WO2008153470A1

WO2008153470A1 - Optical link

Info

Publication number: WO2008153470A1
Application number: PCT/SE2008/000391
Authority: WO
Inventors: Hjalmar Hesselbom
Original assignee: InvOpto AB
Current assignee: InvOpto AB
Priority date: 2007-06-11
Filing date: 2008-06-11
Publication date: 2008-12-18
Anticipated expiration: 2009-12-11
Also published as: EP2156578A1; EP2156578A4

Abstract

An optical link includes an optical fiber (35) for transmission of signals and has from an electrical input terminal to an electrical output terminal a frequency dependent loss and phase linearity characteristics that are equal or similar, for at least a particular or predetermined frequency region, to a specific length of an electrical cable such as a twisted pair cable regardless of the actual length of the optical fiber, naturally within certain but rather wide limits. The optical link may for this purpose include specially selected components or compensating devices such as filters (41) or other devices configured in a suitable way.

Description

OPTICAL LINK

RELATED APPLICATION

This application claims priority and benefit from Swedish patent application No. 0101300-2, filed April 11, 2001 and U.S. provisional patent application No. 0701485-5, filed June 11₅2007, the entire teachings of which are incorporated herein by reference. TECHNICAL FIELD

The present invention relates to a basically electrical transmission link having a portion thereof designed as an optical link or/and a optical only transmission link using electrical protocols and signaling (electrical PHY-circuits). BACKGROUND

A bidirectional link using a standard twisted pair (TP) cable for transmission of electrical digital signals is illustrated in Fig. 2a. The electrical signals at the ends of the transmission link are input to/output from end equipments 1 connected at the ends of the twisted pair cable 3. The twisted pair cable contains at least two wire pairs 5, one pair for each direction. Each of the end equipments 1 includes a transmitter circuit (TX) 7 and a receiver circuit (RX) 9. The transmitter and receiver circuits can generally contain pulse transformers, not shown, for providing the balanced electrical signal output to the wire pair used for transmission from the respective end equipment and for receiving the balanced electrical signal that has propagated through the another wire pair of the TP cable 3. For transmission over a twisted pair cable at the rate of 10 Mbit/s according to the Ethernet standard 1 OBASE-T where the balanced electrical signal have propagated through a TP cable 3 having a maximum length of the 100 m, the electrical signals received from the TP cable and received by the respective RX 9 have undergone some distortion but still the original digital signal can be detected using e.g. a simple receiving circuit including a pulse transformer and possibly a smoothing inductor.

For higher transmission rates such as 100 Mbit/s according to the Ethernet standard 100BASE-TX the balanced electrical signals that have propagated through the TP cable 3 have been more distorted, the distortion e.g. including a frequency dependent attenuation as seen in the diagram of Fig. 1. For receiving such a signal an additional active receiver circuit 11 is required in the end equipment 1, the active receiver circuit receiving the electric signal from the RX 9 in the end equipment and providing the electrical digital signal output from the end equipment that agrees with the electrical digital signal provided to the end equipment at the other end of the TP cable 3. The active receiver circuit includes, as seen in Fig. 2b, a dynamic equalizer 13 and a bit detector 15. The dynamic equalizer can basically include an amplifier that has a frequency dependent characteristic allowing that from the unbalanced electrical signal detected by the RX 9 a signal is output that is sufficiently similar to original electrical signal input from the end equipment into the TP cable at the other end of the TP cable 3 to allow a detection of the bit content of the signal by the bit detector 15. For example, the dynamic equalizer can have an amplification characteristic, see Fig. 3, that corresponds to the inverted frequency characteristic of the TP cable length seen in Fig. 1.

The active receiver 11 can include other functions such as a circuit for switching between receiving a 10 Mbit/s signal and a 100 Mbits/s signal, not shown, and a selector circuit 17 for detecting an equivalent length of the wire pair 5 used for the transmission and setting the frequency dependent characteristic of the dynamic equalizer 13 accordingly. ^•

The most common TP cable presently used is the Cat (category) 5 TP cable that allows, according to the ANSI/EIA (American National Standards histitute/Electronic Industries Association) Standard 568, a maximum data rate of 100 Mb/s and usual applications include 100 Mb/s TPDDI and 155 Mb/s ATM. It often holds four twisted pairs, two pairs used for one bidirectional communication channel and the other two pairs being unused or possibly used for another bidirectional communication channel, compare Fig. 2a. It is generally used for twisted pair signaling according to the 100BASE-TX Fast Ethernet standard, is the cable most frequently used in LANs and has nowadays actually replaced the older 1 OBASE-T and 10BASE-2 (coaxial) cables. Signaling and connectors according to the 100BASE-TX standard follow the same wiring patterns as those for 1 OBASE-T.

