Combines the advantages of ICT and FCT in a single test adapter

Generally, a bed of nails is used to test unpowered circuit boards, using technologies such as direct digital synthesis (DDS) and discrete Fourier transform (DFT) to generate stimulus signals for analog measurement analysis, so that the in-circuit tester (ICA) can measure actual data such as inductance, capacitance, impedance and resistance to confirm that the results of all device under test (DUT) test nodes are within the tolerance range, and whether there are open circuits, short circuits, wrong parts or reverse polarity. These are measured without power. Relay multiplexers can be used to connect probe contacts and analog channels or digital drivers/sensors (D/S) of the circuit board (Figure 1). 

In-circuit testing (ICT) is a method of analyzing electronic products in production.

Generally, a bed of nails is used to test unpowered circuit boards, using technologies such as direct digital synthesis (DDS) and discrete Fourier transform (DFT) to generate stimulus signals for analog measurement analysis, so that the in-circuit tester (ICA) can measure actual data such as inductance, capacitance, impedance and resistance to confirm that the results of all device under test (DUT) test nodes are within the tolerance range, and whether there are open circuits, short circuits, wrong parts or reverse polarity. These are measured without power. Relay multiplexers can be used to connect probe contacts and analog channels or digital drivers/sensors (D/S) of the circuit board (Figure 1). 

Figure 1: Typical 2x16 bed-of-nails relay multiplexer (only one channel is shown)

In some more advanced systems, the ICA module can also be used to do a portion of the device functional testing (FCT) by powering up the circuit and measuring the input and output characteristics under load. This testing is usually done separately using another test adapter. The reasons for this are as follows: 

First, the probes of the ICT bed of nails cannot carry the required supply voltage or load current to perform full functional testing of powered devices. The heavy-duty probes of a dedicated FCT test station must be able to withstand high current or high voltage without overheating, arcing, or excessive wear. The disadvantage is that these heavy-duty probes take up more space, so the FCT test adapter can usually only check one DUT at a time. 

Secondly, the programmable power supply, relays, and electronic loads inside the ICA are not suitable for high current testing. If simply replaced with a larger power supply, the higher current may seriously interfere with the analog quantity of sensitive ICT measurements and cause errors, including ground bounce, line voltage drop, and transients generated by inductive loads at the moment of switching. Measurements using dedicated FCT adapters usually have lower resolution and larger filters, so they are not sensitive to interference. In addition, the power supply and relay contacts are more rugged and durable, so they can switch currents exceeding 1 amp. 

Third, relay interface hardware and software control is typically done through a parallel input/output (PIO) controller and relay driver to change the relay configuration (Figure 2). Relay switching speed is usually not an issue in ICT applications because the relay multiplexes the connections to reconfigure from one set of pins to the next after each DUT test. In the case of an FCT test adapter, relays are used to change the functional test setup for each DUT for each test, so the control data throughput of the relays is higher. In a dedicated FCT setup, this is not a problem because only one DUT is checked at a time, but if multiple devices are to be tested with an ICT/FCT adapter, the speed limitation of the relay control can be a bottleneck. 

Figure 2: Test system diagram

Finally, while ICT can complete measurements in milliseconds, the FCT program cannot measure immediately when the device is powered on, so it is much slower than ICT. Therefore, you must first confirm that the FCT program has output before making measurements to obtain reliable data. Typically, the FCT program takes five to ten times longer than ICT to measure the same product. If the tests are combined into one ICT/FCT platform, the FCT portion may hinder production. If the two programs are separated, then one ICT instrument can supply multiple FCT test stations to increase throughput and reduce congestion. 

However, for RECOM Power’s newly developed DC/DC product range, the additional cost and test time of two separate test adapters were unacceptable, and a way had to be found to combine the speed advantage of ICT with the 100% quality assurance of functional testing (all integrated in one test adapter). This was a complex technical challenge: the devices covered by the product range have an output current of up to 6A and an input voltage of 60V. Each PCB board contains 40 semi-finished modules, which means that they need to be tested in parallel using a powerful and durable power supply. Therefore, not only is the data throughput high, but any timing errors can become a problem. RECOM signed a contract with Elmatest in the Czech Republic to jointly build a combined ICT/FCT test adapter for the Teledyne Teststation LH used by the EMS supplier. 

