Back EMF protection

[Dr.EM]

I intend on making a PWM controller for various applications. Occasionally, this may include switching inductive loads, DC motors are a definate possibility. What would be the way to fully protect the 100v max rated MOSFETs (35v max circuit voltage) in the output stage shown (they would be more than 2 in parallel, but I doubt it makes any difference anyhow).

 

[Nigel Goodwin]

I intend on making a PWM controller for various applications. Occasionally, this may include switching inductive loads, DC motors are a definate possibility. What would be the way to fully protect the 100v max rated MOSFETs (35v max circuit voltage) in the output stage shown (they would be more than 2 in parallel, but I doubt it makes any difference anyhow).

'FULLY' protected is a very loaded term, but (obviously) the first essential is a reverse protection diode across the motor.

 

[Styx]

I intend on making a PWM controller for various applications. Occasionally, this may include switching inductive loads, DC motors are a definate possibility. What would be the way to fully protect the 100v max rated MOSFETs (35v max circuit voltage) in the output stage shown (they would be more than 2 in parallel, but I doubt it makes any difference anyhow).

Just to re-itterate what Nigel stated, the free-wheel diode. Without it the FET will blow up very quickly (see drawing).

Now just do debunk yr "but I doubt it makes any difference anyhow"
Paralleling up makes alot of difference!!!

Again please see my diagram, yes it is using an IGBT symbol but same principle.

You see the small inductors at both collectors, this is stray inductance due to layout. it is alweays there no matter how good you think the design is. I use nice big laminant buzzbar made out of faily thick copper and I still get overshoots (this being a professionally made bar).

YOU must minimise this as much as possible, even sacrificing other parts (ie cost ie bigger heatsink) if needs be. The more you put in parallel the higher the chance of you not getting all of the FET's right up close to the diode, if they are not close there will be stray inductance and this inductance will produce a voltage overshoot at turn-off

To give you an idea I have been switching today an IGBT at 600A with a link of 350V my overshoot was reaching 900V and that was with a good bussbar!!!



Also I have said it umteen times before but I will say again, Gate-drive!
You are switching AMP's !!! DO NOT just hook up the gate of the FET's to any old voltage source, the switching behaviour of FET's starts to become "funky" w.r.t. signal switch characteristics when switching of current is involved.

You need to ensure you can drive the FET hard (say with around 1A of gate current) at a high enough voltage to keep it hard-ON

 

[Dr.EM]

Ok, thanks for the responses. I have the gate driving base covered as i'm using smaller MOSFETs as drivers. I have the system on breadboard and using a 4 amp heater (and only one fet) it seems to work well, showing fast switching times, full rail to ground switching and the MOSFET staying nice and cool with a very small heatsink. But then the heater has no inductance issues. I have used a 1A DC motor with it, and although it works properly, there are lots of spikes showing up on the scope. Seeing as how I intend to use 4 IRF540, with a load of 20A hopefully possible, max voltage only 35v and plan on heatsinking to the case, is there any other way to protect the MOSFETs from inductive loads? Can diodes be placed around the FET somehow to protect it, 36v zeners?

 

[Styx]

Ok, thanks for the responses. I have the gate driving base covered as i'm using smaller MOSFETs as drivers. I have the system on breadboard and using a 4 amp heater (and only one fet) it seems to work well, showing fast switching times, full rail to ground switching and the MOSFET staying nice and cool with a very small heatsink. But then the heater has no inductance issues. I have used a 1A DC motor with it, and although it works properly, there are lots of spikes showing up on the scope. Seeing as how I intend to use 4 IRF540, with a load of 20A hopefully possible, max voltage only 35v and plan on heatsinking to the case, is there any other way to protect the MOSFETs from inductive loads? Can diodes be placed around the FET somehow to protect it, 36v zeners?

You could use a Zener, but you will just cook it (dont under estimate the power of the backEMF )

Three options
1) propper power-cct layout taking into concideration.
This should be your main driving force! keep all hte power-components close (but not so close that arcing is a possibilty - you are working at low-voltage so not really an issue). DO NOT just use some leads to connect up, if you can make a busbar, you say you want 20A you are going to start to have some REAL problems if your power cct is not layed out to minimise stray inductance.