If a transmission link of the twisted pair type has to have a length larger than 100 m a middle part of the transmission link can be an optical fiber or a pair of optical fibers, see Fig. 2c, where only the components necessary for transmission in one direction are shown. A first part of the transmission link can then be a TP transmission link 21 of the kind illustrated in Fig. 2a, a second part a fiber optical link 23 including an electrical-to-optical converter 25 at the input end of the optical fiber 35 and an optical-to-electrical converter 29 at the opposite, output of the optical fiber, and a third part 43 another TP transmission link.

Equipment performing conversion tasks for converting signals between different media can generally be called media converters. Conventional such equipment for communication according to the Ethernet standard and converting between signals propagating in a TP cable and an optical fiber contain, for one communication direction, complete TP termination including equalization, clock recovery, bit detection, descrambling, 5b to 4b conversion and serial to parallel conversion on the electrical side followed by parallel to serial conversion, 4b to 5b encoding and optical transmission on the optical side and in the opposite communication direction optical to electrical conversion, clock recovery, bit detection, 5b to 4b decoding and serial to parallel conversion on the optical side followed by parallel to serial conversion, 4b to 5b encoding, scrambling and electrical transmission on the electrical side, see Fig. 2d. This implies that conventional media converters also are protocol converters between the 100BASE-TX and 5 100BASE-FX protocols.

In a mixed transmission link as disclosed in the published International patent application WO 02/23771, a proximate portion of the link, such as at a central switch, can be a twisted pair cable and the .distant portion, such as at end equipment, can be an optical fiber, see Fig. 2e for transmission from the switch and Fig. 2f for transmission to the switch. In the general case,0 obviously, TP cable lengths can be connected at both ends of the optical fiber segment. In said patent application, directly converting a twisted pair signal, this signal being e.g. generally according to the standard implementation 1 OBASE-T of Ethernet on twisted pair wiring according to the Ethernet standard 802.3 or the standard implementation 100BASE-TX according to the Ethernet standard 802.3u using MLT-3 multi-level transmission according to the5 ANSI copper FDDI Physical Layer Dependent sublayer technology, also called as CDDI, to a modulated optical signal and back is disclosed. For the data rate of 10 Mb/s the direct conversion at the detector side is done in a quasi-analog way by using a comparator. This is possible as the Manchester-coded 10 Mb/s signal is very little distorted after several tens of meters of e.g. a Cat 5 TP (Twisted Pair) cable. The small extra phase-jitter created relative to the original electrical0 signal after conversion to optical signal and back does not seem to be a problem for most of the 10 Mb/s TX receivers available.

However, a 100 Mb/s signal according to the 100BASE-TX Fast Ethernet standard is noticeably distorted even after having propagated over only a short length of cable due to frequency dependent attenuation as shown for different cable lengths in the diagram of Fig 1.5 Applying such a signal to an optical transmitter modulating the light in an analog fashion and then receiving it using a (dual) comparator (due to the 3-level signal), as suggested in the cited patent application, creates a signal from the dual comparator that in most cases is quite different from the signal entering the optical transmitter but also strongly distorted relative to the original 3-level signal entering the twisted pair cable preceding the optical transmitter, due to a more or0 less severe phase distortion and even loss of detectable bits introduced by primarily the combination of fixed trigger levels of the dual comparator and the inter symbol interference on the signal received on the fiber replicating the signal originally on the TP cable end at the electrical to optical converter. This makes the succeeding twisted pair receiver fail in interpreting the signal. Only for very short cable lengths between twisted pair ports and the media converters on each side of the fiber a dual comparator would work. For any longer length of a twisted pair cable, analog diode drivers would still have to be used, more or less replicating the now analog (continuous level) electrical signal on the twisted pair cable as an optical signal onto the fiber and on the other side of the fiber, analog optical receivers and cable drivers would have to be used replicating that optical signal on the fiber as an electrical signal onto the twisted pair cable.

