Three QC/PD Multi-Standard Voltage Converter Boards
I've already reviewed voltage converter boards with QC support, but you always want more: more power and support for more standards. So I ordered several different models to try out. Rather than write three separate reviews, I decided to cover them all in one.
The idea was to build these boards into something, and since there are now so many fast-charging standards, I wanted boards that support this whole assortment to make them convenient to use.
I'd like to build one board into another adjustable power supply for various tests, and possibly put another in a multi-port charger.
The review will be split into two parts: an introduction with a brief description and inspection, followed by testing.
I ordered everything from Taobao. Prices there are a little better than on AliExpress, where products often show up much later or at wildly inflated prices. Once people start buying something in earnest, competition kicks in and prices begin to fall. For now, Taobao is the way to go.
I ordered three boards from three different stores. Two arrived in small bags, while the third was also wrapped in some kind of Chinese newspaper.
The boards are very similar, except that two of the three have two connectors (USB-A and USB Type-C), while the other has only USB Type-C.
I can already hear the question: why do you need USB Type-C at all when there are all kinds of USB-A adapters? That's right, adapters exist, but the PD (Power Delivery) protocol requires USB Type-C.
I'll start with the board shown in the title: the ZC823 from the well-known YZX brand.
This board costs about $8 (prices below exclude shipping) — link.
The input voltage is 8–24 V, but since the board can only step the voltage down, you need to supply at least 22 V for it to work properly.
You may have noticed the rather unusual shape of the circuit board. That's because this is a module available in several versions: the ZC822 with an IP6528 chip, the ZC823 with an SW3516, and the ZC826 based on the IP2716.
The ZC826 is the most expensive at about $20, while the ZC822 is the cheapest at just over $5.
The manufacturer offers both a finished product based on these modules and a "kit" that includes an enclosure, a backplane, and a front panel. The converter modules can be installed in any configuration, giving you a three-port charger.
The complete set—enclosure, backplane, and front panel—costs $2.65 + $2.90 + $1.05 = $6.60.
The board is neatly made and has the model and manufacturer marked on it.
Everything is clean. YZX is something of a brand in its own right, albeit a Chinese one. I've always been amused by Chinese manufacturers' fondness for names like these: YZX makes testers and chargers, another company makes ZXY converter boards, and a third makes electronic loads under the name ZKE.
1. The board is controlled by an SW3516 controller, with shunts and a pair of transistors nearby.
2. The converter's power transistors.
3. A two-color LED indicates the operating mode: red in normal mode and blue in PD mode.
4. A branded inductor—or at least it looks like one—with a pair of capacitors nearby. The product page says:
You may have noticed that the board has two shunts and two mysterious transistors. The reason is that the USB ports are connected separately, each through its own switch and shunt, and powered independently. As far as I can tell, when one port is active, the other is disabled. I don't know why it was designed this way, but while chargers often have a bunch of ports with shared protection that can all be used at the same time, this one is the opposite: it has two ports and two separate protection circuits, but you can't use both ports at once even mechanically—the connectors won't fit simultaneously.
Also, it's hard to see in the photo, but the USB-A connector has five contacts: the four familiar ones (ground, power, and two data lines), plus a contact that is connected to ground when a plug is inserted.
A little about the SW3516.
Unfortunately, all the available information was only in Chinese, so here's a Google translation of what this chip can do.
At the same time, product pages often claim support for QC4.0+, which isn't mentioned in the datasheet. However, it does state that the maximum input voltage is 40 V.
There's also a schematic showing both the power transistors and the output switches with their shunts—in other words, the same things I showed above.
The second board comes in a more familiar form factor. The packaging consists of two bags fastened together.
That's because the package includes a pair of alligator clips and an adapter for a 5.5/2.1 plug.
The specs are more or less the same as for the previous one. If I remember correctly, the kit costs about $3.6 — link.
The product page listed a whole bunch of operating modes, but the information was very unclear. They seem to be test results, so here's my interpretation.
