JPH0254186A - Inspecting method of circuit board - Google Patents

Abstract

PURPOSE:To prevent false determination and to make inspection highly precise by a method wherein a measuring error based on a quantization error is added automatically to a tolerance for the irregularity of an impedance to be measured, in accordance with the magnitude of digitally-converted data. CONSTITUTION:A one-cycle alternating-current signal 1 for measurement is given to each measuring point of a circuit board 3 of good quality, an imped ance corresponding to a current flowing for each of positive and negative half waves thereof is measured 20, and mean values ZSAmu and ZSBmu of the impedance are adopted as reference impedances at the measuring point. Mean square values sigmaSA and sigmaSB of the mean values ZSAmu and ZSBmu and the measured data on the impedance are calculated, and values alpha1A = 3sigmaSA + ZSAmu X 2% and alpha1B = 3sigmaSB + ZSBmu X 2%, for instance, are set as tolerances for the irregularity of the test board 3. Moreover, a variable rho relating to a value of digitally- converted data is calculated for the positive and negative half waves, a value alpha2 = Zmu X k3rho (k3 is a constant) is added to a value alpha1 automatically as a measuring error accompanying the quantization of A/D converters 18 and 19, and a value thus obtained is taken as a tolerance for determination of the quality of the board 3.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a circuit board inspection method for inspecting the quality of a circuit board on which electronic components, etc. are mounted. The present invention relates to a testing method suitable for a circuit board on which elements having different impedances for negative half-waves and the like are mounted.

[Conventional example]

BACKGROUND ART Circuit board testing devices called in-circuit testers have come to be used to test circuit boards on which electronic components and the like are mounted.

An example of this is shown in FIG. 7. For example, a signal source 1 emits a measuring AC voltage of a predetermined frequency, and a circuit board to be inspected (hereinafter referred to as "test board") 3 is transmitted via an amplifier 2.
In addition, a current flows through the substrate 3 that is inversely proportional to the magnitude of its impedance Zx. This current is taken in and detected by, for example, a current detector 4, converted into a voltage there, and then digitally converted by an A/D converter 5 and applied to a measuring section 6. The measurement unit 6 calculates the impedance 2X of the test board 3 using this conversion data, and calculates the impedance 2X of the test board 3 using the reference value Z stored in the memory in advance.
, compared with. In this case, as shown in the formula below, if it is within a predetermined tolerance ±α, it is determined to be good, and if it is outside the tolerance, it is determined to be defective, and the determination result is displayed on the display 7 etc. There is.

Judgment of "good" Zs-α≦ZX≦ZS+α Judgment of "defective"Zx<Zs-α or Zx>Zs+α [Problem to be solved by the invention] The advantage of this conventional pass/fail judgment method in board inspection is that it is simple. have. By the way, as is well known in circuit board inspection equipment equipped with a digital measurement system, for example, during A-D conversion, IL is applied to the analog input voltage.
There is a dead margin in which the digital data does not change even if there is a voltage change equivalent to SB, and a so-called quantization error occurs.

In this case, if the current flowing through the test board is relatively large, the input voltage of the A/D converter is also large, so it is not a big problem, but if the current flowing through the test board is small, the A/D
The ratio of the dead width voltage to the input voltage of the D converter increases, and measurement errors increase and cannot be ignored. However, in conventional board inspections, this point is not necessarily taken into account, so the value of Z, for example, is 2. ±α
Even if it is within the range, the measurement error due to quantization error will be added to the measured value and it will deviate from the range, making it difficult to say "good".
It may be incorrectly determined that the product is defective.

This invention was made in view of the above circumstances.

The purpose of this is to automatically add the measurement error based on the quantization error to the tolerance for the original variation in the impedance to be measured ZX according to the size of the digitally converted data, thereby preventing misjudgments. An object of the present invention is to provide a highly accurate circuit board inspection method.

[Means to solve the problem]

Referring to FIG. 1, an embodiment of the invention is shown.
In order to solve the above problems, the following means A to C are provided.

For example, one cycle of measurement AC voltage is applied from a signal source to each measurement point on n circuit boards that have been confirmed to be good in advance, and the current flowing during the ten half-wave period and the negative half-wave period is measured. Impedance 2 corresponding to the current average value during the 10-half wave period and the -half wave period by measuring the average value, respectively.
A reference data memory 20c and a measurement data memory 20d that determine μS^ and Zμss and hold these values as reference impedances ZsA and zsm at the measurement points.

