Transcript
Power for VoIP Networks Per Grandjean-Thomsen Chairman - IEC SC22H -International Electrotechnical Commission for UPS Engineering Manager - Emerson Network Power
[email protected] Abstracts Voice over IP, also known as IP telephony, is convergence of traditional telephony with data networks. While all telephones need signal and power in order to operate, IP phones obtain signal from the computer network and power from either an external AC to DC converter (a ‘powerpack”) connected to an adjacent mains outlet or from a Power over Ethernet (PoE) connection. PoE provides DC power through the existing data cable. The energy required to feed the PoE, for consistency of availability, is generally taken from the Uninterruptible Power System (UPS) that supports the associated computer network. Power availability is classified in tiers ranging from 99.50% to 99.9999%. Such availability is achieved by combining specific UPS configurations with coordinated power distribution techniques and by providing alternative and redundant routes of power to the critical load.
1. Introduction - Power Requirements for IP voice, data and video networks Voice over IP, also known as IP telephony, is convergence of traditional telephony with data networks and refers to the technology solution used to transmit voice over a data network infrastructure by using the Transmission Control Protocol / Internet Protocol (TCP/IP). The (analogue) voice signal received by the microphone of a VoIP telephone is coded into (digital) TCP/IP packets of information suitable for transmission over a computer network. At the other end, the TCP/IP packets are decoded into analogue voice signal for reproduction through the earpiece (speaker) of the VoIP phone. VoIP offers reduction in long distance carrier charges when compared to those of traditional telephony but other truly remarkable advantages of VoIP are offered by integrating telephony with messaging, video conferencing, web-sites and by eliminating the necessity of maintaining a separate transmission media for voice and for data. The VoIP telephone is a data terminal that processes voice packets transmitted within the same network used for data computing, typically through a RJ45 connector and a Cat5 cable in an Ethernet local area network (LAN).
Page 1 of 8
Power Requirements – Power over Ethernet All telephones need signal and power in order to operate. The quality of service of telephony is directly related to the adequate availability of both. Most traditional telephones work just by connecting them to a telephone socket of the public switched telephone network, PSTN. This is because the PSTN provides both voice signal and 48V dc power through the same pair of incoming copper wires.
Power supply
Data
Ethernet Hub
1 2 3 4 5 6 7 8 RJ45
1 2 3 4 5 6 7 8 Cat5 cable
RJ45
Data
VoIP Phone
Fig 1 – IP phone with External Power supply
IP phones obtain their signal from the LAN. The power for the IP phone can be obtained from an external AC to DC converter (a ‘powerpack”) connected to an adjacent mains outlet or, not unlike the PSTN, from a Power over Ethernet (PoE) connection wherein DC power is distributed through the same Cat5 Ethernet cable that transports data in the computer network. PoE is a path to supply reliable, uninterrupted power to network devices that require only modest amounts of power to function, say around 10W. Examples of such devices are Internet VoIP telephones, wireless LAN access points and network cameras. PoE works with the existing cable plant without requiring modification. The PoE principle consists of supplying both power and data requirements by applying DC voltage onto either the unused wires (7-8 and 4-5) or to data wires (1,2,3,6) of a single standard Ethernet cable. IEEE 802.3af PoE standard prescribes the use of 48 V (DC) as the injected voltage and permits up to 350mA of current per port. This provides approximately 15W of power. The typical PoE-compatible terminals, called Powered Devices (PD), consume in the range of 3-5W for IP phones, 6-10W for wireless access points, and 9-12W for IP cameras. IEEE 802.3af also prescribes safety and requirements to minimise the risk of Page 2 of 8
hardware failure if connecting non-compatible devices to a PoE circuit. PDs are supplied from Power Source Equipment (PSE). PSEs are typically 24-port devices, compatible with IP Phones (an other compliant PDs) and implemented as follows: Mid-span PSE - A patch-panel positioned between an ordinary Ethernet switch and the PDs. Mid-spans are also referred to as "Power Hubs" or "Power Injectors". End-span PSE – An Ethernet switch wherein PoE is integrated and supplied directly to the data ports. 2. Uninterruptible Power Systems for VoIP Embedding of PoE into the data wiring facilitates but does not, by itself, guarantee availability of power. For high availability, the power shall come from a reliable source. The obvious and consistent choice is to feed PoE from the same source that supplies the computer network – generally an Uninterruptible Power System (UPS). The requirements for Safety, Electromagnetic Compatibility and Performance of UPS are well defined in the International Electrotechnical Commission (IEC) 62040 series of standards. This series have obtained world-wide acceptance and has been adopted or directly legislated by most IEC members countries including those of the European Community, Australia, New Zealand and China. The performance and availability of power from a UPS is therefore predictable. The availability calculation generally follows a set pattern of Tiers. Refer to 2.3.
