Coil driver design?

[Guest]

Hi

Want to make a PWM circuit to for power DC coil, 24V 3A.

-<24v>-<COIL>-<FET>-<gnd>-

Will a modern N-FET be a good choice?

[Ttelmah]

Probably.
However, multiple 'caveats'.
First, unless you are using a FET, specifically designed so that it's internal diode structure exhibits fast recovery (Hitachi do ones for example), don't rely on the internal diode to provide the trap function, instead provide an external diode for this.
Second, remember that the energy stored in the coil, has to go somewhere when the FET switches off. The trap diode routes it back into the +ve rail, but unless this has significant capacitance present, with good high frequency performance, this can result in significant voltages being induced on the rail, and noise problems.
Third, if this is something like a relay, you might want to consider _reducing_ the power through the coil, once it has switched on.
Fourth, beware that a standard logic output like those on the PIC, is not the ideal source to drive a typical FET. You either need to look at 'logic level' FETs, or a better driver.

Best Wishes

[RLScott]

[Charlie U]

There are many FETs that are capable of handling the flyback energy of a coil. These FETs are what International Rectifier calls "Avalanche Rated". These devices act like a high power zener diode when the drain to source voltage reaches the breakdown voltage rating. They have specific current and energy ratings for avalanche mode. I just happened to grab the IRFIZ24 data sheet.

[FvM]

[RLScott]

[Ttelmah]

Yes. Hence my first comment.
Hitachi, do actually do some FETs that have quite generously rated diodes built in, with good voltage ratings, high peak current ratings, and fast recovery. However it is 'worryingly' common to see people laying out drives using normal FETs, relying on the diode structure. It is then equally common to see these failing....

Best Wishes

[Guest]

Hi

Thanks for the feedback:-)
I will try to find some N-FET, and some fast clamp diode that can fit in to my needs.

-

[asmboy]

i am assuming that you want to regulate the current through the coil in a variable way since you say PWM and there is a very clean way to do this from the hi side with a P mosfet too - and above all with generally much better PWM/current - linearity than when using a low side N fet.

it requires an extra cheapo n-fet driver and a couple 1/8w resistors but is very stable and linear , in case that matters to you

a High side P-chan Mosfet like the IRF 9520 - with source connected to +24V , a pull up resistor of 10K ohms connected between +24 and the gate , and the drain going to the coil , its other end ground.

the IRF9520 DRAIN junction gets a 40 V ( at appropriate amperage) shottky diode CATHODE conected to said drain , and the diode anode to ground.

Now you drive the gate of the IRF9520 with the drain of a 2n7002 , and its source goes to ground.

Lastly the 7002 'S gate gets 10K ohms to ground and a 390 ohm resistor from the gate of said 7002 to the PWM out port of your pic.

if you pick the right PIC divisors you can get 10 bit accuracy form the darned thing . and BTW - this is your quintessential BUCK regulator design for any PWM drive source - done this a million times in my sleep.

the only drawback in any of these is that the coil may "sing" at the switching rate or a harmonic , no matter which method you use

and that unexpected acoustic component may come as a surprise to you .

[asmboy]

If ultimate energy efficiency is not a concern, then you may want to also consider adding a damping resistor of about 330 ohms 1/2w in parallel with your coil. This gives you a bit of cover for excessively hi Q causing nasty HV spikes off di/dt from coil excitation with narrow pulses and thus improves linearity at very low PWM ratios - not to mention reduces singing too.

[mkent]

TI / Burr-Brown make things easy with the DRV 101 series drivers.
http://focus.ti.com/docs/prod/folders/print/drv101.html

[FvM]

@asmboy: I didn't get the point, why an P-FET buck converter should have advantage over the previous suggested N-FET buck-converter? Apart from changing the polarity (which implies a higher rdson with comparable FETs), your mainly suggesting a higher gate drive voltage of 24 V. The value is considerably above allowed maximum ratings and must be feared to damage the FET.

In special cases, there may be reasons to use a PFET switch of course. For example, if a ground referenced current sense resistor is intended.

As another disadvantage, the gate is rather slowly discharged through 10 k resistor. Driving a logic-level NFET directly from a PIC port pin through a 100 ohms resistor gives a better switching performance at any case. For higher load currents (= FET with higher gate capacitance) or FETs requiring a higher gate drive voltage, level converting drivers as Microchip TC427 are available, discrete BJT transistor drivers are also an option.

[Guest]

Now I getting confused...

What will be best if I only want the pic output (5v) to drive the fet?

Is 5v too little? ...all datasheet put max. 24v(gs) and at least 10v for making "real" on(low Rds on)

[Ttelmah]

5v, is too little for a 'normal' FET, if any significant current is being driven. However there are hundreds of 'logic level' designs around now, that only need perhaps 3v or lower. A search for 'logic level power MOSfets', should find loads of examples. Units like the NTP30N06L, have a 'worst case' Vgs, of only 2v, for their Rds, of 46mR, and most examples, need less than this.

Best Wishes

[asmboy]

1- low side PWM drive has the disadvantage of poor linearity at low pwm duty cycles - which you will soon see if you build and run that way

2- low side inductor drive - whether with N fets or NPN sw xtors - can suffer from severe open state high voltage spikes -- generated when the fet switches off, that must be damped , and can still cause source power supply problems as well

3- note to FWM:: the 10K resistor is indeed a bit slow on the turn OFF not the turn on and this is on PURPOSE to again limit the splash dv/dt effect --

what matters is the turn ON time and the
turn ON TIME - is very fast due to the low channel resistance of the 2n7002 -
the 10k has no effect there at all .

4- this hi side buck regulator has inherent topological protection against all these problems

i don't use the extra parts for fun - rather for the net effect of the circuit

low side PWM looks like it would be cheep and simple - but only superficially when you consider that with a significant inductive load ( your 3A coil) - you can much more likely subject your power supply filter capacitors to fault levels of dielectric puncture voltage

now if you were driving a string of LEDS for brightness modulation - low side is fine - and i've done that - no sweat -

but NOT FOR BIG INDUCTORS - ignore this at your peril.

I do this exact sort of thing for a living , and often at much higher power levels for DC motors and magnetic levitation bearings -often with very high L's and peak I , and believe me - low side PWM is a GREAT way to blow your PSU

i say again - is there nobody here who designs ROBUST PWM buck style power supplies?