Driver System and Distributed Transmission Line Network for Driving Devices Under Test

The present invention relates generally to apparatus for use in the art of burning-in and testing circuit components prior to their distribution and use, and, more particularly, to a system for driving circuit components during testing. Still more particularly, the present invention is directed to a distributed transmission line network for delivering test signals from a driver to circuit components under test.

According to present practices, integrated circuit ("IC") packages are mass-produced and installed in electronic circuits within highly sophisticated, complex and costly equipment. As with many mass-produced products, IC packages are prone to failure, in many cases at the beginning of operation. The complexity of the equipment within which such packages are installed makes post-installation failures highly undesirable. For example, when equipment reaches the final inspection stage of production, before failures are detected, the high level skills required for testing and repair add a significant cost to production expenses. Even more significantly, when the product has been installed in the field and a service technician must make warranty repairs, the costs thereby incurred can have a significant effect on profitability. As a result, manufacturers of electronic equipment are demanding ever greater quality and dependability in commercial grade IC packages.

Quality and dependability are enhanced substantially by detection of those IC packages likely to fail in the first few hours of operation, prior to installation of the packages in electronic equipment. Virtually all IC packages manufactured today, before they are sold for use in electronic equipment, are tested or "burned-in" for a predetermined period of time to detect IC packages that are defective or otherwise liable to fail prematurely. The burn-in process includes (1) placing the IC packages in sockets arranged in arrays on printed circuit boards ("burn-in boards"); (2) placing the burn-in boards with the packages, or devices under test ("DUTs"), mounted thereon in a chamber whose environment, particularly temperature, is controllable; (3) applying direct current (dc) biases to each package on each board in such a manner as to forward and reverse bias as many of the package's junctions as possible, and/or actively clocking each package to its maximum rated conditions, such application of dc biases and clock signals being accomplished substantially simultaneously to each package; (4) removing the burn-in boards from the chamber after the IC packages have been subjected to the environmental condition of the chamber and the biases and clock signals for a designated period of time; and (5) removing the IC packages from the burn-in boards. The present invention focuses primarily on step (3), which may also include the application of test signals to the DUTs.

A conventional burn-in board includes a plurality of sockets arranged in rows and columns. Signal lines, or traces etched onto the printed circuit card comprising the burn-in board, connect each pin of each socket to a tab on the connector edge of the board. Typically, the corresponding pins from each socket on each burn-in board are connected in parallel.

A conventional burn-in system or burn-in and test system includes an environmental chamber having means for receiving a plurality of burn-in boards. The connector edge on each burn-in board is received within an edge connector, generally outside the burn-in chamber. The edge connector may be attached to a second printed circuit card, or driver board, or it may be attached to a printed circuit backplane assembly, which also has edge connectors for receiving driver boards, as shown and described in U.S. Pat. No. 4,374,317.

The driver boards generate digital signals for exercising and testing the DUTs on the burn-in boards within the burn-in chamber. The digital signals are propagated through the edge connector and along parallel transmission lines on the burn-in board to the target pin on each DUT. This use of a parallel network of parallel signal lines to connect DUTs to drivers causes certain problems.

Each DUT on a burn-in board is a capacitive load that must be charged by the corresponding drivers. Similarly, the parallel signal lines connecting the drivers to the DUTs include capacitance that must be charged by the drivers. Thus, the use of a network of parallel signal lines to drive the DUTs means that the drivers must have sufficient power capability to deliver the current necessary to charge the capacitive load represented by each DUT and each signal line. One problem, therefore, with prior art burn-in systems is the need for relatively powerful drivers. Therefore, it would be desirable to develop a suitable signal line network employing a low-power driver capable of driving DUTs at high frequencies (above 1 megahertz).

Perhaps the most efficient low power, high frequency transmission line network, at least in theory, includes a single transmission line for each driver, with DUTs connected in series along the transmission line in a so-called "daisy chain." Significant difficulties in implementing such a network, however, have prevented its use thus far in connection with a burn-in chamber.

