Mosfet driver ic v supply query

[BrendanP]

Im waiting for some humidity sensors to arrive so I can't do any more on the incubator for a while.

I want to do some work on my 12V dc motor driver. Regulars will recall the last time I asked about mosfets so it is with some trepidation that I step into the arena once again.

The state of my understanding thus far is as follows:

mosfets are superior to biploars for pwming motors.

a mosfet driver ic should be used with the mosfet

power mosfets require the v on their gate to be at least 10v higher than their source v to fully turn on.

I am using a IRF3710
http://www.datasheetcatalog.org/datasheet/irf/irf3710.pdf

in conjunction with a TC4420 driver.
http://ww1.microchip.com/downloads/en/DeviceDoc/21419c.pdf

to low side pwm a GM (R.I.P) automotive windscreen wiper motor.

I have connected the parts as per the The Power MOSFET Interfacing Circuit in the 3rd picaxe manual. That is, source to ground and drain to the low side of the motor. Correct me if this is wrong.

The gate of the mosfet is connected to the output (pin 7) of the driver ic.

I note from the data sheet that the TC4420 driver can run on a Wide Input Supply Voltage Operating Range:
- 4.5V to 18V.

Question.

Do I need to run the driver on a voltage at least 10v higher than the voltage I am seeing on the low side of motor? Or does the driver ic generate this higher voltage on its output pin by itself?

I ask because at the moment Im only seeing the same v on the output as I have on the supply. The mosfet is getting warm switching only smallish current. I thought this indicated that mosfet is not fully opening because of insufficient v on its gate and is thus generating heat because of its high resistance.

I have found this article very helpful.

http://homepages.which.net/~paul.hills/SpeedControl/MosfetBody.html

 

[ylp88]

power mosfets require the v on their gate to be at least 10v higher than their source v to fully turn on.

10V sounds about right if you are using high currents. With a Vt of up to 4V for your FET, you'll still have an inversion voltage of 6V at the drain. With an Rds of <1 Ohm (at high operating temperatures) and a few amps current, you'll still have a bit of room to move...

Do I need to run the driver on a voltage at least 10v higher than the voltage I am seeing on the low side of motor? Or does the driver ic generate this higher voltage on its output pin by itself?

Page 3 of the driver datasheet gives you the answer. The spec provides an output high voltage to drive your FET gate as VDD&#8211;0.025, thus you are only going to get an output as high (almost) as your supply.

ylp88

 

[BrendanP]

Thanks ylp88.

I have done some breadboarding and have established this to be correct. Im running the driver on 13V and everything appears to be working well.

I was getting audible ringing which emanates from the mosfet I guess. I tried using 20khz pwm freq as per the advice here
http://homepages.which.net/~paul.hills/SpeedControl/MosfetBody.html
and that cured the problem.

As far as I understand it there should be no problem running the driver ic from the same supply as the dc motor?

 

[ylp88]

As far as I understand it there should be no problem running the driver ic from the same supply as the dc motor?

It will probably be fine, but just do the usual rounds of adding a flyback diode and some good supply decoupling.

ylp88

 

[Dippy]

I haven't looked at the Data Sheet for your MOSFET, but 10V is a common figure , just read the Data Sheet and look at the graphs.
Options inlcude Logic level MOSFETs of course, but if your high V supply is 13V then that sounds fine.
(Aside: Always be aware that there is a maximum voltage for this too.)

Experiment with the frequency, I've always found that much lower PWMs give better ooomph.

Read the driver Data Sheet to see the recommended size and type of the decoupling cap that it requires. This should be placed as close to the i.c. as physically possible - remember its switching largeish currents albeit briefly - this capacitor helps it to suck and blow.

"I thought this indicated that mosfet is not fully opening because of insufficient v on its gate and is thus generating heat because of its high resistance."
- yup that's often the case on this Forum

 

[ylp88]

Experiment with the frequency, I've always found that much lower PWMs give better ooomph.