USB-A port
Full Mate20 - SCP, charge 56 - 4.89 V, 4.59 A
P20 pro - SCP, charge 12% - 5.03 V, 4.25 A
V20 - SCP, charge 43% - 5.13 V, 4.5 A
Redmi 8 - QC3.0, 18% - 6.44 V, 2.79 A
OPPO R17 - VOOC, 43% - 4.8 V, 3.8 A
iPhone XS - charge 5%, 5.22 V, 2.79 A
One plus 6T - DASH, charge 15% - 4.7 V, 3.8 A
USB Type-C port
iPhone XR - PD, charge 7% - 9.17 V, 1.91 A
iPhone X - PD, charge 7% - 9.17 V, 1.51 A
Full Mate20Pro - PD, charge 55% - 9.2 V, 1.58 A
Huawei P20 Pro - PD, charge 10% - 9.17 V, 1.69 A
V20 - PD, charge 40% - 9.05 V, 1.7 A
Redmi 8 - PD, 6% - 6.1 V, 2.63 A
Like the previous one, its maximum output power is 60 W in Power Delivery mode at 20 V, 3 A.
The board has no enclosure, but it is wrapped in heat-shrink tubing. It looks quite neat to me, but I still had to remove the tubing to inspect it.
The inductor is simpler here, and the capacitors are lower quality, but you can clearly see that the USB connector also has 5 contacts.
The board wasn't cleaned very well; there are white residue marks.
1. The controller's marking has been erased, though I don't understand why. Judging by the circuit design, it's most likely the same SW3516 as on the previous board, so everything that applies to that board should apply to this one too (protocols, operating modes).
There is one difference, though: this board has a TVS diode not only on the input but apparently on the output as well. It also has three power transistors instead of two. The third one most likely protects the board against reverse polarity, but I haven't specifically tested that.
2. The output also has two shunts and a pair of switches, each serving a different USB connector. But while the previous board had identical 5 mΩ shunts, this one has a 4 mΩ shunt on the USB-A channel, so its maximum output current may be higher than on USB Type-C.
The third board has only one connector, unlike the previous ones, but you can choose USB-A or USB-C when ordering. The board costs $3.77 — link.
I ordered it for comparison because I wanted “more power,” but this board is rated for only up to 45 W, while the previous two are rated for up to 60 W.
The board is small and neatly made.
It's built around the fairly well-known IP6518 controller. The current-sense shunt is 20 mΩ, and the inductor is somewhere between the ones used on the previous boards. Overall, the board looks a little worse than the first one, but better than the second.
The underside is bare, and the solder pad for the input capacitor looks like it was “finished with a file.”
Unfortunately, there's a flaw: the output voltage is specified as up to 20 V, but the output capacitor is rated for 16 V. What a mess...
The following equipment was used for testing:
1. Modified EBC-A10 electronic load
2. Linear power supply for capturing waveforms
3. Switching power supply for load tests
4. USB-A tester
5. USB-C tester
6. Oscilloscope
7. Thermal imaging camera.
The load testing process looked something like this.
The first board supported all available protocols except QC 20 V.
1. Yes, 20 V isn't supported in QC mode; the maximum is 12 V :(
2, 3. The SCP protocol lets you adjust the voltage from 3.06 to 5.46 V. In my opinion, this protocol is a dead end, since the focus here is on high current, not high voltage.
4. FCP protocol, 9 V
5. Samsung AFC.
6. Power Delivery 3.0 is supported.
7. In most of the photos, you can see that a power cable is also connected to the tester. The reason is that in most situations, the board briefly cuts power to the output, most likely because it periodically switches between the USB outputs.
8. But sometimes it works without cutting out; I still haven't figured out the logic behind it.
9. In PD mode, the board outputs up to 20 V, so the QC limit is software-based.
Load tests.
QC: with 20 V at the input, the maximum output current is 3.2–3.3 A, regardless of the output voltage. As the current increases, the output voltage rises too. In all modes, overcurrent protection works correctly: once the overload is removed, the voltage recovers automatically.
For the Power Delivery test, I raised the input voltage to 24 V, since the board can output 20 V.
