New from Fnirsi: FNB48 USB tester

Some time ago, news emerged about a new USB tester, the FNB48, from Fnirsi, a young but fast-growing company. I thought it looked interesting, so I ordered one for review. I was also curious to compare it with other testers, such as the rather expensive ZY1276 from YZXstudio.


The title gives a price of $30, but when it first went on sale, it was available for $21.6, so it may well go for that price during various sales. And to get ahead of myself, I think the price is quite reasonable. But let's get on with the review.

The packaging is in the RuiDeng style: a small foam box containing an antistatic bag with the tester inside. The total weight is 106 g. There's a QR code on the back of the packaging linking to the software download page.


The tester also comes in its own tin box with a window. While looking it over, I noticed something odd: the older version had the manufacturer's logo on the box, but the new one doesn't. In any case, the box is handy and offers better protection, since the tester itself has only a token enclosure.


It's neatly made and, in my opinion, noticeably better than the previous model I've handled.


Specifications. The full manual, with descriptions of all operating modes, is available at this link


There's a large screen on top, taking up almost the entire upper section. The input/output labels and controls are on the bottom. The top section is acrylic, and the bottom is fiberglass laminate.
There's a USB-A plug at the front. The bottom edge has USB-C and microUSB inputs, a USB-C output, and a switch for controlling PD.
On the back there's a USB-A socket. The top edge has a "back" button, a joystick-style mode switch that also clicks in, a spot for a Bluetooth switch (not fitted), and a microUSB port for connecting to a computer.


The labels for the controls didn't come out very well in my photo; they're easier to see in the manufacturer's render.


The product page has a comparison table showing that, at the moment, the FNB48 is the most fully featured model in Fnirsi's tester lineup.

The model numbering isn't entirely clear. The FNB/FNC part makes sense: B = "regular" USB, C = USB-C. But the first digit in the FNC model numbers seems to be all over the place...


The tester is really big. For comparison, here are some other models:
On the left: the ZY1276 from YZXstudio and the FL001T made by Power-Z.
On the right: Fnirsi's FNC88 and the TC66 from RuiDeng.

In my experience, the ZY1276 and TC66 are the most useful and convenient. The FL001T is accurate, but a real pain to use, while the FNC88 is noticeably less robust mechanically and simply not as convenient.


Tester dimensions.


The tester is very easy to take apart: four screws on top and four on the bottom. The top screws are long, and the bottom ones are short. I was pleasantly surprised that the manufacturer included one spare screw of each type. I should also note that the threaded standoffs are very loose.

Like most testers of this kind, the enclosure is open on the sides, which isn't ideal. To be fair, the only fully enclosed testers I've come across are custom-built models from 100MHz.


The second thing that really surprised me was the double-layer top, made of two acrylic plates: a thin one on top and a thicker one underneath. Even more surprising, both plates are labeled. In principle, you could use just one, but the screws are too long to fasten it properly.
The tester is as short as it can be with a screen this size; you can see that the connectors are practically right up against it.
The bottom section also serves as a Bluetooth adapter board, but the module itself isn't populated.


This isn't a new idea; I've seen exactly the same arrangement in the ZY1276 tester, where the module wasn't populated either. But in addition to Bluetooth, that board also allowed the tester to be powered by a battery, with components for a charger already in place.


For some reason, I expected to see more components when I opened it up. The layout is relatively sparse, and the soldering quality is average—mediocre in places, though that's more a matter of neatness.


1. Microcontroller, current-sense shunt (connected in series with the power rail), dedicated INA226 ADC, and voltage regulator.
2. 16 MB W25Q128JVSQ flash memory, a pair of load resistors (presumably) rated at 18 Ω each, and a Power Delivery chip.

It claims to support currents up to 6.5 A. The shunt is rated at 10 mΩ, so at maximum current it will dissipate 0.422 W, which is perfectly acceptable.
Interestingly, the board has two pads marked Vout and GND, which can presumably be used to connect a load without going through the USB socket.




Features.
Before moving on to the operating modes and how they work, a few words about the controls.
The tester has a left/right rocker switch combined with a button, plus a separate button. The controls work as follows:
1. Back button: cancel or go back in the menu. Press and hold to turn off the backlight.
2. The rocker switch button supports short and long presses. A short press activates the selected mode or starts/pauses the oscilloscope; a long press switches modes, for example, on the oscilloscope.
3. Left/right rocker switch. Short and long presses are supported here too. A short press switches display modes or navigates the menu; a long press activates additional functions within the current operating mode.


