Why do I keep frying MOSFET drivers in this motor circuit?

Go To Last Post
92 posts / 0 new

Pages

Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi, Hoping someone can point out why this circuit seems to keep frying the 4427 driver chips. I thought I was killing the MOSFET, but after a closer look they seem to be okay, but the 4427 gets hot and stops working. the circuit seems to work if I measure the voltage at the motor connectors (motor disconnected) with an analog meter - reads 0 to 12 VDC as PWM duty cycle goes up. But if I attach the motor the 4427 burns up. Any help would be greatly appreciated -Tim

Attachment(s): 

Last Edited: Fri. Apr 10, 2015 - 03:08 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Never used that chip so....is it supposed to run from 12V? Do you need some small resistance in series with the Gate?

edit: so I have a Micrel data book for the chip :roll: and it works up to 18V. One thing it says is to ground the unused driver, the input only I presume.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This is a complete guess, based upon the posting by js, but on page 9 of the Data Sheet found at http://ww1.microchip.com/downloads/en/DeviceDoc/21422D.pdf it says:

Quote:
3.4 Supply Input (VDD)
The VDD input is the bias supply for the MOSFET driver
and is rated for 4.5V to 18V with respect to the ground
pin. The VDD input should be bypassed with local
ceramic capacitors. The value of these capacitors
should be chosen based on the capacitive load that is
being driven.
A value of 1.0 μF is suggested.

(Highlights mine own)

Regards,
Steve
(I really am clueless and guessing...)

  • "Give me six hours to chop down a tree and I will spend the first four sharpening the axe."  -- Abraham Lincoln
  • "All right wise guy, where am I?"   -- Daffy Duck
  • "Well, we're safe for now. Thank goodness we're in a bowling alley."  -- Big Bob, Pleasantville
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

By the way where did you get the IRF5303? I have a IRF book but can't find it there. It seems to be the P type from google however have you wired it the right way aroud?

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the help!

I got the MOSFET from digikey, but I just checked my order summary and it says IRF5305 (note last digit is 5). Sorry for the confusion. I read the number from the part when I made the schematic above. I assume I read it wrong, but I'll check when I get home and double check the wiring. I believe the same ciruit except for the diode has worked fine on a breadboard with smaller motors. This motor's "datasheet" is here The IRF5305 datasheet is here It's P type, 31A, 55V

RE resistance at the gate: I got the basic (not exact) idea for this circuit from a robotics book that used gate resistors with MOSFETS in parallel and I think when driving the gate with a transistor, but not with the 4427. If you think it's worth a try (or can't hurt) I'll try it.

RE capacitors bypasing VDD: Yes, those were included in the circuit I based this on, but the explaination made it sound like they were optimizations, so out of laziness, I left them out. Think that could be the problem? I'll kick myself if it is!

I'll also ground the unused driver (input) to cover my bases.

I bought 4 of the 4427's, and I'm down to my last one. I'll order more ASAP. But in the meantime, does any of this seem very likely to solve the problem? I hesistate to burn the last one quickly with the wrong test.

Thanks again!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

BTW, how many amps can one usually run through a solderless breadboard? This motor draws 2-20 Amps. It's under fairly light load (spinning a flywheel), so except for startup, it's probably on the lower end of that range. I've been afraid to build the circuit on a solderless breadboard, but soldering these experimental circuits together has been slow and painful.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Maybe you need to run the driver chip from 5V and use a pullup on the gate to 12V?? Does it need 12V on the in to turn it on?? (AVR wont give it 12V!)

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What frequency is the PWM running at? You really need to look with a scope at the various signals to see whats going on. Parasitic capacitances are injecting charges back into the gate, together with the stray inductances of the board spikes might be generated that kill the driver (there are two parasitic capacitances from the drain to the gate and from the source to the gate; when the FET turns on charge is injected back to the gate, delaying the turn on).

The gate of a MOSFET is high-impedance but that's only when the voltage applied is steady. A large MOSFET has significant capacitance, so large currents are needed to swing the gate around quickly. At high frequencies this can consume quite some power. If not enough current is available the gate voltage will not rise/fall quickly enough and the FET spends too much time in its linear region, dissipating power.

