1. Home
  2. Electronics Design
  3. PWM drive of DC contactor

PWM drive of DC contactor

I have a DC contactor with a coil that draws about 4 amps at 24 VDC and normally uses an economizer resistor of about 75 ohms to reduce the current to about 400 mA when pulled in (controlled by a delayed contact). But I want to use a PWM drive for another similar contactor (actually a 120 VAC unit) so it will work on DC and thus be more tolerant of AC line brownouts.

Previously I used a combination of a 75 ohm resistor in parallel with a

3300 uF capacitor to provide a full current pull-in pulse and then 400 mA to hold it in. This worked OK but the resistors get hot and it is not always reliable. Thus the desire to use PWM.

I made a simple version of this using a PIC12F675 and it worked well with a PWM frequency of about 1 kHz. It applied full voltage for about 100 mSec and then 10% PWM which drew only about 300 mA. Now I am trying a PIC12F1822 which has a real PWM module and a clock frequency of 16 MHz, so I tried a 20 kHz PWM which started with 95% for 100 mSec and then 10%. But the FQD13N06LTM MOSFET (60V 11A 110mOhm) shorted. I thought it was due to a wiring error on the prototype so I replaced it, and it seemed to work about right, but the PWM occurred before the armature pulled in. So I pushed it in manually, and the MOSFET once again exuded magic smoke. The coil inductance changes from about 15 to 40 mH when pulled in.

So, I think it may be the higher frequency PWM, and I plan to try 5 kHz or even 1250 Hz, but first I thought I would run a simulation (at about 96%

duty cycle). I am driving the logic level gate directly from the PIC so I figure that it may be somewhat slow and weak, so I used both a 5 ohm and

50 ohm resistor in series. Also I tried it with a standard diode and a Schottky. The Schottky produced only about 90 watts peak in the MOSFET for about 1.5 uSec, and the 50 ohm gate resistor showed about 3.5 watts average compared to 1.5 watts for 5 ohms. The standard rectifier produced about 325W peak for 30 nSec and then about 80 watts for an additional 1.2 uSec. The

average power was not much different for the two diodes.

The power dissipation with DC on the gate is about 1.25 W. The SOA seems to allow a 100 uSec pulse of 40 amps at 20 volts or 800 watts, but continuous power is about 2.5 watts without heatsinking other than the PCB. The MOSFET I used for the original prototype was a "self-protected" VNP14NV04 (40V

12A 35 mOhm).

I think the lower frequency will help, and also perhaps using DC (100% PWM) until it pulls in, should help greatly. But I wanted to see if anyone had experience with this and perhaps might offer some advice.

Thanks,

Paul

LTSpice: ========================= ========================= ======== Version 4 SHEET 1 1380 680 WIRE 144 0 -144 0 WIRE 368 0 144 0 WIRE 144 96 144 64 WIRE 368 96 368 80 WIRE 368 96 144 96 WIRE 448 96 368 96 WIRE -144 144 -144 0 WIRE 128 176 80 176 WIRE 256 176 208 176 WIRE 320 176 256 176 WIRE 80 208 80 176 WIRE -144 288 -144 224 WIRE 80 288 -144 288 WIRE 368 288 368 192 WIRE 368 288 80 288 FLAG 368 288 0 FLAG 448 96 coil FLAG 256 176 gate SYMBOL voltage -144 128 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 24 SYMBOL voltage 80 192 R0 WINDOW 3 -207 116 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR Value PULSE(0 4.5 10u 50n 50n 600u 500u 2000) SYMATTR InstName V2 SYMBOL nmos 320 96 R0 SYMATTR InstName M1 SYMATTR Value IRL530NS_L SYMBOL ind 352 -16 R0 SYMATTR InstName L1 SYMATTR Value 40m SYMATTR SpiceLine Rser=6 SYMBOL res 224 160 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 50 SYMBOL diode 160 64 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D1 SYMATTR Value MURS120 TEXT 136 264 Left 2 !.tran 300m startup TEXT -128 336 Left 2 ;R1=50 2 kHz 96% MBRS140 IRL530NS_L 87W peak 1.66 uSec

