Hi,

I have to make a buck converter that takes 48v input and converts it to 24v output. Now, if I am to use a MOSFET as the switching element, I will need to give it pulses of at least 50v (for Vgs=2v) to drive it into conduction, right? So, I can't figure out how I would go about producing pulses of such high voltage. Is there any easy way to do that? I was thinking of using a function generator, but I doubt the ones available to me can produce voltages in the range required...

Any ideas?

 

No -do not apply a 50 VP-P pulse to the gate of a MOSFET without checking the specifications first.

Just for starters, here are some links to High Side Drivers
Linear Technology http://www.linear.com/products/high_side_switches_*_mosfet_drivers
ST http://www.st.com/web/catalog/sense_power/FM1965/SC1037
Allegro http://www.allegromicro.com/en/Prod...-Interface-ICs/High-and-Low-Side-Drivers.aspx

 

One way is to use a PFET instead of an NFET. Then you can turn it on by pulling the gate to ground. You will need a zener from gate to source to keep the gate voltage from going to high for the FET.

 

In small quantities, PFETs are more expensive and have less gain and a higher on resistance than "equivalent" NFETs, but they are much easier to deal with in your type of circuit. What is your peak output current requirement and what is your expected efficiency?

ak

 

Hi,

I have to make a buck converter that takes 48v input and converts it to 24v output. Now, if I am to use a MOSFET as the switching element, I will need to give it pulses of at least 50v (for Vgs=2v) to drive it into conduction, right? So, I can't figure out how I would go about producing pulses of such high voltage. Is there any easy way to do that? I was thinking of using a function generator, but I doubt the ones available to me can produce voltages in the range required...

Any ideas?

If you use a P-channel MOSFET, you only need to pull the gate down - you need a Vgs clamp and a current limiting resistor, typical breakdown voltage is only about 18V on the gate oxide.

When I serviced PC monitors for a living, I encountered a Compaq monitor with a very tidy N-channel buck converter using a diode/capacitor pump to obtain the gate voltage - the rest was just capacitor coupled to the PWM chip.

The hand traced schematic is in a tea chest somewhere in the back of the garage - but If you search out service manuals from about mid period SVGA era, you should find some examples.

The very early SVGA monitors used multiple taps on the main SMPSU secondary with transistors to switch in the one that was required - the main failing of the buck regulator is failure usually means a shorted MOSFET and full unregulated B+ fed through.
The last models before the CRT drifted into extinction, had a lower B+ fed into a flyback boost regulator.

 

If you do not need to have a common ground between the input and output supplies, consider having a common positive lead, and putting the MOSFET into the path of the ground, instead. That means that you can use a (more efficient) N-channel transistor, and the driving voltage needed is between the two voltages of your input supply. You could use a small linear regulator to generate (say) 10 volts to drive the gate of the MOSFET. I'd also suggest using a chip specifically designed as a MOSFET gate driver. There are many to choose from at places like Mouser, and they're like a dollar or two dollars. Just trying to save you the painful learning experience that I went through trying to do my own driver circuits and get decent rise and fall times.

 

Thank you everyone. I will consider all your suggestions. I don't really know much about converters, I'm only learning now that it's part of my project.

The specs are:

48v input to 24v output
2A current
Efficiency: I have no idea. I think even a poor efficiency one would do. As of yet, I just have to show that 48v are being converted to 24v at the output end.

A question about PFETs: I understand that to turn a PFET on, I will have to pull the gate voltage down to ground, but what about turning it off? Won't I again need a pulse of around 50v for that? I'm not totally clear on that point still.

Also, please mention any ICs that could make my job easier.

Thanks a lot, people! I don't know what I'd do without your help.

 

If you pull the gate *toward* ground (not *to* ground) with an open collector driver, a resistor from the gate to source is all that is needed to turn the transistor off when the drive transistor is turned off. The simple circuit has the resistor and a 12V zener from gate to source, and a series resistor from the gate to the open collector driver. The series resistor limits the current through the zener. The gate-source resistor is all there is to discharge the gate capacitance (listed on the datasheet) to turn off the transistor. If you need fast turn off times, decrease this resistor. Note that as this resistor decreases, the current through the drive transistor increases.

