IRFP250 MOSFETs: review and a little about how to use them.

Besides the plastic bag, the transistors were wrapped in a soft strip (I can't remember what it's properly called).


What surprised me, though, was that each transistor came in its own little bag.
I ordered a lot of ten because my specs called for eight, and all ten were included.

Not long ago, I published a couple of reviews describing a current-sense amplifier and a small power supply.
This review is the next chapter in my saga of designing a DIY electronic load. I’ve already written about a device like this, but this version was planned even before that one, and was intended to be more powerful, with electronic control and other bells and whistles. I’ll probably describe the “brains” next month, but today I’ll cover the power stage.

I planned the power module for 200–300 W, with a maximum voltage of 60 V and current of 15 A.
The device uses an unusual 8 V supply voltage for the control electronics. The control signal voltage is also 0–250 mV. This isn’t just my whim; it’s what the control unit can provide, so I designed the module to work with it.

The original design called for a single channel with a maximum current of 5 A and a 50 mΩ shunt. However, the device description said that you could add a couple more identical channels and recalibrate it for 15 A.
I decided to take a slightly different approach. To start with, I planned eight channels instead of three.
I based the design on a modular layout, which makes it easier to build and calculate.
The plan was for 8 channels: 2 channels per board, 2 boards per heatsink, and 2 heatsinks per device.

First, here’s the power-stage schematic.
The values of many components can be changed over a wide range, and different MOSFETs can be used as well.
I needed the voltage across the shunt in each channel to reach 250 mV over the full current adjustment range.
That works out to 15/8 = 1.875 A per channel. So, to get 250 mV, the shunt value needs to be 0.25/1.875 = 0.133(3) Ω. It’s better for the shunt value to be slightly lower, but not higher; otherwise, there won’t be enough adjustment voltage (max. 250 mV).
I decided not to bother with custom shunts and simply bought a hundred precision 1.33 Ω, 1% resistors. For assembly, I used 10 in parallel, arranged in two rows of 5.


I laid out the PCB from the schematic, though I later found that the pads for the power wires were a little too small; they’d be better off larger.
When laying out the board, I tried to make the power section as symmetrical as possible around the ground wire and measurement shunt connections.


After making the PCB, I made four boards at once on a single blank. The process is described here.


Parts list:
Resistors:
1.33 Ω, 1% — 80 pcs (1206)
22 Ω — 8 pcs (1206)
560 Ω — 4 pcs (0805)
6.2 kΩ — 8 pcs (1206)
22 kΩ — 8 pcs (1206)
3 MΩ — 8 pcs (0805)

220 µF, 16 V, 105°C capacitor. Samwha RD.
LM358 op-amp — 4 pcs (SO-8)
TL431 adjustable shunt regulator — 4 pcs (SOT23)
MOSFETs — IRFP250 — 8 pcs


The boards are soldered up. As I mentioned, the shunt resistors are mounted in two layers, with 5 resistors in each layer.


The back side has only an electrolytic capacitor. Since the boards are installed near components that generate a lot of heat, it’s best to use capacitors rated for operation at temperatures up to 105°C.


Since the transistors generate a lot of heat during operation (the whole point of an electronic load is to turn everything into heat), I got a couple of heatsinks ready. These heatsinks have already appeared in some of my reviews, for example this one, so now I’ll have to find replacements for them.
I removed the transistors and some almost-antique voltage regulator ICs from the heatsinks.
Then I cleaned the heatsinks with cotton wool and alcohol :)
Finally, I shortened them a little. This was one of the hardest steps. The heatsinks were 88 mm tall, while the enclosure was 84 mm tall. To make room for 80 mm fans, I cut 3 mm off each side. The tricky part was cutting those 3 mm off the length and trying to make the cuts straight :)
The heatsinks are 100 mm long, with 25 mm fins and a 4.5 mm base. They’re black anodized and have 9 fins.


I marked the mounting holes for the fans. I think this photo makes the planned power module layout clear.


I marked and tapped a whole bunch of holes. I didn’t bother figuring out which side would be the top and which the bottom; I just made everything symmetrical so I wouldn’t have to think about it during assembly. In other words, you can install the module upside down if you like—it’ll still mount properly, with the mounting holes in the same places. I’ve been using a small cordless screwdriver for drilling and tapping for a long time; it’s very handy.


The boards are ready to install. The photo shows how they’ll be mounted. I spent a long time deciding whether to install the boards parallel or perpendicular to the heatsink, but settled on the parallel arrangement because it’s more compact.


The heatsinks and everything that will be mounted on them—or almost everything. I’m also planning to add thermal protection components and so on…
Speaking of thermal protection: since the device generates a lot of heat, I installed a bimetallic thermal switch on each heatsink for safety. The trip temperature is 90 degrees, and the contacts are rated for 10 A. Since each switch handles only half the total current, though, I think they’ll work fine at 7.5 A.
The thermal switches have different terminals: one can be soldered to normally, but the other can’t. That was news to me. As it happened, I accidentally installed them in the same orientation, so I soldered the corresponding terminals. For the other terminals, I used lugs and soldered the wires to those. Keep this in mind.


First test fit. No thermal paste yet; I just wanted to see how everything would look together.
To mount the transistors, I used the original holes left by the previous components. It worked out so that each transistor sits roughly in the center of its quarter of the heatsink. If you’re building one yourself, try to position the transistors the same way.


For the connections, I grabbed a bunch of different wires. Some even looked like audiophile cable—apparently silver-plated—but what I liked was that they were made up of a huge number of very fine strands, making them very flexible, while still having a cross-sectional area of 2.5 mm².
I used this cable for the ground connection.


For the connections, I used a star topology: all the ground wires meet at a single point, positioned so that the resistance to each board is identical. This will distribute the current evenly between the modules.


The module is almost assembled. I used holes left over from the old components to route the wires.


The intake fan is a Sunon EE80251S1-A99. I chose it for its low price and high airflow.


The exhaust fan is a Thermaltake S0801512M. I had it on hand, and used it because it was thin. The enclosure is very small, so space is tight.
I plan to use a supply voltage of up to 15 V, but I also tested it at 20 V and it worked fine.


The ground-wire connection is between the heatsinks. This is far from ideal, as is putting any wires there at all. But I had no other options: routing them around would have been too long, and running them underneath or over the top was out of the question. I’d be glad to hear any suggestions for improving the design.
Of course, I’ll close the gaps at the top and bottom between the heatsinks. I haven’t decided how yet; for now, I’m thinking of simply sealing them with a couple of layers of tape.


The power module is assembled and soldered. All that’s left is to test it :)


Just in case anyone decides to build one, here’s a more detailed photo.


And of course, we have to test it :)
For the test, I set the load current to 5 A and applied 40 V (actually, 41 V).
The power dissipation was 204 W. I didn’t push it any further because only one fan was running during the test (the one visible in the photo, which looks like it’s off). It was powered from 8 V, and the gaps between the heatsinks were still open.
I supplied the control voltage from a potentiometer.
That works out to 25 W per transistor. By the way, don’t be misled by the maximum power dissipation listed in the transistor datasheet. In linear mode, it’s best to stay below 25–30% of the rated value, since the transistor’s die cells can start to fail (MOSFETs are essentially made up of a large number of tiny cells).


I think this stage of the project has been a success. I plan to continue soon—or rather, to finish the device completely. I’ll describe the process in one of next month’s reviews.