I would try a resistor of 10 to 100 ohms in series with the gate.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Without a really beefy capacitance on the 12V, there is all kinds of junk as a result of the motor.

In addition to 1uF to assure proper operation of the driver, I would add at least 100uF (and maybe even 1500uF since this is a 20A motor) on the 12V, to protect this circuit as well as anything else on the same 12V.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the additional help!

Bob, sorry if I misunderstood, but the MOSFET does need 12V (at least more than 5V) at the gate to turn off. I experimented with applying 5V at the gate, and it did not turn off much if at all. That's why I'm using the driver chip, to deliver the higher voltage to the MOSFET gate. Sorry if I'm stating the obvious - this is all new to me.

JayJay: It's an 8bit phase correct PWM signal on a chip running at 1MHz. So if my math is correct, the frequency is 1,000,000 / 512 = 1,953Hz. Your explaination makes me think I need to add the capacitor as Steve also suggested above. My limited knowledge doesn't make it obvious why a resistor in series with the gate would help, but I'll try that too. Do you think these things would burn up the 4427 driver chip or just make it more efficient? I ask because if it's a likely fix, I'll try it on my last 4427 chip (before more arrive), but I'd hate to burn that one up if there is something else I should try first or at the same time. Unfortunately, I don't have a scope.

PS: I just saw mneary's comment added as I was Previewing this post. Another vote for adding capacitors. Okay, I'll cross my fingers and try it out. (may not have time tonight :( )

Thanks again!!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You need high-side driver or N-chanel Mosfet.
Alexander.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks Alexander, but can you give me a few more words to explain? What's a "high-side driver"? I do have N-channel MOSFETs, so I could set up the analygous circuit, but shouldn't this work with P-channel?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The circuit is basically correct IF the gate and drain (pins 1 and 2) are wired up correctly. If they are reversed then the motor will turn the mosfet fully on, putting a full 12V onto the drain which would be connected to the driver chip which in turn would try to pulse it low and..................... :( :(

edit: try it out without the motor to start off with, put say a 100R, 5W, in place of it and perhaps a led in parallel (with suitable resistor in series) so you can see the thing working.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The p channel fets are like pnp transistors... you pull the gate up to vcc to turn it off, you pull it low to turn it on. The part that confuses me is the avr is giving 0-5v to the driver chip input which is running from 12v... does the output look like its switching full 12v ok?

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Bob that's what the driver does, it converts 5v from the input to whatever the chip is running at (up to 18V). That particular chip has also the advantage of being able to deliver 1.5A peak to the load.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I checked (and double/triple/etc. checked) the circuit. It is an IRF5305, and it looks to be wired correctly. So far suggestions are a few capacitors, grounding the unused input to the 4427 and a resistor in series with the gate. Just to make sure I have it correct, I've placed the capacitors in the revised schematic below - look correct? The rest is straight forward enough. I still don't understand the series resistor on the gate, but unless anyone thinks it's a bad idea, I'll try it too.

Any other suggestions?

Thanks,
Tim

Attachment(s): 

Last Edited: Sat. Aug 18, 2007 - 07:35 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you put 100R in series with the gate then the current will be limited to about 120mA which should keep your driver safe. I always start with a current limited power supply when I do something like that.

Quote:
It is an IRF5305, and it looks to be wired correctly
So your output to the motor is wired up to pin 2 (centre pin or the tab), the driver chip's output is going to the gate (pin 1 or l/h side) and 12V goes to pin 3 (r/h side this is looking at the front, plastic part of the fet)

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Rehashing comments by jayjay1974 and js.

The 4427 driver current paths are:

(1) Between Vcc and GND of the driver - transition dissipation.

(2) Between OUTA and alternating Vcc/GND - MOSFET gate capacitance charging and discharging.

Adding a resistor between driver output and MOSFET gate will move some of the power dissipation from the driver to the resistor when charging and discharging the MOSFET gate capacitance of 1200pF. The downside is increased power dissipation at the MOSFET, it's designed for that.

The PWM rate or frequency is directly related to how often the MOSFET gate capacitance is charged and discharged by the driver. If 1 kHz works, don't use 100 kHz.

