TW472150B - Improved test method of test apparatus - Google Patents

Abstract

The present application is an improved test method of test apparatus. Because the test machine is expensive due to the high precision of the test machine, the present application provides a test method capable of calibrating the automatic test equipment (ATE) interface. By using the digital signal skill, the parasitic effect on the test path and the tiny variation of the signal source are completely removed, thereby obtaining the intrinsic response of the object under test. Because the quality of the device can be easily distinguished by using this intrinsic response, the conventional test equipment with a very high precision can be replaced with test equipment with a relative low precision. In addition, because the interface is automatically calibrated, the time required for setting the machine is decreased, thereby reducing the test cost.

Description

472150 V. Description of the invention (1) Yuelu Lu · This case is an improved measurement device measurement method. It is used on a measurement device to use mathematical processing techniques on digital signals to eliminate parasitic effects and signal sources in the measurement process. And other error sources, it improves the accuracy of the measurement equipment and reduces measurement costs. Background: Automated test equipment (Ate, Automatic Test Equipment) is an indispensable measurement equipment for the second generation of ultra-large integrated circuit package testing. For the month 2 please refer to the schematic diagram of the conventional ATE architecture in Figure 1 for its basic characteristics. It consists of 3 components (test device 1), signal source 16 and other components. It can be seen from this figure that the measurement signal on the automatic test instrument basically passes through (l) pogo Pln n_ connected to the contact device of the device under test 15 (DUT, device under test); (2) connected by howling 〇pin Signal transmission line 13 to the external terminal of the ATE device; (3) Signal transmission line between the external terminal of the ATE device and the 1 or 16 source signal transmission line and other components. From the above 彳 § said that the path that must be passed, it can be found that because of the parasitic effect on the path that the signal passes, the results observed on the instrument are not connected to the automatic test equipment because of these parasitic effects. 2. The results of direct measurement of the two ends of the tested component are somewhat different. Since these parasitic effects only affect the AC signal, no special treatment is required for DC signal measurement. In addition, another issue is the official nature of the test data. Because when ate is actually tested, many ATEs usually test the same product at the same time, but the parasitic effects caused by each ATE are not

472150 V. Description of the invention (2) Same as 'Therefore, there will be some gaps in the measurement results of each ATE, and the signal sources of each machine are not the same'. This will cause how to judge the quality of the measurement. Trouble. In order to briefly explain the impact of Figure 1 during measurement, we use the test signal X (t) 21 on the equivalent block diagram of Figure 2 and enter the response wl (t) 22 'the response of the device under test h (t ) 23, the output response ¥ 2 (t) 24 and the output response y (t) 25 are described as observation results. From the test environment shown in Figure 2: The conventional ATE functional block diagram, we can see that the observed output response y (t) will be affected by parasitic effects and cause distortion. The relationship between the time domain and the s domain is as follows. .y (〇 = x (t) * (/) * h (t) Y (S) = X (S) -Wl (S) -H (S) -W2 (S) ... ⑴ If we use To measure using this method, you must first use a standard DUT to perform the test. Because the DUT's response is known, so in order to obtain the same result, we must appropriately adjust the input signal to make the output respond It can be the same as the known result, and then perform the actual test. When using this method for testing, we must face the problem that “the signal source must be accurately adjusted to compensate for the gap caused by parasitic effects” and this This kind of fine adjustment is impossible for generally cheap equipment, so it must be adjusted with very expensive equipment. To solve these problems, the current ATE uses very expensive pi η