But with a cycle time of ~50ns, we should make it realise its potential! Go forth and run it at 20MHz!!!

ylp88

 

[BrendanP]

Ahh long suffering Dippy, &quot;tech support of the last resort&quot;

Ahh long suffering Dippy I new yould show up soon to be further tormented.....

It is fortuitous that you observe
"...Experiment with the frequency, I've always found that much lower PWMs give better ooomph...." because that is exactly the problem now.

I had the motor under control of a off the shelf 'bot motor controller comd serially by a picaxe. With it I could stall the motor only with a fair application of force by grabbing the opening mechanism on the feeder by hand.

With the above parts and pwmout 2 , 49, 173 @ 20khz I can stop the mechanism with ease. I will try a different freq. in the morning as you suggest, bedtime now. I do need a reasonable amount of torque but at low speed.

I presume that raising the freq to 20khz stops the ringing noise because the human ear can no longer hear it at that freq?

 

[Dippy]

I'm suffering now as I've just had a huge lunch with the family... burp.

In my current condition "I presume that raising the freq to 20khz stops the ringing noise because the human ear can no longer hear it at that freq? " sounds fine to me.
No doubt some pedant will mention harmonics - but as my guts ache I have other torments to worry about

All I can say is 'try it out', obv motor dependent. Expreimenting is fun, never take people's suggestions as carved in stone. Playing is a good way to learn, unless you are floating outside the ISS.

 

[BeanieBots]

As Dippy says, experiment.
It can be a bit of a trade off.
High frequencies have the benefit of being less audible. Lower frequencies tend to give more torque at lower speeds.
However, don't go too low or you will get significant vibration which might not be obvious and cause harm.

Fairly recently I was asked to fault find a robotic arm which kept having position failures. These were very high spec (and expensive @ £75ea) servo pots. On some machines they would last for years, on others they would fail within a few days. The failures were quickly found to be open tracks right at the two common 'rest' positions. The pots were rated for >1e6 sweeps and failing after only a few hundred.

Cut a long story short, there were two guys who set up the robots.
One set the PWM frequency high so that it didn't whistle, the other set them low so that slow speed control was 'crisper'.
The low frequency caused a very small oscillation at the two holding positions. Although it was not noticeable to the eye or by touch, it was enough to cut a groove in the servo pots through mechanical wear.

 

[manie]

Every one has said "yes, 10V seems fine..." BUT the question was about " 10V HIGHER than the mosfet drain side voltage". Yes the datasheets ALL show 10V as a good "on" voltage but does it need to be 10V HIGHER ? Example:

I almost understand YLP88's answer, but not completely...

If the voltage at the motor low side (mosfet drain connection) is still say 4V, must the driver voltage on the gate then be 14V ? Or is 10V still sufficient to drive the mosfet FULLY ON ??

 

[Andrew Cowan]

@Manie. The voltage is Vgs - between the gate and source, not the drain and gate as you suggest.

Remember the source is the leg connected to the 0V side - the drain is connected to the motor.

A

 

[manie]

Andrew: Sure, you are correct off course. But please look at the original question, asking about the voltage arriving at the drain which is connected to the low side of the motor. Must the gate driving voltage be 10V higher than THAT voltage ? Or is a simple 10V on the gate (6V higher than the gate threshold V) sufficient to turn on the mosfet to its stated Rds-On resistance ?

 

[Andrew Cowan]

If the source is connected to 0V, then you just need 10V on the gate.

Regardless of what voltage is on the drain - that can be 60V if you want.

 

[manie]

Andrew: Thank You ! Now the question has been answered so clearly that even I can understand it. That was my understanding of the datasheets also.

But plainly, MOSFETS ARE A PAIN to everyone using them ! All mosfets seem to run extremely hot regardless of Drain-Source voltage, regardless of Gate driver voltage and regardless especially of mosfet's current capability. No amount of heatsinking seems to make a difference. Even forced air-cooling is meaningless.