Stan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If I may...

I wonder whether the OP realises the fact that with the combination of driver and FET as specified the PWM output needs to be INVERTED since a m inimum outptu on the PWM will cause a MAXIMUM MOTOR CURRENT and vice versa...

Unless the driver chip incorporates an inverter stage ( have not checked the data sheet.. not in the spirit of giving advice...)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Don't forget grounding the unused input.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Unless the driver chip incorporates an inverter stage
The 4427 is NOT inverting but the 4426 IS inverting.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here some other tips that might be helpfull:

I think the driver is fried due to routing of the power-lines.
Tim, you're working with very high currents. Keep those away from the driver and the Mega16

In the sketch a suggestion. I forgot to draw the schottky (but scanned my artwork already :wink: ): place that directly across the motor-connections.

A few other things:
- Breadboards are not made for high currents. And defenitely not the currents that can flow in this circuit.
- A series-resistor between driver and fet: IMO not necessary; the driver is made for this purpose
- The powersupply for the motor: use adequate filtering, the Cm is the minimal filter
- Check the datasheet for C1 and C2
- Although AVR's are quite forgiving when abused .... I would suggest a 470ohms series resistor between AVR-output and driver4427; it will have hardly any influence on the signal, but it protects the AVR in case something goes wrong.

A picture of your set-up would help to determine the frying-cause.
What is the motor-power-supply? Batteries ?
And how gets the AVR its power?

Plons

A GIF is worth a thousend words   She is called Rosa, lives at Mint17.3 https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks, thanks, and thanks!

Quote:
So your output to the motor is wired up to pin 2 (centre pin or the tab), the driver chip's output is going to the gate (pin 1 or l/h side) and 12V goes to pin 3 (r/h side this is looking at the front, plastic part of the fet)

Yes, if I'm remembering correctly from checking last night. I'll check once more tonight when I get home.

Quote:
I wonder whether the OP realizes the fact that with the combination of driver and FET as specified the PWM output needs to be INVERTED since a minimum output on the PWM will cause a MAXIMUM MOTOR CURRENT and vice versa...

Thanks for pointing it out. I did discover that, but it was a surprise. As a result I've compensated in SW by using a high duty cycle for slow and low duty cycle for fast. Let me know if there's something wrong with that approach.

Plons:
Thanks for the effort with the drawing! It has taken me awhile to do mine in PowerPoint. I need to learn to use a real schematic drawing application, but perhaps I should have done it by hand for now.

Does Cm = 1000uF?

What's the part at the top above the motor (looks like a resistor) labled "R or L"

Yes, the power for everything starts with a 12v battery. The AVR get's it's power through a 5v voltage regulator, which does have a couple capacitors per the data sheet.

Is part of what you are teaching me that routing is important? I have to take a closer look over lunch (have to get to work now), but at first glance, I suspect your circuit and mine have similar connections if you ignore relative placement, what's near/far to what, etc.

I'd be embarassed to show a picture of this point-to-point soldering mess, and it would be hard to interpret as the wires and components are on opposite sides. But I could doctor up a picture of the component side to show where the wires run, or I just might draw it to show placement and routing.

Thanks again,
Tim

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tim, no need for embarassement ... you are amongst fellow-"artists" ... 8) a picture is IMO essential. And yes, routing of the wires is crucial.
Pitty that the datasheet does not tell the DC-resistance of the motor ... maybe you can tell us.

Quote:
Does Cm = 1000uF?
At least, it's a start. Add some ceramics of 100nF as well. I hope other members can do a more precise recommendation.
Quote:
What's the part at the top above the motor (looks like a resistor) labled "R or L"
It's an attempt to draw a european resistor ... :wink: A 1 Ohm resistor is OK, if you add the 1uF and 0.1uF as pointed out in the datasheet. But if you have a 100uH inductor, that is even better. The "R or L" create in combination with C1 and C2 a filter.
Quote:
....I suspect your circuit and mine have similar connections if you ignore relative placement
Indeed. My goal was to show the routing.

I adapted the sketch a bit to the now-known facts. And replaced the euro-resistors ...

