Used FDA59N30 transistors and a comparison with new ones

I’ve covered buying used components from AliExpress/Taobao on my blog several times, but I decided to play roulette once again, and it looks like I got lucky this time.

To keep up with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about every new article there, and might also write about something interesting from time to time: link to my account

You can also subscribe to updates and new articles on my Telegram channel - https://t.me/KirichBlog


But I’ll start today’s review not with the item I received. Instead, let’s go back about two years. That was so long ago—it feels like a past life...
So, back then I bought some transistors for a new electronic load from Kosmodrom, a store in Kharkiv. I still haven’t built the load, though. First I got sick, then there was work, then we went to the seaside, and then the war started. In short, it didn’t work out.



FDA59N30 transistors. There’s nothing particularly remarkable about them, but they do have a few features that make them safe to use in electronic loads.
First, the rated power dissipation is 500 W, though you shouldn’t take that figure at face value: you’ll only get it if you can keep the temperature at 25 degrees. The datasheet explains that above 25 degrees, you need to derate the power by 4 W for every degree over that.
I expect the transistors could easily reach 100 degrees, which means they could dissipate around 500-300=200 W. In reality, I think it’ll be closer to 150 W, since even keeping them at 100 degrees isn’t that easy.


The second thing young bloggers often forget when they build something out of whatever components they happen to have on hand is the safe operating area. Most transistors are designed for pulsed operation, but some have a curve on the SoA graph for DC operation as well.
Here, DC operation is allowed across the full power range, subject to the derating I mentioned above.


Anyway, I was interested in the FDA59N30 transistors: 300 V, 59 A, 47 mΩ, DC operation, high power—what’s not to like? So I bought 13 of them.
You’ll probably ask why such an odd number. It’s simply because I originally planned to open one up.


They look new, and they’re supposed to be genuine. I once spoke with the store’s management about this, and they assured me that they order components like these in large batches, and not from China.


First, I checked the gate capacitance. This is where many buyers make a serious mistake: they measure it with a multimeter or some super-mega-giga tester by simply touching the gate with the probes.
According to the datasheet, these transistors have a typical gate capacitance of 3590 pF and a maximum of 4670 pF. As you can see on the meter’s display above, it reads 5978 pF.


But people often overlook the conditions under which it was measured. Yes, the datasheet specifies a frequency of 1 MHz. Unfortunately, I can't measure at that frequency, but in practice it doesn't make much difference. What does matter is the voltage between the transistor's drain and source, which in this case should be 25 V.
I apply the voltage and measure the capacitance again. For the sake of statistics, I tested six transistors, and all of them fall within the specified range.


The second test is also commonly performed: measuring the on-resistance. But first, I'll check how the resistance depends on the gate voltage.


The test was performed in six modes, with gate voltages of 4, 5, 6, 7, 8, and 15 V. You can see that at just 6 V, the resistance is practically the same as the measurements at 15 V, when the transistor is guaranteed to be fully on.




But there's another measurement caveat here. The datasheet says to measure the resistance with a gate voltage of 10 V and a current of 29.5 A, which is 50% of the maximum current.
It wasn't convenient for me to run nearly 30 A through it; I'd have had to rig up a battery and an electronic load, so I tested it at 18 A, or 30% of the maximum current.
There's another wrinkle: the specs are given for a temperature of 25 degrees, but once current is applied, the temperature naturally starts to rise and the readings begin to drift.
A rough calculation gives a voltage drop of 0.805 V, which means the on-resistance is 0.805/18 = 0.045 Ω, or 45 mΩ—fully in line with the stated specs.


Then I spotted these same transistors on AliExpress for around $10 for 10 pieces, including shipping — link.
Actually, no: ordering 10 pieces would cost more than $10. You could get 10 for that price if you ordered 20, since the seller's shipping charge doesn't increase proportionally.
There are lots of good reviews, both for this listing and others. The seller says the components are genuine, salvaged parts, and some listings even show the transistor dies. People say the seller is reliable; some received a few faulty ones and were compensated for the difference without any hassle.

This kind of thing is fairly common. You really can sometimes buy salvaged components that are genuine branded parts.

I ordered 20 pieces and, a little over a month later, received a small bag in a consolidated shipment.


They do look used: some have marks from mounting, lacquer, desoldering, and so on.


The transistors are from different manufacturers: some are Fairchild, others Onsemi. Interestingly, they have the same datasheet, but the Onsemi version lists both manufacturers, while the Fairchild one lists only one.
What's more, most of the components are from different batches. In the photo, I've grouped them by their markings.


The font differs slightly on some of them, but they all look absolutely identical on the back. The surface with the markings is glossy and doesn't look like it's been sanded. The marking in the small round recess in the middle is still there, too.
Out of curiosity, I compared a couple of Onsemi parts with an original transistor (the one in the middle in the photos below). There's a slight difference, but it also varies a little from one part to another. I think that if they had been remarked, they'd all have been done using the same template.


Let's test them using the same method. I did an initial screening with an ordinary multimeter, checking for shorts, gate capacitance, and whether the transistor turned on when I applied voltage to the gate using the same probes. There were some minor differences, but all the transistors work.

First, I'll measure the gate capacitance, again with no voltage applied. In practice, we get the same 5900 pF as with the originals.


I started by testing three Fairchild parts; they all basically meet the datasheet specs.