1. The manufacturer's logo appears when the tester is switched on.
2, 3. The first mode is a basic volt-amp-watt meter, with a choice of display precision.
4. Press and hold “left” to open the settings menu.
5–9. There are lots of settings, though most are simple on/off options. There are also adjustments for things like screen brightness, backlight timeout, current thresholds, and recording modes.


The second screen shows voltage and current measurements, calculated power, the temperature inside the tester, minimum/maximum current, voltage and power, and capacity in Ah and Wh.
A short press of the switch displays a notification that the min/max counters have been reset.
Press and hold left to access a memory slot (10 in total) for capacity in Ah and energy in Wh.

Press and hold right for a rather interesting, if somewhat complicated, mode. As I understand it, it lets you measure the capacity of a power bank's battery by entering the battery voltage and converter efficiency. In my opinion, though, the feature is too complicated and requires knowing the efficiency, which isn't always easy to find out.
The Ah and Wh counter readings depend on the output voltage (3.7–5, 3.7–12, etc.) and the efficiency.


Out of curiosity, I tested the power bank capacity measurement mode.
1. I connected a load of about 0.5 A and set the recording time to 30 minutes in the settings under Energy REC time.
2. After 30 minutes, I got 0.255 Ah at the output with a 5 V supply, and a calculated value of 0.311 Ah.

I didn't quite understand the calculation here. It should be 0.255 × 1.35 (conversion from 3.7 to 5.0) × 1.1 (90% efficiency) = 0.378 Ah.
I can't figure out why it came out to 0.311 instead of 0.378. The 1.22 ratio between 0.311 and 0.255 doesn't match either the 3.7-to-5 conversion or 90% efficiency. I'll investigate further, since this is a really interesting mode.

Later, I tried setting the efficiency to 100% and got the correct 3.7-to-5 conversion. So the efficiency is being calculated incorrectly—or rather, in the wrong direction. Currently, 5/3.7 = 1.35, then 1.35 × 0.9 = 1.22, but it should be 1.35/0.9.



The third screen shows the same basic measurements as before, but here you can work with fast-charging protocols, see the current mode, and check the voltage on the data lines.
Press and hold right to automatically detect supported protocols and control them directly.

The implementation is convenient and fast, but, as sometimes happens, it doesn't work properly with every power supply. This is because a power supply can sometimes drop the voltage too low, causing the tester to shut down. In that case, it's better to power it through the PC connector. I did notice this happened less often here than with other testers.


Press and hold right to check a cable's PD capabilities, test Dash cables, and emulate DASH and Apple 2.4 A in software.

For some reason, in PD listener mode it sent a command to disable PD on the power supply, causing the voltage to drop to zero. I had to power it from a separate power supply to take the photo.
The PD E-marker works: it correctly identified the cable's maximum current and data speed capabilities, as well as its length.


With the full setup and a Gen 2 cable, it should look something like this. Interestingly, my cable showed an upper limit of 50 V. I hadn't realized before that the limit can apply to voltage as well as maximum current.


I don't have a Dash cable. According to the developer, it should be identified as shown in this picture.


1. an example of automatic detection of fast-charging protocol support.
2. I tried to emulate DASH in software, but either it doesn't emulate it, or I'm doing something wrong. My main tester says it isn't supported.


Fourth screen.
One of the most interesting modes is essentially an oscilloscope emulation. There are four options in total, switched by a long press on the button.

1–4. Current and voltage measurement, with a sweep speed of 0.1 to 5 seconds per division.
5–6. Voltage measurement on the data lines, also with a sweep speed of 0.1 to 5 seconds per division.

A short press on the button pauses the graph, which is handy for taking photos.


This one can actually be called an oscilloscope mode: it measures the ripple amplitude at power supply outputs.
1, 2. The sweep rate can be set from 3.1 to 200 µs per division. For example, when measuring ripple, I usually use 5 µs/div.
3. The trigger mode is automatic and works reasonably well, although it can sometimes be unstable. The pulse frequency is displayed at the top, but sometimes it may show 0 or nonsense, so I wouldn't rely on its accuracy.
4–6. Example waveforms from the output of my USB hub at sweep rates of 3.1, 10, and 20 µs. For convenience, the measured ripple amplitude is shown in the lower-right corner.