Two of the more significant problems are improper timing of input signals caused by propagation delay and signal distortion due to reflection. See Ching-Wen Hsue, "Clock Signal Distribution Network for High Speed Testers," 1989 International Test Conference, paper 8.2, pages 199-207. Propagation delay arises from a number of factors, but is more frequently a concern as the length of signal lines increases. Thus, using long signal lines may mean that the digital signals propagated along those lines will not arrive simultaneously at each DUT.

Most DUTs have a plurality of input terminals necessitating a plurality of digital input signals. For example, a DUT with 32 pins may require input signals at numerous input terminals, such as the clock input, the data inputs, the address inputs the enable input, and the reset input, each of which is connected to a separate driver. To test such a DUT properly, the digital signals at each of these terminals must arrive in a properly coordinated sequence. If the signals do not arrive in a properly coordinated sequence at each input terminal (a condition referred to as "signal skew"), the DUT will not be exercised in accordance with the desired test pattern, and the results of the test will be meaningless. Furthermore, as the frequency of the digital signals increases, the effect of skew becomes proportionately greater. Thus, any new transmission network employing transmission lines to connect DUTs sequentially must account for the problem of signal skew as a consequence of propagation delays in the transmission lines.

Reflection occurs when the original test signal spawns reflected signals on the transmission line. Reflection of the test signal arises because of impedance discontinuities in the transmission line. A reflected test signal causes waveshape distortion of subsequent digital signals as they propagate along the transmission line.

Signal distortion may also be caused by cross-talk between transmission lines. Because the transmission lines are located within a close proximity of each other, capacitive and/or inductive coupling between adjacent signal lines may cause interference known as cross-talk.

Conventional wisdom in the burn-in industry teaches that so-called "daisy chain" distribution of signals from drivers to DUTs is not feasible. It would be advantageous to develop a suitable signal distribution network capable of employing low-power drivers operating at high frequencies.

Accordingly, there is provided a system for connecting DUT's to a driver, comprising the use of a plurality of long distributed transmission line networks linking each driver to a plurality of DUT's distributed along each transmission line network. In this manner, the capacitance represented by each DUT can be charged successively rather than simultaneously, and the driver need not be capable of delivering the substantial power needed to charge all capacitances at once. By adding distributed capacitance at key points along the distributed transmission line network, propagation delays can be introduced into the distributed transmission line to coordinate the timing among the plurality of distributed transmission lines connected to different drivers.

The present invention provides a distributed transmission line network which can provide high frequency test signals with very fast rise times to a plurality of DUT's from a single low-power driver. The distributed transmission line network minimizes cross-talk between current paths on the circuit board, while controlling the rise and fall times and the voltage level of the test signals. The plurality of distributed transmission lines connect sequentially to a plurality of DUT's. Each distributed transmission line connects to the same input terminal of each of the plurality of DUT's, thus providing the same signal, with a predetermined propagation delay, to the same input terminal of a plurality of DUT's.

The driver system preferably includes a driver with a low internal resistance, which reduces the voltage drop across the driver. The distributed transmission line network constructed in accordance with the principles of the present invention also includes a front end termination circuit, which comprises a series RC circuit. The front end termination circuit provides a voltage overshoot, which compensates for the voltage drop in the test signal caused by the internal resistance in the driver and adds resistive damping to reduce ringing.

The distributed transmission line network also includes distributed discrete capacitances positioned along the length of the distributed transmission line to control the amount of propagation delay that occurs in each individual distributed transmission line. Proper selection of the distributed capacitances makes the propagation delay on all distributed transmission lines equal so that the test signals arrive simultaneously to all the input terminals of a particular DUT, thereby eliminating skewed input signals. The distributed capacitances also reduce capacitive cross-talk between the plurality of distributed transmission lines.

In addition, the distributed transmission line network includes a far end termination circuit for minimizing reflections on the distributed transmission line. The far end termination circuit consists of an RC circuit connected to ground, which approximately matches the impedance of the distributed transmission line.