Zdenek Martinek, Application Engineer at Elmatest, realized from the beginning that this was no ordinary project. There were several important issues to be solved: how to combine ICT/FCT on a single board; how to handle such a high throughput of relay control data; how to speed up the FCT program and how not to damage the sensitive probes at high power. In close cooperation with Markus Stöger from RECOM R&D, they found ways to solve these problems. 

The first problem to be solved is how to combine ICT/FCT in the product's board design. Each PCB contains 40 independent circuits. These modules are not parts but a finished product that has been manufactured, packaged and silk-screened, so not all internal nodes can be connected to the ICT pin panel. This is intentional. The DC/DC converter switches at a high internal frequency, and the metal housing and multi-layer PCB form a complete six-sided Faraday cage to avoid EMI problems. Any external connection to the internal high-frequency switching node will form a path for EMI to pass through the EMC shield and emit radiation, which may cause errors in measurement. 

The solution to "how to do ICT testing on sealed products" is to make a test module for each link board. The test module can verify whether each link board is normal by connecting to all necessary ICT nodes. Once the test module passes the regular ICT procedure, the remaining modules only need to do FCT inspection. 

Figure 3: Front and back view of the PCB board, with the ICT test module located in the corner

The code required to execute a test and measurement procedure is called a test vector. The configurations of the inputs, outputs and analog channels required for the measurement are transmitted in the form of "bursts". These configurations are loaded into local on-board memory and then activated simultaneously by a timing trigger signal. The configuration is latched until the test is completed and the measurement data is transferred back to the CPU. At the same time, the next burst of data is preloaded into registers to wait for the next trigger signal. This approach allows ICT to achieve a very high data throughput of about 4µs per vector. 

But the standard relay drivers used by the GenRad Teststation are driven by a parallel input/output port (PIO) controller that receives commands from the control PC via the MXIbus (Figure 2). This configuration was too slow for our project, because we wanted to use a high-speed system controller to control the relay configuration and process different FCT measurements in one test vector. To increase the relay switching rate, RECOM's test adapter uses an "active burst" technique to implement a new relay driver topology. 

When performing an active burst, some relays are not driven by the PIO control card but directly by the D/S outputs, which remain active until the ICA measurement is completed. Each D/S can be set to 9 independent functions (idle, low or high level drive, low or high level sense, hold, drive deep serial memory, sense deep serial memory and collect CRC data). In this example, we used the drive loop to power the relay. The D/S driver output is limited to TTL voltage and current levels, which are usually not enough to drive the relay without a separate drive loop, but if a Darlington transistor current amplifier relay coil is used to make a test adapter, the D/S module can bypass the PIO controller and directly operate the relay, making the relay control instant and making the coding simpler. 

The second problem that needed to be solved was how to speed up the FCT test, since waiting for analog levels to stabilize would make the overall test time too long. The trick was to take advantage of the existing processing power of the ICA system and use waveform generation and analysis techniques such as direct digital synthesis (DDS) and discrete Fourier transform (DFT), because they are inherently faster than any analog bridge balance measurement technology. This breakthrough led us to realize that these advanced techniques could be used to determine the power-on functional test. Instead of applying a stable load and waiting for the output to stabilize before measuring the input and output currents and voltages, the output load pulse is delayed for a few milliseconds and the results are processed to derive the final output characteristics. This can reduce the measurement time by 80%. 

Figure 4: 6-terminal impedance measurement

A significant development problem was matching this dynamic load and power switching with the old Spaghetti software used by GenRad’s test equipment, which was a combination of Pascal, Assembler, and Basic. Although GenRad ceased to exist as an independent company back in 2003, it deserves credit for its durable design, and even today, it can be used with the latest operating systems based on its original hardware.