MOSFET not working with digital input gate signal, Help please

[Tesladude]

Ok, so i have a picaxe microcontroller putting out a pwm signal to the gate of an irlb3034 "logic level" n-channel mosfet.

The mosfet is for grounding on and off about 8v of 20 to 40 amps.

When i take my circuit and jump 8v directly to the gate of my mosfet everything works perfectly and the circuit works great. But when i use the digital 5v signal (even at 25% duty cycle) from my picaxe the mosfet transistor gets VERY HOT and i have already blown a few of them although i do read the pwm signal on my scope from the load which means it is technically working.
Any ideas??

 

[Tom.G]

Check the pulse rise and fall times on the gate. They are probably rather slow, keeping the FET in the linear region for extended periods. The gate acts as a capacitor of 1 to 10nF, thus requiring signifcant drive current for fast switching.

Also, what is the pulse rate and duty cycle? Or equivalently, ON time and OFF time.

Can you post a schematic, at least the FET gate drive and the load?

 

[AlienOMG]

You may need a proper resistor to lower the current to the gate as the fet should be only voltage sensitive. Heat implies current where it shouldn't be. 1k ohm maybe 5k? Jim

 

[davenn]

Ok, so i have a picaxe microcontroller putting out a pwm signal to the gate of an irlb3034 "logic level" n-channel mosfet.

The mosfet is for grounding on and off about 8v of 20 to 40 amps.

showing us your circuit would be very helpful and would stop people playing the 20question game till we figure out what you are doing

 

[Tesladude]

As for a schematic is there a program that you guys use? I agree it would benefit everyone especially me. I normally just draw mine but this is important to me and ill give you whatever is needed.My scope can't seem to latch onto the frequency but every time i turn it on i can hear almost exactly 4khz coming from my circuity and i have it set at roughly 50% duty as of right now. I measured the voltage directly on the gate pin and it hits exactly 5.0vI did put a 220 ohm resistor between signal and the fet gate, as i have seen used in other schematics, showing no improvment but ill probably leave it thereHonetly it seems as if the fet is acting like the signal voltage is bellow threshold, and yet the datasheet says threshold voltage is well bellow 5v, because oddly enough 8vdc works perfectly fine for switching it on.

 

[davenn]

As for a schematic is there a program that you guys use? I agree it would benefit everyone especially me. I normally just draw mine but this is important to me and ill give you whatever is needed.

just draw a tidy image in MS paint or what ever your computer has , save as a GIF and upload it.
keep the longest dimension to a max of ~ 800 pixels

 

[CWatters]

If the gate voltage looks ok what about the waveform on the drain?

 

[Tesladude]

I didnt get to it today but I am still here!

 

[Svein]

Hm. The MOSFET gate capacitance is 10nF. I expect the driving gate to be a CMOS output, which usually has an impedance of about 1kΩ in the "on" state. The RC time constant will then be about 10μs. Assume 3 time constants to get close to 5V: 30μs. You are switching at 4kHz, which means a period of 250μs. Due to the time constants, you will be in the linear region for about 60μs out of 250μs or 24% of the time. So:

• Either insert a low impedance driver in front of the MOSFET
• or lower the switching frequency drastically

 

[CWatters]

Perhaps also check the output isn't in open drain mode?

 

[jim hardy]

Find the datasheet for your mosfet.
It should have a graph that looks something like this: (annotations mine)

https://www.infineon.com/dgdl/mosfet.pdf?fileId=5546d462533600a4015357444e913f4f

Then measure drain voltage when it's conducting. You need to be on the right hand side of that 5 gate volts curve. Else it'll get hot.

Could be you need a little more gate voltage. You said 8 works fine.

 

[Tesladude]

I will admit I am relatively new to fets and although I have read the datasheet there may be somthing you guys see that I don't.
In terms of the frequency, I will definitely try lowering the frequency next chance I get to make up for that gate capacitance and hopfully improve the overheating. any possible way to make it quieter? :)

Find the datasheet for your mosfet.
It should have a graph that looks something like this: (annotations mine)
View attachment 209364
https://www.infineon.com/dgdl/mosfet.pdf?fileId=5546d462533600a4015357444e913f4f

Then measure drain voltage when it's conducting. You need to be on the right hand side of that 5 gate volts curve. Else it'll get hot.