Everything here is exactly the same as in the previous tests.
To test the ripple level, I used a linear transformer power supply. The electronic load also has no switching power supply, and the computer was connected through an isolation transformer. The only device with a switching power supply was the oscilloscope itself.
With no load, the input voltage was 24 V and the output voltage was 5 and 12 V; the peak-to-peak ripple was about 60 and 30 mV, respectively.
Under load
1, 2. 5 V at 1.5 and 3 A: 40 and 45 mV
3, 4. 12 V at 1.5 and 3 A: about 65 mV
5, 6. 20 V at 1.5 and 2.5 A: about 45 and 25 mV
Temperature test: 24 V input, about 20 minutes of operation under a 3 A load at 5 and 20 V. Moderate heating.
Measuring efficiency and losses at different load currents and output voltages.
Table: blue = 5 V, red = 12 V, green = 20 V; the input voltage was 24 V in all cases.
And as a graph, with colors matching the table.
The second board also lacked QC 20 V and, for some reason, Samsung AFC.
Everything here is exactly the same as with the previous board, and Samsung AFC is indeed supported. By the way, in the third photo you can see the converter board trying to shut off the output: with 9 V set, the voltage periodically drops. This only happens when operating with no load. It doesn't consider the tester itself a load, though, and even then it happens inconsistently.
The test modes are identical to those in the previous test, and the results are almost the same too.
In QC mode with a USB-A connection, the maximum current is 3.6 A (the first board managed 3.2–3.3 A). The cutoff is just as abrupt, and the current is also independent of the output voltage.
PD mode, so this means connecting via USB-C.
The current is limited to the same level as on the previous board, i.e. 3.2 A. Above I mentioned that the second board has different shunts, 4 and 5 mΩ, while the first has identical 5 mΩ shunts.
Here I remembered that this board had been tested, and the output current was around 4.8 A in SCP mode through the USB-A output.
We set the required mode and start the test. The board delivered 5 A without any problems, but interestingly, the output voltage dropped as the current increased, whereas in all the previous tests it did the opposite. So the relationship between voltage and current is probably controlled by software, not hardware.
Ripple with no load is a little higher here: about 80 mV at 5 V and 30 mV at 12 V.
Under load
1, 2. 5 V at 1.5 and 3 A, 60 and 65 mV
3, 4. 12 V at 1.5 and 3 A, about 110 mV
5, 6. 20 V at 1.5 and 2.5 A, about 70 and 50 mV
Yes, the results are worse than with the first version; the cheap output capacitors are clearly to blame.
The temperature rise at 5 V, 3 A is the same; at 20 V, 2.5 A it's a little lower. Most likely, it simply “migrated” to the other side of the board. It was very inconvenient to photograph both sides, but I think the overall picture is still the same: it gets moderately warm.
Efficiency and power loss measurements at different load currents and output voltages.
Table: blue = 5 V, red = 12 V, green = 20 V; the input voltage is 24 V in all cases.
And as a graph, with the colors matching the table.
This board is based on the IP6518. The tester detected everything it possibly could, including QC at 20 V. That looks especially impressive given that the output capacitor is rated for 16 V. The Chinese guys are true to form.
But this is Power Delivery version 2.0, not 3.0 as on the previous boards, and the maximum load currents are also noticeably different.
Voltage - IP6518 - SW3516
5 V - 3.03 A - 2 A
9 V - 2.39 A - 3 A
12 V - 2.31 A - 3 A
15 V - 1.91 A - 3 A
20 V - 1.83 A - 3 A
Comparison of all three boards when used with the ZY1276 tester, in the same order. Power Delivery isn't detected because that requires a USB Type-C connection. I don't have that kind of cable, and in the third case I had to use a USB-A/USB-C cable and power the tester from the output.
The results are completely different here. As the current increases, the output voltage initially rises, as it does with the previous boards, but after the protection threshold it starts to drop gradually, which I think is less appropriate. The maximum current at 5 V was 3.5–3.6 A.
