MOSFET

[jwax]

Playing around with driving a power MOSFET to dim a lamp running off a 12 volt battery.
Interested to know how to expand the range of gate voltage to drive the lamp from OFF to full ON.
Lamp begins to turn on at 4.8 volts at the gate (dim), and is full on at 5.5 volts (bright).
What circuitry to add to expand that gate voltage range to maybe 4.8 to 10 volts, to go from dim to full bright?
Thanks!
John

[haklesup]

A voltage divider. I can think of a few configurations. One way is one resistor from gate to ground (or VDD) and one from gate to the output of your variable voltage source.

A few more resistors and you should be able to design a network to allow you to input 0 to 10V and get 4.8 to 5.5 on the gate for example

[Externet]

Hello haklesup. Are you able to see/use direct/personal messages/mail from this forum ?

[sghioto]

Pulse Width Modulation is the best way to use a mosfet as a dimmer. Runs more efficiently, use a 555 or micro to drive.

Steve G

[jwax]

Thanks guys!
Can't use the PWM method, as the load is not a lamp and won't tolerate pulses.
So, what I need is a circuit to convert a DC input voltage of 0 to 5 volts DC to an output of 4.5 to 6.5 volts. DC scaling, if you will.
Looking now like an op amp will do the job. In all my years, never learned how to design around an op amp.
Can anybody direct me toward a basic tutorial/course of study to explore this analog lovechild?

[MrAl]

Hi,

First, using PWM is not more efficient than using a linear control or even a series resistor. It just gives you more control. The reason it is not more efficient is because the peak current is higher and this equates to the SAME power lost as with a series resistor. If you do the math you can prove this, and i did that several years ago but will repeat here if there are any non believers left.
PWM itself does not work better for efficiency, but a switching regulator does. The reason for this is because a switching regulator uses an inductor and that inductor allows a true power conversion to occur and that makes it much more efficient for most settings. Without the inductor though you just end up with pure PWM which does not improve efficiency.

You say you cant use PWM, but you can use a buck regulator. This would be best for efficiency, but of course it is a more complicated circuit. There are devices called "Simple Switchers" that may work for your application, or maybe other devices too can work.

If you want to stick with the single MOSFET design, then i suggest that you dont try to control the gate directly. This is because the gate voltage vs drain current is not as stable as we would like it to be, so the current could change for any given setting of gate voltage.
One way around this is to control the actual current through the MOSFET, while driving the gate. To do this you would use a current sense in either the source or drain, and using an op amp as the error amplifier, and that would drive the gate.

You say you having used an op amp before. Well the math behind this can be simplified in many ways. The simplest is probably the idea where you view the op amp as a difference amplifier with high DC gain. This turns it into a "voltage controlled voltage source", and greatly simplifies the analysis even for complicated circuits.
The main equation is simply:
Vout=(vp-vn)*Aol

where
Vout is the output voltage,
vp is the non inverting input voltage,
vn is the inverting input voltage,
Aol is the open loop gain of the amp, assumed to be high like 10000 or more.
All voltages referenced to ground (0v).

To use that equation, you first calculate the voltage at vn and at vp, then assume some high gain like 10000, then calculate Vout.

If you'd like to see an example or two i can provide some.

One question i have is what current does the MOSFET have to handle, and what is the voltage used for the device you intend to drive with the MOSFET?

[jwax]

Hi Al,
Here's the project: I want to monitor the output, in voltage and current, of a PV solar panel. The idea here is to sweep a resistive load across the panel from zero ohms (short circuit current) to open (no load), while monitoring (and logging) E and I along the curve.
The panel is connected in series with the MOSFET, and I hope to drive the MOSFET from "OFF" to "ON" with a 0-2.5 volt ramping voltage. Max current will be about 6 amps, and Voc (Open Circuit Voltage) about 60 VDC.

Any added resistance, such as a shunt, may keep me from getting a true short-circuit current reading, but that error could be offset in software.

Thanks, and I will be pursuing the op amp approach.

[sghioto]

I'm confused

Playing around with driving a power MOSFET to dim a lamp running off a 12 volt battery.

You said lamp now you say it's not a lamp

Can't use the PWM method, as the load is not a lamp and won't tolerate pulses.

Now you said,

The idea here is to sweep a resistive load across the panel

A resistive load, so what's the difference?

Steve G

[jwax]

My apologies. Originally said lamp to simplify the solution of how to scale driving voltage of the MOSFET. I didn't think MOSFET driving would be load-dependent, but as Mr Al pointed out, I can use the current through the load to change the driving of the MOSFET to be more linear. The "load" is a solar panel, and is certainly not a resistive load.
In actuality, the "load" in this case is the source of power, and the device-under-test. The MOSFET is acting as the "load"!
Weird, eh?

[MrAl]

Hi John,

Do you need to do this measurement repeatedly or just once or twice?
If just once or twice you can probably just use meters.

So the MOSFET and amplifier would simply be configured as a linear current regulator, like a precision current sink which you can probably find schematics for on the web. You dont need speed so you can use a high quality very low input offset op amp for better control.

[sghioto]

First, using PWM is not more efficient than using a linear control or even a series resistor.

MrAL the more "efficient" I was referring to was the fact that the mosfet will consume less power in the form of heat when driven using PWM in a lamp dimming application, as was the original request. That's why those little LED dimmers can control 8 amps without a heatsink. Try that with a linear drive control.

Steve G

[jwax]

Ideally, the measurements will be made every 50 msec or so, over the sweep of several seconds. An Arduino will record the E and I values, to eventually plot a curve of the panel performance.
Thanks guys!

[gerty]

If you use the MOSFET in the linear region plan on a healthy heat sink, they do not like it there..

[jwax]

Thanks. During the sweeping of the curve, the MOSFET will go from fully OFF to fully ON, and yes, will be warming up due to that transition.
The curve will be swept every minute, but since the curve only takes a few seconds, our current heat sink (large- 4"X6") should be adequate to dissipate the solar panels full output for those few seconds.


As an aside, we have some nearby test equipment that went crazy (electrically) during the time we were testing the solar panel curve tracer. After some oscilloscope observations, we found the panel/MOSFET/wiring combination, at one point in the curve, broke into a beautiful sine wave oscillation at around 100 MHz. That radiation upset the nearby electronics.
Mounting the MOSFET with an insulator off the heatsink provided the solution to stop the oscillation.
RF can be crazy!

[jwax]

OK, now we're getting into some power!
4 solar panels in series at 280 watts each. Voltage of the string is 120 VDC, and a current of about 8 Amps DC in full sun. Almost a kilowatt.
Used a few different MOSFETS:
IXFK160N30T, rated 300 volts, 160 Amps
IXTK180N15P, rated 150 volts, 180 Amps
12N50, rated 500 volts, 12 Amps

As well as a few smaller devices. They all go gate-short after a few minutes of operation. All well heat sunk, but they don't last long enough for much heat to get to the sink.
I'm sweeping the gate through a 10K series resistor with a 4-8 VDC ramp, in about 50 msec, once a minute.
The solar array is connected directly Source to Drain of the device.

Although well within specs, it's obvious I'm missing something in this design. What can I do to prevent the frying of fets?