First, calculate the root mean square value σ from the current data and the average value in the ten and a half wave period and the -half wave period at each of the above measurement points, and for example, calculate the ten half wave period and the -half wave based on this root mean square value σ. Tolerance α0 for variations in impedances to be measured ZXA and Z811 during the period. , α11I are set, and as a measurement error due to quantization of the impedance Z, that is, a reading error, for example, tolerances α2^ and α2+1 related to the magnitude of digital conversion data of the measured value of Z are set. Tolerance setting means 20a.

C9 For example, in the impedance measurement during the ten-half wave period and the -half wave period, the above-mentioned variation tolerance α and reading error α
Comparing means 20b automatically adds 2 and performs impedance comparison according to the following formula to determine the quality of the test board 3.
.

(Judgment formula) % formula % By providing the above means, the quality of the test board can be correctly determined even if the impedance measurement value includes a measurement error due to quantization.

〔Example〕

In this example, first, multiple boards that have been previously confirmed to be good are measured and their data is collected, then a test board is measured using the same method, and the data from both are compared to determine pass/fail. It is supposed to be done.

Referring to FIG. 2 in conjunction with FIG. 1 above, for example, one cycle of measurement AC voltage as shown in FIG. Added. As a result, a current flows through the substrate 3 in inverse proportion to the magnitude of the impedance Z8. In this case, if a diode or the like is installed on the board 3, it will have a low impedance for the positive half wave of the AC voltage for measurement.
For negative half-waves, the impedance is high, so the current that flows is as shown in Figure 2 (b). This current is detected by a current/voltage converter 13, for example, and then converted into a voltage with a predetermined gain as shown in FIG. 2(c), and applied to a switch 14.

In this embodiment, the operations of the signal source 11 and the current/voltage converter 13 are controlled, for example, by the measuring section 20, and the signal source 11 sends out an alternating current voltage of a predetermined frequency for one cycle as described above. Further, the current/voltage converter 13 detects a positive current and a negative current flowing through the substrate, and converts them into positive or negative voltages of the same polarity with a constant gain. Note that the switching controller 15 is configured to switch between signal sources 1 and 1, for example.
The positive half wave and negative half wave of the AC voltage for measurement emitted from 1 are waveform-shaped into square wave voltages, respectively, and the output of the waveform is used to turn the switch 14 into a positive half wave as shown in FIG. 2 (d). During the period, the contact is driven toward the A side, and during the negative half-wave period, the contact is driven toward the B side.

As a result, the converted voltage during one half-wave period O to π of the current flowing through the substrate 3 is integrated by the integrator 16 as shown in FIG.
The converted voltage at ~2π is integrated by an integrator 17 as shown in FIG.

If the integrated voltage of these good boards is VS^*V8B, then
These two integrated voltages are digitally converted by A/D converters 18 and 19, respectively, as shown in FIGS. The measurement unit 20 uses these integrated voltage data to determine the center impedance ZSA of the good board for each positive and negative half-wave of the AC voltage for measurement.
In addition to calculating μ and ZSaμ, the tolerances α, ^, α1B for impedance variations of the test board are determined, for example, by the tolerance setting means 20a, and at the same time, the tolerance additions α2^, α2B due to quantization errors are calculated. When the test board 3 is finished, the impedance ZXA+ Z at the positive half wave and negative half wave is determined using the same method as above.
After the XB is measured and the data are compared using equations (la) and (Lb), the results of judgment of good or bad and the measurement data are sent to the recording/display section 21 to be displayed and recorded.

In this case, the tolerance α1 for impedance variations in equations (la) and (lb) and the addition value α of measurement error due to quantization are calculated using, for example, the following equation (%). There is. In addition, in the above formula, k l l
k g −k 3 is a constant, and ρ is a variable related to the magnitude of digitally converted data of each positive and negative half wave.

Hereinafter, the above α1. We will explain how to find α.

Now, when the full-scale input voltage vps is digitally converted by an A/D converter of N pictographs, 2N pieces of digital data are obtained. In this case, there is a one-digit difference in the minimum bit, or LSH, between adjacent data. When converting this magnitude difference into an analog input voltage ΔV, as is well known, ΔV = V ps X 2'
--(4), and is a constant value determined by the full-scale input voltage ■F8 and the number of conversion bits N.

Therefore, even if the converted analog voltage input to the A/D converter changes within a range of ΔV or less, the digital data does not change, and this voltage ΔV becomes the dead width of the A/D converter. The above 2-N is generally called the resolution of the A/D converter, and the digital data is the dead width voltage Δ
Ideal A/D adjusted to match the median value of V
In the converter, ±ILSB/2 corresponding to ±ΔV/2
is called the quantization error.