Uninterruptible Power System
Data & Power
PoE Ethernet Switch (Endspan Power Source Equipment)
1 2 3 4 5 6 7 8
1 2 3 4 5 6 7 8 Cat5
RJ45
Data & Power
IP Phone (Powered Device)
Fig 2 – PoE Endspan PSE powered by UPS
Most modern UPS are compatible with industry communication protocols. It is therefore possible to optimise the use of UPS power during a commercial mains failure by letting Page 3 of 8
the UPS inform the status of remaining battery power to a management console that selectively decides what priority different loads will be given with respect to receiving UPS power. There could be one profile for when mains power is available, another profile for when generator power is available and a third for when the UPS operates in battery mode. Security cameras could be given first priority, IP phones second priority and general work stations third priority. The management console could then send a message to network users that there has been an interruption of commercial power and giving users a limited time to log out of the network. The console could subsequently initiate an automatic shut down of servers and shut power off to the lowest priority devices. 2.1 UPS Classification (Vaa bb ccc) UPS are classified by IEC 62040-3 with a 8-digit code that determines the quality of the output power supplied to the critical load and the to what extent the UPS depends on the input source characteristics to supply such output power. 2.1.1 UPS output dependency on input source VFI = Voltage and Frequency both independent of input VI_ = Voltage only independent of input VDF = Voltage and Frequency both dependent on input 2.1.2 UPS output voltage waveform S = sine-wave output voltage at linear and non-linear load(“computer load”) X = sine-wave output voltage at linear load Y = non-sine-wave output voltage Note: The output voltage waveform is classified in Normal and in Battery modes with a double letter. E.g. “SY” means Sine-wave in Normal Mode and Non Sine-wave in Battery Mode
2.1.3 UPS response to transients 1 = low transient voltage variation and zero voltage interruption 2 = moderate transient voltage variation and max 1 msec voltage interruption 3 = moderate transient voltage variation and max 10 msec voltage interruption Note: The transient response is classified with a triple digit in accordance with the behaviour upon Change of Mode of Operation, application of a linear load step and application of a non-linear load step. E.g. “312” means 10msec voltage interruption upon Mains Failure, low voltage transient upon a linear load step and moderate voltage transient upon a non-linear load step
Page 4 of 8
2.2 UPS Topology UPS can broadly be divided into three categories: Standby
Double-conversion UPS rectify the incoming AC voltage into a DC voltage that is subsequently inverted into a new AC voltage. The output voltage and frequency can both be tightly regulated within a broad AC input window without drawing stored energy from the battery. Upon AC input failure the inverter continues to operate from the DC / Battery source without interruption. A standby bypass switch provides redundancy and takes care of overload conditions.
By Pass Inverter
Rectifier
AC Input
AC Output Critical Load Battery
Fig 3 – Double-Conversion UPS
IEC62040-3 VFI classification
Standby UPS operate like a doubleconversion UPS in reverse. The AC input voltage when available is normally bypassed into the critical load while the battery is kept charged. Input voltage and frequency abnormalities are corrected by permitting the UPS to transfer the load into inverter mode and drawing stored energy from the battery. The battery charger is not designed to supply the Inverter.
By Pass
AC Input
AC Output Critical Load
Battery Charger
Stand by Inverter
Battery
Fig 4 – Stand-by UPS
IEC62040-3 VFD classification
AC Input
Volt Regulator
AC Output Critical Load
Bi-directional Converter
Battery
Fig 5 - Line-Interactive UPS
Line-Interactive UPS operate as a voltage regulator of the incoming AC voltage when available. The battery is kept charged through AC to DC operation of a bi-directional converter. Input frequency abnormalities are corrected by permitting the UPS to transfer into inverter mode and draw stored energy from the battery.