2) slow down yr switching!!! you will be seeing spikes when switching an inductive load due to turn-off voltage overshoot as stary inductance try's to keep the current flowing through them flowing. By slowing down the speed that the FET turns-off (normally by having a different turn-OFF gate resistor) it allows the strays to dissipate their stored energy by the time the millar-region has been reached.
This however, results in more losses (increased switching losses) in the FET

#1 is by far the prefered option BUT even the best layout will still have stray inductance in very bad places, this is where #2 comes in to "tune" the switching characteristics of the switch's to match the present implementation

now we get onto #3

3) Snubbers!! There was a time when these were needed for forced-commutated devices (1st generation IGBT's) but that time has now gone with 4th being fantastic, Snubbers are still used main due to extreamly poor power-layout resulting in either too much switching-loss or too high of a voltage overshoot.

a snubber's job is basically to shift some of the switching losses from the switch to the snubber dissipating element.
in its simplest form a snubber is a RC in parallel with the FET/IGBT and get more elaborate


#1 with #2 is by far the best way

 

[zevon8]

If you are going to parallel a few FET's, often a series gate resistor of say 10 Ohms is often a good idea. It helps with spurious oscillations during switching. Miller effect can cause problems during fast switching, which can get complicated when paralleling FET's if they don't all share the load fairly evenly.

Check the IR site for application notes ( since you are using HexFET's ) they have alot of good information and design help that will keep you from detonating too many FET's

 

[Dr.EM]

Hmm, paralleling fets seems to cause a lot of problems. I'm now thinking it will probably be more sucessful to use one higher rated fet. How about one of these:

https://www.electro-tech-online.com/custompdfs/2006/05/4367.pdf

Or to be a bit silly, how about:

https://www.alldatasheet.com/datasheet-pdf/pdf/24132/STMICROELECTRONICS/STE180N10.html

Though the second one probably needs much larger gate voltages to saurate and has a large input capacitance.

As a hobbyist and not someone who makes PCBs, careful layout as you've detailed seems a bit impossible for me, I looked at those app notes too, confirming it. But with just one higher power fet, a lot of these problems are more manageble?

 

[zevon8]

Paralleling FET's is fairly straight-forward as long as you stick to some basic rules, such as keeping all the gate traces as similar as possible, provide solid drive, and maybe some gate resistors.

I do it all the time at work, using ST's OmniFet, VNP49N04, a logic level TO-220 device. The FET's are mounted back to back on opposite sides of a vertical heatsink, in sets of 4. I use a 10 Ohm gate resistor. This particular FET is pretty indestructable, see the datasheet:

https://www.st.com/stonline/products/literature/ds/1609.pdf

the same design was originally using IRFZ44N devices, but we switched to the ST device since we were using it elsewhere in large quantity.

It is apparently very suitable for PIC or uP control, since there is feedback on the gate pin to show the device status. I have never used this feature, but have had them "self-protect" successfully.

 

[Hero999]

Is the inductive load a big contactor?

Did you know adding a freewheel diode can actually reduce the life of contacts when they're switching high power loads?

Because the decay of the current in the coil is slower the contacts take longer to open, this can result in increases arcing, to avoid this it's good practice to add a resistor of the same value as the coil resistance in series with the diode. This will make the current decay more quickly and the back EMF will be no higher than the normal operating voltage, you can use a higher value resistor if you like just as long as the transistor can hande voltage.

 

[Dr.EM]

Ok, thanks for all the replies, I've learnt a lot from this, and am not sure whether to go ahead with the project now, but I would probably understand better doing it. If I decided to go with a single lower rds on mosfet such as:

https://www.electro-tech-online.com/custompdfs/2006/05/RFP50N06.pdf

How would protection from inductive loads be achieved. With just one, I don't have the issues with stray inductance unbalancing? Would a zener really not work? They all seem to have one built into them, but I would probably add something like these:

**broken link removed**

Also heatsinked to the case, would it survive the EMF from a DC motor rated at 20A, 12v, short term?

 

[TKS]

Join the club see topic : https://www.electro-tech-online.com/threads/mosfet-driver-ir2112-help-needed.23196/

building a starter pwm board well from my point of vieuw i'm gonna get allot of inductance problems... the best thing i have got is that i use a series wound motor sow i have no magnets / generator thing...sow less back emf right??

Tks

 

[rini_v]

Halo,
I have a problem. i use 12V. my control switch and motor are so far with a relay to switch. thr is an led to indicate switch on. and a diode for its protection. either the diode or the led blows off at times. i suppose it must be due to back emf from relay. so wat device id most suitable for it? a TRANSORB or DIODE WITH SERIES RESISTOR? plz reply if u know

 

[crutschow]

Because the decay of the current in the coil is slower the contacts take longer to open, this can result in increases arcing, to avoid this it's good practice to add a resistor of the same value as the coil resistance in series with the diode. This will make the current decay more quickly and the back EMF will be no higher than the normal operating voltage, you can use a higher value resistor if you like just as long as the transistor can hande voltage.

With a resistor, the voltage seen by the transistor will be the supply voltage plus the back EMF voltage. For a resistor is equal to the coil resistance, the transient transistor voltage will thus be twice the supply voltage.

 

[crutschow]

You could use a Zener, but you will just cook it (dont under estimate the power of the backEMF )

A zener is fine if you use it in conjunction with a diode across the coil so the zener will only see transient current due to stray line inductance. These spikes are low energy and should be readily absorbed by even a small zener.