A USB 2.0 High-Speed (480 Mbps) Galvanic Isolator.
Yifeng Li 562622a73c README.md: add deprecation notice. | 3 jaren geleden | |
---|---|---|
bom | 4 jaren geleden | |
gerber | 4 jaren geleden | |
lib | 4 jaren geleden | |
photos | 4 jaren geleden | |
schematics | 4 jaren geleden | |
symbol | 4 jaren geleden | |
.gitignore | 4 jaren geleden | |
COPYING | 4 jaren geleden | |
README.md | 3 jaren geleden | |
device.sch | 4 jaren geleden | |
fp-info-cache | 4 jaren geleden | |
fp-lib-table | 4 jaren geleden | |
hisolator-cache.lib | 4 jaren geleden | |
hisolator.kicad_pcb | 4 jaren geleden | |
hisolator.net | 4 jaren geleden | |
hisolator.pro | 4 jaren geleden | |
hisolator.sch | 4 jaren geleden | |
host.sch | 4 jaren geleden | |
sym-lib-table | 4 jaren geleden |
If you're looking for a High-Speed USB galvanic isolator, do not waste your time looking at this repository. This project uses the WCH CH317 ASIC, which has many performance and compatibility limitations, making it unsuitable for many applications.
A newer design, using the new TI ISOUSB211 ASIC, is a much better solution and does not have these limitations. It's available at: https://notabug.org/niconiconi/isousb211
This repository at this point is useful as a reference design for the WCH CH317 ASIC.
USB 2.0 is one of the most commonly used data interfaces. Often, implementing galvanic isolation (electrical isolation) is desirable. First, galvanic isolation protects the computer from destruction by high voltage transients or faults in an industrial environment. In the context of computing, it can be used by embedded system developers and hardware hackers to protect their computers from unexpected faults from the development boards, such as a dangerous short-circuit from +12 V to +5 V. Given the widespread knowledge of USB Killers, infosec researchers, too, can make use of an isolator. Also, galvanic isolation is useful to stop unwanted conductive electromagnetic interference between the computer and the USB device. Hence, amateur radio operators with cheap, low-EMI-immunity USB software defined radios may also find reliefs from noise with an USB isolator. And potentially, audio and video engineers may find an isolator helpful for breaking a ground loop.
Unfortunately, although USB 2.0 isolators for Low Speed (1.5 Mbps) and Full Speed (12 Mbps) are readily available, there are few isolators for High-Speed USB (480 Mbps) interfaces due to the difficulty of its high data rate. Currently, the available isolators are either expensive (FPGA-based solutions), difficult to buy (ASIC isolators), or inconvenient to use (USB over CAT-5 or fiber optics extenders, which requires a receiver and a transmitter).
This project is an experimental project that explores the possibility of (ab)using the existing CH317 USB-to-Ethernet controller to implement a single-board, low-cost, compact High-Speed USB 2.0 galvanic isolator, without the high cost and complexity of a FPGA.
The single-board PCB is physically separated into the primary (host) side and secondary (device) side with an isolation barrier in between. Both sides has independent power supplies and ground planes, and connected only by an Ethernet transformer and nothing else. The transformer is designed for Power-over-Ethernet application, thus it has a more robust Hi-Pot rating (4 kV) than conventional transformers (1.5 kV). The transformer couples the Ethernet traffic to the device side of the board. As proof-of-concept, both sides use external power supplies and linear regulators to facilitate debugging and development, no isolated DC-DC converters are used, yet, but they will be introduced after the initial development is completed.
On the host side, the incoming USB traffic from the Micro-USB connector is handled by CH317, acting as an USB PHY and USB controller. Subsequently, CH317 acts a Gigabit Ethernet MAC and translate the USB traffic from the computer to Ethernet frames, and drives a RTL8211F Gigabit Ethernet PHY via RGMII to transmit the information on-the-wire. CH317 also controls (read/write registers) on the RTL8211F PHY via a low-speed MDIO serial interface.
The transformer couples the Ethernet traffic from the primary to the secondary side of the board.
On the device side, the incoming Ethernet traffic is handled by another RTL8211F and forwarded to another CH317, acting as an Ethernet MAC. The Ethernet traffic is translated back to USB traffic and sent to the USB device via an USB-A connector. This process is bidirectional, which means the data can flow from host to device, or from device to host, thus establishing bidirectional High-Speed USB communication.
To ensure the communication link can recover and reset if an unexpected error has occurred (or if the USB is disconnected). A separate watchdog, CH9317G, is used to monitor the USB traffic (via D+ and D- lines), CH317 (via MDIO interface) and RTL8211F (via MDIO interface) simultaneously.
A four-layer PCB with controlled impedance is used to route high-speed USB, RGMII and Ethernet signals. Unlike conventional four-layer PCBs, there's no dedicated power plane but two ground planes - the stackup is Signal/Power, Ground, Ground, Signal/Power with an extensive use of stitching vias to provide a low-impedance current return path, thus minimizes its interference to amateur radio receivers.
After some debugging, board v0.02 (see git tag) is functional and can successfully establish a high-speed USB 2.0 link. Unfortunately, the data rate is limited to less than 10 MB/s (80 Mbps) in benchmarks. Therefore, while High-Speed USB 2.0 is technically supported, but its usefulness is limited, and largely unusable by many USB 2.0 High-Speed devices.
It was suspected that it's an inherent limitation of CH317Q - which has already been replaced by the second generation - CH317L. Thus, another prototype (currently unrealesed) has been designed based on the newer CH317L to see whether it could work as intended. Extra efforts were also made to produce a significantly cleaner and better PCB layout to improve signal integrity.
Unfortunately, tests have showned that CH317L has essentially the same performance. I can 100% confirm that the low performance was not a problem of my design, but an inherent performance limitation of the CH317L ASIC - because I found the official product from the manufacturer has the same performance limitations!
Based on the observed low performance, the only feasible application is isolating a USB flash drive, or something similar that doesn't require the full bandwidth of USB 2.0 High Speed.
The new design based on CH317L is not released, since there are still minor bugs in the PCB layout that requires manual patching, and I currently don't have time to finalize it - because at this point it's useless for me at the personal level. Nevertheless, I do realize that releasing my working design may still help the community, especially when one considers that the relevant documentation is not publicly available. Thus, I plan to design and test a final prototype, release it for readers' reference, then permanently discontinue this project.
But there is still hope. I'm now experimenting with an alternative chip for USB isolation.
This is a free hardware design: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
This hardware design is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with this hardware design. If not, see https://www.gnu.org/licenses/