Correction For ALL my Logic FET Posts

[gramo]

If there’s anyone out there that has used the IRL2203 FET in the past because of my guidance, I'm sorry!

I have recommended it in the past with Micro Controller projects, but things may not work as they intended too

Just tonight in a practical circuit I ran into some hick-ups. The IRL2203 is not logic voltage driven - as I have suggested in the past

However - The following available from digikey.com is a logic voltage driven FET;

RFD14N05L-ND
https://www.electro-tech-online.com/custompdfs/2007/03/RFD14N05LSM.pdf

The above model is a TO-251AA package (similar to the TO-220, but without the heat sink hole on top)

 

[gramo]

Just on that, If you want to put a heat sink on the RFD14N05L-ND then simply leave enough room to bend the component backwards on the PCB so that it lays flat, and with some heat compound glue, throw your heat sink under it so that its flush between the component and the PCB

 

[gramo]

Heres how you would control the IRL2203/RFD14N05L-ND

**broken link removed**

**broken link removed**

 

[justDIY]

no base resistors for the totempole?

 

[Hero999]

Why would base resistors be required on a pair of emitter followers?

 

[Nigel Goodwin]

Why would base resistors be required on a pair of emitter followers?

For when the 2N3904 goes collector/base short and sticks 12V up the PIC?.

 

[gramo]

For when the 2N3904 goes collector/base short and sticks 12V up the PIC?.

When does that happen

 

[gramo]

I've just tried simulating the circuit again, and threw in a diode just so that there was some isolation from the source to see if 12V ever develops at the base of the 3904, but it’s operating as normal

Video of simulation

 

[Nigel Goodwin]

When does that happen

When things go badly wrong! - it might NEVER happen, but there have been plenty of commercial examples where 'non-essential' components have been left out, only to cause much more damage when things do go wrong.

In this particular case the FET could easily go S/C, this could very possibly damage the top transistor, which in turn could stick 12V down the PIC pin. Not a problem in this case, as a PIC is cheap - but if this was a commercial product, with a protected PIC, and the manufacturer has gone bust - then your entire device is scrap, all for the cost of a simple resistor?.

It's good practice, it costs very little, and it 'might' prove of great value?.

 

[gramo]

Ahh, I see

Well the whole idea about not having a resistor there is to improve switching speeds of the FET. If the FET becomes biased (not fully on or off) for any period of time, it develops a lot of heat in high current circuits.

So, keeping that in mind, you would need a 480ohm resistor to limit the sink current (max 25mA) on the PIC from a 12V potential, should the transistor break. But that would decrease the bias current on the transistors down to 10mA.

Would that still be suffice to switch the FET's fast enough?

 

[Nigel Goodwin]

You wouldn't need that high a resistor to stop the PIC getting killed, they stand a fair amount of abuse - but I wouldn't expect 480 ohm to be a problem?.

 

[gramo]

The FET's gate acts like a capacitor, and charges to its potential, with the transistors only being driven by 1/2 the PIC's maximum, wont that decrease the amount of current they will switch to the FET's gate?

 

[Nigel Goodwin]

The FET's gate acts like a capacitor, and charges to its potential, with the transistors only being driven by 1/2 the PIC's maximum, wont that decrease the amount of current they will switch to the FET's gate?

How much current do you need? - probably not as much as you think?.

 

[geko]

The totempole circuit shown in gramo's post isn't going to help a non-logic level MOSFET since it doesn't increase the gate drive voltage to the MOSFET anyway.

In fact it's slightly worse than driving it directly from the PIC output because you get the 0.6v base-emitter drop as well.

You need another NPN transistor and resistor to 12V rail between the PIC and the totempole transistors to make it work.

 

[gramo]

The totempole circuit shown in gramo's post isn't going to help a non-logic level MOSFET since it doesn't increase the gate drive voltage to the MOSFET anyway.

The PIC's 5V output will bias the transistors on and off, allowing much more voltage (in this case 12V) and current to switch the FET's Gate.

This allows the Gate to charge and discharge much faster, as the Gate acts like a capacitor, requiring time to charge/discharge.
During this time the FET becomes biased, and will drop a voltage across it, instantly dissipating much more heat as P = V * I

Now you can use PWM signals to control the output safely, or for switching of high current loads without burning out your FET.

In fact it's slightly worse than driving it directly from the PIC output because you get the 0.6v base-emitter drop as well.

The base-emitter voltage drop is not the issue, infact, its across Vce. Ideally Vce = 0, and the FET's gate can charge at 200mA (Ic max of the 3904/3905, 800% more than the PIC's maximum current output). And the opposite with a low output from the PIC - when the PNP ties the FET's gate to earth.

You need another NPN transistor and resistor to 12V rail between the PIC and the totempole transistors to make it work.

I don’t understand why, as the transistors will drive hard on with 5V and no base resistor wont they?

 

[geko]

I don’t understand why, as the transistors will drive hard on with 5V and no base resistor wont they?

The 2N3904 is an NPN transistor and you're using it as an emitter follower. If you drive its base to 5V, its emitter has to be 0.6V less. (unless its failed with an emitter-collector short)

The PIC's 5V output will bias the transistors on and off, allowing much more voltage (in this case 12V) and current to switch the FET's Gate

Not with the circuit shown. Put it on a bread board and try it.

 

[gramo]

Now I see what you mean, so with another NPN driving the other two, the FET would be driven by 11.4 Volts instead of 4.4?

But wouldnt that make the voltage at the base of the totempole 4.4 volts, and thus making the voltage at the gate of the FET 3.8 volts?

Hmm, perhaps a quick piccy is in order

 

[gramo]

Wouldnt a PNP be the better choice?

**broken link removed**

Edit, forgot to link the Sim of the above circuit


What do you think?

 

[geko]

The PNP won't ever turn off will it.

The simulation is flawed because you're not simulating the output of a PIC. Except for the open drain outputs, a PIC will drive the output either to near Vss or Vdd. Your simulation is opening a switch, leaving the base of the transistor floating with no signal on it. It needs to drive it like a real PIC will.

You need an NPN transistor like the attached schematic shows

 

[geko]

Sorry, that schematic has the PNP transistor the wrong way round.

The attached one here is correct.