In a project (detailed elsewhere) I need to control some 12v LED strips. While researching, the RFP30N06LE is recommended as a suitable device in several places and offered by certain hobby electronics outlets for the purpose. I checked the datasheet and the Gate to threshold voltage is shown as 2v max, which should work. The big houses (eg: Mouser, Digi-key) list them as obsolete and discontinued, so I ordered a couple of 5-packs through Amazon.
On arrival, I bread-boarded the circuit to verify and check for heat dissipation, etc. The circuit didn't work. After the usual head-scratching and debugging, I discovered that the MOSFET was not turning on, despite having 3.3 v to the gate.
Further testing showed none of them would. The very lowest of the 10 started to turn on at about 3.6vdc. Here is a screenshot of the curve tracer output:
(note: DSOs struggle with X-Y mode due to the way they capture the data. The curve tracer outputs actual curves when viewed on a proper analog scope, but that doesn't have a print screen function.)
Clearly these would work fine with a 5 volt signal. And they do handle the couple amps of current I need when I use an NPN bipolar to interface with the gate. But that requires the PWM output to be inverted -- ie: max LED brightness = 0, off = 255, which is inelegant.
I don't know if these are counterfeit, culls left over from production or what. But if your circuit isn't working, it may not be the design.
Is the mosfet source connected directly to ground?
The threshold votage is where it just begins to conduct, but at 3.3V you should get at least some current. Does it in fact work fine with 5V at the gate?
I'm not sure the bipolar drive needs to have reverse polarity. Could you connect your PWM output to the emitter of the NPN instead of the base, and connect the base to your 3.3V rail? You wouldn't have any NPN gain, but it still might work. And it would be pretty elegant.
I should have said, for future readers, that you would need a base resistor too.
How fast is your PWM? Maybe the Arduino can't drive the gate capacitance hard enough. So if you connect the gate directly to the 3.3V rail, it still doesn't turn on? If that's the case, it seems you've got some dodgy mosfets, as you suspected.
This was even without any PWM input applied at all. I breadboarded a test circuit -- driving the MOSFET with one channel of the benchtop power supply to gate and another channel supplying + voltage to the load (~1.25 A, depending on voltage). I can vary supply and gate voltages +/- 1mVDC independently and monitor current draw +/- 1mA on each channel independently from the supply display. The lowest gate voltage that any of them would start conducting was 3.5 volts. Typical was 3.8v. Fully switched on (at that load current) by about 4 v (so yes, they do work with a 5v signal). That's when I got out the curve tracer to see what was going on (bearing in mind that my curve tracer is really designed for BJT power transistors and limits current to about 250mA).
But looking at the datasheet, these should be capable of handling 20 amps at 3v to the gate at 25°C. The Rds(ON) rating looks to be taken at 80A -- WAAAAY more than I need to drive some LEDs.
For now, I've ordered something else from someone else, so we'll see what happens there. I'm on the fence about returning these -- they're cheap enough to toss in the parts bin pending some future (5v) project.
Ok FET data sheets need to be read as carefully as a budget airline's terms and conditions.
The at 25°C bit means you have to keep the case at that temperature, not that this is the ambient temperature they will work at. This will be using an infinite heat sink, which is a mythical one that keeps he case at the same temperature as he ambient.
Look again at that peak current you will probably find that is only for a very short pulse over a set period of time and is not the continuous current.
You are miss understanding the 3V gate voltage as well, this is not a quoted Ron figure.
And as @Grumpy_Mike said, the current is being tested for very short pulse durations (here only 80us and 0.5% duty cycle!). And even then, the device goes into the saturation region at 10A with Vgs at 3V. Then, the mosfet forward voltage drop increases rapidly with a small increase in current.
Yeah, I get the pulse thing. That's one reason why I tried the curve tracer. It's running at 682 hz and sends something like 30 pulses per cycle. Call it roughly 50uS pulse duration? A bit fast (orders of magnitude faster than an Arduino PWM frequency), but the results were the same -- 3.6 volts before the best one turned on.
Badly phrased
Looking at Figure 7 above, beyond the knee, increase in Vds leads to minimal current increase. It is in saturation mode. The graph shows it as a horizontal line but I expect there will be a minimal slope to it.
Is it wired correctly? With the metal housing facing AWAY, the left pin is the gate (Arduino pin), the middle is the drain (device GND), and the right is the source (GND).
Just for fun, I dug around in the parts bin and found a couple of RFP50N06. The datasheet gives a gate threshold of 2 to 4v. Drain to Source on resistance value is specified with a Gate voltage of 10V. Pulse duration also 80uS on the Saturation Characteristics curves.
They work perfectly. In the breadboarded DC circuit they start to conduct at 2.6v, full on at 3.2v. Similar performance on the curve tracer:
While a curve tracer tells you about the individual component you are testing, it takes no account of the spread of parameters across many device. The danger here is that you might say "this type numbers works" where you actually mean "I found this type number to work".
If you are going on to publish this work, in any way, even a forum post, this will fool beginners into thinking if they get the same part number it will work. This can result in frustration and waste of money and leave you with a typical unreliable Instructables type project report.
You are absolutely correct, my apologies for any confusion. I did not mean to suggest that anyone should seek out the RFP50N06 for use with a 3.3v logic level. As I noted, the datasheet shows the gate threshold may be anywhere between a minimum of 2v and a maximum of 4v -- ie: designed for 5v logic level. The fact that I happened to have a pair that operated at the low end of that range -- within the 3.3v logic level -- was luck, not design.
The intent of the post was to illustrate the low-voltage gate performance characterized by both the pulsed curve tracer output and the manufacturer's recommended straight DC test circuit (see figure 18 on the respective datasheets) on a device within the same product family. It was in response to questions raised about pulse duration and drain current further along the gate current curve.
The intent of the thread was to educate others who have bought or are thinking of buying these devices:
They have been discontinued by the manufacturer.
They are no longer available from the large suppliers, where stock levels are typically numbered in the thousands.
Therefore small retailers are not getting replacement stock through normal supply chains.
Posts in various internet forums such as this indicate they are suitable, creating a demand.
There are unscrupulous persons in this world who are willing to dumpster dive and sell devices that failed testing.
There are unscrupulous persons in this world who are willing to counterfeit devices by labeling an inferior product with a popular (see #4, above) number.
Therefore if you are having unexpected results after buying one of these, the problem may be inherent in the device you bought, not the circuit design you used. Test the device as per the manufacturer's recommended test circuit.
You normally see that 2-4volt threshold with a non-liogic level 10volt fet.
Logic level fets usually have a 1-2volt threshold. Adafruit sells this one.
There are very few TO-220 fets that work with 3.3volt and high current.
If you want to use a TO-220 fet, and want it 'nomally off' then use an opto coupler between Arduino and fet.
Leo..
The gate threshold is not the important parameter. The important parameter is the minimum gate voltage that allows your project to function properly, over a specified range of operating conditions.