I think it's pretty good. My tester already measured current and voltage and showed the status of the data lines, but it didn't have an oscilloscope like this. It's more than enough for quick diagnostics.


The strangest mode is the offline logger. As I understand it, the idea is to connect the tester to a power supply, connect a load to the tester, and have it save current and voltage readings to memory, then display them as a stretched-out waveform.

1. Recording is off by default; the label Mem no appears at the top.
2. A short press on the button prompts you to clear the recording.
3. To start recording, go into the settings and enable Curve REC time—or, more precisely, set the recording duration.
4. You can choose anything from 30 minutes to 9 hours. The sampling interval is therefore between once per second and three times per second. The first option gives 1,800 readings, while the maximum gives up to 10,800. I think it simply can't record more often than once per second.
5. You can also enable energy measurement recording here, but that has nothing to do with the oscilloscope.
6. After I enabled recording, Mem 0.5H appeared at the top, but nothing else happened. I tried disconnecting and reconnecting the load and restarting the tester, but all I got was a grid screen...


The manual includes a photo of this mode, but how to get it to work is a mystery.


Later, quite by chance, I figured out how recording works and where to turn it on.
1, 2. At the bottom of the current, voltage, and capacity measurement screen, there are two indicators. In the photo, both show 0.5. The upper one indicates graph recording, and the lower one indicates recording of the energy that has passed through the tester. A short press on the button starts graph recording.
3. There's also a progress bar at the top that gradually counts down. When it reaches 0%, the upper graph-recording time indicator at the bottom goes out.
4. If you then open the recorded graphs screen, you can see what happened during the half-hour recording. Apparently, the idea was to let you “stretch” the graph, like changing the sweep time, but it doesn't work: a warning that the limit has been exceeded appears at the edges, and nothing on the screen changes.
5. You can stop recording at any time, of course, but you'll only get a partial graph.
6. If you change the time limit in the settings from half an hour to an hour, the corresponding box in the lower-right corner will show 1.0.


While testing, I remembered that I wanted to update the firmware. It turned out that version 0.6 was available; I had 0.5 installed. According to the description, support for PE+1.1 and PE+2.0 detection was added.
Let's check. First, it reports support for MTK-PE.

The flashing process is simple: download the archive from here with the flashing program and the firmware itself.
Extract it. I recommend replacing any garbled filenames with something readable to avoid problems.
Launch the program.
Hold down the button on the tester and connect it to a USB port on your computer. If something goes wrong, first try simply connecting it so the computer can detect it and install the driver.
The tester will then display the bootloader version.
In the program, select the downloaded firmware and click Start.
The tester screen shows the firmware update in progress. Needless to say, it's best not to interrupt it.
The firmware has been updated successfully; version 0.6.


I tested it with a couple of power supplies and a couple of converters. The results are, of course, different, but you can see that PE+ has not only appeared, but is also displayed for the compatible converters.


The fifth screen.
Measuring cable resistance. Nothing new here; the principle is very simple. First, connect the tester with the load directly to the power supply output and press the switch. The tester records the reading as a reference.
Turn off the tester and connect it through the cable. You can then see the measured voltage drop and cable resistance.


One of the tester's “tricks” is its built-in gyroscope, which makes the screen image orientation depend on the tester's position.
1, 2. In voltage, current, and power measurement mode, the image rotates by 90, 180, or 270 degrees.
3, 4. Since the screen is rectangular, in other modes the image can only rotate by 180 degrees.

Overall, it works well. It can take some getting used to and may occasionally get in the way, but you can disable the feature in the menu if needed.


I also found a rather unusual, but potentially useful, Power Delivery emulation feature.
1, 2. You can select a maximum power from 2 to 60 W, as well as choose from a list of output voltages, with the corresponding current values calculated for each one.
3. I connect the tester to a QC power supply with an output voltage of up to 12 V. A second tester, set to control PD power supplies, is connected to the first tester's output. I set the voltage to 9 V; the QC power supply outputs 9 V, but the tester at the output thinks it's connected to a PD source.
4. I use the second tester to raise the voltage to 12 V, and that works perfectly too.
5. When I try to set 20 V, which the power supply can't provide, the output drops back to the original 5 V. The second tester still shows 20 V as selected, though, while the 20 V mode disappears from the first tester's screen.
6. I try raising the voltage to 12 V again, and everything works normally.