1.32W Avg TEXT -128 368 Left 2 ;R1=50 20 kHz 96% MBRS140 IRL530NS_L 87W peak 1.42 uSec 3.55W Avg TEXT -128 400 Left 2 ;R1=5 20 kHz 96% MBRS140 IRL530NS_L 92W peak 261 nSec 1.49W Avg TEXT -128 432 Left 2 ;R1=5 20 kHz 96% MURS120 IRL530NS_L 325W peak 30 nSec 1.57W Avg TEXT -128 464 Left 2 ;R1=50 2 kHz 96% MURS120 IRL530NS_L 325W peak 30 nSec 1.15W Avg TEXT -128 496 Left 2 ;R1=50 DC MURS120 IRL530NS_L 1.25W Avg
Reply to
P E Schoen
Loading thread data ...

If you have any way of measuring the capacitance of the contactor coil, do so -- that would cause increased (possibly vastly increased) dissipation in the FET.

I agree that if you're driving the gate directly from the PIC that it may not be very fast. Look at the current rating for the pins on the PIC -- your 50 ohm resistor in the simulation may be barely accurate.

Do you have an oscilloscope, that you can measure actual circuit behavior? When one of the most important components in your circuit is a box labeled "unknown -- dragons inside" you want to rely even less on simulation than usual.

--
Tim Wescott 
Control system and signal processing consulting 
www.wescottdesign.com
Reply to
Tim Wescott

V2 in your sim, the gate drive pulse, is zero impedance. The PIC will be much wimpier, so the mosfet will switch slowly and get hot. You need a gate driver chip, and may as well go for more gate drive voltage.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators
Reply to
John Larkin

Is it possible that your coil kickback higher than 40V and damaging the MOSFET? I use a 400V MOSFET for 12V relay.

Reply to
edward.ming.lee

John, what do you think the 50 ohm resistor between V2 and the gate is simulating? Particularly given what the OP says in the text?

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com
Reply to
Tim Wescott

I just noticed your final question -- yes, a 100% duty cycle in the "on" phase will help greatly -- with your wimpy PIC pins the FET dissipates more when its switching because of slow gate drive, so that will help.

Then your biggest worry will be the 10ms or so when you drop to low duty cycle but the coil current is dropping -- the FET will experience a bunch of little heat pulses unless you want to get fancy and turn the PWM off for three or four ms, then turn it on again.

If you really want to save on the gate driver I'd suggest doing two things: first, see if you can find a FET with lower gate charge -- that'll help your wimpy "gate drive" charge it up; second, parallel up every available spare pin on the device to increase the oomph. That means that you can't use the software PWM, and it means that you have to pay attention to the chip's overall current limit -- but it'll give you more drive capability.

Some of the little single-channel gate drive chips are pretty cheap, you may want to take a gander at what's available.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com
Reply to
Tim Wescott

50 ohms simulates 50 ohms. And I really doubt that a PIC port can source and sink 100 mA. But all he needs is an oscilloscope on the gate to find out what the slew rate is, and then estimate power dissipation.

Clearly something happens between 1 KHz and 20 KHz PWM rates. If it's not switching loss, what is it? (Well, aside from a bad breadboard layout.)

There's a reason that people make and sell mosfet gate driver chips. In his case, at 20 KHz, he could probably parallel all the sections of an HC04 and fix things.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators
Reply to
John Larkin

I don't have an impedance analyzer, so that would be difficult. And I don't have the exact contactor (120 VAC) here, so I am using a similar size with

24 VDC coil.

You are correct. The pins are rated for about 25 mA. So I tried the simulation with 200 ohms. Now at 2 kHz it shows only 100 watts peak but

2.1 watts average for the 98% duty cycle. For 10% it shows 11 watts peak and

only 41 mW average.

For 20 kHz and 90% duty cycle (which I now see is actually what I used), it gives 9.9 watts average and 96 watts peak, so that would fully explain the failure. Even at 2 kHz it is 1.8 watts average. And at 20 kHz with 96% duty cycle it is 2.3 watts.

I have a 60 MHz storage scope and I was able to observe the waveforms with a resistive load, but with the intended load the MOSFET popped before I could get any waveform information.

"on"

That may be best, and it is easily accomplished.

That is also easily done, and a good idea.

