Choetech PD6008 power supply (charger) with an output power of up to 100 W

I finally got my hands on a PD power supply that can deliver up to 100 W through a single output. As usual, I’ll take a look, test it, and of course take it apart.


First, let me explain: 100 W isn’t as much power as it might seem, and there are plenty of chargers like this around. But when it comes to delivering that much power through a single port, the options become much more limited.
The list of such models is gradually growing, which is good news: 3 × 33 W or 2 × 50 W isn’t quite the same as 1 × 100 W, even though the total power is the same. For example, you can power a laptop, quickly charge devices with large batteries, or even just power something out in the field.

But let’s get on with the review.
The packaging is identical to that of the previous models I tested. As it turns out, I’ve collected three of them.
There’s a picture on the front, technical specifications on the back, and a tray with the power supply inside.


The contents are no different from those of the previous models: the power supply, a manual, and a little information sheet.


There’s nothing new in the manual either, apart from the differences in the technical specifications, of course.
Power: 100 W
Outputs: Two USB Type-C ports
Load: USB 1/2 - 5 V/3 A, 9 V/3 A, 12 V/3 A, 15 V/3 A, 20 V/5 A - single output in use
Load: USB 1+2 - 5 V/3 A, 9 V/3 A, 12 V/3 A, 15 V/3 A, 20 V/2.25 A - both outputs in use



Externally, it’s the same square case with rounded corners. The mains plug is on one end, and two USB ports and a power indicator are on the other.
The mains plug isn’t detachable, which makes it less convenient for people who travel a lot.


For a sense of scale, the two “ordinary” chargers at the top are rated at 18 W and 60 W; at the bottom are a 61 W charger and the 100 W model under review. Both use components built with GaN technology.


The mains plug pins angle slightly toward each other. It stays in the socket just fine, but there is one minor issue: in some situations, the charger’s case can get in the way of “neighbors,” such as a switch.
The USB sockets grip plugs very tightly. If you pull on the cable, you’ll probably pull the charger out of the socket before the cable comes loose.


Unfortunately, the list of supported protocols is very short. It basically includes only QC (Quick Charge) and PD (Power Delivery).
1, 2. Both ports work identically when tested separately.
3. In single-output mode, either port can deliver up to 20 V at 5 A.
4, 5, 6. QC starts at 3.75 V and goes only up to 12 V. That’s a big mystery to me, since the power supply can physically deliver 20 V, so this must be a software limitation.
7, 8, 9. The other modes aren’t supported, either during automatic detection or in manual mode.


The second tester basically showed the same thing.
1, 2. In single-output mode, the protocols are the same, and power goes up to 100 W.
3. When both outputs are used, the list doesn’t change, but in PD mode the power is limited to 45 W per channel (20 V at 2.25 A).
4, 5. The channels work independently: for example, one can deliver 5 V while the other delivers 20 V, or vice versa.
6. If you disconnect the load from one output, the power supply automatically switches to single-output mode, which allows higher power. Either output can be used this way, so they’re identical in this respect too.

In short:
One output connected, whichever one - up to 100 W if the load can request it, or 60 W if it can’t.
Both outputs connected - up to 45 W each, with the voltage and current settings for each load independent of the other.

Just a reminder: to get 100 W, you need either a special cable or a device with the appropriate negotiation chip. Otherwise, power will be limited to 60 W (20 V at 3 A).



The product page had a table listing the protocols, but for some reason I couldn’t find FCP or AFC. The seller later said the table was incorrect.


For testing, I used an EBC-A10H electronic load rated for up to 150 W. I had to rig up a way to connect it to the charger “with sticks and string,” though, so I took apart one of my USB testers.


First, a note: I ran quite a few tests, but I’ll show only some of them, since it turned out the load characteristics are exactly the same in QC and PD modes up to and including 12 V.

At 5 V, the power supply delivers up to 3.3–3.4 A, after which the protection kicks in. There is no increase in output voltage as the load rises; this applies to all operating modes.


I noticed an interesting behavior: if the power supply trips its protection at 5 V and you immediately run the test again, the current is limited to 1.6–1.7 A instead. I assume this is intentional.


Also, since I ran some of the tests with both outputs in use, I noticed that protection tripping on one output has no effect on the other. Below are two photos, before and after the protection trips. The second tester was set to 15 V and continues to show that voltage even after the first output resets.


12 V, current up to 3.4 A


20 V, both outputs in use, current up to 2.4 A with a 2.25 A limit


20 V, one output in use; the power supply can deliver up to 5.7–5.8 A with a 5 A limit.


But it’s not all smooth sailing. Disconnecting or reconnecting the second output causes two things:
1. The power supply switches the PD profile from 2.25 A to 5 A and back.
2. When switching, the outputs reset to 5 V. The GIF shows that it was initially set to 20 V, but then dropped to 5 V.