At 9 and 12 V, the results are roughly similar, so I'll hide them under a spoiler.
In QC mode at an output voltage of 20 V, I got 2 A, or about 40 W. Beyond that, the voltage dropped linearly as the current increased, which is especially clear in the graph.
There won't be any more tests, though. During testing, I frequently switched between boards, the load, testers, and so on. In the end, I accidentally shorted the board's output contacts. The board itself was powered by a linear power supply, but the current surge was enough to kill it. Apparently, one of the chip's transistors responsible for synchronous rectification blew, since the board tries to start up but ends up with a short circuit at the output. If I apply 5 V directly to the board's output, the power supply hits its 5 A limit and the chip starts heating up slowly. :(
Now for the conclusions.
Personally, I liked the boards based on the SW3516 better. Of course, the chip marking on the second board had been erased, but it really looks like an SW3516 too. The only thing to criticize is the lack of support for QC mode at 20 V. That seemed rather odd to me, especially since the board can output that voltage in PD mode without any problems.
Speaking of PD, this is already version 3.0, with output power up to 60 W, which is certainly nice. Both boards work well. The first one (ZC823) has better build quality and better components, which has a positive effect on ripple and should also improve reliability. Interestingly, though, at output voltages of 5 and 20 V, its efficiency was slightly lower than that of the cheaper option.
The second board is clearly from the "budget" range, especially judging by its appearance and cheap capacitors. Still, it has its place, particularly if you replace at least the output capacitors. It works quite well; the only significant complaint is the ripple level.
The third board, based on the IP6518, is a different story. It seems to use decent components, but for some reason they fitted a 16 V output capacitor instead of a 25 V one. It supports Power Delivery version 2.0, and its power is limited to just 40–45 W. The one thing in its favor is that it can output up to 20 V in QC mode. This board also doesn't cut power during switching, since it has only one output connector and doesn't need to switch between them as the first two boards do.
For objective reasons, I wasn't able to run the full load tests or measure ripple. I think you understand that it's technically impossible for now, but if I get my hands on another board like this, I'll certainly measure everything.
In short, the SW3516 focuses on higher power and PD 3.0 support, while the IP6518's only real advantage is QC at 20 V. So the first chip is a little newer.
That's all. I hope you found this information useful.
The idea was to build these boards into something, and since there are now so many fast-charging standards, I wanted boards that support this whole assortment to make them convenient to use.
I'd like to build one board into another adjustable power supply for various tests, and possibly put another in a multi-port charger.
The review will be split into two parts: an introduction with a brief description and inspection, followed by testing.
I ordered everything from Taobao. Prices there are a little better than on AliExpress, where products often show up much later or at wildly inflated prices. Once people start buying something in earnest, competition kicks in and prices begin to fall. For now, Taobao is the way to go.
I ordered three boards from three different stores. Two arrived in small bags, while the third was also wrapped in some kind of Chinese newspaper.
The boards are very similar, except that two of the three have two connectors (USB-A and USB Type-C), while the other has only USB Type-C.
I can already hear the question: why do you need USB Type-C at all when there are all kinds of USB-A adapters? That's right, adapters exist, but the PD (Power Delivery) protocol requires USB Type-C.
I'll start with the board shown in the title: the ZC823 from the well-known YZX brand.
This board costs about $8 (prices below exclude shipping) — link.
The input voltage is 8–24 V, but since the board can only step the voltage down, you need to supply at least 22 V for it to work properly.
You may have noticed the rather unusual shape of the circuit board. That's because this is a module available in several versions: the ZC822 with an IP6528 chip, the ZC823 with an SW3516, and the ZC826 based on the IP2716.
The ZC826 is the most expensive at about $20, while the ZC822 is the cheapest at just over $5.
The manufacturer offers both a finished product based on these modules and a "kit" that includes an enclosure, a backplane, and a front panel. The converter modules can be installed in any configuration, giving you a three-port charger.
The complete set—enclosure, backplane, and front panel—costs $2.65 + $2.90 + $1.05 = $6.60.
The board is neatly made and has the model and manufacturer marked on it.
