Fnb58 инструкция на русском языке скачать

For many years, I have tested USB power consumption using lab-bench power supplies and electronic loads. While this suffices for ordinary USB 5V, various fast-charge flavours require quite a few “hoops” to enable (e.g. qc3control, USB-C PD trigger modules). But when these things don’t work, I’m pretty much without a clue as to what is going on. Likewise, these setups aren’t exactly convenient for a quick test, and while the Charger Doctor, USB Detector and Keweisi meters have been good enough for basic voltage and capacity checks, it is completely powerless to diagnose the quick charging protocol negotiation process.

Legacy proprietary fast-charging protocols are numerous and considered a ripe pain-in-the-butt to understand why certain combinations of chargers, cables and devices just work and others are slow. Since we are now in a USB-C PD world, understanding the power profiles are similarly important to understand whether a given charger and device are optimal, especially given capabilities such as PPS. To enlighten myself as to what is going on, I need a modern tester that can “sniff” and/or “spoof” the transactions.

By pure coincidence, an element14 member by the name of scottiebabe posted a thread about the Fnirsi FNB58 USB Fast Charge Tester showing some rather good accuracy figures. Being USB-C PD capable, it would fill a capability gap of mine, so I decided to splurge as a “gift to myself”. The FNB58 can be found from eBay sellers listing at about AU$67 to AU$94, although this can be made a little less painful through stacking discount codes, gift cards and cashback offers. That being said, it’s not all that inexpensive, so let’s see if it’s worth the price and if I got a good unit.

Unboxing

The unit arrives in a shrink-wrapped colour-cardboard box, featuring an image of the unit.

The device claims the ability to monitor 4-28V, 0-7A, 0-120W and can integrate capacity/consumption. Resolution is claimed to be 10uV/uA/uW. It supports USB-A, micro-B and type-C connections. The device is Made in China and this version is with Bluetooth-LE connectivity.

The box is rather small and thin …

… because it really only contains the tester inside a foam insert and nothing else.

The body of the tester is metallic and feels rather nicely finished. The unit feels solid with a bit of weight. A colour LCD screen occupies the majority of the front.

Torx screws are at all four corners of the rear, with a plastic window on the rear, likely acting as a radome for the Bluetooth radio in the unit.

The top edge of the unit has a micro-B PC-connection port that can allow for the unit to operate down to 0-volts input and provides the ability to log data to their PC app. An LED indicator shines through the hole to indicate the operational status. A left-right-click toggle switch and a back button is used to navigate the menus, similar to an MP3 player from the early 2000’s.

The yellow-coloured USB-A connector feels especially solid and has an additional contact to support a specific fast-charge protocol that relies on an extra contact. A female receptacle is located on the opposite end.

The bottom side has both USB-C and micro-B inputs, and a USB-C output. A small switch allows for the isolation of USB-C PD communications – this feels slightly fragile, but depending on how it is used, it may only be needed very rarely.

Unfortunately, since Fnirsi’s website is being rebuilt and the English version is currently broken, the software, firmware, manual can be downloaded from a mirror.

Standalone User Experience

The FNB58 is quite the advanced tester, but for simple functions, it’s pretty easy to get started.

Plugging it into power and a load, the initial screen functions as a basic meter with a hold feature. Sliding the toggle switch allows us to move between UI pages.

The next page allows you to select one of ten slots for energy integration statistics to be recorded and to start offline recordings of waveform.

Sliding again allows for a view of voltage and current in an oscilloscope-like display. The modes can be changed to display D+/D- voltages or voltage ripple in “high speed” sampling mode. Otherwise, you reach the final page which is the Application page where the majority of the advanced tools reside.

The fast charge screen requires acknowledging a warning, as it will potentially cause higher voltage modes to be triggered which can damage downstream equipment if it is not expecting it. Automatic detection is a very useful tool as it auto-detects charger capabilities. Trigger modes below that are protocol-specific, but allow for manual selection and activation of various current/voltage profiles.

The statistics page allows for reading the recorded energy integration statistics, clearing them, viewing offline recording records and making a simple battery capacity calculation based on nominal battery voltage and conversion efficiency.

In the toolbox are features for USB cable resistance detection, snooping into USB-C PD negotiations, converting a QC2.0 charger into USB-C PD, reading USB-C eMarker chips, reading DASH cables and emulation of DASH cables and Apple 2.4A modes. The cable resistance detection tool may seem useful, but requires an external load to function. I’m personally quite wary of it, because the connector resistance may be a big variable.