D: \ A \ case \ c60 苏朝琴 \ p6〇-〇〇3.ptd page 5 still 150 5. Description of the invention (3) driver (p〇g0) device (such as adding a waveform reforming circuit at the input end) , So that the amplifier amplifier can be minimized, plus the correct use; ^: the parasitic effect of the source correction, so that the measurement results can fall to two lines of periodic product testing. However, using this method basically has j and then the signal source may be subject to a risk due to a change in some factors, which is effective. Therefore, the general approach is adopted if the product: it is judged that it takes a lot of time for each correction1 This is jealous. Because non-expensive signal sources will be used to reduce: If you want to avoid the cost of parasitic effect === parasitic effect, the biggest challenge is how to eliminate "2" results and how to eliminate ^ Under the change of 楗 1, the signal can still be properly processed. This known test is measured as f 纴. ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, and, `,` `,,,, and ATE, ATE For this reason, in view of the lack of know-how, the author created the spirit of perseverance and research, and finally created:; The following is a brief description of the case. Purpose of this case This case is an improved measurement method for measuring equipment. It is used in a signal transmission system, especially the signal transmission system on a measurement device, to achieve D: \ A \ case \ c60 苏朝琴 \ p60-003. ptd 472150 V. Description of the invention (4) The purpose of improving the measurement accuracy of the signal transmission system. It includes the following steps: providing a signal transmission system with a position to be tested on the signal transmission system; providing a unit element to be placed in the position to be tested to obtain a first measurement effect; Placed in the position to be measured to obtain a second measurement effect; and performing a normalized operation on the first measurement result and the second measurement result to obtain a third measurement effect, and according to the third measurement effect 'to Determine the measurement result of the DUT. As described above, the measuring device is an automatic measuring instrument ate (A u t o m a t i c T e s t E q u i p m e n t). As described above, the measurement device provides a test signal to enter the signal transmission system, so that the input signal should be input to the person and should be output; the number is used as the observation result of the measurement. It is stated that the gain of the unit of 5 Hai unit is 1. As mentioned above, where the standard .... ^ is divided by the first measurement effect, the second measurement effect / second measurement effect of Yunyunhai is its quotient. The picture does not explain: This case can be illustrated by the following diagrams; 5 1 Λ & Solution: and 5 sheep detailed explanations, we can get a deeper picture. Figure 1: Schematic diagram of the conventional ΑΕΕ. " Functional block diagram of provincial A Τ E. Removal of the preferred embodiment of the present case The preferred embodiment of the present case m method is not considered. This case better implements the example of a pure pulse under different input signals at the gate ^ J: D: \ A \ case \ c60 Su Chaoqin \ p60-〇〇3. Ptd 472150 V. Description of the invention (5) Intrinsic Response . Figure 6: Intrinsic Response Extraction Example 1 of the preferred embodiment of this case. Figure 7: Intrinsic Response Extraction Embodiment 2 of the preferred embodiment of the present case. The meaning of each code in the picture: 11: pogo pin 12: output signal transmission line 1 3: input signal transmission line 1 4: measurement unit 1 5: device under test 1 6: signal source 21: test signal x (t) 22: Input response wl (t) 23: DUT response 24: Output response w2 (t) 25: Output response observation result 3 1: Unit element 3 2: First measurement effect 33: Second measurement effect 41: Calibration memory 42 : Deconvolution 43: Standard signal source 44: DUT Gold Copy response 45: DUT 'device under test 46: Convolution 47: Pure impulse response ( Intrinsic Response) Detailed description: This case is an improved measurement equipment measurement method, which is used to improve the transmission effect of a signal transmission system on the -f measurement equipment. To achieve such a goal

D: \ A \ case \ c60 苏朝琴 \ p60-003. Ptd page 8 472150 V. Description of the invention (6) =, I use a three-step measurement method. The first step is to measure the parasitic effect on the transmission path (referred to as “Measurement”). The second step is to measure the response of the component under test (called “Measure Measurement”). A three-step process that uses a signal processing technique called Iterative Deconvution to remove the effects of parasitics from the response of the device under test and restore it to the original component response that you want to be undisturbed. Here again It is worth mentioning that if there is a slight change in the source of 疒, between different subjects, because in this process ^, the same changes will occur in the calibration measurement and component measurement, so this can be changed. The changes cancel each other out and return to the original response. The signal change between the calibration measurement and the component measurement can be ignored because the time is too short. To clearly explain the above steps, I can refer to Figure 3: Remove A schematic diagram of the parasitic effect method and use mathematics to explain this process. On the calibration test day, the parasitic effect measured by the gain one amplifier, that is, the unit element 31, is as follows: p (t) = x ( i) * (ί) * b (t) * w2 (t) P (S) = X (S) Wx (S) B (S) W2 {S) get a first measurement effect 32, which is p (t) Or P (S). Then we use the same signal to test the test object, the results are as follows:

D: \ A \ case \ c60 苏朝琴 \ p6〇-〇〇3. Ptd page 9 ^ 15〇 5. Description of the invention * w2 (t) Y (S) = X (S) -W {(S) -H (S) W2 (S) knows a second measurement effect 3 3, which is y (t) or Y (S). In this way, the transfer function of the test object can be obtained in the following ways: κο = Μ ρ (0 H (S) = ns) ns) x (t) * (/) * h (t) * w2 (/) x (t ) * w {(/) * b (t) * w2 (/) _ XOS) ·, #))) _ ΡΓ20) When we obtain the conversion function h (t), we can use the ideal test result hg (t) and The following method to verify the obtained answer. y0 (t) = h (t) ^ u (t) A⑺ ~ ⑺ From the above mathematical derivation, we can know that if the unit element 3 1 with a gain of 1 can maintain a very stable response within the measured frequency range, then The turn-out response yo (t) has nothing to do with the input response phantom 22) and output response w2 (t) 24 of the parasitic effect. Based on the above analysis, we built the measurement system into a measurement architecture as shown in Figure 4. The standard signal source u (t) 4 3 and the standard DUT respond to hg (t) 44 via the convolution product 4 6 to produce yg (t). Convolution product

D: \ A \ case \ c60 Su Chaoqin \ p60-003. Ptd page 10 472150 V. Description of the invention (8) y〇 (t). In the block diagram of Figure 4: Signal Reconstruction function, the result P (s) of the second positive measurement is corrected. Because of this measurement :: the signal path used in the interposition measurement, the measured time is measured with the actual measurement. The effect is the same. When it enters the real world, we use the data stored in the calibration memory p (i ^ ^ 45 to move Y 反 to perform the convolution product deconvolution. Using this inverse ^ results parasitic effects and traces of the input signal source The change is eliminated, and π is used to get the net impulse response h (t) 47 (Intrinsic RespOnse) to obtain a pure proof of the aforementioned method, so first change the signal source to test whether to obtain the same results. In this experiment Change the measured rise and fall times. The experimental results are shown in Figure i: Different input cores; Net impulse response (Intrinsic Response). The first block in the pure figure 5 below is the input signal source. This signal source is divided by In addition to the original square wave, it also includes the different rise / fall slopes and different delay times generated by 丨% Xiangli 2M cycles due to the relationship of the signal generator. The output is obtained during the calibration measurement. As a result, this result is ^ = in the calibration memory. The third column is the result obtained by actually measuring the device under test. The fourth column is the result obtained by using the inverse convolution method. From these results: to find that, Parasitic using this method The removal should be very effective. In addition to the changes in the input waveform itself, another condition to consider is whether the same pure impulse response can be obtained on different machines. In order to prove this view, Figures 6 and 7 show the results obtained on different sources.

D: \ A \ case \ c60 Su Chaoqin \ p6〇-〇〇3. Ptd page 11 47215ο 5. Description of the invention (9) Refer to Figure 6: The implementation number of this Response Extract impulse response extraction and the use of gain are ideal for different situations The lower and second bars use the first bar in the pure pulse. In the figure, the output of the same signal can be slightly different compared to the output of the second column. It needs to be adjusted. There is no cause calculation (inverse product and no parasitic effect to make a judgment. In addition, the signal obtained in this figure is noise-free, so the result is that if the input signal is used to reduce the noise, the measurement method, Therefore, the change in the results from the above can be used to reduce the auto-amplification and use it to separate the pure impulse response extraction (Intri ο η) of the first embodiment, as well as the first example of Figure 7. The two figures first The actual response of the buffer signal displayed on the display is extracted from the impulse response. Obviously, the output produced by the machine is different. The difference is that the two are pure.) The error causes the rotation to be roughly measured. The parasitic effect is produced by the method. The result will be sent, but the noise of the net pulse is the same, because the incoming signal should result in a significant parasitic effect. The ones in Figure 6 and 7 will respond to the extraction. It is easy to send this 1 ns ic: The case is purely an input letter, and the test results are divided in the second column. No. The effects of these two responses are different; if the output is different from the ideal cause result, the results of the elimination effect of the result are obtained from the letter 6 and figure 7 themselves. The results are taken as periodic news. The original is omitted here. The result of pure pulse positive measurement analysis is that the result is the measurement of the actual signal. Because the signal outside contains the measurement environment, the general noise reduction technique is used. No., the noise can be averaged by the M period. Because this technique is common in the proof of various techniques. Based on the signal source's minor changes and the technique of response extraction between machines, the signal can be restored. This method can make the test environment conditions correct. The database will become unique.