These and various other characteristics and advantages of the present invention will become apparent to those skilled in the art upon reading the following detailed description.

For a detailed description of the preferred embodiment of the invention, reference will be made now to the accompanying drawings, wherein:

FIG. 1 illustrates schematically the prior art transmission line network for connecting devices under test on the burn-in board to drivers on the driver board;

FIG. 2 is a layout of a driver board and burn-in board configuration constructed in accordance with the principles of the invention;

FIG. 3 illustrates schematically the preferred embodiment of the driver system and distributed transmission line network shown in FIG. 2;

FIG. 4 shows an equivalent circuit diagram of the driver system and distributed transmission line network of FIG. 3;

FIG. 5 illustrates schematically an alternative embodiment of an input network for the distributed transmission line of FIG. 3.

"Burn-in" refers generally to any one of several techniques whereby integrated circuit ("IC") packages are stressed within their physical and electrical limits prior to their sale or distribution so that those packages likely to become early failures in completed equipment can be discovered. The burn-in techniques include (1) loading the IC packages into sockets on burn-in boards; (2) placing the burn-in boards in a chamber whose environment, particularly temperature, is controllable; (3) applying electrical signals to the boards while subjecting the IC packages to the maximum temperature rating therefor (in some cases), monitoring the signals (in some cases), functionally testing the IC packages (in some cases); (4) removing the burn-in boards from the chamber; and (5) unloading the IC packages from the burn-in boards. The present invention focuses on the third step and presents an improved driver system and distributed transmission line network for delivering signals to the devices under test ("DUTs").

A typical IC package or DUT, as are referred to herein, comprises a dual in-line package ("DIP") having a body portion which is generally a parallelpiped with from four to sixty-four electrical leads or terminals with a generally L-shaped cros-ssection extending out and down from the opposing sides of the body. The overall width of the DIP may be, for example, 0.3, 0.4 or 0.6 inch. Sockets mounted on the burn-in board include socket contacts defining slots for receiving electrical leads on DIPs.

Unless specified otherwise herein, the terms IC package and DUT may also refer to a surface mounted device (a "chip carrier") including small outline integrated circuits (SOICs), plastic leaded chip carriers (PLCCs), ceramic leaded chip carriers (CLCCs), leadless chip carriers (LCCs), pin grid arrays (PGAs) or any other circuit made by processing semiconductor material.

In a conventional burn-in system integrated with a burn-in chamber, as shown in FIG. 1, the IC packages or DUTs are connected directly to the drivers 10, with all DUTs 11 arranged on parallel transmission lines 13, so that the signal generated by a particular driver 10 reaches each individual DUT substantially simultaneously. Parallel transmission lines, however, require a high-power driver to drive the DUTs.

Referring now to FIG. 2, the driver system of the present invention preferably includes a plurality of driver circuits 25 on a driver printed circuit board 47 providing digital electrical signals to a plurality of DUTs 100 on a burn-in board 49. According to conventional techniques, the driver board 47 includes a plurality of signal lines 40, which terminate in a driver board edge connector 29.

The burn-in board 49 includes a connector edge 36, which is received within the driver board connector 29, and a plurality of signal lines, comprising distributed transmission line networks 50, which provide electrical connections between the connector edge 36 and the DUTs 100. For purposes of explaining the preferred embodiment of the invention, the DUTs 100 shown in FIG. 2 are depicted as having four pins. It will be understood by one skilled in the art that the same principles described herein can be extended to DUTs having any number of pins.

The number of pins on any particular DUT varies depending on the type of device being burned-in or tested. Each type of device requires different signals delivered to each of its pins in a coordinated sequence in order to operate that device in its intended manner. Thus, the signal at each pin, typically, must be generated by a different driver, and the signals generated by different drivers must be coordinated with one another to cause the device to operate.