Everything is clean. YZX is something of a brand in its own right, albeit a Chinese one. I've always been amused by Chinese manufacturers' fondness for names like these: YZX makes testers and chargers, another company makes ZXY converter boards, and a third makes electronic loads under the name ZKE.
1. The board is controlled by an SW3516 controller, with shunts and a pair of transistors nearby.
2. The converter's power transistors.
3. A two-color LED indicates the operating mode: red in normal mode and blue in PD mode.
4. A branded inductor—or at least it looks like one—with a pair of capacitors nearby. The product page says:
solid-state capacitor, Japanese Sibi Audio HIFI capacitor
You may have noticed that the board has two shunts and two mysterious transistors. The reason is that the USB ports are connected separately, each through its own switch and shunt, and powered independently. As far as I can tell, when one port is active, the other is disabled. I don't know why it was designed this way, but while chargers often have a bunch of ports with shared protection that can all be used at the same time, this one is the opposite: it has two ports and two separate protection circuits, but you can't use both ports at once even mechanically—the connectors won't fit simultaneously.
Also, it's hard to see in the photo, but the USB-A connector has five contacts: the four familiar ones (ground, power, and two data lines), plus a contact that is connected to ground when a plug is inserted.
A little about the SW3516.
Unfortunately, all the available information was only in Chinese, so here's a Google translation of what this chip can do.
The SW3516 supports dual-port Type-A + Type-C output, and either port can provide fast charging.
The Type-A port supports low-voltage fast charging with direct charging: QC3.0 / QC2.0 / AFC / FCP / SCP / PE2.0 / PE1.1 / SFCP / VOOC.
The Type-C port supports PPS / PD3.0 / PD2.0 / QC3.0 / QC2.0 / AFC / FCP / SCP / PE2.0 / PE1.1 / SFCP fast charging output with direct charging.
The Type-A port supports low-voltage fast charging with direct charging: QC3.0 / QC2.0 / AFC / FCP / SCP / PE2.0 / PE1.1 / SFCP / VOOC.
The Type-C port supports PPS / PD3.0 / PD2.0 / QC3.0 / QC2.0 / AFC / FCP / SCP / PE2.0 / PE1.1 / SFCP fast charging output with direct charging.
At the same time, product pages often claim support for QC4.0+, which isn't mentioned in the datasheet. However, it does state that the maximum input voltage is 40 V.
There's also a schematic showing both the power transistors and the output switches with their shunts—in other words, the same things I showed above.
The second board comes in a more familiar form factor. The packaging consists of two bags fastened together.
That's because the package includes a pair of alligator clips and an adapter for a 5.5/2.1 plug.
The specs are more or less the same as for the previous one. If I remember correctly, the kit costs about $3.6 — link.
The product page listed a whole bunch of operating modes, but the information was very unclear. They seem to be test results, so here's my interpretation.
USB-A port
Full Mate20 - SCP, charge 56 - 4.89 V, 4.59 A
P20 pro - SCP, charge 12% - 5.03 V, 4.25 A
V20 - SCP, charge 43% - 5.13 V, 4.5 A
Redmi 8 - QC3.0, 18% - 6.44 V, 2.79 A
OPPO R17 - VOOC, 43% - 4.8 V, 3.8 A
iPhone XS - charge 5%, 5.22 V, 2.79 A
One plus 6T - DASH, charge 15% - 4.7 V, 3.8 A
USB Type-C port
iPhone XR - PD, charge 7% - 9.17 V, 1.91 A
iPhone X - PD, charge 7% - 9.17 V, 1.51 A
Full Mate20Pro - PD, charge 55% - 9.2 V, 1.58 A
Huawei P20 Pro - PD, charge 10% - 9.17 V, 1.69 A
V20 - PD, charge 40% - 9.05 V, 1.7 A
Redmi 8 - PD, 6% - 6.1 V, 2.63 A
Like the previous one, its maximum output power is 60 W in Power Delivery mode at 20 V, 3 A.
