Help with mosfets / mosfet drivers

[saburo]

Hello,
I am trying to understand how to use mosftets and drivers, I am still quite at a basic stage and my electronics knowledge is limited (I am more into the "software" world, but I'd like to change that).

Problem :
I have some DC source (i.e solar panels) which can give me about 80V (could vary, but to make it simpler let's assume 80V ) and about 5A.
ideally I would like to create a DC switching circuit (buck, boost, sepic... does not matter, just trying to focus on the mosfet part for now).
I want to use a mosfet, will probably use an IRF NChannel, 100V about 10A+ with the lowest Rdson possible.
This one IRF3710 is a bit an overkill, but looks nice and should run pretty cool.
The issue is that those kind of mosfet don't have a "logic level" Gate, so I need a gate driver to "amplifiy" my pwm signal.
At 10-12V Rds should be pretty low (20 is normally the maximum for the FETs I spotted).
I could go for transistors, but investigated dedicated devices instead as they seem to offer interesting configurations and fast switching, and finally they contain transistors too.

So I had a look at these two :

PMD9050D

PMD2001D

An I was thinking to test them with a circuit like the following :



In this circuit I used the PMD2001D, which is a totem pole driver (just learnt that name and the basics of how it works )... as I did not figure out how the other one works so far.

Now the calculations and relative questions:
1) the PMD2001D has a collector current of 0.6A with 1A max for short cycles (that would be likely my case I guess, using PWM from 30KHz up), I assume even the 0.6 should be enough for the IRF (could not find a Igs value in the datasheet) I guess that depends on how fast you want to switch it since it is voltage controlled, correct?
2) It's V ratings go up to 40V, so should do to carry my 12V in the configuration I chose (load is on the high side of the mosfet)
3) hfe is around 200, so even a tiny 5mA current provided by the pwm signal should reach the maximum output, however I saw in the datasheet (table 7) values for Ib = 20 and Ib= 50mA which does somehow conflicts with what I find in Fig 6.. wouldn't that be way too high since the device can deliver a max of 1A? With that value in mind I calculated the 120 Ohm value for R1, bu it does not make a lot of sense to me, I would have sized it for about 5-15mA (Fig 6) instead depending on the needed Gate current of the fet
Plus, if I have to provide more than 20mA with an MCU I would need to "drive the driver", which kind of defeats the purpose, no?
4) Even accounting for the voltage drop in the driver the 12V should be enough (11.something V) to switch cleanly the Mosfet, right? anything over 7-8V should bring down Rdson to very low values according to the datasheet, still 12 is way less than the 20V max allowed, so I figured it's a safe value.
5) I added a pulldown resistor (R2) with an arbitrary 10K value to ground the driver's base when the pwm signal is low, not sure it is needed, but found it in some schematics and suggestions around the web. Is that correct?
6) the cap across the two collectors of the driver is added as a filter, found that in a**broken link removed** (which I still need to digest, just exploring for now) on the Ti website.

I am sorry it's all really messy, but that's actually the way it is in my head right now
I am hoping for someone to help me in shredding some light on this fascinating subject.

Thanks in advance.

 

[FvM]

How do you think to get 12 V output from this driver with maximum 5 V logic voltage input? That doesn't work. Refer to a driver with voltage translatiion capability like TC426, it also solves the problem of sufficient gate current.

 

[alexan_e]

Check TC4427A (1.5A) https://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010669
and TC4424A (3A) https://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en027087

 

[saburo]

Hello FvM, thanks for your suggestion.
my understanding is that I can switch a 12V between Emitter and Collector if I have a lower (3.3V , 5V) Voltage on the base, provided that enough current is provided to saturate the transistor (and the Vce, Vbe etc are in the limits of the device).
Something like this , isn't basically what the higher part of my driver does in my circuit, where the load is actually replaced by the gate of the mosfet?

I m trying to understand things (learn), so sorry if I come back to this, and I will definitely check the device you suggested, but I am really trying to understand the "why" part now, not just replacing a device to solve a real problem.
Anyway I will compare the datasheet of the device you suggested and try to understand the differences, thanks!

Thanks for your time!

- - - Updated - - -

Thanks Alexan_e,
will definitely check them!

 

[FvM]

my understanding is that I can switch a 12V between Emitter and Collector if I have a lower (3.3V , 5V) Voltage on the base, provided that enough current is provided to saturate the transistor (and the Vce, Vbe etc are in the limits of the device).

You can, but surely not with a complementary emitter follower.

 

[alexan_e]

Your totem pole driver is essentially two emitter followers connected together and in an emitter follower the output voltage follows the base voltage.

If you apply 0v the output will be about 0.6v and if you apply 5v the output will be about 4.4v

If you add a level translator (mosfet or transistor) then you can drive it with 3-5v (the following is just a rough example)



A better version of this is the following because it limits the consumption of the level translator



All that being said it will be much easier to use a mosfet driver like the ones mentioned in the previous post.

 

[rohitkhanna]

IMHO you do NOT need to amplify your 5-v logic levels.

The IRF3710 Vgs is 2-4v, and hence it should switch quite nicely with a direct pwm signal from your TTL compatible signal. The only other point to note is the input capacitance of the mosfet - around 3nF !! So you need to be able to drive THAT well. The better you charge/ discharge THIS capacitance on the MOSFET input, the better it will switch. That is all.

cheers!
-r

 

[saburo]

Hello again,
I checked the devices you suggested and I have to admit the seem more appropriate or at least their specs are higher than the driver I initially found.
The main difference is that those you suggested (TC) are mosfet based, while the ones I found (PMD) are based on bipolar transistors. Another big difference is that I tend to understand the datasheet of the TC ones and have serious issues with the one of the PMD ones
The bipolar based ones seem to provide less current, but switch much faster (i.e. 6ns rise time vs 12 pr 23 of the TCs).
My logic signal is 3.3V and the TCxxx ones have a logic input "high" at 2.8V, so it should be enough.
Shouldn't it be a matter of current -which I limit with a serie resistor- on the base to open the PMDs, even with a 3.3V?
When it comes to gate current for the high power mosfet, my understanding that's similar to charging a cap so, assuming a constant V (I know there are different phases, but just to keep it simple), the more current I can pump in, the quicker it charges, correct? So that depends on the "input capacitance" parameter of the power mosfet : devices with a higher input capacitance will require higher gate current, did I get that right?

Thanks!

- - - Updated - - -

Oh, I saw you already answered some of my points while I was writing.

I understand now, things are starting coming together finally

@rohitkhanna : I think that would be possible, but Rdson would be pretty high with such low voltages, causing the mosfet to dissipate a lot of heat.

Thanks everybody for your explanations, will need to study them a bit now, but they definitely helped!

 

[alexan_e]

IMHO you do NOT need to amplify your 5-v logic levels.

The IRF3710 Vgs is 2-4v, and hence it should switch quite nicely with a direct pwm signal from your TTL compatible signal. The only other point to note is the input capacitance of the mosfet - around 3nF !! So you need to be able to drive THAT well. The better you charge/ discharge THIS capacitance on the MOSFET input, the better it will switch. That is all.

cheers!
-r

The given Vgs-th is given for a drain current of 250uA , I'm not sure you want to operate the mosfet in a high current fast switching situation so close to that voltage range.

------------added-------------
As an additional note the OP has mentioned an mcu driving this circuit so even if the mcu is 5v the output voltage drops quite fast as soon as you start pulling current so you could get 4v or less easily, add to that the voltage drop because of the totem pole driver and you may get in trouble sooner than you think.

 

[FvM]

Apparently you're focussing much on high speed gate drive. Actually is doesn't make sense without a switcher circuit that can take advantage of it. I doubt that the single ended PWM switcher in post #1 can.

 

[saburo]

Thanks for the comment FvM, which introduces another aspect I am trying to understand.

The maximum switching frequency I can use is limited by the rise and fall time of all the devices I have on the path of my pwm signal I suppose.
Normally low power devices (the drivers) are supposed to switch faster than high power devices, does it make sense to have them switch much faster?
I guess that would not improve the maximum pwm frequency or resolution, but it should affect the speed with which the power mosftet is charged / discharged, right?
So, if my driver commutes very quickly it should reduce also the time in which the power mosfet gate is at an "intermediate" voltage, reducing the energy dissipation in the device, isn't that correct?

That basically should justify the need of a driver that can switch much faster than the power device in my mind.
Increasing the pwm frequency this becomes relevant as we increase the number of time we switch the circuit, hence the transitions.

With an MSP430G2 I can easily produce a PWM signal with a 1MHz frequency (8MHz theoretically, but the resolution is pretty lame with those frequencies : 100% or 0% at 8MHz ), faster MCUs can deliver much better there.
That's probably an overkill and I don't think the IRF3710 could go that high but at this point my considerations are just theoretical.
For practical use I would stay in the range of 10KHz-300KHz I suppose.