For long term reliability, it is best to have all devices rated for approx. twice the voltage they will see in operation. So a 50 V PFET will handle the 24 V normal voltage across it, but not the 48 V across it at startup when the output filter caps are not yet charged up to 24 V; better to use a 75 V part as a compromise between ratings and money. The driver transistor will see 48 V every time it is turned off, so a 60 V part might be ok, but an 80 V or 100 V part would be better.

ak

 

If you pull the gate *toward* ground (not *to* ground) with an open collector driver, a resistor from the gate to source is all that is needed to turn the transistor off when the drive transistor is turned off. The simple circuit has the resistor and a 12V zener from gate to source, and a series resistor from the gate to the open collector driver. The series resistor limits the current through the zener. The gate-source resistor is all there is to discharge the gate capacitance (listed on the datasheet) to turn off the transistor. If you need fast turn off times, decrease this resistor. Note that as this resistor decreases, the current through the drive transistor increases.

For long term reliability, it is best to have all devices rated for approx. twice the voltage they will see in operation. So a 50 V PFET will handle the 24 V normal voltage across it, but not the 48 V across it at startup when the output filter caps are not yet charged up to 24 V; better to use a 75 V part as a compromise between ratings and money. The driver transistor will see 48 V every time it is turned off, so a 60 V part might be ok, but an 80 V or 100 V part would be better.

ak

I don't know anything about drivers. Guess I should read up. What does it do exactly?

Thank you!

 

While there are specific ICs and circuits called gate drivers, I was speaking more generally about the circuit that controls the gate of the pass transistor. In your case it might be as simple as a small NPN transistor.

Backing up a bit, is it your intent to grow a buck regulator from scratch, or are you able to use a buck controller IC? Linear Technology and others have many such controllers to choose from.

ak

 

If you do not need to have a common ground between the input and output supplies, consider having a common positive lead, and putting the MOSFET into the path of the ground, instead. That means that you can use a (more efficient) N-channel transistor, and the driving voltage needed is between the two voltages of your input supply. You could use a small linear regulator to generate (say) 10 volts to drive the gate of the MOSFET. I'd also suggest using a chip specifically designed as a MOSFET gate driver. There are many to choose from at places like Mouser, and they're like a dollar or two dollars. Just trying to save you the painful learning experience that I went through trying to do my own driver circuits and get decent rise and fall times.

With an N-channel MOSFET you need the gate to be at about 6 - 8V more than where you want the source to end up - otherwise the source will be that much less than the voltage into the buck, and you will have I2R losses.

Some high side drivers include capacitor/diode charge pump circuitry to develop a voltage higher than the supply for biasing the gate - one particular Compaq monitor I had on the bench had very tidy discrete component circuitry to do that.

At the time, I hand traced the circuit which is in my notes somewhere at the back of the garage - it was also published in my monitor servicing notes submitted to Television and consumer electronics magazine.

There are probably schematics and service manual archives on line that an example can be found - but it would take a bit of effort to go through them.

 

I don't know anything about drivers. Guess I should read up. What does it do exactly?

Thank you!

You need to nail down once and for all whether you're going for the P-channel or N-channel option and stick to whatever you decide.

The P-channel option is simpler to drive, and easier to design the drive circuit - but the P-channel MOSFET will cost more and be less efficient.

The N-channel option has more complex drive and takes more effort to design - but most MOSFETs are N-channel, so huge choice and lower prices.

 

While there are specific ICs and circuits called gate drivers, I was speaking more generally about the circuit that controls the gate of the pass transistor. In your case it might be as simple as a small NPN transistor.

Backing up a bit, is it your intent to grow a buck regulator from scratch, or are you able to use a buck controller IC? Linear Technology and others have many such controllers to choose from.

ak

While making the converter from scratch is more preferable, I don't think I have the expertise to do that. So I'm open to the use of ICs. I did some searching and found some buck regulator ICs, but they only output small usual levels of voltage i.e 5v, 9v etc. I couldn't find anything that could give 24v.

If at all possible, could you draw up a schematic of your suggested design from your previous post? I really do not want to impose, so please don't go out of your way to do it for me.

And any helpful links you could give me for learning a thing or two about gate drivers would be nice.

Thank you so much! I reallyyyy appreciate it!

 

You need to nail down once and for all whether you're going for the P-channel or N-channel option and stick to whatever you decide.

The P-channel option is simpler to drive, and easier to design the drive circuit - but the P-channel MOSFET will cost more and be less efficient.

The N-channel option has more complex drive and takes more effort to design - but most MOSFETs are N-channel, so huge choice and lower prices.