The settings section allows for configuration of brightnesses, standby times, refresh rates, temperature units, language, whether the screen rotates on accelerometer sensing, various recording options, resetting to factory defaults and information about the device and firmware. A count of number of runs is kept as well. Key features which I find most useful are examined in the following sub-sections.

Experiment: Charger Detection

Charger detection allows for sensing what modes chargers are capable of providing. During detection, short power outages may occur and the FNB58 can handle this. Longer transients may cause the FNB58 to reset and restart the detection routine in an infinite loop, thus providing power through the PC connection allows for this to be avoided.

The above is a sampling of detect tool outputs – the top left is from an Anker PowerPort III 65W Pod charger, showing it has 5 USB-C profiles, support for Apple, BC1.2 as a dedicated charging port and Qualcomm QuickCharge 2.0, 3.0 and 4.0. I guess this is part of their “Power IQ” technology which means that both legacy and PD protocols are supported on the same port.

The top right shows the result from a USB-microB connection to an office desk integrated USB-A charge port which claims to support Samsung AFC as well but of course, none of the PD protocols.

The lower left detection is for an office desk integrated USB-C charge port that is marked 25W but actually advertises 30W of capability. Unfortunately, it doesn’t support any of the legacy protocols, but this is actually what is recommended. The right is for a ZMI zPower Trio 65W and it actually supports 65W with 5 power profiles, but none of the legacy protocols. I think this makes it quite clear how useful it is to have around to see what a given port is capable of.

Experiment: USB-C PD Examination

Using the PD Listener tool, we can “sit in” on a USB-C PD negotiation and see what is actually happening. This is an example of a camera charging from a PD source – to get this to work requires two USB-C cables and reversing one in case no negotiation is seen. Here, we can see that the camera is requesting 9V/2A/18W.

Long pressing the center click allows toggling to the next page which shows each of the packets and data. This is probably not quite the protocol-decoder that some would have wished for in terms of being PC-connected, but it is serviceable for basic checks.

Experiment USB-C eMarker Detection

Another useful feature is eMarker detection which allows us to examine whether a given USB-C to USB-C cable has an eMarker chip and if so, what capabilities it is advertising.

This first cable is a Baseus 2m 100W cable, which suprisingly advertises itself as 0-1m. It is a USB 2.0 cable, but the vendor, hardware and firmware are just all zeros. It does do 100W (20V/5A).

This doesn’t entirely seem uncommon either, as this is a Ukiyo 1m USB 3.2 Gen2 (10Gbit/s) 100W cable with all zeroes for IDs. But this one definitely has the speed configured correctly.

These two cables are a little different – Vendor 0x2E87, Hardware 0x0000, Firmware 0x0000 is a Toocki 1m 100W cable which claims to be 2-3m. Vendor 0x2BBE, Hardware 0x0001, Firmware 0x0001 is a Heymix 1.5m 100W cable which is overstating its speed as a USB 4 Gen 3 cable when it is just a USB 2.0 cable. It would be a very advanced cable if that were true!

PC-Connected Features

When the PC port is connected, the device appears to enumerate as a USB-serial device, a USB mass storage device and a USB input device (HID class). The VID 0x2E3C, PID 0x5555 and REV 0x0200 is used for the device. The means of interacting with the device is not obvious, although baryluk seems to have done a good job of reverse engineering most of it, releasing a Python-based logging software that can be run under Linux.

The software can be downloaded from Fnirsi’s website and is a ZIP file that requires no installation. The first thing I chose to do was to update the firmware which requires holding a button during insertion to place the device into bootloader mode.

The main page of the software allows for logging of data at a range of sample rates. The data is saved into a binary .cfn file which I have no idea how to export.

Triggers can be accessed on the second tab.

The third tab is used for firmware upgrades. Alarms are on the right sidebar along with real-time readings. The app appears to support multiple devices attached to a single computer. The app seems quite basic but it does work. Unfortunately, as it doesn’t work with automation, the reverse-engineered Python code is perhaps more useful for me.

The above is an example of examining a top-off charging process under PC logging.

Bluetooth Mobile App Features

The mobile app is not available on the app store, but is instead, an APK that can be downloaded from Fnirsi’s website (when it’s back up). I found it wouldn’t work correctly on my Android 7.0 device, force-closing on start-up, while it works fine on my Android 8.0 device. Location permissions are requested, because Bluetooth Low-Energy scanning and connecting requires this.

The app starts up with Chinese as the default language. Pressing the hamburger menu on the top right allows for changing the language. Then you can connect to the device and then tap on the “enter” button to be able to view the device data.

The app behaves as a nice remote screen for the meter, showing the current readings and waveforms. Energy integrators can be cleared and battery capacity calculation are available. Manual triggering is also possible, while alarms allow for the phone to act as an alert device when certain criteria are met.