It would be difficult to use more pins on this 8 pin PIC, and I would like to use the built-in PWM module.

I might also be able to use a smaller version of the protected MOSFET (some of which are actually listed as a gate driver). Here is a 60V 10A "OmniFET" for about a dollar:

formatting link

The least expensive gate driver I found on a quick search is about $0.35, but is limited to 25V. My raw supply is 24V, and I will have a 5V regulator for the PIC. However, 5V would be OK for the logic level MOSFETs.

formatting link

Thanks,

Paul

Reply to
P E Schoen

the

The commutating diode across the coil should take care of any kickback. But it might need to be faster than the 1N4004 presently used for this application.

A 400V MOSFET with sufficiently low ON resistance may be hard to find and/or expensive. Among 400V MOSFETs rated 5 to 10 amps continuous the lowest ON resistance is about 550 mOhms and cost is over $1 while the MOSFET I plan to use is only $0.50 and is smaller. But more importantly, if the 12V relay

coil is generating inductive spikes requiring a 400V device, there may be other problems such as insulation breakdown and flashover on the PCB or wiring.

Paul

Reply to
P E Schoen

Make sure the gate driver can work at 5V! Most of them expect 12-ish.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com
Reply to
Tim Wescott

On Friday, October 18, 2013 2:19:00 AM UTC-7, P E Schoen wrote: ...

)

... The contactor intended to be driven from AC will probably have a shading ri ng (a shorted turn on part of the magnetic circuit) to avoid chattering. I t will not act like a pure inductance but like a transformer with high-leak age inductance and a shorted secondary (or a secondary loaded with a low re sistor).

kevin

Reply to
kevin93

Oh no, i don't expect much more than 50V. But a 400V 13001 (TO-92) is only nickle or dime in volume. Anyway, i am dealing with much less current than you are.

Reply to
edward.ming.lee

"P E Schoen"

The commutating diode across the coil should take care of any kickback. But it might need to be faster than the 1N4004 presently used for this application.

** Betcha your MOSFET is dying from overvoltage spikes.

A Schottky across the coil is good but you also need zener protection across the FET.

The wiring to and from the coil and FET may have enough inductance to do the job.

I have some experience with DC motors and PWM powered by a 12 cell NiCd pack.

Despite a nice Schottky across the motor, BUZ11 FETs failed until I added a

24V, 5W zener across them.

.... Phil

Reply to
Phil Allison

...

shading

chattering.

So in that case I think it should act like a very small inductor on DC and the current should quickly peak to the value determined by the coil resistance. That might actually work better as far as the inductive energy causing high power during switching. I'll try simulating it by adding a second small coil with a coupling factor K between 1 and 0.95 to see how it looks.

Thanks,

Paul

Reply to
P E Schoen

If you use a high side current monitor in combination with a hysteretic con troller, you may find you're switching at sub-kHz with those time constants (L/R), and you will also know exactly when the contactor is pulled in as w ell as by how much margin you have it pulled in when you back off to the lo w power mode. The hysteretic controller can be your PIC if you're determine d to use it.

Reply to
bloggs.fredbloggs.fred

I didn't know about them thar shading rings.

It could very well make it lossy -- try it out.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com
Reply to
Tim Wescott

I simulated it with a 1 mH coil and a 1 mOhm series resistance with K=0.99. Here are the results:

R1=200 20 kHz 90% MURS120 IRL530NS_L 95W peak 6 uSec 9.8W Avg R1=200 20 kHz 10% MURS120 IRL530NS_L 6W peak 8 uSec 419 mW Avg

R1=200 2 kHz 90% MURS120 IRL530NS_L 94W peak 6 uSec 1.70 W Avg R1=200 2 kHz 10% MURS120 IRL530NS_L 30W peak 3.8 uSec 128 mW Avg

I don't know if this is really any better. But the most apparent result is that there is no significant build-up of current as there was previously, so that seems to indicate much lower inductance and a much faster response. And ~2 kHz seems like the way to go. It may make more audible noise but these are inside a steel box along with big transformers and powerful fans and

used in a typically noisy shop or factory environment, so a little more industrial music won't be an issue. :)

