Mosfet Circuit

[Marksd]

I have set up a op amp driving a IRF9510 P Channel mosfet. I have it regulated to 11.2 volts and it is rock steady for a Dynaco PAS 3. I get .12% regulation based on the current draws below.

Here is my question. I am at the limit of my skill set and have the following data to determine if this device will live long term and whether the mosfet is in saturation or linear region. I cannot tell how to really read the VGS data and correlate that to the IDS.

I have listed 3 scenarios at different current levels below. I can definitely say at 1.1 amps the heat sink I am using is not touchable. At .435A the heatsink is 51C at ambient of 25C. I have paralleled two of these mosfets and they definitely split the current fairly evenly.

Lastly, I am having the most trouble figuring out how to read the IRF9510 SOA of the device under the following conditions.

Amps = .220
VGS = 2.33V
VGD = 1.40V
VDS = 3.71V

Amps = .435
VGS = 2.40V
VDS = 1.25V
VDS = 3.67V

Amps = 1.1
VGS = 2.58
VGD = .733V
VDS = 3.391V


Thanks for any assistance tying to better understand how the device really works from the data sheet.

 

[PRR]

The 1.1A 3.4V condition is under 4 Watts. That IRF is a 43W part.

If the heatsink is too hot to touch, it "may" be OK. The part is rated for 175 deg C, far past the boiling point of water. However _I_ would not want it run that hot. If this is really your op-point, you want a bigger heatsink.

But why three test currents? You apparently want to heat a PAS. The heater current should be known. Not counting the rectifier, 0.6A, right?

Unless you have w-i-d-e wall-voltage variations, I don't see why you need to regulate a PAS's heaters. Add a resistor to compensate for loss of the Selenium rectifier(?). If you do have wide variations, that "3.391V" will be much less at low line, drop-out, no regulation at all. At high line, the IRF will be much hotter.

I can understand wanting to understand. The IRF9510's log-log plots make me dizzy. I have attached my understanding of your three points, curves and SOA. The curves appear to be "drain saturated" (good), though very slight drop of wall voltage puts you on the knee. The SOA would be safe if you had an infinite heatsink, which you don't. You must plot the safe Power for the heatsink and ventilation you have. I plotted a guess; very slight rise of wall voltage might put you over.

 

[AndrewT]

using your heatsink = Ts = 51degC
and assuming you have an Rth s-a of 1.2C/W
then 1.1A @ 3.4Vds gives a Pq=3.74W
The Tc will be 3.74W*1.2C/W = 4.488Cdegrees above the heatsink temp. i.e. Tc=51+4.5=55.5degC

The temperature de-rating for a 175C device is (175-Tc)/(175-Ta) = 119.5/150 = 0.797
your 43W device is de-rated to 0.797*43 = 34W when the heatsink is @ 51degC
you are passing 3.74W through that 34W device so that's only 11% of it's capability.

I would generally advise that the operating Pq use less than 50% of the temperature de-rated power capability.

Repeat the modelling for a 70degC heatsink and see what proportion of the capability you would be predicting.

For the sake of learning (not for actual operation) try repeating for a 100degC heatsink.
And try Ts=100degC, when Pq=2W

 

[indianajo]

PAS is a sealed box with no fan. Better couple the heat sink thermally to the back wall or somewhere else on the case. In fact a 50 mm square aluminum plate stuck to an external wall with BE compound is probably better than any heat sink with fins.

 

[Marksd]

To all who replied. Thanks. The PAS heater current is actually .6amps. I had a current peaking meter on a cold start up and it peaked at 1.1A. It steady stated at the .65 to .6A draw. I have a small Radio Shack 18degreeC/W heatsink on the mosfet. I can purchase one from mouser for $1.44US that is 9.8degreeC/W. I like to test many iterations over time to make sure I get this right. I started down the path with a LM317 regulator, but that requires 3 volts extra. The op amp option is a much better option. When finished I was going to fasten the aluminum heat sink to the rear of the chassis with mosfet leads extended to a terminal strip. I do appreciate the SOA operating area information. I had an estimate of the SOA I was in and you helped confirm I am reading this correctly. I am mostly self taught so thanks. Lastly, I paralleled two mosfets to split the current and it worked as perfect as the books say. it really does help to put them both on the same heatsink.