I think I'm going with the P-channel option, but I am presuming it won't cost significantly more than the N-channel option?

Also, I'm totally ignorant of capacitor/diode pump circuit you are speaking of. I understand you might find it annoying that I'm trying to do something I'm not entirely qualified to do without trying to get through the requisite knowledge to build the device, but it is only what's required of me and there's not much time left for experimentation.

I hope you guys don't lose your patience with me lol.

Thanks again for your kind help!

 

While making the converter from scratch is more preferable, I don't think I have the expertise to do that. So I'm open to the use of ICs. I did some searching and found some buck regulator ICs, but they only output small usual levels of voltage i.e 5v, 9v etc. I couldn't find anything that could give 24v.

!

Offhand - I think the MC34063 has an uncommitted output transistor that can drive either buck or boost.

There's various push-pull SMPSU chips with floating output transistors that you can parallel for single ended output, I think; TL494, UC3524 and KA7500B.

Its probably worth having a look on the LT website - but I'm not brushed up on their product lines.

 

The former National Semiconductor (now TI) LM2576HV and 2596HV are ebay favorites. Search for "buck regulator 60V" (without the quotes) to see lotsa options, including this:
http://www.ebay.com/itm/48V-buck-mo...580?pt=LH_DefaultDomain_0&hash=item43cb449e04

Thermal management is an issue with these small modules, but they give you an idea of how simple the circuit can be with the right part. If you start with just the IC in a traditional TO-220 5-lead package so you can screw on a small heatsink, and follow the datasheet app notes, it really is a "simple switcher".

What is this circuit for? What is the load it is supplying power to? Do you need a regulated output that maintains a constant voltage as the load current varies, or do you just need to hold the output to 24 V max., and it can sag a bit with load? As much as it hurts to say this, you might be able to get by with a 555 circuit plus a few parts.

ak

 

The former National Semiconductor (now TI) LM2576HV and 2596HV are ebay favorites. Search for "buck regulator 60V" (without the quotes) to see lotsa options, including this:
http://www.ebay.com/itm/48V-buck-mo...580?pt=LH_DefaultDomain_0&hash=item43cb449e04

Thermal management is an issue with these small modules, but they give you an idea of how simple the circuit can be with the right part. If you start with just the IC in a traditional TO-220 5-lead package so you can screw on a small heatsink, and follow the datasheet app notes, it really is a "simple switcher".

What is this circuit for? What is the load it is supplying power to? Do you need a regulated output that maintains a constant voltage as the load current varies, or do you just need to hold the output to 24 V max., and it can sag a bit with load? As much as it hurts to say this, you might be able to get by with a 555 circuit plus a few parts.

ak

You can make a PWM generator with as little as a unijunction (preferably PUT), a single op-amp and a few assorted passives.

 

The former National Semiconductor (now TI) LM2576HV and 2596HV are ebay favorites. Search for "buck regulator 60V" (without the quotes) to see lotsa options, including this:
http://www.ebay.com/itm/48V-buck-mo...580?pt=LH_DefaultDomain_0&hash=item43cb449e04

Thermal management is an issue with these small modules, but they give you an idea of how simple the circuit can be with the right part. If you start with just in a traditional TO-220 5-lead package so you can screw on a small heatsink, and follow the datasheet app notes, it really is a "simple switcher".

What is this circuit for? What is the load it is supplying power to? Do you need a regulated output that maintains a constant voltage as the load current varies, or do you just need to hold the output to 24 V max., and it can sag a bit with load? As much as it hurts to say this, you might be able to get by with a 555 circuit plus a few parts.

ak

Right now it's just a power supply and a buck converter without any load. But later I'll need to incorporate it into a power conditioning unit in a solar panel. For that purpose, I'll probably have to use a complete IC such as the one you linked, as I don't trust myself enough to build one that could drive a load. But don't worry about that part now.

I thought of using a 555 timer to generate the gate pulses for the MOSFET. How do you want me to use it?

Thanks for your help!

 

@Vorador

If you decide to build one here is about as simple as it gets from scratch.

 

@Vorador

If you decide to build one here is about as simple as it gets from scratch.

I've seen a simpler buck regulator than that - it was built around the sharp knee of a TL431.

Mainly of the hysteretic topology, the run of the mill examples use an op-amp or comparator - the TL431 is a comparator, it just has less pins and a built in Vref.