A 100 Mb/s receiver that is adapted to the 100BASE-TX Ethernet standard mentioned above is designed to be capable of interpreting signals that have passed a Cat 5 TP cable having a length, from 0 to 100 m. The signal received after 100 m is heavily distorted compared to the 3- level digital signal entering the TP cable as the higher frequency components are much more attenuated than the lower frequency components, this resulting in severe inter symbol modulation The attenuation in dB increases with the square root of the frequency and linearly, also in dB, with the length of the cable or rather, the attenuation in dB/m increases with the square root of the frequency at least for frequencies over 1 MHz and up to well above 100 MHz as illustrated in Fig. 1. In order to handle this, all 100 Mb/s TX receivers or rather their so called PHY-circuits ("physical layer circuits"), are, as already mentioned, equipped with a dynamic equalizer 13, see Fig. 2b, performing an amplification with a weighted opposite frequency response to that of the cable attenuation, see Fig. 3, substantially correcting or compensating for the cable distortion or at least, in the most general case, providing a signal from which a bit stream can be derived that agrees with that issued into the input end at the end equipment of the respective twister pair cable. Several different schemes exist to perform this task and to determine the required amount of compensation/equalization depending on the length of the cable.

A 100 Mb/s TX signal according to the Ethernet standard specification mentioned above is a differential three-level signal coded such that every new 1 in the sequence changes the state of the output in the sequence 0 V, +V, 0 V, -V, 0 V, etc., whereas every new 0 in the sequence does not change the state.. In this way the effective physical band width is reduced by a factor of two. Due to the 4b to 5b coding performed before transmitting to avoid long periods of successive ones or zeroes the signal is transmitted at 125 Mb/s. The highest fundamental physical frequency for any bit sequence without the three level coding would be 62.5 MHz, but due to the three level coding (MLT-3) it is reduced to 31.25 MHz. In order to have a good detectability of the signal, the third harmonic at 93.75 MHz should not be reduced more than a few dB. However, for a 100 m Cat 5 TP cable this frequency has been reduced by approximately 9 dB in relation to the fundamental frequency and it has been reduced by 20 dB in relation to the lowest used frequency. This 20 dB relative reduction is compensated for by the dynamic equalizer at the receiver before bit detection. Of course, also lower frequencies exist since the 4b to 5b coding in combination with the MLT-3 can result in up to four times 8 ns = 32 ns of consecutive +V, 0 V or -V output. The idle signal, used in order to phase lock the clock retrieval and for link integrity purposes in the bit detector 11, also contains symbols up to 12 bits long which also need to be transmitted. This is equivalent to a physical frequency of approximately 2.5 MHz.

A 100 Mb/s optical signal according to the 100BASE-FX Ethernet standard or the Ethernet standard 802.3u using the ANSI X3T9.5 FDDI Physical Layer Dependent (fiber PMD) standard is also 4b to 5b coded resulting in a line bit rate of 125 Mb/s. This is however a usual NRZ signal and thus the highest physical fundamental frequency is 62.5 MHz. In order to have a reasonably detectable signal the third harmonic, in this case 187.5 MHz, can be reduced by more than several dB, but the detectability (the "openness of the eye") would be increased if this harmonic was not reduced by more than 7 dB.

Contrary to the solution for twisted pair cables, an optical receiver has no dynamic equalizer, as an optical fiber does not impose any frequency dependent attenuation. Instead, an optical receiver usually has an automatic gain control (AGC) function that compensates for variable frequency independent losses due to different lengths of optical fiber and different numbers of connections/connectors in the transmission path. SUMMARY OF THE INVENTION

It is an object of the invention to provide an optical link including an optical fiber that can have an extended length.

As said above a direct conversion between Ethernet 100 Mb/s signaling on more than a few meters Cat 5 TP cables and optical fiber would have to use analog diode drivers, more or less replicating the now analog (continuous level due to frequency dependent loss resulting in inter symbol interference) electrical signal on the twisted pair cable as an optical signal onto the fiber and on the other side of the fiber analog optical receivers and cable drivers would have to be used replicating mat optical signal on the fiber as an electrical signal onto the twisted pair cable.

The optical link as described herein provides a solution for allowing significant further distortion of the electrical signal from point where it enters the first converter (converting from electrical to optical signaling) to the point where it exits from the second converter (converting from optical to electrical signaling).

Generally, an optical link includes an optical fiber for transmission of signals and has from an electrical input terminal to an electrical output terminal a frequency dependent loss and phase linearity characteristics that are equal or similar, for at least a particular or predetermined frequency region, to a specific length of an electrical cable such as a twisted pair cable regardless of the actual length of the optical fiber, naturally within certain but wide limits. The optical link may for this purpose include specially selected components or compensating devices such as filters or other devices configured in a suitable way.