Next, I tested three Onsemi parts. With no power, they also measured 5900 pF.


With power applied, the capacitance is slightly lower than on the originals, but still within spec.


Just out of curiosity, I tested one transistor bought from a local store under the same conditions. The difference is probably within the normal spread of component characteristics.


I checked how the on-resistance of some transistors depends on the gate voltage, and I didn't just pick them at random. During the initial check, I measured the channel resistance by turning the transistor on using the multimeter's test voltage. That is, Basically, in diode-test mode, I touched the gate with the probes first, then measured the voltage drop across the transistor once it had turned on. Most were around 1.9 V, but a few stood out: one was 2.4 V, another 1.6 V. For the test, I picked those two and one with the usual 1.9 V.


I’ll put the results in a table later. Here’s a photo from this stage for one of the samples, which I think is from Fairchild. I’ll include the manufacturer and batch in the table. You can see that its resistance is higher than the original I bought locally.


A table of the four transistors tested, with their batch numbers. The noticeable difference between the Fairchild samples and the Onsemi ones is a bit concerning; the latter’s results are very similar to those of the transistors I bought locally.


Naturally, I also tested them at 18 A, just like the ones I bought at Kosmodrom. I could have gone up to 24 A, but even at 18 A the cables were struggling.


Results for the same four samples measured above, in the same order—you can tell from their batch numbers.

The two Fairchild samples are at the top, and the two Onsemi samples at the bottom. The difference is noticeable here too, and it turns out the results are consistent within each manufacturer.
Let’s do the math.
For the first two, let’s say the voltage drop is 1 V. That gives a resistance of 1/18 = 0.055 Ω, or 55 mΩ.
For the second pair, the drop is about 0.85 V, so 0.85/18 = 0.047 Ω, or 47 mΩ.

The datasheet specifies a typical resistance of 47 mΩ and a maximum of 56 mΩ. The second pair is right on target, but the first is at the upper limit, which is rather unpleasant.


More out of curiosity than anything, I tested all the remaining transistors too. Most were made by Fairchild, but there’s a photo of three Onsemi samples at the end.
So, out of 20 pieces, 15 were Fairchild and 5 were Onsemi.
But that’s not the point. Looking at the first 13 photos—the Fairchild transistor measurements—you can see that almost all the results are more or less the same. But two are clearly different, and their results match those of the Onsemi samples (the last three).



Those two Fairchild samples also look different on the outside: they have traces of lacquer on the case, flange, and leads, and they’re from the same batch.


As I wrote earlier, I originally planned to open up one of the transistors I bought locally, but when I saw its current price, I was a bit stunned. When I bought them, they cost 51.44, which I think was the price in the second column and came to about $1.9.
Now the price in the second column is 141.21, or about $3.8—that’s twice as much. I don’t know, maybe I’ll still open one up; a single transistor is only good as a replacement anyway.


I may open up the ones I received, but possibly after the second batch arrives. In the meantime, I came up with another way to estimate the die area.
As I wrote earlier, a transistor’s resistance increases as it heats up, which is normal. But if resistance depends on temperature, and temperature depends on how well heat is carried away—which in turn is directly related to the die area—then we can try comparing them some time after applying current.
In other words, the larger the die, the more effectively it transfers heat to the flange, and from there to the heatsink. Naturally, this only works for the same type of component, since other types may have a different response curve.

So, I took three transistors: one bought locally and two from AliExpress, one from Onsemi and one from Fairchild.
Then I tried to give them identical conditions. The only difference was that the Fairchild transistor was mounted on a warm heatsink, which introduced some error into the initial reading, but shouldn’t affect the relative measurements.

The test was roughly time-based, with each cycle lasting about 4–5 minutes. Since the increase was very small toward the end, we can compare the starting and ending values.

1. Transistor bought locally: initial 0.84 V, final 1 V, a change of 0.16 V, or +19%.
2. Onsemi from AliExpress: initial 0.82 V, final 0.976 V, a change of 0.156 V, or the same +19%.
3. Fairchild: initial 1.055 V, final 1.3 V, a change of 0.245 V, or +23%.

The Onsemi results match those of the locally bought sample exactly. Fairchild’s are higher, but we need to account for the fact that its temperature was proportionally higher at the end of the test: the first two dissipated about 18 W at the end, while the Fairchild dissipated 23.4 W.


Now for the conclusions.
Subjectively, the transistors are genuine Fairchild and Onsemi parts. The gate capacitance measurements suggest that, and they do fit within the datasheet specifications. I have no doubts at all about the Onsemi parts or the pair of Fairchild samples with lacquer, but unfortunately I can’t be as confident about the rest. The difference is especially noticeable in how the resistance depends on the gate voltage, and that makes me uneasy.
They don't look much like reworked parts or counterfeits, though of course anything is possible. I plan to test how they behave at high power dissipation in linear mode a little later.

For now, I've ordered some more transistors: three types, 20 pcs. of each, including the ones from the review, HY3708 and SPA20N60C3. The price was fairly low, and I also saved on shipping. I was going to order a couple more types, but changed my mind for now. I'll wait and see what arrives; maybe I'll open them up while I'm at it.

What do you think: are these genuine or counterfeit? I'd especially like to hear your arguments for and against.

That's all for today. I hope you found it useful, and happy May 9. It's a shame my grandfather didn't live to see this day, though given the current situation, perhaps that's for the best. I wouldn't want him to see this.