Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Mosfet gets hot in 3 phase inverter circuit

Status
Not open for further replies.

Nihaludeen

Member level 2
Joined
Oct 29, 2017
Messages
46
Helped
0
Reputation
0
Reaction score
0
Trophy points
6
Activity points
460
I am trying to build a 1KW bldc inverter circuit.
I used atmel328p processor, gate driver and 12 mosfets. The mosfets swiches good, but when i connected to 1kW motor (at 20v and 10amp supply) the mosfet gets heated up above 60C. The circuit work fine for low watts motors but not for high power motor(1KW).
I attached the schematic below, correct me friends.
 

Attachments

  • IMG_20180815_121014.jpg
    IMG_20180815_121014.jpg
    503.2 KB · Views: 178
  • ir2110.jpg
    ir2110.jpg
    483.5 KB · Views: 150

what is the part number of your Mosfet ? are you musing heat sink?
does nosfet have internal flyback diode? you have used d1n4007 as floyback diode for mosfet and also speedup for discharging resistors, that's not a good choice at all , you need a fast diode for fly back,
and for fast discharging diode you can use D1n4148 instead

by the way 60c is not big problem for your mosfets
 
-MOSFET=irf3205
-Yes I used heat sink
-Ok friend sure i will change to fast switching diode
-mosfet increasing gradually , and increases even above 60C goes on . So I switched off my supply to prevent the MOSFET.
 

Is it possible you have power factor error? What waveform does your inverter produce? (Sinewave, square wave, or modified sine?) For a square-ish type the mosfets are turned off suddenly.

However a heavy motor load may introduce power factor error. It is common when you have an inductive load. This means it continues trying to pull current as a mosfet shuts off. (A resistive load allows sudden shut-off, even a heavy resistive load, so it does not cause power factor error.)

A slight amount of inductance causes slight power factor error, which your mosfets may be able to tolerate.
However a large inductance with large Ampere load is likely to stress your switching devices.
 

-inverter output is Square wave
-yes my load is inductive load( 1kw motor )
-To reduce power factor error by the inductance, I used a resistance at the gate. I think it will minimize it, or correct me friend if I am wrong
- Then how to control or such a type of power factor error.
- Is this is the reason which causes MOSFET to heat?
 

Usually a capacitor is installed across the inductive load. The value needs to be selected so C & L values create a resonant frequency which matches the AC supply frequency.

It may be difficult to find the proper solution. This problem is a cousin of the voltage spike generated by the inductor when current is abruptly shut off. The spike can be strong enough to destroy mosfets. Some kind of snubber network reduces the problem. The AC waveform requires a snubber which reduces spikes in both directions.
 

You need some large caps across the power supply rails near your mosfet totem poles. 4148 is a bit small for the app, the 4007 should be 45V 1A schottkies, 10 ohm gate drive may be too small to stop high dv/dt affecting your controller...
 

you can check the PWM waveform on the output with ohmic load and be sure that your driver and mosfets work correctly (output pulses are correct and sharp enough)
you can use bigger heat-sink and heat transfer pads or silicon glue between mos and heat sink and let it warm up
 

Usually a capacitor is installed across the inductive load. The value needs to be selected so C & L values create a resonant frequency which matches the AC supply frequency.

It may be difficult to find the proper solution. This problem is a cousin of the voltage spike generated by the inductor when current is abruptly shut off. The spike can be strong enough to destroy mosfets. Some kind of snubber network reduces the problem. The AC waveform requires a snubber which reduces spikes in both directions.
Ok i will put snubber circuit across mosfet

- - - Updated - - -

You need some large caps across the power supply rails near your mosfet totem poles. 4148 is a bit small for the app, the 4007 should be 45V 1A schottkies, 10 ohm gate drive may be too small to stop high dv/dt affecting your controller...

I replaced capacitance across the power supply but no effect. 1n4148 is fast switching diode is used that, but no change.
I used 10 ohm to allow more current to mosfet gate (to charge the mosfet gate capacitance for fast switching), if i am wrong correct me friend.
 

Something that is missing in all of the discussions above, are your power loss and thermal calculations.

Start from there, rather than blindly attempting a fix to a problem.
 

Hi,

1n4148 = 200mA to 400mA, depending on the datasheets I've seen. Is that enough? Easy Peasy's suggestion will no doubt be correct for your diode needs.
 