Here, as shown in FIG. 3, if the voltage lower than the full-scale voltage vps is V8Aμ and its digital conversion data is 08A, then the width ΔV including this voltage VIAμ is
Similarly, other voltages VSAI VIIA', etc. in data I
) will be converted to sA.

Therefore, the voltage V8Aμe Vl! lAt V8A tΔ
The true value of the impedance corresponding to v is I) sA
μ.

D8A, D8A' *If ΔDA, 108^-D$
In the range of A'l≦ΔD^, it is a matter of course that the measurement is DsA=DsA'=Ds8μ, and an error occurs in the quantum.

Therefore, if the reading error based on digital data conversion, that is, the ratio of the quantization error to the data read value is ε^, then i A: D8A − D$Aμ/DsAμ or (A=IDsA − DSAμl/DIIAμ, which is When expressed as an analog voltage, IA=1VsAVs^μl/VsxJA or 5, WVsA−VIIAμ/VIAμ.

The maximum error occurs when the numerator on the right side of the above equation is equal to Δ■, so in general, EA=ΔV / VsAμ・” ”’ (5)
In this case, the numerator ΔV on the right side is a constant value as described in the explanation of equation (4), so in order to reduce the reading error, the denominators V and Aμ may be increased.

Substituting equation (4) into equation (5) and rewriting it.

i A= Vpg X 2-N/ VSA uwhere,
Since VFg is the full case input voltage, N bits A/
The conversion data of the D converter corresponds to 2N. Also, since vsAμ similarly corresponds to DsAμ, the above formula can be written as i: 2'X 2-N/DgAJ'
, , , , 151).

Therefore, if the reading error ε^ is, for example, within 1%, then D11A/J=1 when εe=0.01 in the above equation.
00 (01100100) is obtained, and the input voltage (K) of the A/D converter can be set to satisfy this.

In this embodiment, for example, the measurement section 20 monitors the conversion data (FIG. 2 (G)) of the A/D converter 18, and its input, that is, the output of the integral 1918 (FIG. 2 (E)) is as shown above. The gain of the current/voltage converter 13 is controlled and fixed in this manner.

Figure 4 (a) shows the integrated voltage vs Ai during the half-wave period O to
For example, the average value Vs^μ of (i=1.2....n
), and the distribution of each integrated voltage is plotted against this average value. However, in the figure, the horizontal axis represents analog voltage, and the vertical axis represents the number of measurement data (number of boards). Here, if n is set to an appropriate size, the distribution state of the number of data approximates a normal distribution that is bilaterally symmetrical with respect to the average value V, 8μ, as shown by the solid line, for example.

Therefore, if we calculate the root mean square value σS^ from, over 99% of the number n of test boards falls within the range of ±3σS^, and as is well known, the number that falls outside this range is less than 1%. .

In this case, since all the test boards used are non-defective products, it is necessary to judge the number of 1% or less outside the range of ±3 deviations as non-defective products.

Here, if each analog voltage data V3Ai is converted into digital data by an A/D converter with an extremely large number of bits, and the impedance is determined by making the measurement error due to quantization negligible, for example, the solid line in (a) in Figure 5 As shown, an impedance distribution diagram similar to that shown in FIG. 4(a) above is obtained. However, in reality, an 8-bit A/D converter is used and a 1% measurement error due to quantization, that is, a reading error, is allowed, so the distribution of the reading value DsAi of the digital conversion data corresponding to the impedance ZsAi is shown by the dotted line in the figure.

Therefore, in this embodiment, measurement errors are taken into account, and if, for example, □ is set to 3 and k3 is set to 2% in equation (2), and the reading error can be ignored, then a L = 3 a 8A + 0 .02 Z SAμ
[Ω]...-(2a) or αz=(3ffaA/ZsAμ+0.02)X100
[%] - Set to (2b).

Next, the measurement error due to quantization, that is, the addition α2 to the tolerance for the reading error will be explained. As shown in FIG. 2 above, for example, when a positive voltage is applied to a diode and when a negative voltage is applied to the diode, its impedance is significantly different. Currently, the digital conversion data on the half-wave side (Figure (G)) is approximately 100.
When the gain of the current/voltage converter 13 in Fig. 1 is controlled so that the count becomes (01100100), the digital conversion data on the -half wave side (Fig. 2 (H)) becomes a value considerably lower than 100 counts, and therefore, , the reading error becomes large.