Each of the DUTs 100 includes four input terminals (labelled 1,2,3,4 on FIG. 2), each of which receives a signal from one of the drivers 25. According to the preferred embodiment, one distributed transmission line network 50 connects to each DUT 10 on the burn-in board 49, providing the same signal to the same input terminal of each DUT. In some cases, depending on the length of the signals lines and the frequency of operation, it may be necessary to employ more than one distributed transmission line and to divide the DUTs 100 among the transmission lines employed. The number of distributed transmission line networks 50 therefore preferably equals the number of drivers necessary to exercise or test the DUTs 100. For purposes of explanation, as shown in FIG. 3, four distributed transmission line networks provide digital signals to DUTs with four input terminals.

Referring now to FIG. 3, two of the distributed drivers 25 and two of the transmission lines 50 shown in FIG. 2 are depicted schematically in connection with the plurality of DUTs 100. As noted above, it should be understood that many more drivers and associated distributed transmission lines may be necessary to provide signals to other pins on each DUT. The drivers 25 each generate a digital signal that is transmitted to the same input terminal of each DUT. The requirements of the drivers 25 are discussed below in relation to FIG. 4.

Each of the transmission line networks 50 on the burn-in boards 49 preferably comprises an electrical conductor 70, an

input RC network 30, a plurality of distributed capacitors, C.sub.A1, C.sub.A2 . . . C.sub.AN (collectively referred to as CA), and a termination network 60.

The electrical conductor 70 preferably comprises a single conductor with a length that typically runs up to several feet. The electrical conductor 70 may be formed by conventional etching techniques on a printed circuit board.

The input RC network 30 connects to the connector edge end of the transmission line network 50. The input network 30 provides a high capacitive load and includes a resistor, R.sub.F, in series with a capacitor, C.sub.F, connected between the electrical conductor 70 and ground.

The DUTs 100 connect to the electrical conductor 70 at discrete points along the transmission line network 50. Conductor 70 connects electrically to the same input terminal of each of the DUTs 100, thereby providing the same signal, to the same input terminal, of each DUT 100.

The distributed capacitors, C.sub.A1, C.sub.A2 . . . C.sub.AN are located at discrete points along the distributed transmission line network 50. The distributed capacitors C.sub.A1, C.sub.A2 . . . C.sub.AN connect the electrical conductor 70 to ground. As will be discussed more fully below in relation to FIG. 4, the distributed capacitors, C.sub.A, are used to equalize the propagation delay on each distributed transmission line 50. The distributed capacitors, C.sub.A, also serve to diminish cross-talk between adjacent transmission line networks 50.

The termination network 60 includes a resistor, RT, in series with a capacitor, C.sub.T connected between the terminal end of the electrical conductor 70 and ground. The termination network 60 minimizes reflection in the distributed transmission line network 50.

The equivalent circuit diagram of a single driver system and its associated distributed transmission line network is shown in FIG. 4. As shown in FIG. 4, each of the DUTs 100, and its accompanying socket, may be represented as a capacitance to ground C.sub.D. In addition, the electrical conductor 70 between the DUT's may be represented as a series inductor, L.sub.C, with a capacitor C.sub.C to ground. The signal lines on the driver board 25 give rise to an inductance, which is represented as a series inductor L.sub.B.

The driver 25 generates a signal that is transmitted to the DUTs 100. As shown in the equivalent circuit diagram of FIG. 4, the driver 25 includes an internal resistance, R.sub.I. Selection of a driver with a minimum internal resistance, R.sub.I, is preferred for proper operation of the invention.

The internal resistance, R.sub.I, of the driver 25 causes a difference between the initial value of the input signal and final value of the input signal, at the input end of the distributed transmission line network. The voltage drop across the driver caused by its internal resistance remains constant until all of the capacitances in the distributed transmission line network have charged. When all of the capacitances in the distributed transmission line network have charged, the voltage drop across the transmission line network will float up to the output voltage of the driver V.sub.OH. By keeping the voltage drop across the driver at a minimum, the subsequent increase in voltage across the distributed transmission line network, as the capacitances on the network charge, is minimal.