472150 V. Description of the invention (10) Basically, the method has prevailed. The method caused by the modification of the intrinsic device is not only the original result of the measurable result. Because the cost of the machine can reduce the line, we will The use of face books to slow down the use of a lifetime dagger to correct the amount of high-frequency details of the method, such as the high-frequency amount of the method required to study how to benefit the known effects ~ set of low test Machines have increasingly used the concept of compensation for many years, but in this article, -extraction caused by micro-gap has been proven to reduce the accuracy of calibration. The main purpose is to send different simulations between the research surfaces and verify it. In addition, the current test cost can be reduced. The result of DC measurement is still lacking a specific method. How can pure pulse response extraction be used to solve the problem? The feasibility of error and trace signal change 'Therefore, by this accuracy, but also improved the measurement and measurement techniques to eliminate the difference, so that the ATE analog module and the actual test theory parasitic effect removal method, Taiwanese manufacturing and product measurement trends. In summary, the available measurement equipment can reduce the quantity T. Conclusion: T, the application of this case may also improve its measurement accuracy. =, But it is not only not reduced, it can be performed automatically during operation. 1. The measurement equipment applied in this case is complicated. Therefore, this case is an accurate correction, which can simplify the artificial control of patentability, and propose a highly competitive invention according to law, which is not enough to cover all the cases and applications of θ. However, the above-mentioned embodiment is still ° 疋The scope of patent application is attached as attached.

472150 V. Description of the invention (11) References: [1] Chauchin Su, Yue-Tsang Chen, Shyh-Jye Jou and Yuan-Τζυ Ting, " Metrology for Analog Module Testing Using Analog Testability Bus ", 1996 Infl Conf. On Computer Aided Design, San Jose, CA, Nov. 1996, pp. 594-599.

[2] Chauchin Su, Yue-Tsang Chen and Shyh-Jye Jou, " Parasitic Effect Removal for Analog Measurement in P1149.4 Environment, " Proc. 1997 Infl Test Conference, Washington DC, Nov. 1997, pp. 499- 508.

[3] Chauchin Su and Yue-Tsang Chen and Shyh-Jye Jou, " Intrinsic Response for Analog Module Testing Using Analog Testability Bus ", 1st revision by ACM TODAES.

[4] Yue-Tsang Chen and Chauchin Su, " Analog Module Metrology Using MNABST-1 P1149.4 Test Chip ", 7th Asia Test Symposium, Singapore, Dec. 1998.

D: \ A \ case \ c60 苏朝琴 \ p60-003.ptd page 14

Claims (1)

472150 VI. Application for Patent Scope Measurement 1. An improved measuring device measurement method is used in a signal transmission system, especially the signal transmission system on a measurement device, to improve the measurement accuracy, which includes the following steps: A signal transmission system is provided, and the signal transmission system is provided with a position to be tested; a unit element is provided at the test position to obtain a first quantity effect; a test element is provided at the test position Measure the position to obtain a second measurement effect; and standardize the first measurement result and the second measurement result to determine the first measurement result and the second measurement result.测量 The measurement result of the DUT. φV ； fa # # ， J ^ ^ ^ ^ > Among them, HH: the measuring equipment is a white gas i T t F, which is an automatic measuring instrument ATE (Automatic i es t tlplpment) 〇 3. If applying for a patent The range is 篦 9 &, where the measurement equipment provides ::, Improve the measurement method of the measurement equipment so that an input signal enters the signal transmission system, and an output signal is used as A $,丨 Γ the second measurement position, and the position to be tested is rotated out, 丄 is, then the view of 1 is 4. As described in the scope of patent application No. 1, $ ,, σ, where the gain of the unit element is described in 丨The improved measuring device measurement method, such as the first item in the scope of the patent application of the 5th Patent, where the standardized operation is described in the face of the improved measuring device measurement method 'effect. , And divide the second measurement effect by the first measurement 472150 D: \ A \ case \ c60 苏朝琴 \ p60-003. Ptd page 16