If the impedance variation on the one-half wave side can be regarded as a normal distribution, the distribution can be plotted, for example, as shown in FIG. 4 (b). However, when digitally converted data of voltage having such distribution characteristics is plotted, if the converted data value is large, it will look like the solid line in Figure 5 (b), but if the data value is small, a reading error will occur. Due to the influence of

This increase in variation is originally caused by α1 when calculating α1 above.
However, when the number of samples of non-defective substrates n is small, it is not always possible to absorb all of it and it cannot be ignored.

In this case, the leading error ε8 has its center value as DS
If IIμ, then according to formula (5'), s @=17 D
H1μ −−−−−−(6) Expressing this as a percentage and using the variable ρ, ε3=ρX100(%)...
...(7) However, p=1/(DssμX 100)
becomes.

This error is defined as the tolerance α2 due to the reading error,
- When formed into a film, α2 = ε8 = ρX100 (%)...
・It becomes (3a). For the center impedance Zμ, α, = 2μ・k,・ρ[Ω] ・・・・・・
(3b), and an equation equivalent to the above equation (3) is obtained.

Here, 3 is usually 1.

In this way, the tolerance α1 for the impedance variation and the tolerance α for the conversion error of the A/D converter are determined based on the actual measurement data of the non-defective board.

After setting, start measuring the impedance of the test board.

Note that if the diodes on the board are connected in the opposite direction to the state explained above, the reading error will increase during the 10-half wave period, so in practice it is necessary to α, (αiAt αtl) and α2
It is necessary to set (α,,αam).

In the above explanation, the values of the constants kl, k, , k, etc. are shown as examples, and may be adjusted as appropriate depending on the actual situation when performing the inspection. The same applies to α2, and if there is no problem, it is desirable to simplify the inspection by rounding off the fraction of the value obtained by calculation.

In addition, FIG. 6 shows a flow diagram of an example in which the measurement section 20 automatically sets the tolerance α1 for the variation and the addition value α2 for the measurement error.

〔effect〕

As explained above in detail, in the present invention, one cycle of a measurement AC signal is applied to each measurement point of a circuit board that has been confirmed to be good, and the current flowing for each positive and negative half wave is Measure the corresponding impedance, and use the average values Z, Aμ, and ZSBμ as the reference impedance of the measurement point, and also calculate the root mean square values σS^, σ, B of the average value ZSAμeZ&Bμ and the above impedance measurement data.
3 times that is, 3σSAp 3σ8B, and add, for example, 2% of the average values Z8Aμ and ZSmμ to this value α
i=3σ0+Zs＾μ×2%, a1B=3σlB
+ZBBμ×2% is set as the tolerance for variations in the test board.

Furthermore, for the positive and negative half waves, a variable ρ related to the value of the digitally converted data is calculated, and the product α2 of this variable ρ, the known constant 3, and the impedance Zμ is α2=ZμX
k and ρ are automatically added to α as measurement errors due to quantization of the A/D converter, and are used as tolerances for determining the quality of the test board.

Therefore, according to this circuit board testing method, it is possible to automatically perform accurate pass/fail judgments, taking into account measurement errors associated with AD conversion as well as variations in the original impedance of the test board.

[Brief explanation of the drawing]

1 to 6 relate to embodiments of the present invention; FIG. 1 is a block diagram showing an example of the configuration of a circuit board inspection device to which the present invention is applied; FIG. 2 is an explanatory diagram of the operation of each part; Third
The figure is an illustration of measurement error due to quantization, Figures 4 and 5 are illustrations of the principle of setting the pass/fail judgment tolerance α1゜α2, and Figure 6 is an illustration of the tolerance αL, α□ automatically in the measurement unit. FIG. 7 is a block diagram showing the configuration of a conventional device. In the figure, 11 is a signal source, 3 is a circuit board to be tested, 13 is a current/voltage converter, 16.17 is an integrator, and 18.19 is an A/V converter.
The D converter 20 is a measuring section.

Claims (1)

[Claims] (1) A response signal obtained by applying a measurement AC signal from a signal source to a circuit board under test is converted into a voltage and integrated, and the impedance of the board is measured in a measurement section using the digital conversion data. In a circuit board inspection method that compares the value with a reference value taken in advance from a non-defective board and determines whether the board is good or bad based on whether it is within a predetermined tolerance, the impedance variation of the board is compared with the standard value taken from a non-defective board. A tolerance is set by adding a predetermined value to 3σ of the impedance distribution characteristic of the board, and the positive half of the AC signal for measurement is used for impedance measurement errors caused by quantization errors associated with digital conversion of the response signal. A circuit board inspection characterized in that a predetermined value is automatically added to the above-mentioned tolerance according to the magnitude of digitally converted data of a response signal in a wave period and a response signal in a negative half-wave period to determine pass/fail. Method.