Levelshifter or totempole?

[PWM]

I saw many circuits on the internet with totempole drivers for FETs. Now I built such a totempole in a SMPS, but it doesn't work correctly. Now someone told me that a levelshifter should ALWAYS be used, even with totempole. The reason: when the mosfet is ON, the gate-drain voltage will both be tied to +12V. That's why a levelshifter should always be used to insure that the gate-source voltage does not drop below the threshold.

Is that the reason why my totempole does not work? After the totempole comes an N-channel FET. For very low currents it works fine (60 mA), but once the current increases, the FET somehow does not pass current to the output.
Source = grounded
Drain = +12V supply
Gate = to totempole
totempole is connected between +12V and ground

Thanx in advance!

 

[BrianG]

You don't say what FET you are using. Assuming that the FET is sinking current through a load from +12V:
1) have you checked that Vgs ON for your FET is less than about +10V at the current you want the FET to control? The totem-pole upper transistor will still exhibit a reasonable voltage drop when turned on, even with virtually no current flowing into the FET gate.
2) Following from the first point, if your FET gate-source resistor is too low in value the totem pole may not be able to supply sufficient current pull the gate voltage up high enough to turn the FET on fully.

 

[PWM]

The FET is of type BUZ11.
The totempole consists of 2N3904 and its complementary.
The main issue is that the FET receives 12V whereas the drain-source voltage is also equal to 12V. I can't use more than 12V because the circuit is intended for a car battery.
Any suggestions?

 

[BrianG]

Although the BUZ11 is a discontinued type with many manufacturers it is a useful device which is good for 20A max current. A quick look at a typical device data sheet shows that 6V gate-source should turn the FET on hard enough to handle 20A with only 5V drain-source. If your totem pole definitely gives a voltage of 6V or greater at the gate then there should not be any problem with switching currents up to 15-20 amps unless the device is faulty.

 

[blcpro]

PWM, what do you mean by "the drain-source voltage is also equal to 12V"? What is your load and did you consider the effect of the voltage drop on the load as current through the FET increases? Also, keep in mind that it is possible the FET or totem pole components are not working. I've gone through similar problems only to find out that it was a component failure, and not the circuit!

 

[sdmays]

What kind of SMPS are you talking about? If you are referring to a classic buck voltage regulator, then your 'totem pole' transistor configuration will not work properly.

If, on the other hand, your SMPS uses some kind of flyback transformer to transfer energy, you could possibly connect the source to ground and the drain to the low side of the primary of your transformer. That way, when the NPN transistor of your 'totem pole' attempts to drive current into the gate, the drain will be at a low enough voltage (ground) to allow current to flow into it (after which the pnp drains the energy back out of the gate to source 'capacitance').

 

[PWM]

Well, I have tried different totempole transistor pairs. Also BD139 type with its complementary.

The circuit is a very simple buck converter with a SG3525 controller.

The drain-source voltage is also equal to 12V. That means, equal to the gate-source voltage.
drain-source = gate-source = +12V.
The gate-source voltage is supplied by the totempole, and the drain-source voltage is supplied by the 12V power supply. This means that when the FET is ON, the gate-source voltage will drop below the threshold voltage, because the source is tied to ground. Is that the reason why it doesn't work?

This buck converter drives a few LEDs in parallell, also a big LED that causes a 10W-resistor to become too hot to handle! But the big LED does not consume current, only the small LEDs (which are up to 50mA in total). The big LED works fine with a LM7805.

 

[sdmays]

PWM,
You asked "Is that the reason why it doesn't work?" The answer is mostly yes. A simple buck converter has the MOSFET source connected to Vcc (+12V in your case), the drain connected to an inductor and to the cathode of a fast diode (schottky), the other side of the inductor is connected to the load and any filtering you might have (i.e. electrolytic capacitor), and the anode of the fast diode (schottky) is connected to ground.

When the buck converter is working, the voltage of the side of the inductor connected to the source and the diode should hover around Vcc - if the source is hovering around Vcc, then Vgs can never be as much as 1V (considerably less than the threshold voltage), which is why a 'level shifter' is used to get the voltage of the collector of the npn of your 'totem pole' greater than Vcc. I would be concerned with possibly exceeding the maximum Vgs at some point.

Would you be willing/able to use a different dc-dc configuration, such as a forward dc-dc converter? Check out the following web-site (look at the "Single Transistor Forward Converter&quot:

http://henry.fbe.fh-darmstadt.de/smps_e/smps_e.asp

This web-site will also calculate the values of your inductors for you!

 

[PWM]

oh sorry...
I gave a wrong explanation.
The website you sent to me, look at the first buck converter. That is exactly what I'm using. So the source of the N-channel BUZ11 is connected directly to the inductor. The gate goes to the totempole, and drain to the +12V source.
The diode is 1N4148. I know it's very small, but it should be able to deliver at least some 100mA. Those fast diodes are hard to get and have to order some. I am not using a transformer, just a HF-core (yellow).
It does deliver 3 to 4V at the output, depending on the tuning of some components. But if the load becomes too big, it shuts off even though the SG3525 keeps working.

So until now I have learned that a levelshifter is a MUST in a buck converter. But, is there an alternative? I really hate levelshifters.

 

[sdmays]

Punch Line: In order to use a simple buck converter, i.e. the first converter shown on that web-site, you have to have a rail at least the Vgs threshold greater than Vdd. For example, Vdd = +12V, Vgs threshold = +4V, Vlevelshifter = Vdd + Vgs threshold = 12 + 4 = 16V (minimum). When you first apply dc power, the gate will see every bit of the 16V, which could destroy your MOSFET ('punch through').

Alternative: Consider the "Single Transistor Forward Converter" on that web-site. You will have to wind the transformer by hand (can be done WONDERFULLY on a ferrite toroidal core), but the converter will be pretty efficient due to the unused energy stored in the core when the MOSFET is turned off will be folded back to your power source (Vdd).

 

[cbarn24050]

Hi, a few pointers for you:- 1. the 3525 is not the best device to use as there are numerous devices designed for that exact function. 2 the 3525 has high current drivers and so will drive a mosfet directly easily. 3.For this application rearrange the output curcuit, replace the buz with a pnp transistor (emitter to supply, collector to inductor) connect base to resistor 22R other end resistor to pin13, connect pin 11 and 14 to ground.470R between coll and base.