My experience (albeit limited) indicates that lower freq's (slower switching) helps a bit. Lower duty cycles (below 50% duty) helps a bit. Some resistance in the current path (above or below the mosfet) helps a bit. Really LOW currents (less than 5A) helps a bit. But none of these REALLY helps MUCH ! The mosfet gets HOT ! The mosfet eventually leaks out ALL of its built-in smoke ! It then stops working altogether !

 

[ylp88]

If the source is connected to 0V, then you just need 10V on the gate.

Regardless of what voltage is on the drain - that can be 60V if you want.

Actually, if you have 60V at the drain then you're back running the MOSFET in saturated mode, rather than the desired triode/linear region. To the first order, you need to keep a sufficiently high enough voltage at the gate such that a channel exists all the way from the drain to the source.

If your source is grounded, then the voltage at the gate will always support a channel at then end of the MOSFET (Vgs is constant).

Passing a current through the MOSFET results in a voltage drop across the channel, that is, the voltage at the drain is non-zero and the channel starts to "taper" at that end (as the voltage, and thus the field, between the gate and the drain decreases). If the voltage at the drain becomes too large, then the channel completely tapers off at the drain end (often termed "pinch-off") resulting in the onset of saturation. Unwanted heat production follows.

A nice high voltage (but keep in mind the warnings Dippy mentioned about an upper limit) at the gate maintains a solid (read: low resistance, equals: lower power dissipation) channel all the way from the drain to the source.

Apologies if I lost you along the way: I've tried my best to keep it simple...

EDIT:

But plainly, MOSFETS ARE A PAIN to everyone using them ! All mosfets seem to run extremely hot regardless of Drain-Source voltage, regardless of Gate driver voltage and regardless especially of mosfet's current capability. No amount of heatsinking seems to make a difference. Even forced air-cooling is meaningless.

Remember that MOSFETs are just like any other device - basic power dissipation can be calculated. You should find that a high gate voltage will reduce the heat produced (as explained above, it should produce a better - low resistance - channel) while the drain-source voltage should be kept at a minimum (eg. by reducing the channel resistance = high Vg).

Power dissipation (to the first order, which is usually sufficient in high power applications) in the MOSFET can be calculated as P=Vds*Id. Again, reducing the drain-source voltage, Vds, will reduce power consumption. And simplistically, Vds=Rdson*Id, thus we want a good channel to reduce Rdson.

My experience (albeit limited) indicates that lower freq's (slower switching) helps a bit. Lower duty cycles (below 50% duty) helps a bit. Some resistance in the current path (above or below the mosfet) helps a bit. Really LOW currents (less than 5A) helps a bit. But none of these REALLY helps MUCH ! The mosfet gets HOT ! The mosfet eventually leaks out ALL of its built-in smoke ! It then stops working altogether !

P = I^2 * R = 5^2 * 0.1 = 2.5W, which should be very managable with a modest heatsink.

The spec states a Rdson of 0.023 ohm for Vgs=10V @ 28A. With Vs=0V, then Vd=0.644V which still gives a large Vgd to prevent saturation.

ylp88

 

[Dippy]

"MOSFETS ARE A PAIN"
- possibly only to those who don't understand... (I don't mean to sound cheeky but its easy to rubbish something before you get to grips with it)
MOSFETs are great when used correctly and in an appropriate app and when used within spec.

Here's a pretty thing to show things... may take a few seconds to load.
http://jas.eng.buffalo.edu/education/mos/mosfet/mosfet.html

There will be a test after you have read it

 

[ylp88]

"MOSFETS ARE A PAIN"
- possibly only to those who don't understand... (I don't mean to sound cheeky but its easy to rubbish something before you get to grips with it)
MOSFETs are great when used correctly and in an appropriate app and when used within spec.

Here's a pretty thing to show things... may take a few seconds to load.
http://jas.eng.buffalo.edu/education/mos/mosfet/mosfet.html

There will be a test after you have read it

That's a nice find, Dippy... Now why didn't my explanation have pictu... wait... interactive applets? Makes it look sub-standard now, right?...

ylp88

 

[Dippy]

Well, it covers the basics don't it.