Additional objects and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objects and advantages of the invention may be realized and obtained by means of the methods, processes, instrumentalities and combinations particularly pointed out in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS While the novel features of the invention are set forth with particularly in the appended claims, a complete understanding of the invention, both as .to organization and content, and of the above and other features thereof may be gained from and the invention will be better appreciated from a consideration of the following detailed description of non-limiting embodiments presented hereinbelow with reference to the accompanying drawings, in which: - Fig. 1 is a diagram showing graphs of frequency dependent loss for a 30 m and a 60 m long Cat 5 TP cable (one pair),

- Fig. 2a is a schematic of a complete twisted pair link according to the 100BASE-TX standard, - Fig. 2b is a schematic of the receiver portion of end equipment used for terminating communication over a twisted pair cable according to the 100BASE-TX standard, - Fig. 2c is a schematic of a composite communication link for one-directional transmission having an extended total length using an optical portion e.g. according to U.S. patent 4,691,386,

- Fig. 2d is a schematic of a composite communication link for one-directional transmission having an extended total length, where TP end equipment and fiber end equipment are integrated to form media converters, - Figs. 2e and 2f are schematics of a composite communication link for one-directional transmission having an extended total length, the optical portion using direct converters according to the published International patent application WO 02/23771,

- Fig. 2g is a schematic of a composite communication link for one-directional transmission having an extended total length using an ideal analog optical link, - Fig. 2h is similar to Fig. 2g of a composite communication link for one-directional transmission including non-ideal components and possibly a correcting circuit,

- Fig. 2i is a schematic of an optical link where the direct converting converters are directly, without using any length of TP cable, connected to TP end equipment including TP PHY circuitry for detection of signals, - Fig. 3 is similar to Fig. 1 showing graphs of the characteristics of an equalizer function in the receiving portion of end equipment,

- Figs. 4a and 4b are schematics of optical-to-electrical conversion circuits having an AGC- function, - Fig. 5 is a diagram where the graphs A and B represent the frequency characteristics (attenuation) of 30 m and 60 m Cat 5 TP cables, respectively, C and D represent the frequency characteristics of non ideal optical links having a first order low pass characteristic for emulating frequency characteristics within the required frequency range for TP cables of 30 m and 60 m lengths, respectively, and E and F are the required frequency compensations for L_eq = 30 m and L_eq = 60 m, respectively,

- Fig. 6 is similar to Fig. 5 but for a logarithmic frequency scale,

- Fig. 7 is a diagram of a correcting circuit,

- Fig. 8 is a diagram of graphs of the characteristic of the correcting circuit of Fig. 7 and of ideal characteristics and the difference between them, - Fig. 9 is a diagram of graphs of the frequency dependence of a signal having passed a non-ideal optical link and of a signal having passed a TP cable, and the correction characteristics and the resulting characteristics after applying the correction

- Fig. 10 is a graph of the group delay as a function of frequency caused by the correcting circuit of Fig. 7, - Fig. 1 Ia is a schematic of a complete unit for direct electrical-to-optical conversion, and

- Fig. 1 Ib is a schematic of a complete unit for direct optical-to-electrical conversion.

DETAILED DESCRIPTION

End equipment for terminating communication over a twisted pair cable according to the

100BASE-TX standard has built-in facilities for receiving twisted pair signals that are designed to be capable of compensating for the signal distortion appearing after propagating through a Cat

5 TP cable having a length of up to 100 m. This compensating capability is favorably used in the methods and devices described herein.

In the schematic of Fig. 2b conventional components at the end of a twisted pair cable 3 for digital transmission at 100 Mb/s are illustrated. The receiver 9 receives the electrical signals through a connector 31 from one wire pair 5 of the cable. The received signals are input to a dynamic equalizer 13 compensating for the frequency dependent loss of the cable 3 to render an undistorted signal or at least a signal that sufficiently agrees with the electrical signal input to the cable. The term "dynamic" implies that the compensation is adjusted according to Fig. 3 to the correct length of TP cable used. Choosing the correct amount of compensation is done in many ways, e.g. by a selection unit 17, such as either by analyzing the frequency content of the received signal, and therefrom determining the length of the cable 3, or by applying a variable compensation that increases from a low value until the best possible detectability is reached, i.e. until a signal is obtained from which the digital content can be detected in the most reliable way. Hence, the dynamic equalizer 13 may apply a frequency dependent amplification to the received signal. The graphs of Fig. 3 are amplification characteristics that are appropriate for a 30 meter and a 60 meter long Cat 5 TP cable, respectively. Obviously, such amplification characteristics are substantially .the opposite to the characteristic of the frequency dependent attenuation of the used cable length. The signal from the dynamic equalizer 13 is provided to a bit detector 15 from which the obtained digital signal is output to other devices, not shown, for further processing or use.

In the case where transmission over a twisted pair cable is desired and the physical transmission length exceeds 100 m, a portion of such a transmission link can be replaced by an optical link including optical fibers. In such a combined transmission link ideal, directly converting analog converters 33 can be used on either side of a piece of optical fiber 35, as illustrated in Fig. 2g. The addition of the optical link segment is, using an ideal analog optical link, not seen from the electrical signal point of view and hence the total added length of the two TP cables, only one wire pair 5', 5" thereof shown in Fig. 2g, is maximally 100 m, this length being arbitrarily divided between the two TP cable parts. The system would however see a further delay which is of no concern using full duplex which is anyway required when electrical or optical links having a delay more than that of a 100 m TP Cat 5 cable are used.

However, ideal, directly converting analog converters may be costly and/or they may require a relatively high amount of supplied electrical power for driving them.

In the case where an optical fiber 33 is used as part of a communication link based on twisted pair technology, there are only in rare cases a need for long TP cables at either end of the fiber part as the major length of the communication distance required advantageously is met by the optical fiber. A typical installation based on TP cables may include a switch, not shown, that is connected to a converter pair via a patch cable of the twisted pair type having a length up to 2 m. The optical fiber is at one end connected to such a converter pair and at the other end another converter pair is attached which is connected to user equipment via a TP cable having a length less than 30 m. Therefore, if conventional converters 37 including complete end equipment for communication according to the 100BASE-TX standard is used at each end of the two TP cables, or correspondingly, wire pairs 5', 5", compare Fig. 2d in this typical example, 98 and 70 m at each side respectively of the compensation capacity for TP frequency dependent losses of the receiver parts of the respective end equipment is not used.

Instead, if ideal, directly converting analog converters 33 are used, see Fig. 2g, 68 m of the compensation capacity of the receiver parts of the respective end equipment is not used.

Therefore, in this specific case an optical analog link, in which the electrical signal is directly converted to optical form and back as described above and where the converters themselves have no conventional receivers or transmitters for communication according to the 100BASE-TX standard, between the TP cable ends, in this example, does not require a frequency dependent attenuation better than that of a 68 m long TP cable, i.e. it can have a frequency dependent attenuation smaller than or equal to that of a 68 m long TP cable and the combined transmission -link will still work. Thus, the optical link portion can have the same frequency and phase linearity characteristics as a 68 m long Cat 5 TP cable and will hence behave as a 68 m long Cat 5 TP cable in the total transmission link, and when connected to a total of 32 m long Cat 5 TP cable arbitrarily divided between the two ends of the optical portion, the total transmission link will work as a 100 m long Cat 5 TP cable. Similarly, an optical link having the same frequency and phase linearity characteristics as a 30 m long Cat 5 TP cable will work as a 30 m long Cat 5 TP cable allowing for up to a total of 70 m Cat 5 TP cable to be arbitrarily divided between the two ends, and so on. As the frequency dependent losses (in dB) per length segment are additive it does not matter how the lengths of the Cat 5 TP cables are divided between the ends as long as the total Cat 5 TP cable length does not exceed 100 m minus the equivalent Cat 5 length, "L_eq", for the optical link portion. However, this is true only if the optical link portion does not contribute significant amounts of noise and other types of distortion not caused by TP cables.

Also, in order that the optical link of arbitrary length, within some limits that may be rather wide, will simulate a Cat 5 TP cable of a specific length, the output amplitude within the used physical bandwidth should be independent of the optical (frequency independent) loss/fiber length. This can be achieved in two different ways, or as a combination of these two different ways, depending on the sensitivity to the absolute level of (the lowest frequency component of) the signal of the dynamic equalizer in the end equipment. In the first case where the receiver is not critically sensitive to the absolute level of the signal, an AGC function keeping the amplitude of (the lowest frequency component of) the output signal into the TP cable to a set value within the specified input range of the TP receiver, usually an amplitude in the range of 0.5 - 1 V, independent of the fiber loss is sufficient. This is the case for many TP Ethernet receivers of today. In the second case, where the receiver is critically sensitive to the absolute amplitude of (the lowest frequency component of) the signal, the AGC would not be dependent on the amplitude of the signal (the light modulation) but on the average optical signal, i.e. the DC component of the light entering the optical receiver, compensating for the light loss of the fiber connection, resulting in a total signal (AC) amplification (or rather loss in this case) of (the lowest frequency component of) the signal equal to that of a cable of the equivalent length of the fiber link independent of the loss of the fiber (within the AGC limits), see Figs. 4a and 4b.