If the input signal generated by the driver 25 is treated as a step signal, with an initial voltage of zero volts and a peak voltage of E volts, then the initial value of the input signal, V.sub.i, at the input end of the distributed transmission line network may be calculated as follows:

where Z.sub.D is the characteristic impedance of the distributed transmission line as defined below in formula (4).

The peak voltage of the input signal, V.sub.f, is:

To minimize the voltage fluctuation at the input end of the distributed transmission line network, the internal resistance R.sub.I must be much smaller than the magnitude of the characteristic impedance Z.sub.D. As R.sub.I in formula (1) approaches zero, the initial voltage, V.sub.i, approaches the peak voltage, V.sub.f. To keep the magnitude of the voltage fluctuation below ten percent of the original signal, the internal resistance R.sub.I must be less than Z.sub.D /10, as follows:

In the preferred embodiment, the driver 25 is selected to maintain the magnitude of the reflected signal to less than 5% of the original signal. This is done by selecting a driver 25 with an internal resistance R.sub.I, which is more than twenty times smaller than the resistance of the characteristic impedance Z.sub.D :

The characteristic impedance of the distributed transmission line network Z.sub.D may be calculated as follows: ##EQU1##

L.sub.C is the inductance of the electrical conductor 70, found by adding L.sub.C1, L.sub.C2, . . . L.sub.CX, as follows:

C.sub.C is the capacitance to ground of the burn-in board electrical conductor 70 found by adding C.sub.C1, C.sub.C2 . . . C.sub.CX, as follows:

C.sub.S is the total capacitance from adding the capacitance of the DUTs and sockets, C.sub.D, to the distributed capacitance, C.sub.A, as follows:

C.sub.D is the total capacitance from the DUTs and sockets and may be found by adding C.sub.D1, C.sub.D2 . . . C.sub.DX, as follows:

C.sub.A is the total distributed capacitance found by adding C.sub.A1, C.sub.A2 . . . C.sub.AN, as follows:

Referring again to FIG. 4, electrical signals propagate from the driver 25 along the signal lines on the driver board and through the edge connector on the driver board onto the burn-in board. This portion of the driver system may be represented as a transmission line with an inductance, L.sub.B.

The input RC network 30 marks the beginning of the distributed transmission line 50. In the preferred embodiment, the input RC network 30 is formed by conventional techniques on the burn-in board.

The input RC network 30 preferably comprises resistive/capacitive load, which, together with the inductance, L.sub.B, from the driver board signal line, compensates for the effects of the voltage drop at the driver, V.sub.R. The RC network causes a slight overshoot in the voltage of the driver output signal on the distributed transmission line network 50. Thus, the input RC network 30 raises the voltage of the signal entering the distributed transmission line network to the desired level. In addition, the resistor, R.sub.F, provides damping to prevent ringing of the L-R-C circuit.

In an alternative embodiment shown in FIG. 5, two Schottky diodes or clamping diodes 101, 102 may be added to the RC network 30, one between the electrical conductor 70 and a dc bias voltage, V.sub.OH, and one between conductor 70 and ground. Diodes 101, 102 prevent any excessive overshoot in the voltage of the signal. The dc bas voltage V.sub.OH is chosen to be the same as the voltage output of the driver 25. Thus, diode 101 prevents the voltage of the input signal from exceeding the voltage output V.sub.OH of the driver. Diode 102 clamps to ground, thereby preventing the voltage of the input signal from being less than zero.

As noted above, the DUTs 100 and corresponding sockets present to the distributed transmission line network 50 a capacitive load to ground. In the equivalent circuit diagram shown in FIG. 4, these capacitive loads are shown as capacitors, C.sub.D.

