Solid-State Switching in Automated Test

Electromechanical relays for testing products ranging from automotive engines to thermocouples are readily available, low-cost devices you can use to route signals originating from almost anywhere. It is no wonder that they have become ubiquitous in the automated test market. But despite industry adoption, electromechanical relays are relatively slow at hundreds of channels per second and have limited lifetime, which makes them less suited for some applications.

More suitable features are prevalent in solid-state switches than in electromechanical relays. With switching speeds as high as 50,000 channels per second and unlimited lifetime, solid-state switches can help minimize overhead in many applications by reducing test times and eliminating the need to frequently replace switching components.

The stark contrast in the strengths and weaknesses of electromechanical vs. solid-state relays is attributed to the construction of these parts. Electromechanical relays are made of a coil, an armature mechanism, and electrical contacts. When the coil is energized, the induced magnetic field moves the armature that opens or closes the contacts.

The contacts on electromechanical relays tend to be larger and more robust than some other relay types. This enables them to withstand unexpected surge currents caused by parasitic capacitances that may be present in the circuit and cables.

One trade-off of the larger contacts is speed since the contacts have more mass and must travel a greater distance to form a closed connection. Typical models can switch and settle in 5 ms to 15 ms, which is too slow for some applications. Mechanical contacts also limit the life of electromechanical relays because moving parts tend to wear and break after significant usage. Consequently, electromechanical relays have a finite lifetime.

FET switches use a series of MOSFETs to connect and disconnect signals. A MOSFET acts like a closed switch in the triode state when the following equation is satisfied:

where: VDS = potential between the drain and source of the MOSFET

VGS = potential difference between the gate and source

Vt = threshold voltage of  MOSFET

In an analog FET switch, the n-type MOSFET will act as a closed circuit when the control signal is high and it is in the triode state, satisfying Equation 1. Similarly, the p-type FET switch acts as a closed circuit when the control signal is low and in the triode state. When the potential difference between the control signal, applied to the gate of the FET, and the input signal is beyond a certain threshold voltage, the FET acts like a closed switch and conducts the signal.

Like analog FET switches, solid-state relays use MOSFETs to transition between states. However, in the case of solid-state relays, the MOSFET is photo-sensitive and actuated using an LED. While the optical isolation provided through LED activation enables solid-state relays to transfer higher voltage signals than analog FET switches, it also makes them slower.

Switching time for solid-state relays is dependent on the time required to power the LED on and off, approximately 0.5 ms to 1 ms, and the time for the light from the LED to charge the MOSFET gate. This is faster than the time required for the mechanical arm in electromechanical relays to move between states but slower than the typical ON/OFF times of analog FET switches.

Solid-state switches certainly provide many benefits such as speed and longer life, but they still are not prominent in the automated test industry. Their lack of popularity in ATE is primarily attributed to high path resistance.

As an example, consider the analog FET switch used in the NI PXI-2501 48×1 FET Multiplexer released more than 10 years ago. The device has a maximum voltage and current rating of ±12 V and 30 mA, a switching speed of 15,000 channels per second, and a path resistance of 50 ?.

This part does not have overvoltage protection. Consider, for example, the case of applying a 24-V signal at the input of the switch when it is closed. This creates a current greater than 30 mA which will damage the switch.

To add overvoltage protection up to 24 V and limit current flow through the switch to 30 mA or less, two 800-? resistors have been placed in series with the FET switch as shown in Figure 1. These resistors increase the path resistance of the signal path in the PXI switch module from 50 ? to 1,650 ?.

Figure 1. Older Analog FET Switches Using Physical Resistors to Provide Overvoltage Protection

Newer FET switch designs provide overvoltage protection using comparators instead of physical resistors. Figure 2 illustrates a newer analog FET switch used in the NI PXI-2535 544 Crosspoint Matrix Switch Module. In the case of this switch, if the input signal voltage level exceeds that of the positive voltage rail (+12 V) or falls below that of the negative rail (-12 V), the comparator circuitry sends a logic 0 to the AND gate and turns off the FET switch.

Figure 2. Newer Analog FET Switches Using Comparators for Overvoltage Protection

By using newer designs that provide built-in control circuitry to turn off the switch in overvoltage conditions, the PXI-2535 can provide overvoltage protection up to 30 V with a total path resistance of 10 ?. This is three orders of magnitude better than its predecessor, the PXI-2501, which supplies 24 V of overvoltage protection with a total path resistance of 1,650 ?.

The capability of newer solid-state relay modules to provide overvoltage protection without increasing path resistance has enabled their use in applications where speed and longer life are crucial for cost savings.

The validation of semiconductor devices is one such example. Consider a system that is used to perform 10 tests on a chip with 500 I/O points. The chip under test is used in numerous devices, and its cumulative sales are estimated at 1 million per month.

The test system shown in Figure 3, built using a single source measure unit (SMU) and a switching front end that routes all 500 points to the SMU, is required to run continuously without interruption. The cost comparison of using an FET-based switch product vs. an electromechanical relay-based product is as follows:

Figure 3. Test System With SMU and FET Switching Matrix

Assuming that the switching speed of the FET switch is 50,000 channels per second, you can test all 1 million chips in less than 12 days. Because FETs have unlimited mechanical lifetime, no switch-module replacement costs are incurred during the process.

If you used an electromechanical relay-based switch module with the same density, the expenses would be much higher. Electromechanical relays have a typical lifetime of 1 million closures and a speed of 250 channels per second. Because each relay is closed 10 million times during the process of testing all 1 million chips, the relay module needs to be replaced 10 times. This would increase the total expenses incurred on maintaining the system.

The slower speed of electromechanical relays also adds costs in comparison to the FET-based solution. It takes 231 days to test 1 million chips using electromechanical relays. Accordingly, using electromechanical relays would incur the cost of maintaining and running a production floor for 219 additional days in comparison to the FET-based module. Alternatively, to keep pace with the 1 million chips/month sales rate would require six or seven additional test setups.

Longer test times also introduce challenges when managing inventory and shipping products to customers. Although hypothetical, this example shows the real cost savings you can achieve due to the benefits of FET and solid-state relay technology.

Summary

While improvements in technology have enabled the use of solid-state switches in semiconductor validation, electromechanical relays still remain the preferred option for many applications, especially those that route high-power signals. Commercial-off-the-shelf solid-state switches typically have a maximum switching current capability of less than 5 A. This is insufficient for many applications such as fault simulation in automotive test, which requires routing signals in the 20-A to 40-A range. Although custom circuitry can enable the use of solid-state switches for routing signals as high as 100 A, this approach can add unnecessary time and complexity to the system development process.

There is no single solution for routing signals in automated test systems. In fact, the number of solutions available in the market is continually increasing. FETs and solid-state relays are examples of switching solutions that always have been available to the market but have become viable options only recently due to advances in transistor technology. With these advances, you now can reap the benefits of solid-state switching, which include faster switching speeds and unlimited mechanical life, to build better, faster, and more economical test systems.

About the Author

Jaideep Jhangiani is a product marketing manager at National Instruments. During his tenure at NI, he has developed and implemented many marketing campaigns, including one for the launch of 11 new RF switches. Mr. Jhangiani holds a B.S. in computer engineering from Texas A&M University. National Instruments, 11500 N. Mopac Expwy., Austin, TX 78759, 512-683-5753, e-mail: [email protected]

November 2008

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!