IEC 62040-3 VI classification
Page 5 of 8
2.3 High Availability UPS Power availability to the critical load is defined as A = MTBF / (MTBF + MTTR) where MTBF = Mean Time Between Failures and MTTR = Mean Time To Repair. High Availability is generally achieved by building fault tolerant systems containing at least one degree of circuit redundancy. This is to minimise the likelihood of failure to the critical load upon failure of a single point of the system or of its sources of supply. There are four tiers of power availability wherein any higher tier may include the characteristics the lower ones. 2.3.1 Tier One – Selective switchboard coordination – No UPS (A=99.50%) Tier one systems contain no UPS and their objective is simply to protect hardware and to ensure that power interruptions are limited to those caused by factors external to the network being protected. The implementation of tier one consists of - selective coordination among circuit-breakers and fuses connected to the critical load so that a load fault causes the nearest protective device to open without disturbing other loads. - selective coordination of transient voltage surge suppressing (TVSS) devices within the network so that both external as well as load created disturbances are attenuated in a cascading manner. The typical availability of a tier one power supply is that of the incoming AC supply, generally the commercial mains supply: Tier-1 - No UPS :
MTBF 200 hrs. Availability: 99.50%
Critical load outage < 44 hours / year
2.3.2 Tier Two – Single UPS with automatic bypass (A = 99.9957%)
By Pass
AC Input Rectifier / Battery Charger
Inverter
To Power Source Equipment
In a tier two UPS an AC Input failure is tolerated during the battery back-up time. Failures of either Rectifier / Inverter / Battery or Bypass are tolerated. Tier two systems are typically built with a single ups that can be rack mounted.
Battery
Fig 6 – Tier-2 UPS :
MTBF 141,000 hrs Availability: 99.9957% Critical load outage < 22 minutes / year
Page 6 of 8
2.3.3 Tier Three – Parallel Redundant UPS with automatic bypass (A=99.9994%)
AC Input
To Power Source Equipment
In a tier three UPS all the failure modes of tier two are tolerated and additionally a whole UPS module can be isolated to permit maintenance while still supplying the load in UPS mode. Tier three systems are typically built with free standing modular UPS.
Fig 7 – Tier-3 UPS :
MTBF 1,000,000 hrs Availability: 99.9994% Critical load outage < 3 minutes / year
2.3.4 Tier Four – Dual Bus UPS (A=99.9999%) In a tier four UPS all the failure modes of tier three are tolerated and an additional fault at any of the output bus bars feeding the Power Source Equipment is also tolerated. Tier four systems are typically built with free standing modular UPS and make use of external Static Transfer Systems (STS) that comply with IEC 62310 series of STS standards.
Bus A
Dual Input
Static Transfer Switch
PSE
Single Input PSE
Bus B Fig 8 – Tier-4 UPS :
MTBF 6,000,000 hrs Availability: 99.9999% Critical load outage < 0.5 minute / year
Page 7 of 8
3. Conclusion Power for VoIP networks is efficiently transmitted through PoE technology that utilises the existing Cat5 Ethernet cabling of a typical computer network. The energy for PoE is obtained through End or Midspan power injectors that are connected to a UPS. The availability of PoE is predictable when the UPS design complies with IEC 62040 series of safety, electromagnetic compatibility and performance standards for standby, line-interactive and double-conversion UPS. Up to 99.9999% of power availability for VoIP networks is achieved by combining prescribed UPS configurations with coordinated power distribution techniques and by including alternative paths for routing of power to the critical load 4. References - IEEE 802.3AF-2003, Data Terminal Equipment Power via Media Dependent Interface (includes Power over Ethernet requirements, end and mid-span) - IEC 62040-3-1999, UPS performance and test requirements (includes UPS topology and system configuration) - You’ve got the power over Ethernet by Joel Conover, Network Computing, 2004 http://www.networkcomputing.com/story/singlePageFormat.jhtml?articleID=19200021 (includes a 802.3af vendor check list) -0-
Page 8 of 8