Similarly, the electrical conductor 70 between the DUTs 100 gives an inductive and capacitive effect. In the equivalent circuit diagram of FIG. 4, these inductive and capacitive effects are shown as inductors L.sub.C and capacitors C.sub.C. The capacitive values of the DUTs 100 and sockets C.sub.D and the capacitive values of the electrical conductor C.sub.C can be measured using conventional techniques, as can the inductance value of the electrical conductor L.sub.C. The values of L.sub.C, C.sub.C and C.sub.D can also be determined as set forth in the following publications: J. R. Birchak and H. K. Haill, "Coupling Coefficients for Signal Lines separated by Ground Lines on PC Boards," 1989 International Test Conference, paper 8.1, pages 190-198; J. R. Birchak and H. K. Haill, "Characteristic Impedence and Coupling Coefficients for Multilayer PC Boards," 1988 International Test Conference, paper 2.3, pages 28-38; H. K. Haill and J. R. Birchak, "Multilayer PC Board Design Approximations for High Speed Functional Testing," 1987 International Test Conference, paper 15.1, pages 354-362.

As noted above, it is essential to the proper operation of the burn-in system that all signals on the plurality of distributed transmission line networks arrive substantially simultaneously at each DUT 100. The period of time required for a signal to travel between any two points on a signal line, commonly called the propagation delay, t.sub.pd, varies directly with the length of the signal line therebetween. Capacitance in the signal line also increases the propagation delay.

The propagation delay, t.sub.pd, on each transmission line may be calculated as follows: ##EQU2## and the values of L.sub.C, C.sub.C and C.sub.S are the same as defined above in relation to formulas (4) and (5).

The propagation delay, t.sub.pd, for each distributed transmission line network is compared to find the line with the longest propagation delay. This value for the propagation delay t.sub.pd then is inserted in formula (6) above and the total capacitance, C.sub.S is treated as a variable. Formula (6) then is solved to determine the total capacitance, C.sub.S, necessary on each distributed transmission line to make the propagation delay on that line equal to the propagation delay for the slowest distributed transmission line. Once C.sub.S is determined, the capacitance which must be added as distributed capacitance, C.sub.A, can be found by subtracting the capacitance due to the DUTs and sockets, C.sub.D, from the total capacitance C.sub.S :

Small distributed capacitors, C.sub.A in FIG. 4, then are added along the transmission line as required.

The preferred number of distributed capacitors, C.sub.A, required be determined as follows:

where t.sub.pd is the propagation delay calculated above and t.sub.r is the rise time of the input signal on the distributed transmission line. In the preferred embodiment, the rise time t.sub.r of the input signal is defined as the time period that it takes for the input signal to go from ten percent (10%) to ninety percent (90%) of its total voltage.

Thus, the total capacitance to be added as distributed capacitance C.sub.A is known, as are the number of capacitors N to add. Dividing the total calculated distributed capacitance C.sub.A by the number of distributed capacitors N gives the required capacitance for each discrete capacitor C.sub.A1, C.sub.A2, . . . C.sub.AN. In the preferred embodiment, the capacitors then are distributed evenly throughout the distributed transmission line network. The distributed capacitors C.sub.A1, C.sub.A2, . . . C.sub.AN also function as shunt capacitors to minimize cross-talk interference between the plurality of distributed transmission line networks.

The termination network 60 connects to the end of the distributed transmission line network and includes a resistor, R.sub.T, in series with a capacitor, C.sub.T, connected between the terminal end of the electrical conductor 70 and ground. The termination network 60 matches the characteristic impedance of the distributed transmission line network to minimize reflections. Thus, the resistor, R.sub.T, should approximately equal the characteristic impedance of the distributed line Z.sub.D. That is,

The capacitance, C.sub.T, preferably is large to minimize the voltage swing on the capacitor during the rise and fall times of the signals on the transmission line. Thus,

In the preferred embodiment,

Alternatively, C.sub.T can be given a somewhat smaller value to create an overshoot in the voltage of the signal on the conductor 70 to compensate partially for voltage drops that occur due to electrical resistance in the electrical conductor 70. This is particularly useful at high frequency (fast rise times), where the skin effect of the electrical conductor becomes a significant factor in determining the electrical resistance of the distributed transmission line network.

While a preferred embodiment of the invention has been shown and described, modifications can be made by one skilled in the art without departing from the principles of the present invention.