Main things are:
1. Keep motor, fet, driver and associated filter {R1 C1 C2} close together. Cm as well, if possible.
2. Starpoint == motor-minus
3. Fat wires in the sketch should be fat wires in the real world. I am not so familiar with AWG, but 2.5 mm2 for the fat ones.

How do you like my kinda circular fet ? Neat huh ?

R2: I put in 1 ohm as default, but I don't know what the load is on the 5Volts, An inductor, if available, would be better.

Quite a motor, btw. At full load 20+ amps ... wouldn't it be better to use two fets, and both drivercircuits in the package?
And how about EMI/EMC ? Anyone tips ?
And I wonder if a low-side drive (using N-ch fet) and the 4427 on 5 V power would be preferable?

Plons

A GIF is worth a thousend words   She is called Rosa, lives at Mint17.3 https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Some other things that popped up: John (Samperi) already suggested to use a current-limited supply in test-phase: good advice, strongly recommended :!:

Add a fuse in Battery + when using battery

Use a small motor in test-phase, not the monster :wink: and before putting in your last driver: take a 100R resistor and a multimeter on 200 mA range; ground the gate of the fet with the 100R and MM in series, and see if the gate-current = zero, and the small testmotor is running @ full speed. Next, take the ground-side of 100R / MM to the +12V, and the motor should stop.

One more suggestion: do a search on this forum for Robotics or so .... There are a few guys that are more familiar with this than I am.

Plons

A GIF is worth a thousend words   She is called Rosa, lives at Mint17.3 https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks again for all the advice. I can post a picture, since I perhaps owe it to all those who took the time to help. That might confirm suspicions. But, it looks like I should start over with a new circuit

I'm wondering what I can use for a current limited supply. Would a battery plus a fuse qualify? I have some salvaged computer power supplies with 12VDC outputs, but I'm not sure if they are limited or will just overheat.

I do have a few small 12 volt motors, which I used when developing a similar circuit on a solderless breadboard, but I'll start testing with them again on these.

I like the multimeter/resistor test of the fet gate idea!

This circuit was meant to be a quick one I could use for testing and experimenting with the motor. The real circuit for my application (ball throwing machine) was also meant to have a break mode (with another fet that would result in both motor leads connected to ground - essentially a half-bridge I believe). Now I'm wondering if I shouldn't build that circuit if I need to start over anyway. On the other hand there's the walk before you run strategy. Perhaps I can design the whole circuit, but only build the existing part first and get it working before adding the components for breaking.

Thanks again!!!!!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

forgot to respond:

Quote:
Pitty that the datasheet does not tell the DC-resistance of the motor ... maybe you can tell us

I guess that's the resistance of the motor when standing still. I tried measuring that, but couldn't seem to get a reading. Then I reasoned that it has to be less than 1 ohm if the motor can draw 20 amps when stalled at 12 VDC. (which is about what I think the datasheet shows)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
.... some salvaged computer power supplies with 12VDC outputs, but I'm not sure if they are limited or will just overheat.

They can be used, but there is something involved: they don't like to operate without a decent load on 5V and 3.3V output. And current-limiting isn't their best property.

Quote:
Would a battery plus a fuse qualify?
Well, ... yes, if you have enough spare fuses ... :D

Do you have a LM317T ? Put that in series with battery and your circuit; connect the control-pin to the output-pin. Now it will act as a current-limiter of appr. 1.5Amp. And I think that I speak for all LM317's when I say: a heatsink would be appreciated :wink:

The resistance of the motor will be even less than 1 ohm. The 20A are when it's running under load ... umpff, I would go for at least 2 fets. But maybe I am too cautious ...

Plons

A GIF is worth a thousend words   She is called Rosa, lives at Mint17.3 https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well, I just added a few LM317T's to my order along with some more mosfet drivers. Seems I have my plans laid out. Now just need to get to it! I'll update with progress, but this might take several days. Thanks again.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I built a regulated supply (hmmm that would be about 20+ years ago...how time flies :) ) that allows me to adjust the voltage up to 80V and the current from a few mA to 3A, you should be able to buy something similar off the shelf pretty cheaply, I do see some supplies that would go up to 30V and have both voltage and current ajustment. This is something that ALL workshops need next to a multimeter :)
hmmm don't know why I'm no longer getting any notification from any threads...is it the forum or is it my ISP AGAIN?!!!

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'll have to check into a regulated supply. This hobby already seems expensive. All the odds and ends I added to yesterday's digikey order was over $100.

Plons thought pictures of the original set up would be helpful, so I'll suppress my embarassment at sloppy soldering and show you what I had put together. Seems like you'll find the routing suspect. In particular I wonder if sharing the ground between the motor and the driver chip before the schottky was a bad idea. The solder to the fet is especially messy (but no shorts I promise) because I twice removed one and soldered in a new one before I discovered that it is the driver chip that's going bad. There are 3 pictures below. The first is the front, the second is the back flipped side to side so it can be overlayed with the front, and the third is the back with the components from the front overlayed so you can see the whole thing.

EDIT: This site is giving me an error when I try to upload a picture, so I'll have to try again later.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

timwhunt wrote:
I'm wondering what I can use for a current limited supply.

Tim,

Check out the Texas Instruments LM317 adjustable voltage regulator datasheet.

Page 8, figure 4 shows an adjustable current source and, page 10, figure 9 shows a fixed current source.

The equation is ~1.2/R

There is also a 5 ampere version of the LM317 but, the number escapes me at the moment.

At least this will keep the current flowing throough the FETs to an ampere or less.

Note: I tryed to attach the datasheet but, for some reason, the system would not allow me to.

Hope this helps.

Edit:

Tim,

Here is the link to the TI LM317 datasheet, in PDF format.

http://focus.ti.com/docs/prod/fo...

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks Carl! That LM317 sure looks useful. I think I should have ordered more of them!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In a world of electronics where the life of most stuff is a year or two, a bench power supply and a meter and a scope should all last 20 years. Durable goods. Better than a car.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well the forum seems to have gone to the dogs again!! I haven't received any topic reply notification on any thread for about 2 days now :( and no more pics either :(

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For what it's worth, I'm getting topic reply notifications, but the pics issue is unfortunate.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

js wrote:
Well the forum seems to have gone to the dogs again!! I haven't received any topic reply notification on any thread for about 2 days now :( and no more pics either :(

I did notice last night that the forum was down for several hours. Since then, I've recieved no noticifications and I can't download pictures. In addition, I tried to attach the LM317 datasheet for Tim but, the system would not let me - no matter how many times I tried.

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Since then, I've recieved no noticifications

hmmm Tim is getting notifications and we are not....are they discriminating against old geisers? :lol:

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

js wrote:
Quote:
Since then, I've recieved no noticifications

hmmm Tim is getting notifications and we are not....are they discriminating against old geisers? :lol:

Might be. But I don't think international law covers that.

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As far as the current source is concerned.. you dont need an active one.. ( no silicone.. it gives up too easily).

Use a light globe ..apropriately dimensioned.. it will act as a self limiting current source ( temp coefficient of the filament resistance is strongly positive )

PS... old geisers never die.. they just stop spouting

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Good idea, Igno ...

Car's light-bulbs are ideal for that purpose: a whole range to choose from. With Tim's 12 battery: perfect match.

Plons

A GIF is worth a thousend words   She is called Rosa, lives at Mint17.3 https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Some time ago I used Harris CPA 7667 as MOSFET drivers
and found them a little susceptible to LATCHUP.
This usually happened in a quite erratic way. Usually without load no problems, but with load sometimes even touching the gate of the MOSFET with a probe the
running system crashed and the driver was destroyed.

I never really found the cause. Perhaps something similar happens in your circuit. A resistor
between driver and gate may help.

If your PWM operates at a few kHz a simple Logic-level
translator 5V ->12 Volt may be sufficent as MOSFET driver.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the suggestions Igno and ossi!

