Introduction
Last month, we discussed how to use an H-field probe to help trace the path of injected ESD current. This month, we’ll use the same tools to troubleshoot a known ESD failure. In this case study, statistics showed that ESD was likely destroying the USB hub of a point of sale (POS) system. Because the system had been deployed widely, the client wanted to determine the least expensive retrofit to reduce the field failure rate.
The hypothesis was that customers using the PIN pad (credit card reader) were discharging ESD energy into the powered USB port (PUSB) and, after repeated discharges, finally destroying the USB hub. Identifying marks on the product have been removed for confidentiality.
Test Method
Because we already know the device that was being stressed, we’ll monitor the injected ESD current through the USB hub while injecting ESD into the most likely customer “touch points”. Various fixes may be tried while monitoring the results. I’ll go through the troubleshooting steps I used, and I hope this will help you with your own products.
Using the test setup in Figure 1, we can monitor the relative ESD pulse measured at the USB hub while trying various mitigations. For each experiment, the peak-to-peak voltage is recorded.
Figure 2 shows a typical oscilloscope display of the injected ESD. The ringing is due to cable resonance and is unimportant for the purpose of this investigation. The measurement used during data collection is the peak-to-peak voltage. The scope was set to 1V/division vertical and 200 ns/division horizontal. Input impedance was set to 50Ω. Normal triggering was used to freeze the capture.
Visual Inspection
For any benchtop troubleshooting, it’s always a good practice to visually inspect the system for obvious or potential EMC design issues. Like most point-of-sale terminals, this one was an integrated system with a touch screen. The PIN pad keyboard and credit card reader were attached using a short cable attached to the terminal. An external printer and bar code scanner were also attached. The system had an internal power supply.
The bottom side of the terminal had rows of USB and Ethernet connectors. As this was an obvious entry point for ESD, the cover was removed and the I/O ports were inspected. Figure 3 shows the row of ports.
An easy test for gaps is to use a business card and try to force it between metal chassis panels. In this case, it slipped in easily (Figure 4).
For best ESD protection, we need the connector ground shells to be well-bonded to the chassis. This will also help reduce radiated emissions. This gap turned out to be the root cause of the ESD field failures. Figure 5 shows a better view of the gap between the connectors mounted to the PC board and rear chassis bracket.
Experiment 1 – Inject into the enclosure near the USB ports
In this first experiment, I wanted to measure the ESD response at the USB hub by using a direct connection to one of the connector ground shells (Figure 6). The ESD simulator was adjusted to 500 volts, which I considered a safe level to avoid component damage.
Copper tape was then added to simulate a good connector to chassis bond and the measurement was repeated (Figure 7). This is not a good long-term solution, but served to confirm the experimental results.
Experiment 2 – Determine the effectiveness of clamp-on ferrite chokes
Because these systems were already deployed in the field, one mitigation suggested by the client was to add clamp-on ferrite chokes to the PIN pad cable, an easy field retrofit. The test setup is shown in Figure 8.
A screw on the case was used to inject ±4kV into the PIN pad, while measuring the resulting p-p voltage on the USB hub. Contact discharge was used for consistency. The scope was set to 1V/division vertical and 200 ns/division horizontal. Input impedance was set to 50Ω.
I also clamped an RF current probe around the cable to monitor resulting RF cable shield currents.
Fair-Rite model 0431164281 (#31 material) chokes were used, and these have excellent impedances of 100 to 500 Ohms from 10 to 500 MHz. Placement is shown in Figure 9. Chokes were added one at a time until all three were installed.
Experiment 3 – Direct injection into the USB data pins and cable shield
Following the ferrite choke experiment, I tried injecting ±500 volt ESD directly into the pins of a generic USB cable. The test setup is shown in Figure 10. This test was done with and without copper tape installed to bond the connector ground shells to the chassis enclosure (Figure 10).
Results
This first set of results was with the existing poor connector bonding and ferrite chokes added to the PIN pad. This was the most desirable from a field retrofit viewpoint (Table 1).
A single ferrite choke reduced the measured ESD at the hub by about 30%. Adding two more for a total of three decreased by about 40%. It’s difficult to judge whether a single ferrite would be enough, but it was probably worth a try in this case and would be a more moderate field retrofit, rather than installing three!
Table 2 shows the results of adding copper tape to bond the connector ground shells to the chassis enclosure. The average improvement was a 50-60% reduction in ESD current at the USB hub. This proved to be a much better mitigation, but would require a complete disassembly of the display.
Suggestions for Future Designs
The goal for any I/O connector is to divert the high ESD current to earth as directly as possible. Figure 11, from my colleague Todd Hubing (www.LearnEMC.com), shows the ideal configuration between I/O connector, board ground return plane and chassis.
When ESD is injected into an I/O connector ground shell or even a data pin, we want the current to be directed in a low-impedance path to chassis and then on to earth. A combination of good bonding to enclosure, plus transient protection, is the best solution.
Good solutions would include a custom shim between connector ground shells (Figure 12) or the use of I/O connectors with built-in bonding gaskets. When the connectors in Figure 13 are soldered to the PC board, they can just press up the chassis or enclosure to form a low-impedance path for ESD current to return to the enclosure and then on to earth.
Summary
Unfortunately, there really was no good solution to the field failure issue the client was experiencing. The copper tape was the best solution, but would require retrofitting failed units at field service locations. Adding three ferrite chokes was also relatively effective and an easy field retrofit, but then you end up with a bunch of chokes on the PIN pad cable, which doesn’t look good.
A single ferrite choke was the easiest to implement, but only time would tell if this mitigation was effective. The combination of added ferrite chokes and better bonding would likely be the best solution.
This is a good teachable moment for any manufacturer who leaves EMC to the last step in the development process or does not understand basic EMC design practices. Poor connector to chassis bonding is an obvious design flaw and should have been caught much earlier when a simple shim could have been developed, as in Figure 12, or a gasketed connector in Figure 13.
