Some Pitfalls in Hardware Testing

When testing hardware circuits, we often encounter some problems that are easy to ignore and difficult to detect. However, we must face up to the existence of these problems and find ways to reduce or eliminate them. Here we call them traps in hardware circuit measurement.

Load effect and filter effect of test instruments and meters

When measuring voltage or current with a multimeter, the multimeter is connected in parallel or in series with the measurement object as a load. If the load size of the measurement object is compared with the size of the equivalent load of the multimeter, if they are of the same order of magnitude, then the multimeter load will definitely have an impact on the measurement object. For example, when measuring voltage, if the size of the measured load is 10K, then if the equivalent load of the multimeter used is also of this order of magnitude, then the test result will definitely have a large error. According to the shunt theory in parallel circuits, if you want to reduce this error, you must choose a multimeter with a large equivalent load, and the larger the better. Generally speaking, when a pointer multimeter measures voltage, its equivalent load is distributed in the order of tens of kilohms to hundreds of kilohms depending on the range. When a digital multimeter measures voltage, because it uses an active circuit as an equivalent load, its value is generally distributed in the order of megaohms to more than ten megaohms depending on the range. Relatively speaking, its impact on the object being measured is much smaller, and the test results are more reliable. However, if the digital multimeter is regarded as a voltage sensor, its high-resistance equivalent load will easily pick up some noise voltage, so it will also introduce some test errors. If you want to reduce the high-resistance probe effect, you must keep the test leads and the meter body as far away from some potential noise interference sources as possible during measurement. The theory of contradiction is once again reflected here.

From a connection point of view, the measuring instrument as an equivalent load actually participates in the work of the measured circuit. If its influence is to be considered, once the measuring instrument intervenes in the test circuit, the working state of the entire circuit changes. If the measuring instrument has a small influence on the circuit, then the influence of the measuring instrument is a perturbation and can be ignored. If the measuring instrument has a large influence on the circuit, then the influence of the measuring instrument is a shock to the circuit system. This is why, sometimes when we do a test, once the test pen is placed on the test object, we see that the test object is self-excited, or does not work, or there is inexplicable noise. At this time, what we need to do is to replace the instrument or test pen probe with a small load effect.

When using a multimeter to measure AC signals, it is also necessary to pay attention to the operating frequency of the measured object. When the multimeter participates in the measurement as a load, if you simply look into the multimeter from the measuring pen, you can think that the multimeter is a filter, because its measurement circuit is nothing more than a measurement circuit composed of some resistors, capacitors, and transistors. Then this circuit must have an operating frequency range (bandwidth). If measured within this frequency range, the test result is valid. If measured outside this frequency range, the test result is inaccurate. Therefore, attention must be paid to the frequency range of the test instrument. This is the filter effect of the multimeter.

Similarly, when using an oscilloscope, AC millivoltmeter, ultra-high frequency microvoltmeter and spectrum analyzer, attention must also be paid to the corresponding load effect and filter effect. The corresponding instrument should be selected according to the load and operating frequency of the object being tested. In the manual of the instrument, there is generally an explanation of the size of the equivalent load and the operating frequency. This is more common and very easy to understand.

Generally speaking, when measuring low-frequency AC signals, if you simply want to measure the size of the signal, you can choose a digital multimeter. If you also want to see the time domain waveform of the signal, then choose an oscilloscope. If the signal is very weak, you can choose to use a millivoltmeter and an oscilloscope together. When measuring audio signals, you can choose an oscilloscope or a millivoltmeter according to the size of the signal. For AC signals in the mV order, an oscilloscope and a millivoltmeter can be used together. For high-frequency signals, you can choose an ultra-high frequency millivoltmeter or a spectrum analyzer. When using these instruments, attention must be paid to the load effect and filter effect. Especially when measuring high-frequency small signal (uV order of magnitude) circuits, if the load of the high-frequency amplifier is a parallel resonant circuit, if a spectrum analyzer (50ohm load effect) is used for measurement, the 50ohm spectrum analyzer and the parallel resonant circuit will inevitably be used as the load of the high-frequency amplifier, which will inevitably lead to a decrease in the gain of the amplifier, and the test results will inevitably be inaccurate. At this time, a differential high-impedance probe can be used in conjunction with a spectrum analyzer for measurement, which can greatly reduce the impact of the load effect.

In addition, when testing crystals, it is common to use an oscilloscope to measure timing, and there are also tests to see if the crystal oscillates. At this time, you must pay attention to the load effect of the oscilloscope probe, because there will be parasitic capacitance on the probe, which is relatively small, generally in the order of pF, but the load capacitance of the crystal is generally also in the order of pF, so the intervention of the probe will cause the frequency of the crystal oscillation circuit to shift, thereby affecting the operation of the crystal oscillation circuit, and in severe cases, it will cause the crystal circuit to fail to oscillate. At this time, you must choose a differential high-impedance probe for measurement.

Test the filter effect and load effect of the wire

Generally, when measuring high-frequency circuits, we usually use RF coaxial lines, such as the common RG-58C. Intuitively, a 50ohm RF coaxial line will not have much impact on the measurement. However, if we think from a practical point of view, the transmission line can be regarded as a network composed of a series of LCRGs. Due to the existence of CL, this network must have an operating frequency range. Due to the existence of RG, this network must have losses, so the transmission line will produce a filter effect on the test system, that is, the transmission line also has an operating frequency range problem. If we use a network analyzer to test an RF coaxial line such as RG-58C, and scan its S21 over a wide frequency range, we will find that this coaxial line is a low-pass filter. From the perspective of the non-ideality of RF wires, this measurement result should be expected. It can be imagined that if we simply let this coaxial line work far away from the corner-frequency of its low-pass filter, then from the perspective of RF matching, this coaxial line is not a 50ohm or 75ohm matched coaxial line. If this line is used, it will inevitably cause a large reflection loss, and other RF wires must be reselected at this time. The table below is a parameter table of an RF coaxial line. It can be seen that the equivalent capacitance of each foot length is about 20~30pF. For high frequencies, this is a relatively large capacitance. This is why we must choose the shortest possible coaxial line when performing RF measurements.

Uploaded on 2012-2-21 15:25:52

Download attachment (35.87 KB)

 

Accordingly, when measuring high-frequency signals (or high-frequency digital signals), the equivalent load effect of the ground wire of the oscilloscope probe must also be considered. The intervention of the probe ground wire changes the characteristics of the test system. As an inductive load element, the probe ground wire will inevitably cause changes in the transmission characteristics of the test object, thereby causing changes in the test results. Severe cases will cause system oscillation and self-excitation.

The above are the filter effect and load effect of the test wire.

If we understand the traps in the above test systems from the perspective of basic circuit theory and signals and systems, we will easily understand and know how to reduce and eliminate the impact of these problems when selecting instruments or tests.