mosfets in linear and switch mode / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

mosfets in linear and switch mode

1 2 next

Move Thread

LAN_403

IamSmooth

Tue Aug 11 2009, 12:27PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

I understand using mosfets as a linear device. I also know they are used as switching devices. Are there any decent books describing some theory and design using these two modes?

I am correct in saying that mosfets, in swich mode, regulate current flow by the frequency of the switch? They are set to conduct maximally, and the frequency of the on/off state controls the flow of current?

HV Enthusiast

Tue Aug 11 2009, 03:13PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

In switchmode, voltage regulation is achieved by varying duty cycle (in most cases) - Frequency is typically constant although there are variable frequency supplies, but in those cases, duty cycle still controls regulation.

A MOSFET used linearly acts just like a variable resistor or pass element. Your standard linear voltage regulators are typically MOSFETs used as pass elements although BJTs are also employed, especially in older linear regulators.

For simple switching regualtor, just look up boost or buck regulator.

Proud Mary

Tue Aug 11 2009, 05:08PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

In linear use, FETs performs very much like a thermionic triode valve - the source being the cathode, the gate the grid, and the drain as the anode.

This is good to know because it means you can take any of the good triode valve circuits which are out there, and adapt them painlessly for use with FETs - almost on a 'drop in' basis - when the different supply voltages have been taken into account.

IamSmooth

Tue Aug 11 2009, 06:47PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

I can use diodes to serve as a means to select the trigger voltage on the gate. What is done to control the duty cycle when using a switch-mode approach? If I want to start shunting at 100v, what is the basic circuit design to start increasing the duty cycle when this target voltage is reached?

HV Enthusiast

Tue Aug 11 2009, 10:40PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

IamSmooth wrote ...

I can use diodes to serve as a means to select the trigger voltage on the gate. What is done to control the duty cycle when using a switch-mode approach? If I want to start shunting at 100v, what is the basic circuit design to start increasing the duty cycle when this target voltage is reached?

You need to provide more information. I'm not sure what you want to do here.

Hazmatt_(The Underdog)

Tue Aug 11 2009, 11:01PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

There are quite a few books out there for college circuit design, and they make it seem simple enough when it comes to doing calculations, but that's because they give you all the information you need.

When I had my classes, we were given Vov, Vth, Vt, for our calculations, but these don't show up on datasheets, so doing it by the book is sometimes a huge waste of time since you need to play with the circuit anyway.

It could be me, but I characterized a driving transformer for the preamplifier I'm working on as nearly a 100R load reflected back to the class-A input stage. I tried to design the driver to that load spec., and it was a pretty miserable situation. So I went back to my previous attempt which more closely resembles the spec. sheet of the amplifier that I'm copying.

I'm not saying give up on trying, its just that from my experience, the calculations are typically less helpful. What I find to be more helpful is watching the current draw, making sure the device is pulling current, make sure there isn't too much heat buildup, look for distortion, those kinds of low tech solutions.

Otherwise, what I would have to do to make sure I'm at the peak of efficiency and proper load line, would be to literally characterize each transistor, and I can't be bothered to do that. I just want to get up and running.

A book I would suggest is: Practical Circuit Analysis of Amplifiers. You will not find a more descriptive book anywhere, I have most of them. Pick yourself up a copy from Amazon and you will not regreat it, I promise.

Proud Mary

Tue Aug 11 2009, 11:35PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Certainly the easiest and cheapest way is to use the famous 556 dual 555 timer, configured to allow variation of the mark/space ratio without affecting the frequency, which is impossible with a single 555 device.

One of the most common circuit errors one sees duplicated over and over on 4HV and copied round the web is to ignore the high gate capacitance of power MOSFETs, and so attempt to run them with insufficient power - directly from a 555 for example, without even a buffer amplifier.

Much the best solution is to use a dedicated MOSFET driver IC as an interface between the 556 (or other PWM) low power switching source and the much larger current needed by the MOSFET.

The driver IC's current rating and the MOSFET's gate charge determine how quickly the MOSFET can be switched on and off.

Now, as you'll see, Smoothie, as the switching transitions are the deciding factor underpinning the switching losses within the MOSFET, to improve the circuit efficiency you've got to choose a driver IC with a current rating adequate to the needs of your power MOSFET.

I believe that the best driver for your needs would be one with a combined bipolar and MOS topolgy, where MOS sections are paralelled with the bipolar devices so you can have genuine rail-to-rail switching between LT+ and Earth. (without 'true' rail-to-rail switching you'll never get anywhere near the always optimistic maximum current claimed in the MOSFET data sheet)

When you've got your pocket money, or emptied your piggy bank, you'll find you have choice between two basic types of MOSFET driver IC -

1. Driver ICs where the drive amplitude is set internally.

2. Those where the adjustment is external

If your PSU isn't all that stiff , you'd be best off buying the internally regulated type, because the chip is designed with a high power supply variation rejection ratio, so the output amplitude remains constant despite variations in your power supply voltage.