The board has no enclosure, but it is wrapped in heat-shrink tubing. It looks quite neat to me, but I still had to remove the tubing to inspect it.
The inductor is simpler here, and the capacitors are lower quality, but you can clearly see that the USB connector also has 5 contacts.
The board wasn't cleaned very well; there are white residue marks.
1. The controller's marking has been erased, though I don't understand why. Judging by the circuit design, it's most likely the same SW3516 as on the previous board, so everything that applies to that board should apply to this one too (protocols, operating modes).
There is one difference, though: this board has a TVS diode not only on the input but apparently on the output as well. It also has three power transistors instead of two. The third one most likely protects the board against reverse polarity, but I haven't specifically tested that.
2. The output also has two shunts and a pair of switches, each serving a different USB connector. But while the previous board had identical 5 mΩ shunts, this one has a 4 mΩ shunt on the USB-A channel, so its maximum output current may be higher than on USB Type-C.
The third board has only one connector, unlike the previous ones, but you can choose USB-A or USB-C when ordering. The board costs $3.77 — link.
I ordered it for comparison because I wanted “more power,” but this board is rated for only up to 45 W, while the previous two are rated for up to 60 W.
The board is small and neatly made.
It's built around the fairly well-known IP6518 controller. The current-sense shunt is 20 mΩ, and the inductor is somewhere between the ones used on the previous boards. Overall, the board looks a little worse than the first one, but better than the second.
The underside is bare, and the solder pad for the input capacitor looks like it was “finished with a file.”
Unfortunately, there's a flaw: the output voltage is specified as up to 20 V, but the output capacitor is rated for 16 V. What a mess...
The following equipment was used for testing:
1. Modified EBC-A10 electronic load
2. Linear power supply for capturing waveforms
3. Switching power supply for load tests
4. USB-A tester
5. USB-C tester
6. Oscilloscope
7. Thermal imaging camera.
The load testing process looked something like this.
The first board supported all available protocols except QC 20 V.
1. Yes, 20 V isn't supported in QC mode; the maximum is 12 V :(
2, 3. The SCP protocol lets you adjust the voltage from 3.06 to 5.46 V. In my opinion, this protocol is a dead end, since the focus here is on high current, not high voltage.
4. FCP protocol, 9 V
5. Samsung AFC.
6. Power Delivery 3.0 is supported.
7. In most of the photos, you can see that a power cable is also connected to the tester. The reason is that in most situations, the board briefly cuts power to the output, most likely because it periodically switches between the USB outputs.
8. But sometimes it works without cutting out; I still haven't figured out the logic behind it.
9. In PD mode, the board outputs up to 20 V, so the QC limit is software-based.
Load tests.
QC: with 20 V at the input, the maximum output current is 3.2–3.3 A, regardless of the output voltage. As the current increases, the output voltage rises too. In all modes, overcurrent protection works correctly: once the overload is removed, the voltage recovers automatically.
For the Power Delivery test, I raised the input voltage to 24 V, since the board can output 20 V.
Everything here is exactly the same as in the previous tests.
To test the ripple level, I used a linear transformer power supply. The electronic load also has no switching power supply, and the computer was connected through an isolation transformer. The only device with a switching power supply was the oscilloscope itself.
With no load, the input voltage was 24 V and the output voltage was 5 and 12 V; the peak-to-peak ripple was about 60 and 30 mV, respectively.
Under load
1, 2. 5 V at 1.5 and 3 A: 40 and 45 mV
3, 4. 12 V at 1.5 and 3 A: about 65 mV
5, 6. 20 V at 1.5 and 2.5 A: about 45 and 25 mV
Temperature test: 24 V input, about 20 minutes of operation under a 3 A load at 5 and 20 V. Moderate heating.
Measuring efficiency and losses at different load currents and output voltages.
Table: blue = 5 V, red = 12 V, green = 20 V; the input voltage was 24 V in all cases.
And as a graph, with colors matching the table.
The second board also lacked QC 20 V and, for some reason, Samsung AFC.
