Ciss = 5.8 nanofarads
Vgate = 5.5 volts
V from opamp = 13

Rmax = 200 ohms

Rmax to get the gate to 5.5 volts in 1 microsecond from an opamp that can deliver 13 of its 15 volt supply is 200 ohms.

You are free to substitute different values of time, gate capacitance, and voltage.
I have given you a tool.

 

If you reduce this resistance to 1 ohm, you get near your 12us tsw time. This is still too long IMHO, which is why I recommended a gate driver (the IRS2117PBF sources 290mA).
~AnnexGR

thanks very much for that .

so I will:

1-reduce gate resistance.
2-Increase operating frequency to 10KHZ.
3-use IRS2110pbf

thanks again for that great forum , I didn't expect that very fast and helpful aid.

 

thanks for your reply.
I didn't notice that before , I could increase freq. to 10KHZ and see what happens , thanks very much for that

The SOA only applies if you are operating in the linear region. Do you mean to tell me that when the MOSFET is on, you have 160V across it AND 15A? I don't think so.

Please realize that when the MOSFET is on the voltage across it is very small (RDSon * Id).

~AnnexGR