I also measured efficiency at 5, 12, and 20 V, with load currents from 0.3 A to 3.3 A at 5 and 12 V, and up to 5.4 A at 20 V. The measurement was taken after the connector, which slightly lowers the result because of contact resistance.
The scale at the bottom is marked in 0.3 A load-current increments.


The thermal test was carried out in several stages and took about two hours in total. I started at 2.5 A (50% load), then moved on to 75% and 100% load. After that, as a bonus, I decided to push it to the maximum, at around 110% load. Each stage lasted 30 minutes.


The first-stage data isn’t interesting. After the second and third stages, the case temperature was 47 and 58 degrees according to the IR thermometer, and 51/62 degrees according to the thermal camera. I’ve mentioned before that there can be a significant difference depending on how and with what you measure temperature. The thermal camera shows the temperature at the hottest point, while the IR thermometer gives the average case temperature.


After two hours, with the last half hour at 110% load, the temperature was 62/68 degrees on top of the case and 74/81 degrees on the bottom, according to the IR thermometer/thermal camera, respectively. Overall, that’s pretty hot. The power supply was noticeably warm. It wouldn’t cause a burn, of course, but it was unpleasant to touch.


Ripple in the various operating modes ranged from 70 to 170 mV, which is perfectly normal.
1, 2. No load, 5 and 20 V
3, 4, 5. At 3 A and 5, 12, and 20 V
6. At 5 A and 20 V


Probably the most interesting part is the teardown. I have to say, the case is glued together really well. At one point, it looked like I’d have to get a milling cutter and cut it open, but I managed with a knife and a small hammer.


The layout is very compact, although there’s a little room near the mains plug. I think a universal mechanism would fit there, with a plug that has flat pins and an adapter attachment.


There’s a copper shield on both sides. As with the models I tested earlier, almost the entire underside of the board is covered with thermally conductive rubber, which is glued in places with something like silicone sealant. It’s almost impossible to remove without damaging it.


I was surprised to find two diode bridges installed. At first, I thought the designers had decided to fully separate the channels, but it turned out they hadn’t—the diode bridges themselves are simply connected in parallel.


Near the output are 9 mΩ current-sense resistors.


The power stage uses controllers from Power Integrations — INN3379C. The synchronous rectifier transistors are located next to them. The channels are completely identical, and essentially the charger consists of two 50 W power supplies.


According to the datasheet, this series includes several controllers. The model used here is rated for 65 to 85 W, depending on the input voltage range and operating conditions. In this power supply, the rating is lower—65 W—but even so, there is a decent power margin, since even in the worst-case mode each channel is limited to 55 W.
The controllers use PowiGaN power transistors and are rated for up to 100 W without a heatsink.


The datasheet also includes several typical circuit designs. I doubt the charger manufacturer deviated much from them, at least in the parts involving the specified controllers.


Taking it apart didn't come without consequences: later I noticed that I'd cracked the body of one of the diodes. I checked it with a multimeter and it was still working. I meant to replace it, but somehow forgot about it.


The input has a proper mains filter, a thermistor, a fuse, and a bunch of electrolytic capacitors.


Based on what I could make out on the capacitor cases, I estimated the total capacitance at 56+56+22+22+22+22=200 µF. It turned out to be less in reality—160 µF—but I can say that even this is more than enough for normal operation on "our" mains voltage.


There are also two transformers and a board with QC/PD controllers mounted on top.


The board has two IP2726 controllers.


Interestingly, the datasheet lists protocols that this charger doesn't support. Perhaps there's another software limitation at play.


In addition, the board has four power transistors. Two are used to connect and disconnect the output, while the middle pair is used to combine the channels.
In other words, when full power is needed and only one output is used, the outputs of the power supply units are connected in parallel.


Of course, you can't just connect power supplies in parallel: the load will always be unevenly shared, with more going to the one with the higher voltage. But I suspect the manufacturer uses the Constant Current function supported by the IP2726 controllers, which also relies on current-sense resistors.
In effect, two current sources are connected in parallel here, with a voltage limit set, so they can be used together.


Having found nothing else of interest, I put everything back together and glued it shut. It didn't turn out quite like before—a small gap remained—but it doesn't affect usability, and you can barely notice it anyway :)



To sum up, the only thing I could fault was the high operating temperature: at full power, the unit gets pretty hot. Still, it not only delivers all the output voltages and currents it claims, but can also handle a 10% overload.
I also had no complaints about the ripple level, or the design and components themselves—especially since it uses genuine Power Integrations controller chips.

I was a little disappointed by the list of supported fast-charging protocols and the lack of QC at 20 V. It's very strange that they're missing, since nothing in the hardware prevents their implementation. By the way, the product page claimed support for AFC and FCP, but either my testers couldn't send the right command to the charger, or it supports them only in very limited modes.

Now, about the price. Almost 38 dollars (there's a 2/20 coupon) is a bit much, of course. But the store periodically offers coupons and promotions that let you buy this and other models at a reasonable price. There was one just recently, so if you're interested, it may be worth waiting a little.