Overall, the feature works. Whether it's useful is debatable, but I think it could come in handy. Just remember to set the output power limit. Essentially, you can power a PD device from a QC power supply.


Since this is a measuring instrument, even if it isn't officially certified, we should check the accuracy of its main measurements: voltage and current.

No issues with voltage: the specified measurement error is 0.2% + 2 counts, and the tester is comfortably within that limit. The tester starts working at 3.8 V and immediately displays correct readings, so I don't think it's possible to get it into a state where it starts “lying” about the readings.


The current readings are similar: the specified error is 0.5% + 2 counts, and the instrument also stayed within that limit, although the error increased noticeably near the top of the range.


It also handles very low currents without any problems. For example, I used 680 Ω and 1.2 kΩ resistors as loads, giving currents of about 4 and 7 mA.


As an additional test, I checked a current of about 1 mA using a 4.7 kΩ resistor as the load. In my opinion, getting 0.83 mA instead of 1.092 mA at such a low current is still pretty good. Also, keep in mind that the measurement uses a single 10 mΩ shunt, so at 1 mA the voltage across it is 10 µV, or 0.00001 V, at the ADC input.


In general, it's worth noting that it's very difficult to get the voltage or current measurement wrong, since the tester uses a dedicated chip that handles the entire measurement section, apart from the current-sense shunt. The same chip is used in the tester models listed above, so instantaneous accuracy depends only on calibration, while long-term accuracy also depends on the quality of the shunt—more specifically, its temperature coefficient of resistance.


Power consumption.
1. At maximum backlight brightness, the tester draws about 34–35 mA.
2. Power consumption is almost independent of the operating mode. In oscilloscope mode, which seems to me to be the most resource-intensive, the difference compared with the other modes was 1 mA.
3. At minimum backlight brightness, the current drops to 23 mA.
4. With the backlight completely off, power consumption is 22 mA.

Power consumption also depends on the input voltage: at 3.8 V it is 45 mA, while at 24 V it is only 9 mA.
Previous models used a linear regulator. It's hard to say whether replacing it with a switching regulator made things better; I'll just say that a 3.8 V cutoff isn't particularly low.

The tests show that there's not much point in turning off the backlight, unless it happens to get in the way.



Conclusions.
This turned out to be a pretty good device, with good voltage and current measurement accuracy. But as I wrote, that's largely thanks to the specialized chip from Texas Instruments.
Functionally, I have no major complaints either. I especially liked the offline logger, the emulation of PD power supplies, and, of course, the oscilloscope. It won't replace a full-fledged instrument, but even in this form it can be a big help. It supports a large number of protocols and gets updates too; for example, the latest update added support for PE+1.1 and PE+2.0.
There's a mode for measuring power bank capacity (which still needs some work) that accounts for voltage conversion and efficiency. The latter is a sticking point, though: not only does it depend on the output voltage, it's usually not specified anywhere. Still, it's a good idea.
And let's not forget that the firmware is only version 0.6, so this isn't really a finished product, but rather a well-debugged beta.

Strictly speaking, it would be more appropriate to compare this tester with the ZY1280M, since it's closer in both software and hardware. The well-known WEB U2P falls behind both of them, as it has a less powerful microcontroller and a much slower “oscilloscope.” Perhaps the U2P's only real advantage is that it can work with Apple cables and devices.
For comparison, here are the microcontroller specs:
U2p - STM32F072CBT6 microcontroller with a Cortex-M0 core, 48 MHz
Fnirsi - STM32F411 with a Cortex®-M4 core, 100 MHz
ZY1280M - 32F330F8P6, Cortex®-M4 Core, 108 MHz

The Fnirsi oscilloscope's shortest time base is 3.1 µs, the ZY1280M's is 5 µs, while the U2p goes down to just 200 µs, which is completely unsuitable for measuring ripple amplitude.

Things that need improvement:
Zooming in on a recorded signal. I've already contacted the manufacturer, and they said this is a software limitation; it will work when viewing the data on a PC.
Viewing recorded data on a PC. They said the software that enables this should be ready in about a month and a half. We'll wait and see.
Bluetooth module. When the tester was sent to me, this module wasn't available yet, but a trial batch has now been made, and one of them is already on its way to me.
A proper enclosure would be nice.
Power bank capacity testing mode. I've already contacted the manufacturer about the problem.

That's all for now. Hopefully, once the Bluetooth module arrives and the PC software is released, I'll be able to continue.