The down-side of this choice is that as the gate-to-source voltage of your MOSFET is increased, the on-resistance will decrease in proportion resulting in lower power dissipation from conduction losses.

Going the other way, as you increase your gate-to-source voltage , power losses caused by gate charge also get bigger.

So if the absolute maximum of efficiency is your goal, Smoothie, you should choose a driver IC which has got external adjustment, so you can tweak up the balance between gate charge losses and conduction losses.

But if, like me, you're all for an easy life, and are prepared to accept a second-best which is nearly as good in most circumstances, I'd go for the driver IC with automatic internally adjusted gate drive amplitude control. Fewer parts!

I hope this helps

Harry

Dr. Slack

Wed Aug 12 2009, 06:26AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Are you still trying to stop your windmill generator overspeeding, or has the application moved on? BTW, if you use the phrase "trigger voltage" with respect to an IGBT or FET, then alarm bells ring with me that you have an innappropriate model of the device

You need to make a clear distinction between how the application has to behave, so really how you control it, and the detailed implementation (voltage, current , power to be handled, what's in your junk box etc) of the power stuff.

When you are not pushing the technology (unlike say SSTC use), then most power elements are pretty much interchangeable between switch and linear mode, and with each other as far as behaviour goes. They "like" different working regimes, tubes high voltages, then IGBTs, then FET for lower voltagess, with BJTs overlapping the FET and IGBT regions, and have Vsat in a similar order. Tubes and BJTs need more power to run in terms of base or grid drive. Electrically, tubes are uber robust compared to fragile silicon, mechanically it's the other way round. Most (all unless it says so) silicon has rubbish power handling in continuous linear operation compared to its pulse (switching) specifications, read the data sheet carefully.

Indepednaently of the power element, choose the control strategy. Tracking a set point in linear more, or keeping a parameter within a range? The latter allows you to do really cheaty things like a simple relaxation oscillator, much simpler in terms of componentns and control stability than anything more complicated.

Then choose a suitable type of component to implement the control, comparator, opamps, 555, uP.

Then finally design an interface between control and power. To switch very fast, a dedicated driver is handy. At a more pedestrian speed (check the data sheet for tolerance of power loss during slow switching edges) then a few mA from an opamp or 15v logic may be all you need.

What is done to control the duty cycle when using a switch-mode approach?

Anything you like, there is not enough information and too many solutions to give definitive advice. However, you could try playing with a sawtooth reference waveform and a comparator

If I want to start shunting at 100v, what is the basic circuit design to start increasing the duty cycle when this target voltage is reached?

If you want 0% shunt conductance at 100v, and 100% shunt conductance at (say) 150v, then arrange the sawtooth to go between 100v and 150v. FET on when voltage above the refreence , off when below. The sawtooth would have a constant frequency, which would be the PWM frequency.

There are plenty of other ways of doing it, you might be able to use a 555, but they're too complicated for me :)

IamSmooth

Wed Aug 12 2009, 05:47PM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

Neil,

My linear shunt controller is working well so far. I was asking about this approach for my own enlightenment since I have never built anything like this. Is there a simple schematic around that can serve as a skeleton reference? I can understand turning conductance ON when i go above a threshold voltage.

If I wanted to redesign my shunt controller to use a switchmode approach, where would I begin. I have a book on switchmode power supplies, but this does not seem to address my application for a shunt device.

For those just coming in on this, I want a device that will start shunting current above a threshold voltage such that the voltage does not go any higher. Right now I am using mosfets linearly. I would like to see how this could be done by increasing the duty cycle of switching them on and off.

Avalanche

Wed Aug 12 2009, 08:23PM

Registered Member #103 Joined: Thu Feb 09 2006, 08:16PM
Location: Derby, UK
Posts: 845

The simplest way is with 'bang-bang' control, this is what I did (quickly!) for a 100kW brake chopper on an electric vehicle at work. Hardware layout was simply a 3 ohm, several kW roof mounted load bank, and a huge lowside IGBT. If DC bus voltage went above 650v (regeneration) then switch IGBT on. Switch IGBT off when DC bus goes below about 630v. It's a bit crude though!

On a solar dumpload controller, I've simply made a 'buck converter' with a mosfet, inductor, and power resistor as the load, it's a bit more EMC friendly because the current in the load bank is constant. The battery voltage is monitored with an ADC, and compared to a target ADC reading (which corresponds to 13.8v). If the reading is high, the duty cycle is incremented. If it is too low, the duty cycle is decreased (right down to zero). It's been running for a few months now.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob