Research on the reliability of power supply equipment

The reliability issues discussed in this article are applicable to almost all electronic systems and mechatronics equipment. As the basic components of electronic systems, power supply equipment, especially AC power supply equipment, is particularly important to maintain normal working ability for a long time and stably. A research report from Bell Laboratories in the United States pointed out that the main cause of damage to precision electronic equipment such as computers is voltage surge, that is, short-term (about 10ms) or long-term overvoltage, accounting for 45.3% of all damage causes. Lightning strikes account for 9.4%. The main causes of abnormal equipment operation and bit errors are low voltage (including short-term pulsation) (sags), accounting for 87%, and pulse spike interference, accounting for 9%. Therefore, many well-known manufacturers in the world have strict site power supply standards and require users to ensure them.

In recent years, power supply equipment has become increasingly complex, and the variety and quantity of components have increased rapidly; the use environment has also become harsh and diverse; and the electronic systems it serves have become more and more important and expensive. Taking AC parameter regulated power supply as an example, it has been widely used in vehicle-mounted, ship-mounted, ground-based military equipment, aerospace departments, railway and transportation signal and communication systems, etc. The power supply needs to run continuously day and night, and must withstand high and low temperatures, high humidity, impact and other tests. During operation, maintenance is often not allowed, or only simple maintenance can be carried out. All this makes the reliability research of power supply equipment urgent and very important. In fact, as early as the 1970s, a paper published by the British Institution of Electrical Engineers pointed out: In the design and development of the British Skynet system that provides military communications, the central topic is reliability!

Internationally, the general definition of reliability is: the ability to complete the specified function under specified environmental conditions and within a specified time. This definition applies to a system, as well as a device or a unit. Due to the random nature of failures, reliability is described mathematically, often expressed as "probability".

Thus, the definition of reliability [R(t)] is derived: the probability that the system can complete the specified function under specified environmental conditions and within a specified time.

For example, N products are tested and checked once every Δt time interval. The number of failed products each time is ni. The reliability R(t) within the time T is: R(t) = [(N-)/N], which can be approximated as: R(t) = (N-)/N

The value range of R(t) is: 0≤R(t)≤1. The closer the value of R(t) is to 1, the higher the reliability. If the system consists of N units (in series), and the R(t) of each unit is R1(t), R2(t)...RN(t), then the RΣ(t) of the entire system = R1(t)·R2(t)...RN(t). It can be seen that the more complex the system, the worse the reliability.

1 Factors affecting system reliability

There are many factors involved in system reliability. At present, the main misunderstanding of people is to attribute reliability completely (or basically) to the reliability of components and the manufacturing and assembly process; ignoring the decisive role of system design in reliability. According to statistics from the US Navy Electronics Laboratory, the causes of machine failure and their respective percentages are listed in Table 1:

lim

Δt→0

N→∞

Table 1 Statistics of machine failure causes

Cause	% of total failures
Reasons in design Reasons in component quality Reasons in operation and maintenance Reasons in manufacturing	40302010

In the field of civilian electronic products, Japanese statistics show that 80% of reliability problems are caused by design. (Japan also attributes the selection of components, the determination of quality levels, and the load rate of components to design reasons.) In short, for system designers, it is necessary to clearly establish the important concept of "reliability", take system reliability as an important technical indicator, pay serious attention to reliability design work, and take sufficient measures to improve reliability, so that the system and products can achieve the goal of stability and quality.

2 Indicators for measuring system reliability and their mathematical relationships

2?1 failure rateλ

λ is defined as: the number of failures of this type of product per unit time. That is:

λ=dn/dt

Relative to the failure rate of each sample that is still working properly,

λ=(1/NS)·dn/dt

Where: NS is the total number of test samples N, which is the number of samples that still work normally after Δt time.

In engineering, an approximate formula is used. If within a certain time interval (t1-t2), the number of normal working samples at the beginning of the test is ns, and the number of faulty samples after (t1-t2) is n, then the failure rate λ for each normal sample in this batch of samples is:

λ=n/[ns(t1－t2)]

The smaller the value of the failure rate λ, the higher the reliability. λ can be used as a reliability characteristic quantity of electronic systems and complete machines, and is more often used as a reliability characteristic quantity of components and contacts. Its dimension is [1/h]. Internationally, [1/109h] is commonly used, called [fit], as the dimension of λ.

For example, the service life of the metallized film capacitors of the 97F8000 series used for AC power supply of GE Company of the United States is: after 60,000 hours of operation, 95 capacitors are normal and 5 capacitors may fail during this period. Then:

λ=n/〔ns(t1－t2)〕

Substituting ns=100, n=5, (t1-t2)=60000h, we have:

λ=0.83·10－6/h=830[fit].

The basic failure rate of components under standard working conditions promulgated by the United States in 1974 is listed in Table 2 (for reference).

2?2Mean time between failures MTBF

MTBF is defined as the average failure-free working time of an electronic system.

For a batch (N) of electronic systems: MTBF = ti/N[h]

Where: ti—failure-free working time of the i-th electronic system [h];

N—Number of electronic systems.

In engineering, for example, when a complete machine is tested, the total test time is T, and n faults occur. The faults are repaired and then tested again (the repair time is not included in the total test time T). Then:

MTBF=T/n[h]

The larger the MTBF value, the higher the reliability of the electronic system. The reference data of MTBF is listed in Table 3:

Table 3 MTBF reference data

Electronic system name	MTBF/(h)
1978 Integrated color television receiver (international level)	≥2000
Apollo spacecraft computer	(2～2.5)×104
UK Skynet satellite system	1000
American "Taikang" long-range navigation equipment (1980s)	150
Simods Digital Frequency Synthesizer	10×104

Let's take the above-mentioned GE 97F8000 series metallized film capacitors for AC power supply as an example:

T=60000h, 5 out of 100 tested capacitors have faults, then for each capacitor:

MTBF=100T/n=120×104h.

Here, it must be made clear that both the failure rate λ and the mean time between failures MTBF are "probabilistic" indicators for measuring the reliability of equipment or components. Do not misunderstand that each of the above capacitors can work for 1.2 million hours before failure. For a specific capacitor, it may fail after one use, and it is more likely that it will still be normal after working for 60,000 hours.

2.3 Mean time to repair (MTTR)

MTTR is defined as the average time of each repair during system maintenance. That is:

Table 2 Component failure rate under standard working conditions promulgated by the United States in 1974

Component Type		λ(fit)
Resistors	Fixed film	4
	Synthesized potentiometer	138
	Wirewound Potentiometer	167
Capacitors	Paper	70
	Aluminum Dielectric	117
	Variable ceramics	393
Relay		6
Semiconductor diode	silicon	20
	Zener	18
Semiconductor transistor	Germanium PNP	56
	Germanium NPN	140
	Silicon PNP	63
	Silicon NPN	33

Table 4 Reference data on R(t) of the international communication satellite system

Electronic system name		R(t)/(%)
Intelsat III	Ground Station	99.7
	antenna	93.5
	power supply	94.2
Intelsat IV transponder electronics	When working for 2 consecutive months	99.9
	After 7 years of continuous work	79.0
	Power supply system at international level	99.95

MTTR=ti/M[h]

Where: Δti—the i-th repair time [h].

M—Number of repairs.

No matter how reliable any equipment is, there will always be maintenance issues. So the smaller the MTTR, the better. Therefore, it is of great value to achieve convenient and fast maintenance or maintenance without downtime.

2?4 Effectiveness (Availability) A

A is defined as the ratio (usually expressed as a percentage) of the time that an electronic system can be used normally during use (especially under uninterrupted continuous use) to the total time. That is:

A=MTBF/(MTBF+MTTR)

The closer the A value is to 100%, the more effectively the electronic system works.

In fact, the MTBF of equipment is limited by many factors such as system complexity and cost, and it is not easy to reach a very high value. Trying to shorten MTTR can also achieve the purpose of increasing A. For units with high failure rates, a redundant design that quickly replaces failed units with backup units can make MTTR close to 0 when MTBF is not very high, so that A can also be close to 100%.

2?5 Reliability R (t)

Reliability R(t) is the most basic indicator for measuring the reliability of electronic systems. The failure probability F(t) can be derived from the definition of reliability R(t). That is:

F(t)=1-R(t), or R(t)=1-F(t).

It can be seen that for R(t) and F(t), their values ​​are functions of the time t. In extreme cases, when t=0, R(t)=1 for any system, [F(t)=0]. When t=∞, R(t)=0 for any system, [F(t)=1]. R(t) and F(t) have specific meaning only within the specified time range. In actual use, annual reliability P is often used to represent it.

The definition of annual reliability P is: the probability that an electronic system can complete the specified function within 1 year under the specified environmental conditions. For example, if P=0.9, it means that the system has a 90% chance of not failing within a year. (That is, there is a 10% chance of failing). If there are 10 similar devices at a location, on average one device may need repair every year.

Reference data on the reliability R(t) of the international communications satellite system are listed in Table 4.

2.6 Mathematical relationship between failure rate λ, mean time between failures MTBF and reliability R(t), failure probability F(t)

Based on the definitions and basic mathematical expressions of λ, MTBF, R(t), and F(t), after mathematical operations, the following mutual mathematical relationships can be obtained (the calculation process is omitted).

(1) MTBF=1/λ or λ=1/MTBF,

That is, λ and MTBF are inversely proportional to each other.

(2) R(t)=e-λt or R(t)=e-t/MTBF=1/et/MTBF,

That is, there is an exponential relationship between R(t) and λ.

(3) F(t)=1-R(t) or R(t)=1-F(t),

In this way, the three indicators of λ, MTBF, and R(t) can be converted from one quantity to the corresponding values ​​of the other two quantities. In different occasions, the above three indicators may be used interchangeably to measure the reliability of electronic systems.

3 Ways to Improve System Reliability

3?1 Seriously engage in system reliability design

The reliability model of electronic systems generally has the following three forms:

(1) Reliability model of series system

The series system model is shown in Figure 1. A series system means that each of its components is necessary and indispensable for the normal operation of the system; the failure of any component will cause the system to malfunction. This is a relatively common and simple system.

If the system has N components, and the failure rate of each component is λi (i=1～N), then the total failure rate of the series system is:

λ?=n1λ1+n2λ2+…nNλN

Total trouble-free working time:

MTBF?=1/λ?=1/[n1λ1＋n2λ2＋……nNλN]

Annual reliability: P = 1/e8760·λ? = 1/e8760/MTBFN. (Because there are 8760 hours per year).

Example (1): The MTBF0 of a high-quality AC parameter stabilized power supply unit is 200,000 hours. If each railway signal panel uses 10 power supply units, then the MTBF of the AC power supply part of each panel is MTBF0/10=20,000 hours. This is equivalent to an annual reliability of P=0.645=64.5%. That is, the annual failure probability F=1-P=35.5%. In other words, there is a 35.5% probability that each power panel needs maintenance every year. If a station has 10 signal panels, it is normal for 3 to 4 AC parameter stabilized power supply units to fail every year. This is also similar to the failure probability of a department with 100 power supply units, most of which are working continuously.

Figure 1 Series system model

It can be seen that although the MTBF0 of each unit AC parameter stabilized power supply is 200,000 hours, which is many times higher than other types of AC power supplies (the MTBF of other types of power supplies is often only a few thousand hours), the reliability of the signal screen of the series system model under continuous working conditions is not very satisfactory.

(2) Reliability model of parallel system

The parallel system model is shown in Figure 2. In the figure: U1 and U2 can realize the functions of the system independently, and if any unit of U1 or U2 fails, it will be automatically (or manually) disconnected from the input and output terminals and connected to another unit that serves as a backup for each other.

Obviously, the failure of any unit in the parallel system will not affect the function of the system. Only when both units fail will the system fail to work properly. Similarly, N units can also be connected in parallel to form a system.

The mathematical relationship is:

Failure probability: F(t) = F1(t)·F2(t)…FN(t)

If F1(t)=F2(t)…=FN(t), then reliability:

R(t)=1-F(t)=1-[F1(t)]n

Example (2): The MTBF0 of a high-quality AC parameter regulated power supply unit is 200,000 h. Each railway signal panel uses 10 power supply units. If each power supply unit has two power supplies that back up each other to form a parallel system, then the annual reliability of each power supply is:

P1=1/e8760/MTBF, P1=0.957

Annual failure probability F1=1－P1=0.043

Therefore, the annual failure rate of each power supply unit (composed of two power supplies backing up each other) is:

F11=[F1]2=1.85×·10－3

Annual reliability of each power supply unit:

P11=1－F11=1－[1－P1]2

=1-1.85×10-3=0.998=99.8%

Each railway signal panel has 10 power supply units, so the annual reliability of each signal panel is:

P = (P11) 10

= (0.998) 10 = 0.98 = 98%,

That is, the annual failure probability F=1-P, which is 2%.

If a station has 10 signal screens, there is only a 2% chance that they will be repaired once a year. Compared with the series system in Example (1), the failure probability is reduced by nearly 18 times.

The conclusion is clear: when the reliability of each unit cannot be too high due to various restrictions, and the system is required to have a very high reliability, the fundamental way to improve the reliability of the electronic system is to use a parallel system instead of a series system. The generator of the Boeing 707 aircraft in the United States uses a 4-unit parallel system (1 in use and 3 in reserve), and the DC power supply of the nuclear power plant uses a 3-unit parallel system (1 in use and 2 in reserve), which are good examples.

The cost of a parallel system will be higher than that of a series system, but in order to ensure the necessary reliability, the cost is necessary and worthwhile.

(3) Hybrid system reliability model

In actual projects, in order to strike a balance between cost and reliability, a series and parallel hybrid system is often used. That is, a parallel system is used for units with lower reliability, while a series system is maintained for units with higher reliability. The model is shown in Figure 3.

Reliability of hybrid system:

R(t)=R1(t)·R2(t)·R3-2(t)·R4(t)

If R1=R2=R4=0.99, R3=0.9

Then R3-2=1-[1-R3]2, R3-2=0.99

R=R1·R2·R3－2·R4

=0.96=96%. (F=4%).

If U3 does not use a parallel system, then R = 0.87 = 87%, (F = 13%). It can be seen that the difference in reliability between the two is still very obvious, and the failure rate is reduced by more than 3 times. The hybrid system is more reliable than the series system and simpler than the parallel system.

3.2 Improve the use environment of electronic systems and reduce the ambient temperature of components

The reliability of electronic systems is closely related to the use environment. The failure rate of components in different use environments is very different from their basic failure rate, and should usually be corrected by environmental factors. The United States announced the environmental factor values ​​of different components in the 1970s. There were originally 9 environmental conditions, and now only the more commonly used and representative 4 are listed as follows:

Figure 2 Parallel system model

Figure 3 Hybrid system model

——GB: Good ground environment. Environmental gravity is close to "0", and engineering operation and maintenance are good.

GF: Ground-fixed use environment. Installed in a permanent rack with adequate ventilation and cooling. Maintained by military personnel, usually installed in a non-heated building.

——NS: Ship cabin environment. Surface ship conditions, similar to GF. But subject to occasional severe shock and vibration.

——GM: Ground mobile and portable environment. Inferior to ground fixed conditions, mainly impact and vibration. Ventilation and cooling may be limited, and only simple maintenance can be performed.

The environmental coefficient πE under the above environmental conditions is listed in Table 5:

Table 5 Environmental coefficient πE

Component Type		GB	GF	NS	GM
integrated circuit		0.2	1.0	4.0	4.0	Note: λp=λb·πE Where: λp in actual use Failure rate λb basic Failure rate πE environmental coefficient
Potentiometer		1.0	2.0	5.0	7.0
Power Film Resistors		1.0	5.0	7.5	12.0
Capacitors	Paper and plastic film	1.0	2.0	4.0	4.0
	ceramics	1.0	2.0	4.0	4.0
	Aluminum Dielectric	1.0	2.0	12.0	12.0
transformer		1.0	2.0	5.0	3.0
Relay	military	1.0	2.0	9	10
	Poor quality	2.0	4.0	twenty four	30
switch		0.3	1.0	1.2	5.0
Connectors	military	1.0	4.0	4.0	8.0
	Poor quality	10	16	12	16

From Table 5, we can see that the use environment has a great impact on the failure rate of components. The failure rate of GM is 4 to 10 times higher than that of GB. The improvement of environmental conditions is often limited by the use occasion. What is relatively easy to do in design and production is to pay attention to and strengthen ventilation and cooling as much as possible.

Excessively high ambient temperature is very harmful to the reliability of components:

(1) Semiconductor devices (including various integrated circuits, diodes, and transistors)

For example, if a silicon transistor is designed with PD/PR=0.5 (PD: operating power, PR: rated power), the impact of ambient temperature on reliability is shown in Table 6.

Table 6 Effect of ambient temperature on the reliability of semiconductor devices

Ambient temperature Ta[℃]	20	50	80
Failure rate λ[1/109h]	500	2500	15000

(2) Capacitors (taking solid tantalum capacitors as an example)

Designed with UD/UR=0.6 (UD: operating voltage, UR: rated voltage), the influence of ambient temperature on reliability is shown in Table 7.

Table 7 Effect of ambient temperature on capacitor reliability

Ambient temperature Ta[℃]	20	50	80
Failure rate λ[1/109h]	5	25	70

(3) Carbon film resistor

Designed with PD/PR=0.5, the impact of ambient temperature on reliability is shown in Table 8.

Table 8 Effect of ambient temperature on the reliability of carbon film resistors

Ambient temperature Ta[℃]	20	50	80
Failure rate λ[1/109h]	1	2	4

A German research report pointed out that the SK-2 color TV set, through the reasonable design of ventilation and cooling conditions, reduced the internal temperature of the machine by about 10°C, resulting in a 2-fold increase in the mean time between failures (MTBF), which significantly improved reliability. The electronic system of the US "Minuteman" intercontinental missile strictly limits the ambient temperature to ≤40°C, achieving the goal of significantly reducing the failure rate.

It can be seen that strengthening ventilation and cooling is very beneficial to the reliability of electronic systems. Some domestic departments (such as railways) require the system to have high reliability, and explicitly prohibit the use of fans for forced ventilation and cooling. As a result, not only the cost of equipment is increased, but also the reliability is difficult to truly guarantee, which artificially causes many problems. In fact, now high-quality fans can guarantee a service life of 50,000 to 60,000 hours (equivalent to more than 6 years of continuous operation). Replacing fans is also much more labor-saving and time-saving than repairing other parts. As long as the system design conditions stipulate that even if the fan does not work, the equipment can still operate normally for a long time. Then, strengthening ventilation and cooling is definitely beneficial to reliability, so why not do it!

3?3 Reducing the load rate of components is a shortcut to improving failure rate

There is a direct relationship between the load rate and failure rate of components in actual operation. Therefore, after the type and value of components are determined, the rated values ​​that the components must meet should be selected from the perspective of reliability. Such as the rated power, rated voltage, rated current of semiconductor devices, the rated voltage of capacitors, the rated power of resistors, etc.

(1) Silicon semiconductor devices

When the ambient temperature is Ta=50℃, the influence of PD/PR on the frequency is shown in Table 9.

Table 9 Effect of PD/PR on failure rate of silicon semiconductor devices

PD/PR	0	0.2	0.3	0.4	0.5	0.6	0.7	0.8
λ[1/109h]	30	50	150	700	2500	7000	20000	70000

It can be seen from Table 9 that when PD/PR=0.8, the failure rate increases by more than 1000 times compared with 0.2.

(2) Capacitor

The UK once published data that the failure rate λ of capacitors is proportional to the fifth power of the operating voltage, which is called the "quintic law", that is, λ∝U5.

When U=UR/2,

λ=λR/25=λR/32 (λR is the rated failure rate)

When U=0.8UR=UR/1.25,

λ = λR/（1.25）5 = λR/3.05

When the capacitor operating voltage is reduced to 50% of the rated value, the failure rate can be reduced by as much as 32 times.

(3) Carbon film resistor

The ambient temperature Ta=50℃, the military product data actually used by the United States in the 1970s are listed in Table 10.

Table 10 Effect of PD/PR on failure rate of carbon film resistors

PD/PR	0	0.2	0.4	0.6	0.8	1.0
λ[1/109h]	0.25	0.5	1.2	2.5	4.0	7.0

It can be seen from Table 10 that when PD/PR=0.8, the failure rate increases by 8 times compared to 0.2.

The above data show that in order to ensure reliability, the load rate of components must be reduced. For example, the electronic system of the US "Minuteman" intercontinental missile stipulates that the load rate of components is 0.2.

The empirical data in actual use are:

——The load factor of semiconductor components should be around 0.3;

——The capacitor load rate (ratio of operating voltage to rated voltage) should be around 0.5, and generally should not exceed 0.8;

——Resistors, potentiometers, load factor ≤ 0.5.

In short, the load factor of various components should be kept at ≤0.3 whenever possible. If necessary, it should usually be ≤0.5.

3?4 Simplifying the circuit, reducing the number of components, integrating as much as possible, and carefully selecting high-reliability components are the most basic ideas for improving reliability.

Electronic system reliability

R=R1·R2·R3...RN (0≤R≤1).

Failure rate of electronic systems

λ=n1·λ1+n2·λ2+n3·λ3……nN·λN. (λ≥0)

Obviously, the more components there are, the less reliable it is.

If each component has Ri=0.999 and there are 5000 components in total, then R=0.9995000=0.01, which is obviously extremely unreliable.

If the number of components is reduced to 1800, then R = 0.9991800 = 0.19, which means that if the number of components can be reduced by 64%, the reliability will increase by 19 times.

Therefore, integrated devices should be used as much as possible. For example, an integrated circuit can replace thousands of semiconductor transistors and diodes, thereby greatly improving reliability.

It is also important to note the importance of selecting high-reliability component types and quality grades. For example, for capacitors with similar functions, the failure rate of mica dielectrics is about 30 times lower than that of glass or ceramic dielectrics. The same type of components, but different quality grades, such as military products and civilian products, high-quality and low-quality, will have a failure rate difference of 3 to 10 times under the same functions and conditions, so you should be very careful when selecting.

It can be said that under the conditions of ensuring the same functions and usage environment, the simpler the circuit and the fewer components, the more reliable the system will be.

For example: A company's 1000VA high-quality AC parameter regulated power supply is used in GM environmental conditions (mobile, vehicle-mounted, poor ventilation, and inconvenient maintenance). It can also guarantee MTBF ≥ 200,000 hours. The main reason is that the circuit is simple and the number of components is small. The entire power supply only includes:

——1 special transformer

The basic failure rate is λ1=300×10－9/h.

——2 metallized film capacitors

The basic failure rate is λ0=830×10－9/h.

The capacitor load factor is 0.8. Therefore,

λ2=(830/3.05)×10-9/h.

——20 welding points

The basic failure rate is λ3=5.7×10－9/h.

Therefore: λΣ=λ1+2λ2+20λ3

=[300＋544＋114]×10－9/h

=958×10－9/h.

Used in GM environment conditions, average πE=4,

λΣP=λΣ·πE=3832×10-9/h.

Mean time between failures

MTBF=1/λΣP=（1/3832）×109/h

=26×104h=260,000h

≥200,000 hours.

Annual reliability: P = 1/eλΣP·8760 = 0.967 = 96.7%

Failure rate: F=1－P=3.3%

Statistics from the company's long-term production practice also prove that the MTBF of this type of power supply is ≥ 200,000 hours.

Of course, the reliability will be better when used in other environmental conditions.

3.5 Pay attention to the aging of components and reduce the early failure rate of the system

The failure rate of components, equipment, and systems is not a constant throughout their service life, and usually there is a "bathtub curve" as shown in Figure 4.

(1) Early failure rate is usually much higher than the failure rate in the stable period. The cause of failure is defects in the manufacturing process of components and errors in installation or imperfect connection points or the mixing of unqualified products that were missed when the components left the factory. Therefore, the equipment must be operated for a period of time to age, so that the early failure problem is exposed during the aging period of the factory. What is provided to users is a reliable product that has entered the stable period.

Figure 4 Failure rate vs. time curve

The aging time of Japanese civilian products (such as televisions) is generally not less than 8 hours. However, the US spacecraft stipulates that each component must be aged for 50 hours before being installed on the spacecraft, and then aged for another 250 hours after being installed on the spacecraft, for a total of 300 hours. This is to eliminate components with hidden dangers and ensure working reliability. In actual work, it is more appropriate to determine the aging time of equipment with higher reliability requirements to be 20 to 50 hours.

(2) Stable period: At this time, the failure rate λ is close to a constant and is used as the normal use period. Other reliability indicators of the equipment can also be estimated based on the failure rate λ. Generally, in a good use environment, if a failure can be repaired promptly and correctly, the stable period of the electronic system should not be shorter than 6 to 8 years.

(3) Wear-out period: At the end of the equipment's service life, the failure rate λ will gradually increase due to the aging and deterioration of the components, or the oxidation corrosion, mechanical wear, fatigue, etc. of the equipment, and the equipment will enter an unreliable service period. The specific time when the wear-out period occurs is affected by various factors and is very inconsistent. The wear-out period of equipment with reasonable design, strict component quality selection, and not too harsh environmental conditions will occur much later.

4 Conclusion

Ensuring the reliability of equipment is a complex system engineering involving a wide range of knowledge. Only by paying full attention and taking various technical measures can we achieve satisfactory results. The basic points are:

(1) For complex systems with high reliability, parallel systems must be used

The system has backup units with sufficient redundancy, which can be switched automatically or manually. If the function allows, switching of cold backup units can better ensure the reliability of long-term work than switching of hot backup units.

(2) No electronic system can be 100% reliable. Design

The modular structure that is easy to maintain off-machine should be used as much as possible, and the necessary number of spare parts (usually 5%) should be reserved in advance to shorten the mean maintenance time MTTR as much as possible and make the effectiveness A close to 100%.

(3) Strengthening ventilation and cooling and improving the use environment can double the efficiency.

The easiest and most economical way to ensure reliability.

(4) Simplify the circuit, reduce the number of components, and reduce the

Load rate and selection of highly reliable components are the basis for ensuring high reliability of the system.

(5) Pay attention to equipment aging and reduce the early failure rate of the system.

We believe that through careful design, conscientious production, strict quality inspection and timely maintenance, we can make the electronic system (including power supply equipment) reach a reliability close to 100%, meet the needs of national defense, scientific research, industry and other aspects, and then go global.