Could be you need a little more gate voltage. You said 8 works fine.

As I said before I am a little new, but if I am understanding you correctly then that is more than likely the problem. I can not seem to find a similar graph on my datasheet (you may have better luck) but If the graph you supplied is anything like the mosfet I am using then i think that might be the issue because I am trying to pass 40A with only a 5v gate input.
But sadly picaxe microcontrollers can not run off of 8v. Npn transistor and p-channel mosfet maybe?

Here is a link to the fet that I am using:
http://html.alldatasheet.com/html-pdf/340846/IRF/IRLB3034PBF/186/3/IRLB3034PBF.html

 

[CWatters]

Then measure drain voltage when it's conducting. You need to be on the right hand side of that 5 gate volts curve. Else it'll get hot.

It's late here so I might be wrong but I think you mean it should be on the left hand side of the chart when conducting (eg Low drain-source voltages).

I can not seem to find a similar graph on my datasheet

I think it's Fig 1 (25C) and 2 (75C) on page 2 (They have a log scale so looks a bit different to the one Jim posted).

Fig 1 says at 25C, with a 4.5V gate voltage and 30A drain current the drain voltage will be <0.1V so power dissipation is going to be < 3W. Pretty much same at same numbers at 75C.

That's also consistent with an Rds_on of 2 milli Ohm at Vgs = 4.5V

0.002 * 30A = 0.06V
0.06V * 30A = 1.8W

+1 to what Svein said about the frequency.

 

[jim hardy]

EDIT i see @CWatters has already solved it too
...................
I HATE that "alldatasheet" site that interjects itself between you and information. No value added just sheer frustration.

Always go to the source, manufacturer's site.

https://www.infineon.com/dgdl/irlb3034pbf.pdf?fileId=5546d462533600a40153566027b22585

If i got the right mosfet i'd say my hypothesized trouble is not what's wrong. That's one burly transistor !

Make sure yours is genuine not an imported counterfeit. Does it have an IR logo ?

Now

going back over the thread i see @Svein has already figured out the next step

Either insert a low impedance driver in front of the MOSFET

ie drive the gate harder

a 555 timer IC makes a stout buffer. Tie TRIG and THR together, use them as input. Output will drive that mosfet gate. But it inverts - so you'll have to change the picaxe output state to accomdate inversion.
You can power the 555 from 8 volts , tie its CONTROL pin to your +5 supply . That way it'll follow the pICAXE's logic level output and give you good 8 volt drive to MOSFET. Tie its RESET pin to its power pin to keep it out of mischief.
Be sure to place a few tens of microfarads adjacent the 555's power pin, they're noisy beasts.

There exist many buffers. Surely others here can suggest some. 555 is so universally available and cheap that i just jumped on it. Its output structure will interface almost anything.

Good Luck - this is how we learn !

old jim

 

[Baluncore]

Once you have a low impedance gate drive, with sufficient current to quickly transition the capacitive gate, you may find that the mosfet still gets hot. You then need a series gate resistor, Rg, to stop ultrasonic gate oscillation. (See test circuits at end of data sheet). Install Rg to raise gate resistance sufficiently to limit gate drive current and prevent oscillation. Compute Rg from about half gate drive voltage divided by the current drive capability. I would expect Rg to be between 2R2 and 12 ohms.

 

[sophiecentaur]

As I remember, to cut off a mosfet, you need the gate to be negative with respect to the source. That circuit shown no components of achieve this (bias).
Am I raving mad or just out of date?

 

[Baluncore]

Am I raving mad or just out of date?

You are out of date. Depletion mode JFETs, now Enhancement mode MOSFETs.

 

[sophiecentaur]

These 'ere new fangled things!