The app doesn’t seem to do any direct waveform recordings which seems to be exclusive to the PC application.

That being said, the device has a lot of BLE characteristics – click on the thumbnail to the left to see the full list.

Accuracy Testing

Tests were performed with the FNB58 attached to a Digilent ADP3450 Analog Discovery Pro as a Linux-based device with LAN and USB connectivity. The reverse-engineered Python code from baryluk is used to log data from the FNB58, with the occasional time-out potentially due to ADP3450 USB-controller quirks.

Voltage accuracy tests were done using the Keithley 2450 SourceMeter as the voltage reference. Five 10PLC readings from the 2450 are taken and used as the average “ground truth”. In the same time, 300 readings are simultaneously taken from the FNB58 (at a fixed 100Hz).

Ignoring the error of the 2450 SMU which is quite small by comparison, it seems that the FNB58 overestimates the voltage with a nearly fixed gain error. At 20V, it is reading about 11mV high. Reading noise post-averaging appears to be about +/- 3mV.

Raw reading noise at 100Hz, however, seems closer to +/- 12mV. This suggests to me an ADC effective bit count of about 11- to 12-bits with oversampling used to create higher accuracy.

Plotted as percentage error, it seems that at higher readings, the error is settling out to 0.06% which is quite excellent given that this is a “gadget” and not a bench-instrument.

Testing of current used a Rohde & Schwarz NGM202 power supply, averaging 100 readings from the FNB58 compared to the average of 10 readings from the NGM202. Testing used the USB-A ports and involved a short being plugged into the output. The test did not use the full 7A range, due to the 6A limitation per channel on the NGM202. However, as USB-C PD ordinarily provides only up to 5A, this is still considered sufficient.

The current error seems to show a bathtub curve shape which I suspect is due to heating of the device due to dissipated power in resistive losses. The FNB58 underestimates current readings, usually by up to +/-0.75mA but up to +/-1.1mA at peak. Averaged reading noise, however, is about +/-0.75mA.

Looking at the raw individual reading noise at 100Hz, it seems a reading noise of about +/-1.3 mA is achieved. This suggests an effective number of bits of about 13- to 14-bits.

Because of the continuously changing error, it seems that the unit was trending towards a zero error toward the high-end of the scale, but mid-scale, it was about 0.07% low. Still a very good result.

Ripple and Noise Testing

A square-wave signal with DC-bias was injected from a Rohde & Schwarz MXO4 signal generator into the USB-A inputs to test the oscilloscope-like display of the FNB58.

The default mode plots both voltage and current but is limited to 0.1s/division. As a result, it seems that waveforms up to about 10Hz can be clearly seen. The mode has to be changed to VBUS mode to activate high-speed sampling.

In this mode, the high-speed sampling can resolve signals of higher frequency but the bandwidth seems quite limited. By 500kHz, we are looking at what looks like a triangular wave.

At 100kHz, the square wave nature comes back out, but with ringing top and bottom. As a result, while this mode will give you some indication of the ripple content of the voltage, it may also introduce its own measurement artifacts due to measurement bandwidth limits. The automatic vertical scaling and trigger can result in unstable displays as well.

Self-Consumption

Powered from the R&S NGM202, the unit consumed around 64.12mA when in high brightness displaying the metering screen. With dimmed brightness, consumption fell to 54.18mA.

Higher consumption is registered when running the Vbus ripple analysis, averaging 71.47mA. When powered via the PC port under that condition, current consumption reaches 100.26mA. This is still quite reasonable, considering all of the work it is doing and the colour LCD screen it is powering.

Conclusion

I’ve got to say that the Fnirsi FNB58 is quite a useful USB charger and cable testing tool. While the manual is a bit difficult to understand and the menus on the device can be a bit quirky at times, it does a lot of things for its size and most things just seem to work. While it may seem a little pricey, its build quality is fairly solid and the support by PC and Android app is a good bonus. Being able to diagnose, trigger, translate and emulate various proprietary and standard quick charging modes can be quite useful for very specific situations. Accuracy is pretty good overall, even though raw 100Hz readings may be a bit noisy.

The main downside is the lack of good documentation about the recording file formats and the device interfaces, relying on reverse-engineered code to allow for more effective use in automation. While there is an oscilloscope-like display for ripple, it is somewhat frequency bandwidth limited and the vertical scaling is automatic which can result in unstable displays.

But keeping in mind that this is a USB gadget, meant for use for quick diagnosis or to get out of a specific slow-charging situation, it really seems to do things more than just well-enough. I think @scottiebabe was right to recommend this!