 

[rohitkhanna]

Hello again,
....
@rohitkhanna : I think that would be possible, but Rdson would be pretty high with such low voltages, causing the mosfet to dissipate a lot of heat.

Thanks everybody for your explanations, will need to study them a bit now, but they definitely helped!

thats true. But nevertheless I believe you would dissipate much more during the transients if your switching isn't clean. Thats my point.

cheers!

 

[FvM]

Fast switching can minimize losses, but the load current can't drop to zero in no time. You need to provide a commutation path, e.g. a free wheeling diode. The fast voltage change at the load can still cause problems like RF interferences. So fast switching isn't enough, the whole circuit has to be designed to support it.

 

[rohitkhanna]

Fast switching can minimize losses, but the load current can't drop to zero in no time. You need to provide a commutation path, e.g. a free wheeling diode. The fast voltage change at the load can still cause problems like RF interferences. So fast switching isn't enough, the whole circuit has to be designed to support it.

Oh ... is that an inductive load ??

 

[FvM]

Oh ... is that an inductive load ??

Almost any load reveals to be inductive when switching fast enough and if the load is purely resistive you still have cables and parasitic circuit inductances.

That's the difference between the simple schematic in post #1 and real electronic circuits.

 

[saburo]

Yes FvM, you are obviously right.
As I tried to explain in my initial post, the goal is to produce a DC switching circuit such as a Buck / boost /sepic and there for sure I would have inductor and freewheeling diode.
I just neglected to represent them because I was only focusing on understanding how to switch the power mosftet with the PWM.
In a real circuit they definitely have to be taken into account and play a major role in the overall result... I'm simply not there yet.
I think what you said (parasitic inductance) does also apply to the driver part, so in a real application, especially if driven at high frequency, I suppose it is vital to design cleanly the traces and keep the driver as close as possible to the power mosfet.
With a smaller magnitude, that also apply for the pwm signal coming out of the MCU and switching the driver.
However I don't think a diode is needed after the driver since, if designed correctly, the parasitic inductance should be negligible there.
However I saw in some schematics a small resistor in series to the gate of the power mosfet, in some literature I found it is regarded as "optional".
What's the purpose of it? Is it to limit the effects of some possible current overshoots during transition phases?

 

[FvM]

Although circuit inductance doesn't mainly affect the driver operation, you'll notice that the source series inductance can play a role, even with a perfect layout.

Gate resistors are used to reduce switching speed.

 

[saburo]

Ah, good to know... but why would I want to reduce the switching speed?
I understood (in a switching converter) we normally put a lot of effort to keep that speed as high as possible to minimize the transition times.

 

[alexan_e]

**broken link removed**.

Question
The IRF4905 data sheet shows a gate resistor, Rg, in Figures 10a & 12a.
I understand that this resistor can suppress instabilities in switching
applications. So, what are the guidelines for sizing Rg?

Answer

Please see Application Note AN-944.

The gate resistor in series with the output impedance of the circuit used to drive the gate in a given application, controls the switching on and off times.

MOSFET gate charge in Coulombs = I t. So if a gate driver having say 0.2A output current capability switches a particular MOSFET in 100ns and by increasing the series gate resistance it reduces the current into the gate to 0.1A, then the MOSFET will switch in 200ns.

The optimum value for Rg is very application dependant. You want the MOSFET to switch as quickly as possible to minimise switching losses, but not so fast that parasitic inductances and capacitances associated with the pcb layout and any wiring to a load, will cause high di/dt voltage spiking or ringing.If you find that an optimised value of Rg controls switch on OK but slows the turn off too much, then a fix is to place a diode across Rg with its cathode towards the gate drive circuit. This will bypass Rg during turn off thereby speeding up the turn off. Placing a resistor in series with the diode will enable you to control turn off time independantly of turn on.

 

[mtwieg]

Yes, the gate drivers need to be much faster than the operating frequency of the converter. For converters operating up to a about a hundred KHz, drivers are generally fast enough (like <100ns switching times) that you don't need to worry much about it. As you go higher, gate drive becomes more important, and then layout and parasitic becomes the dominant issue. For example I'm designing wide operating converters at 2.5MHz, and my duty cycle range is effectively limited to between 2.5% and 90%, just due to parasitic ringing, even though the rise and fall times are just a few nanoseconds.