 

[Tesladude]

Ok!
So i went ahead and changed the frequency in my program, picaxe doesn't really list a way to know what frequency you are actually getting out of the 0-255 variable so I took it from 255 down to 50, (I think it is something like 17hz,) and although it did not completely fix the problem as far as I can tell it has improved slightly and I can no longer hear that nasty 4khz.

So how about this:
I leave my program at that low frequency.
Install a 555 buffer to get my pwm up to 8v.
And switch my n-channel mosfet with the p-channel version to counter act the 180 degree flip caused by the 555.

 

[Baluncore]

And switch my n-channel mosfet with the p-channel version to counter act the 180 degree flip caused by the 555.

You could just change the program to invert the output signal from the picaxe.

 

[Tesladude]

So after taking a look at my program i realized it was simpler to do then I thought.
So I was able to take my word variable holding the 1024 bits of the pwmduty and before It actually fires I wrote "let w11=1024-w11" and changed w11=0 to w11=1024 wherever the mosfet is supposed to shut off.
The simple 555 buffer apears to have worked as well
I am going to give it an official go tomorrow if I can and hopefully I don't run into any problems.

As I said I am here to learn so if anyone was kind enough to correct me on any of my knowledge hopefully I will be ok on my own in the future:

1: If I understand correctly this mosfet is able to switch with a digital signal, but according to the graph switching such a high current caused a relatively high voltage drop (Vds= voltage measured across drain and source?) due to the internal resistance of the mosfet which of course came off as heat.

2: The small amount of capacitance on the gate was not able to charge to 100% of the 5v I was trying to achieve due to a pulse width that was not greater than the capacitor's time time it took to charge completely CxR=(63% charge time in seconds)

 

[Baluncore]

With that said, will putting the p-channel version of my mosfet cause any issues that may be going over my head?

Usually a high side driver is referenced to the +ve power rail, so level shifting is needed. For low voltage supplies the NE555 can do that level shifting, but it inverts the signal.

It costs less, is cooler or more efficient to switch a high current with N-channel mosfets, so you could avoid a P-channel mosfet by using a non-inverting complementary follower in place of the NE555. See the circuit attached.

Rpu is a pull-up resistor, it may be needed to make the output approach the positive supply.
Rg is there to stop oscillation during transition, the peak gate current may be limited by the transistor current gain.
Transistors are selected for a high collector current, over 500 mA.
Output voltage is slightly reduced by ±Vbe, but that is not usually a problem with logic level mosfets.

 

[CWatters]

1: If I understand correctly this mosfet is able to switch with a digital signal, but according to the graph switching such a high current caused a relatively high voltage drop (Vds= voltage measured across drain and source?) due to the internal resistance of the mosfet which of course came off as heat.

That's not my interpretation of the data sheet. See my earlier post..

Fig 1 says at 25C, with a 4.5V gate voltage and 30A drain current the drain voltage will be <0.1V so power dissipation is going to be < 3W. Pretty much same at same numbers at 75C.

That's also consistent with an Rdson of 2 milli Ohm at Vgs = 4.5V

0.002 * 30A = 0.06V
0.06V * 30A = 1.8W

 

[jim hardy]

but according to the graph switching such a high current caused a relatively high voltage drop (Vds= voltage measured across drain and source?) due to the internal resistance of the mosfet which of course came off as heat.

Datasheets are wonderful things !
The graph tells me you will have less than 0.15 volts across your MOSFET if its gate gets to 4.5 volts.

At 40 amps that's six watts, so you WILL need a small to medium sized heatsink.
Junction to Ambient tells how hot it will get with no heatsink, degreesC rise per watt.

Six watts is 372 degrees with no heatsink.
(Did we all miss that ?)
You'll want a heatsink rated less than 10 degrees per watt . Less R_Thermal is better

 

[CWatters]

(Did we all miss that ?)

Hands up I missed it. Particularly the R_JA of 62C/W

 

[jim hardy]

Hands up I missed it.

Me Too !
Hindsight is always 20/20.