Everything here is exactly the same as with the previous board, and Samsung AFC is indeed supported. By the way, in the third photo you can see the converter board trying to shut off the output: with 9 V set, the voltage periodically drops. This only happens when operating with no load. It doesn't consider the tester itself a load, though, and even then it happens inconsistently.
The test modes are identical to those in the previous test, and the results are almost the same too.
In QC mode with a USB-A connection, the maximum current is 3.6 A (the first board managed 3.2–3.3 A). The cutoff is just as abrupt, and the current is also independent of the output voltage.
PD mode, so this means connecting via USB-C.
The current is limited to the same level as on the previous board, i.e. 3.2 A. Above I mentioned that the second board has different shunts, 4 and 5 mΩ, while the first has identical 5 mΩ shunts.
Here I remembered that this board had been tested, and the output current was around 4.8 A in SCP mode through the USB-A output.
We set the required mode and start the test. The board delivered 5 A without any problems, but interestingly, the output voltage dropped as the current increased, whereas in all the previous tests it did the opposite. So the relationship between voltage and current is probably controlled by software, not hardware.
Ripple with no load is a little higher here: about 80 mV at 5 V and 30 mV at 12 V.
Under load
1, 2. 5 V at 1.5 and 3 A, 60 and 65 mV
3, 4. 12 V at 1.5 and 3 A, about 110 mV
5, 6. 20 V at 1.5 and 2.5 A, about 70 and 50 mV
Yes, the results are worse than with the first version; the cheap output capacitors are clearly to blame.
The temperature rise at 5 V, 3 A is the same; at 20 V, 2.5 A it's a little lower. Most likely, it simply “migrated” to the other side of the board. It was very inconvenient to photograph both sides, but I think the overall picture is still the same: it gets moderately warm.
Efficiency and power loss measurements at different load currents and output voltages.
Table: blue = 5 V, red = 12 V, green = 20 V; the input voltage is 24 V in all cases.
And as a graph, with the colors matching the table.
This board is based on the IP6518. The tester detected everything it possibly could, including QC at 20 V. That looks especially impressive given that the output capacitor is rated for 16 V. The Chinese guys are true to form.
But this is Power Delivery version 2.0, not 3.0 as on the previous boards, and the maximum load currents are also noticeably different.
Voltage - IP6518 - SW3516
5 V - 3.03 A - 2 A
9 V - 2.39 A - 3 A
12 V - 2.31 A - 3 A
15 V - 1.91 A - 3 A
20 V - 1.83 A - 3 A
Comparison of all three boards when used with the ZY1276 tester, in the same order. Power Delivery isn't detected because that requires a USB Type-C connection. I don't have that kind of cable, and in the third case I had to use a USB-A/USB-C cable and power the tester from the output.
The results are completely different here. As the current increases, the output voltage initially rises, as it does with the previous boards, but after the protection threshold it starts to drop gradually, which I think is less appropriate. The maximum current at 5 V was 3.5–3.6 A.
At 9 and 12 V, the results are roughly similar, so I'll hide them under a spoiler.
In QC mode at an output voltage of 20 V, I got 2 A, or about 40 W. Beyond that, the voltage dropped linearly as the current increased, which is especially clear in the graph.
There won't be any more tests, though. During testing, I frequently switched between boards, the load, testers, and so on. In the end, I accidentally shorted the board's output contacts. The board itself was powered by a linear power supply, but the current surge was enough to kill it. Apparently, one of the chip's transistors responsible for synchronous rectification blew, since the board tries to start up but ends up with a short circuit at the output. If I apply 5 V directly to the board's output, the power supply hits its 5 A limit and the chip starts heating up slowly. :(
Now for the conclusions.
Personally, I liked the boards based on the SW3516 better. Of course, the chip marking on the second board had been erased, but it really looks like an SW3516 too. The only thing to criticize is the lack of support for QC mode at 20 V. That seemed rather odd to me, especially since the board can output that voltage in PD mode without any problems.