But also I finally found a contactor with a 120 VAC coil, and it measures

575 mH open and 3.72 H closed, with 136 ohms resistance. Also, it will not even actuate with 24 VDC, and that;s barely enough to hold it in (with about 170 mA). So now I am thinking that perhaps the contactors in the unit are 24 VAC. I may get a chance to test this circuit on the actual machine on Monday. It's been a while since I worked on it, and my memory is foggy, but that makes more sense. Here is the relay we use:
formatting link

The relay I found is not this exact part but a similar 3 pole HVAC type. It seems that 24 VAC and 120 VAC coils are most common. We decided to use these because they are inexpensive and available from most electrical supply houses such as Grainger. Otherwise, for 24 VDC devices, we have used American Solenoid Karaus & Naihmer as well as ABB, but they are more expensive and harder to obtain.

Thanks,

Paul

Reply to
P E Schoen

the

There very well could be damaging spikes. I was having strange problems where the relay would pull in and then drop out before the PWM kicked in, and I found that the PIC was resetting. I had replaced the 24 VDC contactor (6 ohm coil) with a smaller relay (280 ohm coil) that draws only about

100 mA, and even that had problems, but it cleared up when I put a 0.047 uF capacitor across the coil and diode. I had to use 20%-40% PWM to keep it

pulled in, however.

But when I reconnected the larger contactor, even with 10% PWM, it drew about 2.5-3 amps. That was too much for the MOSFET so I could not leave it on long enough to look at the waveform, but with the smaller relay the voltage took about 100 uSec to rise after the PWM turned off. The smaller relay coil has an inductance of about 1.1 H and a resistance of 280 ohms. Oddly, the inductance only increased slightly when the armature is pulled in.

I think the larger contactor may have taken even longer to stop conducting and thus caused this effect. When I redid the simulation with a 100 mSec

full-on pulse, followed by 200 mSec off, and then the PWM, the relay current was still 85 mA after 90 mSec. However, the simulation showed the coil current should drop to about 320 mA at 10% PWM after about 100 mSec so it may be something else.

I may still add a 30V or so TVS diode across the MOSFET. And I should probably add like a one ohm resistor in series with the coil capacitor to limit the turn-on current.

Thanks for the tip.

Paul

Reply to
P E Schoen

problems

Well, I tried the circuit in the unit where it is to be used, and I had the same problem with the contactors drawing too much current. I think the circuit was also resetting. So I got one of the actual contactors and brought it home to do better testing.

The 24VAC coil is 4.3 ohms and has an inductance of 9.05 mH, which increases to 20.4 mH when pulled in. I tried it here and it acted the same, and then the MOSFET blew. So, I replaced it with a VNP14NV04-E protected OmniFET, and it works fine. It draws 5 or 6 amps for 100 mSec to pull in, and then throttles back to 90 mA with 10% PWM.

Now I will need to observe the operation more closely and see if I can detect a high voltage spike or oscillations or other behavior that could

have overstressed the MOSFET. I think a 28V TVS diode across it will take care of it, and I may just spec in the OmniFET. I think I will use this

60V 10A version which is only about $1.00:
formatting link

The VNP14NV04 is about $2. The VNL5050N3TR-3 is another possibility with

19A and 41V in a SOT-223 package and about $1.
formatting link

But I'd really rather not have to depend on a self-protected MOSFET, since it is rather special and could easily become obsolete. So I do need to make sure I understand what's going on and deal with it properly. The protected device could then just be an extra layer of protection.

Paul

Reply to
P E Schoen

On a sunny day (Thu, 24 Oct 2013 03:21:18 -0400) it happened "P E Schoen" wrote in :

I had oscillation problems at about 20 MHz with 400 V power MOSFETs switching light bulbs. The cure was a 2.2 nF capacitor from drain to source. The other thing to watch is your gate drive, perhaps use a series resistor. The third thing is in my case I needed some sort of current limiter setup, basically a source resistor driving a NPN open that shorted the gate:

| load |------- d | -- R ------------g === | s | 2n2 c | /// b---R ---- | e Rs | | /// /// peak current limiter

The MOSFET was on a heatsink.

Not sure this is the solution for your case, but it worked in my case.

Reply to
Jan Panteltje

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.