Can i use UF4007 which is Ultra-fast reverse recovery time diode.
Friend what causes my mosfet to heat too much, any suggestion. I calculated theoretically power loss and power dissipation.
Power loss in single mosfet,
- 0.4 Watts (major loss: conduction and switching loss)
Power dissipation,
T = 62C * 0.4W + 30C
= 54.8C (Without heat-sink)

But my mosfet heat above 60C and gradually increasing (including heat-sink)
 

Hi,

1n4148 = 200mA to 400mA, depending on the datasheets I've seen. Is that enough?
The 200mA to 400mA is DC current.
But in this application there are only narrow peaks in the range of some 100ns.
This is easy to widthstand for the 4148.
*******

but when i connected to 1kW motor (at 20v and 10amp supply) the mosfet gets heated up above 60C.
You say 1kW...
But if you have 1kW and 20V you need at least 50A.
For a motor the MOSFFET current is additionally increased by the motor efficiency, the commutation scheme and the formfactor.
So for 1kW output don´t expect less than 100A peak current.

--> what is your commutaion scheme?

--> Measure the MOSFET current with a scope and show us the picture.

*****
How do you come to 0.4W?
In my eyes this is unrealistic.
* switching loss
* conduction loss MOSFET
* loss of internal diode

****
A capacitor directely connected to a "switched" output is no good idea.

Klaus
 
I supplied only 20volt to the inverter for 1kw motor.
I didn't increase my supply because i think there is a mistake in my circuit.
Calculation,
Vin= 20v, I= 5amp, rds= 0.008
I calculated for 5amp because I paralleled the MOSFET and I calculate for per mosfet.
- conduction loss
P = i^2 * RDS(on) * duty cycle
= 5*5 * 0.008 * 0.5
= 0.1W
- switching loss
Fsw= switching frequency, Qs= total gate charge of mosfet, Ig= gate current
P= V * I * Fsw * Qs/Ig
= 20 * 5 * 20K * 100ns/1A
= 0.29W
Ploss= 0.1+0.29+ other gate loss
= 0.4W ( loss on each MOSFET)

I think the above calculation is correct to my knowledge, sorry friend if I am wrong

****
A capacitor directely connected to a "switched" output is no good idea??
I didn't get it.
 

Hi,

Again:
--> what is your commutaion scheme?

--> Measure the MOSFET current with a scope and show us the picture.

Klaus

- - - Updated - - -

Hi,

A capacitor directely connected to a "switched" output is no good idea??
I didn't get it.
From your above informations it seems you switch 20V/100ns.
Now what capacitance do you intend to use?
Usually 100nF are not enough to compensate for the motor inductance.

Can you calculate the current for a 100nF capacitor and 20V/100ns?

Now for 1uF...
and then you have 3 phases...

With this new current re-calculate the switching loss. And the radited EMI will cause new problems.

Klaus
 
Hi,

Again:


Klaus

- - - Updated - - -

Hi,


From your above informations it seems you switch 20V/100ns.
Now what capacitance do you intend to use?
Usually 100nF are not enough to compensate for the motor inductance.

Can you calculate the current for a 100nF capacitor and 20V/100ns?

Now for 1uF...
and then you have 3 phases...

With this new current re-calculate the switching loss. And the radited EMI will cause new problems.

Klaus

Now what capacitance do you intend to use? i didn't get you, in which place you are saying.
How do you choose 100nF , friend could you explain.

- sorry i am not having scope.

Commutation scheme,

hall sensor commutation scheme .jpg
 

Hi,

Please note:
* You said you want to use a capacitor
* But you did not give any value. Thus I asked you for the value. 100nF is just an example. As said: to compensate for the motor (whiich is unknown to us) inductance you usually need an even higher (than the 100nF example) capacitance value
* I said it's no good idea (independent of value)

Commutation scheme:
Now we know that it is hall sensor controlled, which us a very essential information.
Did you verify that all the connection (channels, wiring, polarities...) are correct?
How do you start the motor from 0 RPM? Maybe the rate of duty cycle rise is too fast?

Klaus
 
Your commutation scheme needs to account for motor acceleration / deceleration and over-speed (field weakening conditions) usually one has a mag sensor (hall effect) to "see" the spinning magnets and use this to change the pulses applied to the motor drive - are you doing this correctly...?

- - - Updated - - -

also, at low speed you need to PWM the fet drive to keep the current in bounds for long ON times of a mosfet ...
 
Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top