I would imagine that as soon as you said 'saturation' and 'triode' regions then most people here would say "Aaaarggh MOSFETs are a pain" and promptly glaze over.

For newbies, being saturated with jargon can often twist their triodes to beyond their elastic limit. In string theory we call this sloblocks.

And it has to be said (oh yes it does) that I have only rarely ever seen the word 'saturation' used in a MOSFET data sheet. So, if Newbie starts looking for "Saturation Characteristics" in a Data Sheet they may be looking for quite a while.

I think for most people saturation = switch and triode = amplifier (or VCR) will suffice. Not thorough but makes it stick in the bean.

 

[manie]

MOSFETs are great when used correctly and in an appropriate app and when used within spec.


There will be a test after you have read it

Test !? AGAIN !?? Dippy, REALLY ! One thing I've learnt is that "within spec" also means "do not totally OVER spec" because higher current capability usually means slightly higher Rds-On etc. But an IRF540 (28A & 50V) is not really over spec'ing is it ? In my last application I switched the IRF540 by PWM at 42KHz,28KHz,8KHz via a TC4420 driving the Gate at 14.8V (20V upper limit). I fed 14-15V through the "famous" Toroidal in an earlier thread. The mosfet switched the low side. It drew maximum 6 Amps (8KHz), 2 Amps (28KHz) and 0.7 to 1 Amp (42KHz). Heatsinked on natural aluminium of 90mm x 75mm x 3mm (twin fins down both sides). It was COLD ! 13 deg C ambient. The mosfet plastic body (front) ran at 57^C (42KHz) to 87-95^C (8KHz @ 6 Amps). The latter heatsink ranged between 48-55^C. The mosfets are still working but hell ! They're not doing MUCH WORK are they ? I 'scoped all signals, from 08M's through TC4420 output's, Gates were very sharp ON/OFF. The mosfet drain signal showed some oscillation after SHARP on and some again after SHARP off. (+- 1V pp oscillation). Adding clamping diodes,cathode to +14V, across the induction took Amps to 12-15 A !!! Removed diodes, the mosfets ran cooler !! Diodes actually got VERY hot ??? Why ?

All TC4420 de-coupled with 100nF close to (1.5mm from) power pins.
So where is the problem ? What do I NOT understand ?

And: SORRY, I'm NOT trying to hi-jack the thread !

 

[BeanieBots]

A MOSFET can be compared to a Formula One race care.
If not driven properly (and very few know how until they try it) it will simply stall on the grid. Once driven properly, it will out perfom most others.

If you are using a proper driver, proper gate voltages, currents well within spec, sensible PWM frequencies but still experiencing excessive heat. (over and above what you calculated would occur), then you more than likely have one of two common errors.

1. Long current path between output of driver chip through gate through source and back to driver chip 0v. A common mistake is to make the ouptut pin to gate short and forget about the return path. Both are equally important. The 0v path MUST NOT share the load current. (except for the little bit of leg on the FET itself).

2. Over voltage. If the load is inductive, the FET could easily be over-volted. This can have many undesirable effects. Several of the undesirables (besides simply popping it) relate back to not being able to switch it off quickly and hence dissipating large amounts of heat.

Have a look at some commercial hobby radio control speed controllers.
I have some which pass >100A, have about 1" square of heatsink and only get warm. All done with FETs.

 

[jglenn]

Manie: Any transistor can be hard to use, due to subtle things that sometimes
have to be learned the hard way. My first big (20HP) dc motor controls would
blow 4" jets of blue flame out of the TO3 case, finally developed a good current
limit circuit.

Sounds like you are heating up due to switching loss. What is the turn on and turn off time of your waveform? Zoom in with the scope, how many uS?. This
slewing time puts you in the active region, where heating occurs. Besides having enough gate voltage and a driver, tuning the gate drive by adjusting the gate resistor and sometimes adding a ferrite bead to it, will clean up the waveform. About 5-20 ohms might be right.