Speaking of PD, this is already version 3.0, with output power up to 60 W, which is certainly nice. Both boards work well. The first one (ZC823) has better build quality and better components, which has a positive effect on ripple and should also improve reliability. Interestingly, though, at output voltages of 5 and 20 V, its efficiency was slightly lower than that of the cheaper option.
The second board is clearly from the "budget" range, especially judging by its appearance and cheap capacitors. Still, it has its place, particularly if you replace at least the output capacitors. It works quite well; the only significant complaint is the ripple level.
The third board, based on the IP6518, is a different story. It seems to use decent components, but for some reason they fitted a 16 V output capacitor instead of a 25 V one. It supports Power Delivery version 2.0, and its power is limited to just 40–45 W. The one thing in its favor is that it can output up to 20 V in QC mode. This board also doesn't cut power during switching, since it has only one output connector and doesn't need to switch between them as the first two boards do.
For objective reasons, I wasn't able to run the full load tests or measure ripple. I think you understand that it's technically impossible for now, but if I get my hands on another board like this, I'll certainly measure everything.
In short, the SW3516 focuses on higher power and PD 3.0 support, while the IP6518's only real advantage is QC at 20 V. So the first chip is a little newer.
That's all. I hope you found this information useful.
This page was found when searching for:
sw3516p распиновка, чип управления зарядкой по протоколу pd, usb плата на 2 sw3516p как подключить, sw3516p, конвертер 12 в протокол pd, преобразователь с 19 вольт на протокол pd, понижающий преобразователь pd, qc 18.6.5 плата, разница преобразователей t25a9qb и 7z25d, модуль sw3516 купить на алиэкспресс, PD / QC / AFC триггер с таблицей напряжений, sw3516 модуль, sw3516h., pps pd чип прошиваемый, платка преобразователя напряжения, ac dimmer signals, поддержка протоколом pd qc2, sw3516h, sw3516, rbfoo v2.2, pd qc преобразователь, плата преобразователя напряжения, понижающий pd преобразователь, что значит pps поддерживает 5 11 вольт, регулируемый повышающий 18650 5v 3,7 v 9v 2a, преобразователи напряжения, технические обзоры, купить преобразователь напряжения, тест преобразователей напряжения
sw3516p распиновка, чип управления зарядкой по протоколу pd, usb плата на 2 sw3516p как подключить, sw3516p, конвертер 12 в протокол pd, преобразователь с 19 вольт на протокол pd, понижающий преобразователь pd, qc 18.6.5 плата, разница преобразователей t25a9qb и 7z25d, модуль sw3516 купить на алиэкспресс, PD / QC / AFC триггер с таблицей напряжений, sw3516 модуль, sw3516h., pps pd чип прошиваемый, платка преобразователя напряжения, ac dimmer signals, поддержка протоколом pd qc2, sw3516h, sw3516, rbfoo v2.2, pd qc преобразователь, плата преобразователя напряжения, понижающий pd преобразователь, что значит pps поддерживает 5 11 вольт, регулируемый повышающий 18650 5v 3,7 v 9v 2a, преобразователи напряжения, технические обзоры, купить преобразователь напряжения, тест преобразователей напряжения
Questions and answers
What are the specs of the ZC823 board with SW3516?
It supports PD 3.0 and output power up to 60 W. The author says the buck converter needs an input of at least 22 V for full operation.
How does the IP6518 board differ from the SW3516 boards?
The IP6518 board is rated for PD 2.0 and up to 40–45 W, but can output up to 20 V in QC mode. It has one output connector, so it does not need to reset power to switch between ports.
Were the QC/PD boards fully tested?
No. The author could not carry out full load tests or measure ripple. In the comparison, the ZC823 has better build quality and lower ripple, but was slightly less efficient than the cheaper board at 5 V and 20 V output.
The figures from this review are in the measurement database: Converters
Calculate manuallyAll calculators
Discounted products
You may be interested in
Discounted products
Support the project
All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!
Comments2
Discuss on the forumHave a question or some experience to share? Discuss the article with the author and readers: the forum is a convenient place to share schematics and photos.
Go to the forumOnly registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.
Log inRegister
Comments are published in their original language.