I got tired of waiting for the forum to accept pics again, so I'll temporarily link to them on another site. I had intended to add them to a previous post, but since that's gotten a bit old, I'll put them here. Here's the text from the earlier post that was meant to go with them:

Plons thought pictures of the original set up would be helpful, so I'll suppress my embarassment at sloppy soldering and show you what I had put together. Seems like you'll find the routing suspect. The solder to the fet is especially messy (but no shorts I promise) because I twice removed one and soldered in a new one before I discovered that it is the driver chip that's going bad. There are 3 pictures below. The first is the front, the second is the back flipped side to side so it can be overlayed with the front, and the third is the back with the components from the front overlayed so you can see the whole thing.

Revised post to put pictures here.

Attachment(s): 

Last Edited: Sat. Aug 18, 2007 - 07:40 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ossi mentioned latchup, which can be a real problem.
Another very serious situation is when you have
a special function ic intended to handle higher
voltages and/or currents. A *rapid* application
of supply voltage can cause a sort of latchup
condition that can cause *spectacular* failures.
One needs a filter cap on the supply bus to limit
how *fast* the voltage can rise when the power is
switched on. I notice your board does not have
one. I believe the problem is that when power
instantaneously appears, the various internal
transistors etc. cannot respond quickly enough to
set up properly, so BOOM! Been there, done it, got
the t-shirt.

Tom Pappano
Tulsa, Oklahoma

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The PWM return isn't shown, therefore it probably has a common current path with Motor-.

From the PWM, try running (PWM-) directly from the microcontroller to the driver chip.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tom, thanks for the analysis and suggestion. Sounds like it could be very important. I'm sorry to have to ask, but can you tell me more specifically where I need to connect the cap, and about what size you'd recommend?

Thanks mneary. I'll try that too. (I believe that would be running the ground from the uC to the ground for the driver chip instead of sharing the motor ground with the driver chip as I've currently done)

I sure seem to have done a lot of things wrong with such a seemingly simple circuit!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

timwhunt wrote:
I sure seem to have done a lot of things wrong with such a seemingly simple circuit!

Tim,

Believe me, most of us have had many failures while in route to our successes. You are learning from those failures! Nothing comes easy - for me, at least...

Hang in there!!!

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You would have had great success with something like an opamp filter or something that doesnt deal with hi current and hi speed square waves. You just skipped the easy stuff and went right for the problems. Keep on debugging!

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
...can you tell me more specifically where I need to connect the cap, and about what size you'd recommend?

I would place maybe a 500 to 1000 uf cap right on
your board where +batt and -batt are connected.
That should help, if indeed it was a dv/dt problem,
and I just think it is a good idea to include one in
any case.

Tom Pappano
Tulsa, Oklahoma

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I sure seem to have done a lot of things wrong with such a seemingly simple circuit!

Not true. The high amp motor controllers go for big bucks. So it must not be so easy.

FYI there is a short article on Mosfets in the May issue of Nuts and Volts. There is a rebuttal on this article in the reader comments in the July issue, along with a heatsink article, which will be useful at sometime.

Interested in your outcome with the mosfet driver chip. Tried doing this with an analog input, and ended up smoking inverter outputs and fets both. So probably was not getting a full on voltage to the mosfets, they can get REALLY HOT!

Keep up the good work.

EDIT: Meant to say that I tried a mosfet h-bridge, but without the mosfet driver chip. The con-figuration was solar cell to op-amp to logic gate(s) to opto-isolator to inverter to diode logic to mosfets. No wonder this project has been shelved.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As several others have suggested, the problem with your circuit is the lack of filtering capacitors.

When a high inductance load such as a motor is switched, the voltage at the output (in this case, the MOSFET drain) very easily over- and undershoots. This would cause severe spikes if it weren't for the free-wheeling diodes in the circuit - the Schottky and the anti-parallel diode integrated in the 5305. These diodes will clamp the output to the DC line levels, but without filtering capacitors the DC line will also over-/ undershoot due to circuit inductance and battery ESR.

I'd add an electrolyte of maybe 1 mF to the battery poles, a ceramic of 220 nF between the Schottky's anode and the MOSFET source, and a 470 nF between VDD and GND on the 4427.

Another thing: Do you have a heatsink for the MOSFET? With an on resistance of 0.06 ohm (from the datasheet) and a current of 20 A, the dissipation will be 24 W even if switching losses are ignored. Without a heatsink the MOSFET won't survive even 3 W.

/ Grimmy

Pages