If you were putting a catch diode on the output, and the diodes got hot, they
must not have been fast recovery type. Do you really need them with only
one volt of overshoot? The diode is only needed with inductive loads. Try an
RC snubber instead. .01-.1 uF and 20 ohm in series across whatever is
causing the trouble.

Another solution is to reduce your switching freq for testing, is 5KHz ok?
This way you can play with the circuit and not be on the edge of destruction.

 

[BrendanP]

Once again thank you everyone for taking the time to write detailed replies. You guys are right, there is tendency for the eyes to glaze over when trying to absorb new information. It takes time and determination to learn.

I understand where Manie is coming from in feeling frustrated and discouraged with mosfets. I try and remember in such situations that the devices must work and be useful or the billion dollars corporations who design and manufacture them would not do so.

When it comes to electronics R&D I have adopted the philosophy of not giving up. The answer/solution is there, I just have to find it. If I put the hours in I will solve the problem.

I'm going to do some more work today based on what has been written so far.

 

[manie]

Brendan: Hopefully we've both learnt something here. At least I got the 0V return right on the last circuit, separate 0V return (not main current path) back to the gate driver. Seems we're getting there !

 

[BrendanP]

"....The 0v path MUST NOT share the load current. (except for the little bit of leg on the FET itself)."

First thing I have wrong on my bread boarded design. Thanks BB.


"....2. Over voltage. If the load is inductive, the FET could easily be over-volted. This can have many undesirable effects. Several of the undesirables (besides simply popping it) relate back to not being able to switch it off quickly and hence dissipating large amounts of heat...."

BB et al ,Is the answer to this simply to use a higher v rated part? (if my scope shows this to be so)


"...Have a look at some commercial hobby radio control speed controllers...."
The SyRen unit I bought for just that reason can handle 25 amps , it doesn't even raise a sweat driving the wiper motor. Its all smd in on a small pcb thats fits in the palm of my hand.

 

[Dippy]

Please check what BB has suggested and do some measurements at various points on the power side of your circuit.
When triggering big inductive loads you will have some big pulses kicking about.
As JGlenn suggests, try a bit of snubbering.

Check the supply to your driver with a 'scope, maybe it needs more decoupling?

And often those 'ordinary' diodes aren't fast enough. And often a diode isn't enough. You sometimes have to move on from the simple low-power schoolboy text-book circuits and very often you will have to pay attention to the physical layout of your circuit.
Check for big nasties getting back to the supply line. I had a similar over-voltage pulse problem years ago when trying to design an electronic ignition - I was was in reinvent-the-wheel mode

I can't tell you what to try next as I don't have your circuit in front of me.
As suggested, try and find a commercial design. But take care, just because something is posted on the Internet doesn't make it Gospel.

"I try and remember in such situations that the devices must work and be useful or the billion dollars corporations who design and manufacture them would not do so."
- how true.
And the lads designing on kitchen tables who have experience wouldn't be able to make such good circuits with them ....?

Anyway, good luck. You'll get it sorted and will have gained some useful experience.

 

[BrendanP]

Thanks Dippy. I will act on everything that has been suggested here and in the previous thread I started on fets.

I have to get a motor driver sorted out both for my stock feeder and for future projects. Almost any mechatronic project will require motor speed and direction control.

Im using a DPDT relay for direction and the fet for speed.

I haver ordered some logic level fets to experiment with as well.

Im sure Ill get it happening, already I'm getting more adept at reading the data sheets and recognising the key parameters.

Im ordering this kit from Ozitronics K67 - DC Motor Speed Controller to play around with and use as bench mark as well.

 

[BeanieBots]

As Dippy points out, make sure any catch diodes are fast recovery and not bog-standard rectifier types. In many on/off situations, a simple 1N4002 (or similar) is fine, in a high speed, highly inductive PWM design, it won't cope.

Also check the path of the high current spikes which flow through that diode.
They are very fast and very high current. Hence, they will induce other spikes in nearby tracks and volt-drops down their connection leads. Take note of how these might effect other parts of your circuit.

It's not all black art, it just requires a little extra thought, care and attention.