Secrets of testing Ethernet channels. Switches

Facebook

The ComputerPress testing laboratory tested 10 unmanaged gigabit switches: Allied Telesyn AT-GS908GB, ASUS GigaX 1108, D-Link DGS-1008D, D-Link DGS-1008TL, Gigabyte GN-GT8S, HP ProCurve 2708, Surecom EP-808DG -S, TRENDnet TEG-S80TXD, TRENDnet TEG-S80TXE and MultiCo 5-Port Gigabit Switch.

The appearance on the market of a large number of highly integrated chips, which are ready-made 4-8-port gigabit switches, sometimes even with an integrated physical layer, has significantly simplified the development and manufacture of switches, which has led to a rapid reduction in cost and the emergence of gigabit switches in the network infrastructure of even small companies .

Let us remember that technology Gigabit Ethernet This is an extension of the IEEE 802.3 Ethernet standard, using the same packet structure, transmission protocols, flow control, etc., but at the same time theoretically providing a tenfold increase in performance. Compatibility with 10Mbps and 100Mbps Ethernet makes it easy to migrate to this technology without investing heavily in software or staff training, and you don’t even need to change the cabling structure if you have installed twisted pair category 5.

In this testing, we decided to test the capabilities of eight-port unmanaged gigabit switches. The technical characteristics of the test participants are given in table. 1.

The term "unmanaged" means that the switches do not support management by network management protocols such as SNMP. At the same time, unmanaged switches can be customized - they are called Smart switches. Such switches offer the possibility of some settings, for example, VLAN configuration, port aggregation, etc., carried out through the management port available on the switch.

Testing methodology

During testing, we checked the performance of the switches in six modes, which, in our opinion, correspond to the main operating modes of the switch in a real network:

Communication between two workstations duplex mode.
Unidirectional transmission from the server to all workstations.
Unidirectional transmission from all workstations to the server.
Simultaneous operation of all workstations with the server in duplex mode.
Operation of the server and all workstations in the “everyone with everyone” mode.
Server operation with five workgroups.

In the fifth and sixth modes, the load on the switch gradually increased due to the daisy chain connection of workstations or workgroups.

NetIQ Chariot 5.0 software package was used to generate network traffic and measure performance. The transmission was carried out using the TCP protocol, the most common in small networks. Each test was performed three times and the average performance was calculated.

For testing, a local network was assembled, the clients of which were workstations based on an Intel Pentium 4 3.0 GHz processor, equipped with an integrated motherboard Marvel Yukon Gigabit Ethernet 10Base-T/100Base-TX/1000Base-T Adapter, as well as a dual-processor server with two processors Intel Xeon 2.66 GHz with integrated Intel 82546EB Gigabit Ethernet network adapter.

When performing tests in the first five modes, the server and workstations were connected directly to the switch under test, which acted as a high-performance switch working group.

When testing in the sixth mode, a local network was used, the configuration of which is shown in Fig. 1. The switch under test in this case worked as the central switch of a small enterprise. The test results are shown in table. 2, and it shows the average performance during the entire test.

Table 2. Results of measuring the performance of switches, Mbit/s

Test results

Testing shows that using gigabit switches is a good way to improve the performance of small and medium-sized enterprise networks. At the same time, all the presented switches have some basic set of functions, greatly facilitating their operation.

Thus, all ports support automatic MDI/MDIX polarity detection, that is, any port can be connected to a server, router or switch using any Category 5 twisted pair cable. This eliminates the need to select cables (straight or crossover) and use uplink ports.

Auto-detection of 10/100/1000 Mbit/s speeds and duplex/half-duplex connection mode makes it very easy to install the switch on the network; you just need to connect network cables to the switch so that the network is up and running in minutes.

Switching is carried out using Store and forward technology, which allows you to filter packets and remove damaged ones. Flow control is performed using IEEE 802.3x based on frames in full-duplex mode and backpressure in half-duplex mode.

Despite the wide variety of elements used in the design of switches (base chips manufactured by ASIX, Broadcom, IC Plus and Marvell), in most tests the spread of results was insignificant, and the overall network performance was determined by the performance of the network adapters of the computers and server used.

In the first test, the switch was lightly loaded, so the result obtained was almost equal to the maximum performance value of the network adapters installed in the computers. The data obtained turned out to be quite good for a gigabit adapter operating on a 32-bit PCI bus.

The results of the next three tests were determined by the performance of the server adapter in three operating modes. The slight increase in overall performance with the 5-port switch in tests using transfer from the server was due to the reduced load on the server adapter (it worked with four computers instead of seven). The relatively poor performance of the Allied Telesyn AT-GS908GB switch in this test was likely a consequence of poor performance of the switch with the Intel 82546EB adapter.

In tests with a large number of concurrent connections, differences in the performance of the switches were noticeable. The HP ProCurve 2708 switch experienced significant lag, especially in the workgroup test. TRENDnet TEG-S80TXE also lagged behind.

It is worth noting the behavior of the Allied Telesyn AT-GS908GB switch in the test with daisy-chain workgroup connections, which showed an increase in overall network performance with the connection of each group. For the remaining switches, at a certain point when a new group was connected, the total network performance did not change or even dropped slightly. When the group connection sequence was changed, the behavior of the switches did not change.

It should be said that during testing there were no failures in the operation of the switches or errors during packet transmission.

We present the results of the MultiCo 5-Port Gigabit Switch for comparison. It should be taken into account that when testing it, a server and four computers (instead of seven) or a server and four workgroups (instead of five) were connected to the switch.

Based on the results of the tests performed, we can conclude that even under extreme network load conditions, consisting of computers equipped with gigabit adapters and a dual-processor server, the gigabit switch will not become a bottleneck limiting its performance.

Editor's Choice

a significant difference in the performance of switches even under extreme loads, which becomes even smaller under normal loads, allows you to pay more attention when choosing equipment to such parameters as price, customization, ease of placement, security thermal regime during long-term work, etc.

We decided to award the “Editor's Choice” sign to switches ASUS GigaX 1108 And Gigabyte GN-GT8S. The first of them, at a relatively low price, showed high performance in all modes, and in addition, its design used an original solution that facilitates placement; The second switch, with high performance, had good configuration capabilities.

Test participants

The AT-GS908GB switch, made in a metal case, can be installed in a standard 19-inch rack using the included brackets.

Main feature this switch What sets it apart from other test participants is the presence of two slots for installing GBIC modules, which are connected instead of the seventh and eighth ports.

The switch is based on chips manufactured by Broadcom: single-port transceivers BCM5421, 4-port transceiver BCM5404 and 8-port switch BCM5680. The configuration capabilities available with the BCM5680 are not used.

The switch showed very good results when working under high load. Thus, in the test with the connection of workgroups, only this switch ensured an increase in the total network performance with the connection of each group (Fig. 2).

Rice. 2. Test results of the Allied Telesyn AT-GS908GB switch with a gradual increase in the number of active workgroups

A slight lag from other switches was noted only in the duplex operation mode of all computers with the server, which is most likely due to poor performance of the switch with the Intel 82546EB adapter. The disadvantage of this switch is its rather high price.

ASUS GigaX 1108

The GigaX 1108 switch is made in a miniature dark blue plastic case, allowing both desktop placement and wall mounting. In addition, two pieces of magnetic rubber are built into the base of the switch, which make it easy to mount the switch on any steel surface, such as the side wall of a server case or PC, without scratching the surface.

On the front panel, smoothly turning into top cover, there is a power connection indicator and port indicators (three for each port), allowing you to determine network activity, the set connection speed and transmission mode (duplex or half-duplex).

On the rear panel there are eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports) and a socket for connecting an external power supply.

The switch has passive cooling and therefore works silently. A large number of holes promotes air circulation inside the device and ensures good cooling. Normal thermal conditions are also maintained thanks to the presence of an intelligent energy management circuit.

The switch is based on chips manufactured by Marvell: 4-port transceivers 88E1145 and 8-port switch 88E6181. The 88E6181's existing configuration options are not used.

In all test modes The switch showed high performance and was awarded the "Editor's Choice" sign. Like most switches tested, the daisy-chain workgroup test saw a drop in performance when connecting a fourth group and a sharp increase in performance when connecting a fifth group (Figure 3).

Rice. 3. Results of testing the ASUS GigaX 1108 switch with a gradual increase in the number of active workgroups

Its high performance, low price, and easy placement make this switch a good choice for small workgroups.

The DGS-1008D switch is made in a plastic case of traditional D-Link shapes and colors. The design of the case allows both desktop placement of the switch and its mounting to a wall.

On the front panel, behind a translucent insert, there is a power connection indicator and port indicators (two for each port), allowing you to determine network activity and the set connection speed.

On the rear panel there are eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports) and a socket for connecting an external power supply.

The switch is passively cooled and therefore operates silently, and the large number of holes promotes air circulation inside the device and provides good cooling.

The switch is based on chips manufactured by Marvell: 4-port transceivers 88E1145 and 8-port switch 88E6181. The customization options available with the 88E6181 are not used.

When the switch is turned on, the Cable Diagnostic function is activated, which detects three main problems in the cable connection: an open circuit on any of the contacts, a short circuit between the conductors and an increase in terminator resistance (more than 100 Ohms). The function only works if a gigabit device is connected to the switch.

The switch showed good performance in all tests. In the workgroup test, like most switches tested, there was a drop in performance when a fourth group was connected (Figure 4).

Rice. 4. Test results D-Link switch DGS-1008D with a gradual increase in the number of active working groups

The DGS-1008TL switch is made in a metal case. The included brackets allow you to mount it in a standard 19-inch rack.

The switch has a built-in power supply without a switch and turns on immediately after connecting to the power supply. A fan installed in the case is used for cooling.

The switch showed high performance in all tests. In the test with working groups, there was also a drop in performance when the fourth group was connected (Fig. 5).

Gigabyte GN-GT8S

The GN-GT8S switch belongs to the category of Smart switches that allow local configuration. The model is made in a miniature stylish silver plastic case. Behind the translucent front panel are the power indicator and port indicators (three for each port), allowing you to determine network activity, the set connection speed and transmission mode (duplex or half-duplex).

On the rear panel there are eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports), a socket for connecting an external power supply and a D-SUB DB-9F control connector.

To configure the switch, you can use any terminal program (for example, Hyper Terminal), and the switch is connected to the PC using a standard RS232 serial interface cable (not included in the package).

The switch's configuration options are not as extensive as those of managed switches, but they fully satisfy the needs of a wide range of users.

For increase bandwidth communication channel, you can create a trunk group based on combining the first and second ports, and you can choose one of three load balancing methods.

The basic element of the design is the 8-port Tamarack TC9208M switch manufactured by IC Plus. Physical layer chips are covered with radiators.

The switch has passive cooling and therefore operates silently, however, the number of holes for air circulation, in our opinion, is clearly not enough; under high load, the switch noticeably heats up. The temperature decreases if the switch is positioned vertically (up one of the side faces), but there are no means for mounting in this position.

The switch performed well in all tests and earned an Editors' Choice designation. Maximum performance the switch was observed when working with workgroups (Fig. 6), however, like most other test participants, when connecting the fourth workgroup there was a drop in performance.

Rice. 6. Test results of the Gigabyte GN-GT8S switch with a gradual increase in the number of active workgroups

Featuring a metal casing, the ProCurve 2708 switch can be mounted in a standard 19-inch rack or wall-mounted using the included brackets.

In terms of the composition of the indication and control elements available on the front panel, the switch differs significantly from the other test participants. Large power and failure indicators are installed at the left edge of the panel. Nearby is the Reset button, which starts the internal test of the switch. Eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports) are well spaced for easy cable management. The status of each port is indicated by two LEDs located next to the ports. One LED indicating the presence of a connection is made in the form of a number corresponding to the port number, which is convenient for maintenance. Another LED displays network activity, the set connection speed, or the transmission mode (duplex or half-duplex). The indication mode of this LED is switched by a special button for all ports at once.

There is a connection socket on the rear panel standard cable power supply networks.

The switch has a built-in power supply without a switch and turns on immediately after connecting to the power supply. Two fans are used for cooling.

The switch is based on chips manufactured by Broadcom: 4-port BCM5404 transceivers and 8-port BCM5682 switch. The Smart functions available with the BCM5682 are not used.

In tests with a small number of simultaneously working pairs, the switch showed good results, but as the load grew, it began to lag behind its competitors. The lag was especially noticeable in the test with working groups (Fig. 7), starting with the connection of the second working group.

Rice. 7. Results of testing the HP ProCurve 2708 switch with a gradual increase in the number of active workgroups

The EP-808DG-S switch belongs to the category of Smart switches, which provide the possibility of local configuration. The model is made in a metal case and has a built-in power supply. The included brackets allow you to mount the switch in a standard 19-inch rack.

The front panel contains eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports), a power indicator, and port indicators (four per port) that allow you to determine network activity, set connection speed, and transmission mode ( duplex or half duplex).

On the rear panel there is a D-SUB DB-9F control connector, a socket for connecting a standard power cable and a power switch.

To provide the necessary cooling, two fans are installed in the case.

To configure the switch, you can use any terminal program (for example, Hyper Terminal), and the switch is connected to the PC using a standard RS232 serial interface cable (included in the delivery package).

The options for configuring the switch are not as extensive as in the case of managed switches, but they fully meet the needs of most users.

To increase the throughput of the communication channel, you can create a trunk group based on combining the first and second ports, and you can choose one of three load balancing methods.

If necessary, you can hardcode the speed for each port and connection mode, enable or disable flow control, and select one of four priority levels.

To increase the reliability of critical connections, port mirroring is provided.

It is possible to organize VLAN networks based on port aggregation.

The basic element of the design is the 8-port Tamarack TC9208M switch manufactured by IC Plus. Physical layer chips are covered with radiators. The customization capabilities available with the TC9208M are not used.

The switch showed good results in all tests. In the workgroup test, like most switches tested, there was a drop in performance when a fourth group was connected (Figure 8).

Rice. 8. Results of testing the Surecom EP-808DG-S switch with a gradual increase in the number of active workgroups

The TEG-S80TXD switch is made in a metal case. The included brackets allow you to mount it in a standard 19-inch rack.

This model has a built-in power supply without a switch and turns on immediately after connecting to the power supply. Thanks to passive cooling, the switch operates silently.

The front panel contains eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports), a power indicator and port indicators (three for each port), allowing you to determine network activity and the set connection speed.

On the rear panel there is a socket for connecting a standard power cable.

The basic element of the design is the 8-port Tamarack TC9208M switch manufactured by IC Plus. 4-port B5014A1KRB transceivers are used as the physical layer. The TC9208M's existing configuration options are not used.

The switch showed good performance in all tests. In the test with working groups, there was also a drop in performance when the fourth group was connected (Fig. 9).

Rice. 9. Test results of the TRENDnet TEG-S80TXD switch with a gradual increase in the number of active workgroups

TRENDnet TEG-S80TXE

The TEG-S80TXE switch is made in a miniature metal case for desktop placement (wall mounting is not provided).

On the rear panel there is a socket for connecting an external power supply.

The basic element of the design is the 8-port switch AX88658AB manufactured by ASIX. 4-port B5014RA1KFB transceivers are used as the physical layer. The customization options available with the AX88658AB are not used.

The switch is passively cooled, so it operates silently.

In tests with a small number of simultaneously working pairs, the switch showed good results, but as the load increased, it lost to its competitors. The lag began when connecting the sixth computer in the test with gradual connection of workstations and when connecting the third workgroup (Fig. 10).

Rice. 10. Test results of the TRENDnet TEG-S80TXE switch with a gradual increase in the number of active workgroups

The MultiCo 5-Port Gigabit Switch is made in a metal case designed for desktop or wall mounting. The chassis has holes for mounting brackets (not included) for mounting in a 10- or 19-inch rack.

The switch has a built-in power supply without a switch and turns on immediately after connecting to the power supply. Thanks to passive cooling, the switch operates silently.

The front panel contains five RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports), a power indicator and port indicators (two for each port), allowing you to determine network activity and the set connection speed.

On the rear panel there is a socket for connecting a standard power cable.

The basic element of the design is the 5-port BCM5385 switch manufactured by Broadcom. For four ports, all physical layer elements are built into the BCM5385, and for the fifth, an additional single-channel transceiver B5011A1KQM is used. The configuration options available with the BCM5385 are not used.

We tested the MultiCo 5-Port Gigabit Switch outside the competition and present the results for comparison. Interestingly, during the test with daisy chaining of workgroups under some load, there was also a drop in performance (Figure 11).

Rice. 11. Test results of the MultiCo 5-Port Gigabit Switch with a gradual increase in the number of active workgroups

Instructions for using the cable testing function in QTECH QSW series switches

To use the VCT (Virtual Copper Cable Test) function on the interface command line(CLI) you need to switch to enable mode (when connecting via the Telnet protocol, this mode is already installed) by typing the command of the same name in the CLI.

Once executed, the command returns us data about the cable in the form of a table (see the figure above), which states:

Cable pairs – serial numbers pairs of wires being tested, the pairs are separated by brackets. (2)

Cable Status – shows the status of the pair being tested.

There are statuses:

well – cable in good condition;

open – pair is open;

short – pair is short-circuited;

no cable – the pair is not connected to the port (or the cable length is less than a meter). (3)

Length (meters) – length of the cable pair being tested in meters. It must be distinguished depending on the status of the cable:

Status well – the length of the entire cable pair (often determined incorrectly, as a result of which in most firmware the value N/A is set - meaning it is impossible to determine);

Status open – length of the pair before breaking;

Status short – cable length before shorting;

Status no cable – returns either 0 or N/A (2).

Notes:

(1) - the port numbers used in the CLI of this switch can be clarified by running the show running-config command.

(2) - serial numbers of cables in the connector, starting with white-orange and further according to standard B.

(3) - not implemented in most firmware. Usually, if there is no cable in the port, the port shows a break at a length of 0 or 1 meter.

To use the VCT (virtual cable test) function in the WEB interface, you must select the following items in the WEB interface menu: Port configuration → Port virtual-cable-test config → virtual-cable-test configuration as shown in the figure below.

The output data is presented in the same table as in the CLI (for a description, see the instructions about VCT in the CLI).

If you see the message "% Invalid input detected at "^" marker." see picture below.

This means there is an error in the WEB interface of the switch that does not allow you to test the cable in the WEB interface. This error has been fixed in a newer version of the WEB interface or new firmware switch.

FAQ:

What does a cable test tell you?

This function provides diagnostic information to help determine the subscriber's problem. For example, if VCT shows that the cable is connected, but there is no link, then this indicates that either the port on the switch or the subscriber’s network interface is disabled.

What happened if, when connecting to a subscriber’s network interface, VCT shows that the cable is shorted at this end?

This indicates that the network interface on the subscriber side has failed (burnt out), or that the cable crimping procedure is incorrect.

If one pair of cable shows the open status, and the other the well status, then by disconnecting the cable from the subscriber’s side, you can see the total length of the cable. Then, using a simple formula<общая длина кабеля> - <обрыв/закорочение пары>= distance to break/short from the subscriber connector.

Why does one switch show information about only two pairs, and another about four?

The first switch port supports standards no higher than Fast Ethernet(100Mbit/s), and the second Gigabit Ethernet (1 Gbit/s). Accordingly, different numbers of copper pairs are used to connect the interfaces.

Why is the cable test not supported on some switch ports?

Because the port is optical, and specialized equipment such as an “Optical reflectometer” is used to test fiber-optic communication lines.

11.03.2015

Testing gigabit switches.

For testing, we selected six models of gigabit switches of different price ranges, available in almost any computer store, and tested them. Let's see how good they are.

As a rule, modern routers designed for creating a home network support connecting four devices, although now this is clearly not enough. “Smart” TVs and refrigerators, storage devices and multimedia processors, computers, phones, printers..., requiring network connections, literally filled the apartments and houses of our compatriots.It’s impossible to count everything.

Switches will come to the rescue - devices that allow you to increase the number of available ports while maintaining maximum network speed. They are inexpensive, since the appearance on the market of a large number of highly integrated chips, which are ready-made gigabit solutions, has significantly simplified the development and production of switches, which has led to a rapid reduction in the cost of such products.

Of course, you can get by with either a cheaper 100 Mbit switch, or use wireless connection, so popular now. However, in reality worthy alternative wired connections not yet, especially when it comes to high-speed transmission lines that can withstand significant loads, such as copying large files or watching high-quality videos.

For testing, we selected six models of gigabit switches of different price ranges, available in almost any computer store. Let's see how good they are.

TP-LINK TL-SG1008D (average price- 1500 rub.)

This switch is housed in a black plastic casing with numerous holes to improve ventilation. The build quality of the device is high enough, and the design is good enough to place it in a prominent place.

The front panel contains eight port indicator connectors that allow you to determine network activity and the established connection speed, as well as a power LED. And on the back of the device there are eight RJ-45 ports (10Base-T/100Base-TX/1000Base-T ports), as well as a connector for an external power supply..

The switch does not have a separate switch; it turns on immediately after power is applied. Cooling is passive - there are no fans installed in the case.

The basic element of the design is the 8-port switch BCM5398KPBG from Broadcom. 4-port G48201SNG transceivers are used as the physical layer. The configuration options available with the BCM5398KPBG are not used.

The switch showed high data transfer speeds in all tests, although performance dropped slightly as the number of connected groups increased.

ASUS GX-D1051(average price - 1500 rub.)

This miniature model ASUS GX-D1051 has a light gray plastic body. It can be placed on a table or attached to a wall. In addition, two pieces of magnetic rubber are built into the base of the switch, making it easy to attach to any steel surface, such as the side wall of a server case or PC, without scratching the surface.

The front panel has a power indicator and port indicators that help determine network activity, the set connection speed, and the transmission mode (duplex or half-duplex).

On the rear panel there are five RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports) and a socket for connecting an external power supply.

The switch is passively cooled and therefore operates silently. A large number of holes promote better air circulation inside the device and provide good cooling. Normal thermal conditions are also maintained through an intelligent energy management circuit.

The device is based on a chip manufactured by ASIX Electronics - the AX88178 switch, but the customization options are not used.

In all test modes, the switch showed fairly high performance. Like most of the switches reviewed here, as the number of connected workgroups increased, the overall data transfer rate decreased.

D-Link DGS-1008D(average price - 2000 rub.)

The DGS-1008D device is made in a plastic case with a traditional shape and color for D-Link products. The design of the case allows for both desktop placement and wall mounting.

On the front panel, behind a translucent insert, there is a power connection indicator and port indicators (two for each port), allowing you to recognize network activity and the set connection speed. On the rear panel there are eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports) and a socket for connecting an external power supply.

The switch, which has passive cooling, operates silently, and a significant number of holes improves air circulation inside the device.

The model is based on chips manufactured by Marvell: 4-port transceivers 88E1145 and 8-port switch 88E6181. The customization options available with the 88E6181 are not used.

When the device is turned on, the Cable Diagnostic function is activated, which determines the main problems in the cable connection, such as an open circuit on any of the contacts, a short circuit between conductors and an increase in terminator resistance (more than 100 Ohms). The function only works when a gigabit device is connected to the device.

The switch showed good data transfer speeds, although it could not get ahead of the other participants in this review.

TRENDnet TEG-S80TXE(average price - 1900 rub.)

The switch, made in a miniature metal case, can only be placed on a table; it cannot be hung on the wall - there is no mount for this.

On the front panel of the device there are eight RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports), a power indicator and port indicators (three for each port), showing network activity and the set connection speed. On the rear panel there is only a socket for connecting an external power supply.

Thanks to passive cooling, the device operates silently. In tests with a small number of simultaneously working pairs, it showed good results, but as the load increased, it lost slightly to its competitors. In the test with gradual connection of workstations about The lag began when the sixth computer was connected.

Multico EW-2005V(average price - 2000 rub.)

This small device in a plastic case can be either installed on a table or suspended. Its body has holes for installing brackets (although they are not included in the package) for mounting in a 10- or 19-inch rack.

The switch does not have a built-in power supply, a connector for connecting external source placed on the side. Thanks to passive cooling, the device operates absolutely silently.

The front panel contains five port indicators (two LEDs per port) to identify network activity and set speed connections, on the rear there are five RJ-45 connectors (10Base-T/100Base-TX/1000Base-T ports).

The basic element of the design is the 5-port controller BCM5385 from Broadcom. For four ports, all physical layer elements are built into the BCM5385, and for the fifth, an additional single-channel transceiver B5011A1KQM is used. The configuration options available with the BCM5385 are not used.

The switch showed good speed performance, but it turned out to be slightly lower than that of other participants in the review.

Linksys SE2500(average price - 2500 rub.)

The plastic body of this model has a rounded shape, which is somewhat unusual. The logo is embossed in the center of the black lid. On the front panel, behind the ventilation slots, there is a row of indicators - one for each of the eight ports and a separate one for power. The ports themselves occupy the entire rear panel, and there is also a slot for an external adapter. Holes for ventilation are made onsidesand at the base, in the center of which there is an information sticker duplicating the information from the sticker on the package.

The device can only be placed on a table. The screws for securing the cover are hidden under the rubber feet. The basic element of the design is the 8-port switch RTL8169S-32 from Realtek. Physical layer chips are covered with radiators. The device, which has passive cooling, operates silently. But the number of holes for air circulation, in our opinion, was clearly not enough - under high load it noticeably heated up.

Results

We have to admit that we liked all the models that participated in the tests. Undoubtedly, each of them will be able to significantly expand the capabilities of a home network by combining various devices that require a wired connection.

Of course, all the tested models had both advantages and disadvantages, the main one being a decrease in operating speed as the number of user groups increases. Therefore, this time we decided to refrain from awarding the traditional “Editor's Choice” award.

We considered the “Best Buy” model TP-LINK TL-SG1008D, which has good functionality at a more than reasonable price.

Introduction

When the Ethernet standard was developed in 1973, its throughput was defined as 10 Mbit/s. It was a technology with a shared data transmission medium, that is, all traffic was, as it were, divided between all users.

Naturally, as their number increased, the share of each in the common channel decreased. Meanwhile, the needs were constantly growing. And Ethernet was replaced by Fast Ethernet, providing a throughput of 100 Mbit/s.

If the network bandwidth based on 10 Mbit/s Ethernet is not enough, then 100 Mbit/s Ethernet will successfully cope with all user needs. But this is only for now, and provided that your network is not large. What will happen in a year or two? Network traffic will only grow, and soon the 100 Mbit/s that we have today will not be enough. And this perspective applies not only to corporate, but to home networks. The much-anticipated video-on-demand - is this an example that will require high bandwidth? For one user, 10 Mbit/s may be enough, but what if there are five, ten or a hundred users on one channel who want to watch different programs?

Let's compare what it will take to build a network with a throughput of 100 Mbit/s and 1 Gbit/s. Are changes to the cable structure of 100-megabit Ethernet necessary to move to gigabit? We will also look at the TigerSwitch gigabit Ethernet switch from SMC and network adapters from the same manufacturer.

Cables

As you know, today most networks are wired. This means that information is transmitted through wires - cables. All cables can be divided into two large groups: electrical and optical.

We will not touch upon optical ones too much, let’s just say that they differ from each other in design (sheath, braid, etc.) and the number of fibers contained in the cable. There are single-mode (light pulses pass along the axis of the optical fiber) and multimode (light pulses are reflected from the walls of the optical fiber) cables. Single-mode cable can provide higher capacity and longer distance. Multimode is divided into two types: step refractive index and smooth refractive index.

Electrical cables can also be divided into two more groups: coaxial and twisted pair. Coaxial cables have one central conductor enclosed in a braid. There are thin and thick coaxial cables.

Thick coaxial cable designed for Ethernet networks 10Base5. It got its name for a reason: its thickness is about 12 mm. Sometimes it is called yellow, which is also quite justified - its shell has that color. Characteristic impedance 50 Ohm.

Thin coaxial cable is used in 10Base2 networks. The characteristic impedance is also 50 Ohms. The cable thickness is about 5 mm.

Twisted pair is divided into categories. The main characteristics of each of them are given below:

CAT 1
Cables of the first category can be used where the requirements for data transfer speed are minimal. It was and is used for voice transmission, both analogue and digital, and data transmission at speeds up to 20 Kbps. Previously, it was actively used in telephony.

CAT 2
A higher quality type of cable, can operate at frequencies up to 1 MHz. Used in ISDN lines.

CAT 3
This standard was adopted in 1991. Can be used for both voice and data transmission. Operates with 16 MHz bandwidth. Used in standards such as 10BaseT, 100BaseT4. Quite widespread in cable systems built before 1996.

CAT 4
It is an improved version of the third category, capable of operating at frequencies up to 20 MHz. Used in Token Ring networks with a throughput of 16 Mbit/s.

CAT 5
Standardized for the range up to 100 MHz. Works with protocols such as Fast Ethernet (100 Mbps), 100VG-AnyLAN (100 Mbps), ATM (155 Mbps) and Gigabit Ethernet (1000 Mbps). He came to replace the third category. Today it is the most common type of cable.

CAT 6
This cable is characterized by operation with frequencies up to 200 MHz. Designed to support high-speed protocols over longer lengths than using Category 5 cable.

CAT 7
This cable can operate at frequencies up to 600 MHz. Designed for the same purposes as category 6 cable. It has a significantly higher cost.

Ethernet Standards

In general, consideration of Ethernet should begin with the physical topology of the standard. Everyone knows such types of topologies as star, bus, ring, and cellular. Which ones apply to Ethernet? Physically, Ethernet can use a bus or star topology, but the logical topology is bus.

The next important parameter is the method of multiple access to the transmission medium. The Ethernet standard uses CSMA/CD (Carrier Sense Multiple Access and Error Control). This means that only one device on the network can transmit data at a time. The bandwidth of standard Ethernet is 10 Mbit/s, but today there is not only standard Ethernet, but also Fast Ethernet with a bandwidth of 100 Mbit/s and Gigabit Ethernet with a bandwidth of 1 Gbit/s.

There are quite a few types of Ethernet, which differ in the type of cable used. Let's look at some of them:

Ten Megabit Ethernet

10Base5
This standard specifies a maximum throughput of 10 Mbps when using thick coaxial cable. The maximum segment length was 500 meters. Thanks to the long segment length, it has found its niche in the market network equipment. Today it is practically not used.

10Base2
Here the throughput was also 10 Mbit/s, but a thin coaxial cable (diameter about 6 mm) was used. The physical topology is a bus, which causes relatively low cable costs, but if one node fails, the entire network ceases to function. It was more convenient for installation than thick coaxial cable and became quite widespread. The maximum segment length is 185 meters. Today, like thick coaxial cable, it is practically not used.

10BaseT
This is probably the most convenient standard of the entire 10Base group, which can be explained by the fact that it uses twisted pair (UTP) cable of the third category and higher. This type of cable is more convenient for installation than coaxial cable, but due to the fact that the physical star topology is used (it is necessary central device), cable consumption is greater than when using coaxial cable and a physical bus topology. Although this disadvantage is compensated by the higher reliability of the network - if one communication channel or node fails, the rest of the network continues to work. The maximum segment length is 100 meters. Today it is used only in old networks and is gradually disappearing.

10BaseF
This standard is absolutely irrelevant today; now no one would even think of using fiber optics to create a network with a throughput of only 10 Mbit/s, although various specifications support distances of up to 2000 meters.

We can conclude that ten-megabit Ethernet is practically no longer used and today one should not even think about building a network based on it. Its speed is insufficient for most of today's applications.

Stamegabit Ethernet

Now we will try to figure out how things are going with hundred-megabit Ethernet. Is its bandwidth sufficient for today's applications and what will the situation be like tomorrow? To do this, consider the standards of hundred-megabit Ethernet. Like the ten-megabit one, there are several of them.

100BaseT4
As you can see, the maximum throughput is determined to be 100 Mbit/s when using twisted pair cable of the third and higher category. For communication, the standard uses all four pairs of cable. This implementation is quite profitable. At a time when most networks operate on category 3 cable, it is very convenient to increase network throughput to 100 Mbit/s without changing the cable structure. However, all other parameters remain the same as before (maximum segment length 100 meters).

100BaseTX
This standard uses two pairs of cable of category 5 and higher. Although the third category of cable is not suitable for building a network using this standard, today this is not a problem in the case of building a local network from scratch. Category 5 cable has a fairly low cost and can provide a throughput of 100 Mbit/s when using two pairs of conductors. The maximum segment length is also 100 meters. This standard is the most popular today.

100BaseFX
This specification specifies operation over two multimode fibers at 100 Mbps. It also defines the specifications 100BaseSX and 100BaseLX, operating at different wavelengths; in addition, 100BaseLX can also operate on single-mode fiber.

We have just looked at the Ethernet specifications for speeds of 100 Mbit/s (Fast Ethernet). The use of each of them can be quite accurately determined: 100BaseT4 Used in networks where the cable structure is based on category 3 cable, there are no possibilities for further expansion. 100BaseTX uses Category 5 cable (two pairs out of four), which makes it possible to further increase network throughput. 100BaseFX - runs on fiber optics, so its use can be considered limited to server rooms.

A little about gigabit Ethernet

What is Gigabit Ethernet? You may also have heard "gigabit over copper" or 1000BaseT, which are basically the same thing. This is essentially the same Ethernet, but operating at speeds of up to 1,000 megabits per second, which is 100 times faster than regular Ethernet (10BaseT), or 10 times faster than Fast Ethernet (100BaseT) over twisted pair cable of the fifth category and higher. The maximum segment length is 100 meters. The basis of this standard, approved in 1998, is the IEEE 802.3z standard. This standard was used only for backbone channels until its addition for Gigabit Ethernet over copper, 1000BaseT, was released in 1999. It was this addition that helped this standard enter almost the same market as 10/100 Mbit Ethernet.

The resulting standard is ten times faster than 100BaseT and backward compatible with earlier Ethernet equipment. It uses the traditional CAT-5 (or better) cable found on today's networks.

About its necessity

Previously, when gigabit Ethernet just began to appear (as it once was with 100-megabit Ethernet), it was aimed at the corporate market - connecting file servers and database servers. This was a completely natural use of it, since the capacity of hundred-megabit channels was no longer enough. As a result, it became possible to meet gigabit between high-speed switches and on highways. Today it has already become a necessity.

As prices for gigabit Ethernet equipment fell, its scope expanded. Today it can be found in workstations where bandwidth requirements are high. But there are still few of them.

Why is gigabit Ethernet relevant and why not? Let's try to find the answer to this question, which seems simple at first glance.

It is quite natural that for network applications designed with the ability to work over the Internet, such speed is absolutely not needed. Examples of such applications include online games. Considering that most of them can be played using a dial-up connection, then even a ten-megabit channel will be more than enough for more than a hundred of these players!

For multimedia streaming, you should also not have a gigabit network. The maximum you will need is 40 Mbit/s, this is for professional quality, but for the average user this number can be safely divided by two, and when using MPEG-4 - even by ten. Therefore, here it is worth thinking about switching to gigabit only if you need to simultaneously process several streams of professional quality.

The relevance of gigabit Ethernet for copying and backup is worth considering in a little more detail. When copying large files over a hundred-megabit network, and it is precisely such copying that more clearly shows throughput, you may encounter a lack of throughput. And while speeds above fifty megabits per second are very difficult to achieve on a Fast Ethernet network, switching to Gigabit Ethernet can improve the situation. When backing up over a network, as well as when copying from several computers to one simultaneously, switching to gigabit will only help if the performance of the computers and their disk systems will be sufficient. Remember that there is no benefit from a gigabit network on a 486 computer, or on a computer with a slow disk subsystem.

Client-server business applications used in enterprise networks can be quite critical to network throughput, but here everything depends on the specific application and to determine the need to switch to gigabit, you should analyze the load on the network.

What do you need to prepare to switch to gigabit?

If you already have a ready-made cabling structure using Category 5 twisted pair cable, then that's all you need to make the transition. The only requirement is that all four pairs of wires must be connected. When using only two pairs, the equipment itself will be able to determine this and operate in 100 Mbit/s mode. All other requirements remained almost the same as for 100BaseT (segment length, etc.). Particular attention should be paid to the quality of the cable, connectors and crimps.

In addition, it is worth paying attention to the performance of computers and operating system, used by them. When using computers with a processor speed of less than 700 MHz, even with an OS such as Windows 2000 or XP (compared to Windows 9X), you will not be able to experience the full capabilities of a gigabit network, as well as when using even more powerful machines under Windows control 9X.

If you have a cable system, all that remains is to buy active network equipment: network adapters and a switch or router.

Testing the SMC8606T TigerSwitch 1000 gigabit switch and gigabit network cards

Equipment

Let's take a closer look at the so-called heart of the network. Devices such as hubs or switches can be used here. When using hubs, the maximum distance between stations is limited (the rule of four), between two workstations there can be a maximum of four hubs, for Fast Ethernet - two. When using switches, this rule does not apply, so switches remove this limitation. Switches come in managed and transparent types. It is clear that managed ones are significantly more expensive; naturally, they provide higher functionality. Good news for budget-conscious users is the emergence of transparent models of gigabit switches at a lower cost.

SMC

Today we will look at the following equipment from SMC: the SMC8606T TigerSwitch 1000 switch, the SMC9452TX network adapter for working over twisted pair cables, and the SMC9462SX TigerCard 1000 network adapter for working over fiber optics.

TigerSwitch

Let's start with the switch. Its delivery set includes the following components: SMC8606T TigerSwitch 1000 switch, power cable, cable for console connection via the COM port, installation and management instructions. It is worth noting that there are no GBIC modules included in the kit.

TigerSwitch 1000 is a high-performance Gigabit Ethernet switch. It has eight Gigabit ports to increase the throughput of your network backbones. Six 1000BaseT ports with RG-45 twisted pair connectors and two GBIC slots. With a total performance of 16 gigabits, it is a fairly convenient solution for quickly expanding your network.

TigerSwitch 1000 included

TigerSwitch 1000

Possibilities

To provide greater flexibility, it has a built-in management agent that allows management through the built-in management program or through SNMP/RMON applications. In addition, there is the possibility of console connection to the device. And that's not all, you can also connect to control using Telnet or Web interfaces.

Ports

1000BaseT ports can operate at speeds of 10 Mbps, 100 Mbps, and 1000 Mbps in full-duplex or half-duplex modes. All ports support both straight and crossed cables. Implemented on all ports automatic control streams, which prevents port buffers from overflowing.

GBIC ports are designed to use five-volt GBIC modules for both multimode and single-mode fiber. They support hot-swapping of GBIC modules, so you don't have to turn off the switch to insert or remove an optical module.

On the front panel of the device there are indicators that display the status of both the switch as a whole and each port individually. Three indicators located separately display the status of the switch itself: power, power from a redundant power supply, and diagnostic mode. The remaining group of indicators displays the status of the ports: the top row (FDX) shows what mode the port is operating in (full-duplex or half-duplex), the middle row (ACT) indicates data transfer, and the bottom row (Link) indicates the transfer speed (green flashing - speed port 1000 Mbit/s, green - 100 Mbit/s, orange - 10 Mbit/s)

Standards Compliant

Control and Security

The switch has a built-in management agent that includes extensive management capabilities and allows operation through the console (when using a serial port), Telnet and Web interfaces, as well as through the SNMP and RMON protocols.

In general, to work with any device interface, with the exception of work via a serial port, the device must have its own IP address, which is used to access the device and further work with it. Therefore, you need to specify the IP parameters (IP address and subnet mask). We also strongly recommend changing your administrative account settings.

Telnet interface

When Telneting to the switch, the first thing we will see is a rather nice Telnet prompt window, which prompts you to enter a username and password. You'll also see the SMC logo, device model, and firmware version, which is especially useful if you have multiple devices that you connect to using Telnet.

After entering the correct username and password, you will be taken to the main menu screen.

As you can see, here you can view information about the device, as well as configure any device settings.

The Switch Information menu item displays switch parameters: hardware, firmware version, POST ROM version, serial number, number of ports and power supply status.

The Switch Information item contains data such as system description, ID (same as MAC address), operating time, name, etc.

We will not dwell on management in detail; we will consider only the most important points in our opinion.

In the Management Setup item, you can configure many different parameters, including the IP parameters of the device, enable or disable the built-in Web server for administration, serial port parameters, SMNP configuration, set Accounts To work with the switch, the firmware update is performed from the same point.

The next item we will look at is Switch Configuration. This item configures parameters such as port configuration, port status, Spanning Tree configuration and information, port mirroring, grouping, IGMP configuration, setting broadcast storm protection and setting port security.

The most interesting item we found was Network Monitor, in which you can view statistics of various parameters (error report, amount of received data, etc.) for each port.

The penultimate item suggests resetting parameters and settings. And the last one is to end the current session.

Thus, we are convinced that using the Telnet interface you can configure almost everything, from IP parameters to accounts and STAs.

Web interface

The Web interface is more pleasant and user friendly than Telnet. It is for this reason that we will dwell on it in a little more detail.

By typing the switch address in any Web browser window and entering the correct credentials, you will be presented with something like the following window:

As you can see, everything here is intuitive and simple. Also displayed are names, IP address, ID (as you remember, it coincides with the MAC address) and operating time of the device, in addition, on home page there are several additional buttons, allowing you to connect to the Telnet interface, send an email to SMC technical support, or go to the SMC website. From this page you can get to the help page. Conveniently enough, the help is stored in the device’s firmware, thereby saving you from the problems associated with losing the documentation CD; help is available whenever the switch is available. But there are also problems with this solution, for example, you will not be able to get help when the switch is turned off or inaccessible.

The next item is Switch. Everything displayed here is the same as in the Switch Information item when using Telnet.

The IP item displays following parameters devices: assigning an IP address (BootP or manual), IP address, subnet mask, default gateway, MAC address and maximum amount Telnet sessions. It is quite natural that all parameters except the MAC address can be redefined.

The SNMP setup screen is shown below. The switch supports SNMP management over UDP transport.

With the Security screen, everything is clear even to a baby - it is intended for changing the password. By the way, the password can be empty or consist of 1 - 11 characters, and it is not case sensitive.

The next Upgrade screen consists of two parts - for saving current version firmware and to write a new one. To update the firmware version, you need to download the binary version of the firmware for this switch model from the manufacturer’s website. After this, the switch will automatically restart.

On the Configure screen, parameters such as the server and storage location for the current firmware, and the server and storage location for the new firmware are configured.

By clicking on the next button, you will go to the screen for setting up the address table for virtual networks.

It is also worth stopping at the Port button. After pressing it, information about connections is displayed on the screen.

The last button shows port usage statistics for various parameters.

SMC9452TX

SMC9452TX network adapters are sold in such beautiful packaging. The adapter comes with a floppy disk with drivers for the following operating systems: Windows 9X/Me, Windows NT/2000 and NetWare.

Here, unlike the switch, everything is simple. We had no problems installing adapters on computers, Windows XP, and Windows 2000 were able to automatically detect them and install drivers. The adapter is designed for the 32-bit PCI bus. It supports 1000BaseT, 100BaseTX and 10 BaseT standards. In full duplex mode it can provide throughput up to 2 Gbps.

After installing the drivers and connecting the cables, you can admire the beautiful, but still unusual picture of the connection status:

We were a little disappointed by the fact that the adapter gets very hot during operation, but we did not experience any performance problems related to overheating.

To reduce processor load, the adapter uses Jumbo Frame technology, which allows you to overcome the limitation on the maximum Ethernet frame size of -1514 bytes, thanks to which the transmission bandwidth large volumes data can increase by 300%.

The adapter supports the 802.1Q standard, which describes the use of VLAN. Support for traffic priorities described in IEEE 802.1p is implemented, including eight levels of traffic priorities.

SMC9462SX

The packaging of this adapter is made in a different style. The adapter comes with a disk with drivers for the following operating systems: Windows 9X/Me, Windows NT/2000 and NetWare and Linux.

We did not encounter any serious problems installing adapters on computers. In both Windows XP and Windows 2000, we had to install the driver from the included disk. The adapter can be used with 32-bit PCI bus, as well as with 64-bit. Can provide a maximum throughput of 2 Gbps.

After installing the drivers and connecting the cables, you can see an excellent, but still unusual picture of the connection status:

Here, unlike the SMC9452TX adapter, we were disappointed that we encountered network losses, which may be due to the cable. Temperature The adapter did not cause any complaints.

Just like the SMC9452TX, it implements Jumbo Frame technology, thanks to which it becomes possible work with frames up to 9014 bytes in length, it is natural that other devices should also support this technology.

VLAN support and eight-level traffic priority support are implemented.

Performance

As we expected, the performance of this device turned out to be significantly higher than the capabilities of the computers with which we tested. We were never able to get the switch to work at full capacity. All our efforts to prevent the switch from coping with the data flows were in vain.

As you know, testing begins with connecting all devices to the network. We had no problems with this. We used both straight and crossover cables, all switch ports have automatic MDI/MDIX detection, so there were no problems here either. During testing, twisted pair cable of the fifth category and multimode optical fiber with the corresponding ports were used.

When turned on, the switch reads data from permanent memory and checks ports and connections to them. It takes a lot of time to prepare the switch for operation. We got results ranging from 1 minute 54 seconds to 3 minutes 4 seconds. This may seem like a long time, but think about the capabilities this device has! And its capabilities are impressive! And, besides, it is usually better to wait once at system startup than to experience constant delays when transferring data over the network.

Testing was carried out test system NetIQ Chariot on Windows 2000 machines. Hardware: AMD processor Duron 750 MHz, memory 256 MB. Taking into account the fact that 1 bit per second requires approximately one hertz of a processor, on such a system we expected to see a speed of about 700 - 800 gigabits per second, but we will dwell on the results obtained a little later.

To compare throughput with standard Fast Ethernet, we performed speed measurements on it as well. This measurement was carried out using Realtek hundred-megabit network adapters and an SMC8606T TigerSwitch 1000 switch. Here's what we got:

Optics-2xUTP

Here we will summarize the total values. Minimum 64.5 Mbit/s, maximum - 200 Mbit/s, average value - 184.5 Mbit/s, while the average throughput was calculated on the optical segment of the network.

2xUTP optics

Here the result of the total throughput turned out to be worse, as well as the minimum throughput - 29.4 Mbit/s, but the maximum was 222.2 Mbit/s. As a result, we received an average throughput of 167.7 Mbps.

Fast and Giga Ethernet

We also performed a test connecting two machines with a speed of 100 Mbit/s to a gigabit segment. The results obtained indicate that such an implementation will also bring an increase in throughput compared to the hundred-megabit segment. Minimum - 9.8 Mbit/s, maximum - 95.2 Mbit/s, average - 141.3 Mbit/s. Don't be surprised - the average value is shown for a gigabit connection, and the minimum and maximum are shown for 100-megabit connections.

	Average, Mbit/s	Minimum, Mbit/s	Maximum, Mbit/s
Fast Ethernet	76	30,8	80
Optics-UTP	200,7	200,7	235,2
UTP optics	215,4	150,9	216,2
UTP-lUTP	184,7	121,2	205,1
Optics-2xUTP	184,5	64,5	200
2xUTP optics	167,7	29,4	222,2
Fast and Giga Ethernet	141,3	9,8	95,5

We saw the highest average throughput in the UTP-optics test (from a gigabit twisted-pair segment to a gigabit fiber-optic segment), and the highest peak throughput in the UTP-Optics test (from a gigabit fiber-optic segment to a gigabit twisted-pair segment).

Compared to Fast Ethernet, average speed gigabit connections was more than 2.5 times higher under the test conditions described above.

We also looked at processor load. When performing tests, it was 100%, which allows us to hope that as the processor speed increases, the network speed will also increase.

conclusions

This switch model, together with the considered network adapters Perfect for use as a central device in almost any network: from small to large corporate. It is worth considering that when used in a small enterprise network, the switch will be able to provide a sufficient performance reserve, which will eliminate problems with upgrading network equipment for several more years. And provided you have a cable system based on category 5 cable, all you need to migrate to gigabit Ethernet is a switch (central device) and network adapters.

The speed of the network is highly dependent on the performance of computers. Therefore, it is worth thinking about it too. The only time you shouldn’t even think about switching to gigabit is when using weak computers (with clock frequency less than 500 MHz). Even with all your desire, you will not be able to get something very different from 100 Mbit/s. On computers with a 750 MHz processor, we got just over 200 Mbps, which is more than twice the Fast Ethernet result.

If the bandwidth of Fast Ethernet is no longer enough for you, then today you can think about switching to Gigabit Ethernet. It’s up to you to decide what to use for it: optical fiber or twisted pair. During testing, we did not notice a big difference, although this can be explained by the insufficient performance of computers, but not by the limitations of the switch.

The switch worked reliably, and during our testing we never encountered any malfunctions. I was pleased with the possibility of hot-swapping GBIC ports. The device's wide configuration capabilities allow you to use it for almost any purpose, and the presence of several administration interfaces (console, Telnet, Web) allows you to use the one that is most convenient for you.

Among the variety of car parts, there are many elements on the serviceability of which depends normal operation power unit. One of them is the well-known switch, which is an integral part of electrical equipment. Its main purpose is to provide normal functioning, therefore, in the event of an element failure, problems with starting the engine cannot be avoided. In most situations, this unit is reliable and wear-resistant, but sometimes troubles happen with it.

1. Signs of a faulty switch

Problems in the operation of the switch are one of the most common causes of operational failures of the power unit (of course, provided that the fuel system Everything is fine). Most often, switch malfunctions manifest themselves in the form of a drop in acceleration dynamics, failure to start the engine, “failures” during sudden acceleration, as well as “triple” of the engine. An experienced driver will immediately notice that something is wrong, and to be convinced of his theory, it will be enough to carry out simple diagnostics.

The most common cause of switch failure is a “bad ground.”, which most often appears after lengthy repair work or due to oxidation of contacts. As a result of this, the device simply cannot send the appropriate impulses to the ignition coil, and without them the engine will not start and the car will not start.

In some cases, the pulses do not reach the switch itself, which is caused by problems with the contactless sensor. In both versions of this situation, devices require more detailed diagnostics followed by repair or replacement.

For example, to check the condition of a contactless sensor, you need to measure the voltage at the output of the sensor-distributor. In good condition, cranking the crankshaft with a key should cause a sharp change (often ranging from 0.2 - 0.4 V to 5 - 11 V). If this does not happen, the sensor most likely needs to be replaced. Diagnosing the status of the switch also does not take much time, and most often it does not require any special equipment.

2. Algorithm of actions for checking the switch

Many motorists prefer not to waste time troubleshooting, but to immediately replace the switch with a new unit. In principle, there is logic in this decision: Firstly, you don’t have to waste time checking, Secondly, having installed a new part, it will immediately become clear whether there is a problem with it or not. True, there is no need to worry about wasted time, because diagnosing the switch will not delay you for long.

There are two main ways to test your switch. The first one is a little simpler and requires a carrying lamp. The procedure algorithm in this case is as follows:

1. Disconnect the wire coming from the switch terminal from the ignition coil;

2. We connect the freed tip of the wire to the control lamp, and the second terminal of the lamp to the terminal of the ignition coil;

3. Turn on the ignition and turn the engine crankshaft using the starter.

If the control lamp does not flash when the crankshaft rotates, it means that the corresponding current pulses are not emanating from the switch to the ignition coil. That is, the part is faulty and requires replacement. For the second method of diagnosing the switch, you will need more tools, including a soldering iron and a metal sheet as a “ground”.

In this case, the sequence of verification actions is slightly different. First, you should know certain symbols located on the device body. Often these designations are presented in the form of Latin letters (for example, B, C, T, K).

Secondly, you need to understand that when checking a device, you need to pay attention to the reliability of the connections, and there should be a good “minus” on the case itself. Quite often, after lengthy repair work or under the influence of oxidative processes, certain problems arise in the operation of the switch, which are explained by a “bad ground”.

To check, the switch together with the ignition coil must be placed on one metal sheet acting as a “ground”, and once again check the reliability of all connections, as well as the distance from the ignition coil (more precisely, the “output on it”) to the metal sheet. The value of this distance should correspond to 15-25 mm.

At the next diagnostic stage, you should alternately close and open the wire that should go to the switch contact. It is connected to a 12-volt control lamp, and general rules The block signal should not exceed 5 V.

However, if these same 12 V are supplied for a short time, then you can check the switch in two directions at once: for its serviceability and for the quality of operation in “extreme” conditions.

As in the first option, for the test work to be effective, you need to crank the engine with the starter, and if the switch is in working order, you will see a spark hitting the metal sheet (the light bulb begins to glow). The use of a metal sheet is completely optional, and you can do without it (as described in the first option); the main thing is that there is a good “mass”.

To check the functionality of the internal switch, perform all the same steps, only this time you remove and replace another contact.

3. Materials for testing the switch

To test the switch, in both of the above cases you will need a test lamp with a nominal voltage of 12 V and standard set keys with which you can verify the presence or absence of pulses(that is, in the serviceability of the device itself). The second diagnostic option involves the presence of other elements: the most common button, a soldering iron and a metal sheet as a “ground”.

However, if you do not want to waste time at all, then to check the health of the switch you just need to buy a new device. If after installation everything is fine, then the problem really was in the old element.

Subscribe to our feeds at