The case where a non-ideal optical link is used is illustrated in Fig. 2h, including e.g. non- ideal analog converters 39 for converting between electrical and optical signals, these converters e.g. having a reduced band-width, compare the discussion above. A compensator 41 may be included at the end of the optical link to restore the signal to be sufficiently similar to a TP signal. FL_max designates the maximum fiber length for which the AGC- of the optical receivers can compensate, see the discussion herein, and/or where other phenomena like path length or wave length dispersion limits the maximum fiber length. L_max is the maximum length of TP cable for which the equalizers in the TP receivers of the end equipment for 100Base-TX communication can compensate and L_eq is the length of the TP cable that the optical link emulates. A total of up to L_max - L_eq can be arbitrarily divided between the left and the right TP cables. The end equipment TP receivers are now aware of the inclusion of the optical link, but interprets this as an added length (equal to L_eq) of TP cable for which the equalizers can compensate.

What does this imply in comparison with a conventional optical link solution, as illustrated in Fig. 2d? hi this case both the transmitter and the receiver should have an approx. -7 dB bandwidth of 187 MHz. A link as described herein, however, corresponding to a 60 m long Cat 5 TP cable needs a 95 MHz transmission roughly 6 dB below the 32 MHz signal level which can be 5.5 dB below the 2.55 MHz (corresponding to the longest symbol of the idle signal) signal level, i.e. the signal at 95 MHz can be roughly 11.5 dB attenuated relative the lowest used frequency. Assuming the over all electro-to-opto-to-electro conversion to have a frequency response of a first order low pass filter (which is common for single amplifiers), the above implies allowing for a -3 dB point at only approx. 33 MHz. An optical link of this kind designed for direct optical communication between two end equipments, i.e. without using any TP cables, see Fig. 2i, could use the full compensating ability of the equalizers of the TX circuit, i. e. the equivalent length would be 100 m. hi this case the signal at 95 MHz could be approx. 20 dB attenuated relative to the lowest used frequency and the corresponding -3 dB point of the amplifier with a first order low pass characteristic as low as approx. 20 MHz. However, a frequency and possibly phase characteristics compensation network as generally seen in e.g. Fig. 1 Ib is required to more or less equate the frequency and phase linearity characteristic to that of a Cat 5 TP cable between 2.5 MHz and 95 MHz, as for this kind of signal, there is substantially no necessary information above 95 MHz. Therefore this compensation network only has to work between the lowest used frequency, in this case approx. 2.5 MHz, and the highest used frequency, 95 MHz. This can in principle be accomplished using simple passive components. Depending on the specific properties of different PHY circuits, varying degrees of non perfect frequency and phase linearity characteristics the correction/simulation will still work. The differences between the frequency characteristics of a cable, an amplifier with a first order low pass characteristic and the compensation characteristics to equate the first order low pass characteristics to the cable characteristics within the required frequency range is shown in Fig. 5, using a linear frequency scale, and Fig. 6, using a logarithmic frequency scale.

One example of a compensating circuit working as a simple correction filter is seen in Fig. 7. It has been designed to convert a signal that has passed a non-ideal optical link of limited bandwidth working as a first order low pass filter, which e.g. can be assumed to have the same attenuation at 100 MHz as a 60 m Cat 5 TP cable, to become more similar, over the whole frequency range used, to a signal which has passed a real Cat 5 TP cable of the equivalent length. For this particular example the component values can be Rl = 500 Ohm, R2 = 900 Ohm, C = 30 pF and L = 4.7 μH. A graph of the characteristic of the filter is shown in Fig. 8.

As seen in Fig. 8 the characteristic A of the correcting filter of Fig. 7 is close to the characteristic B of an ideal correction circuit line. The deviation from the ideal characteristic seen at C is only significant for frequencies lower than those used according to the 100BASE-TX standard and for third harmonics where for the highest at 93.75 MHz a reduction of a few dB is fully acceptable. In Fig. 9 other graphs of the correcting operation are shown. The correction filter according to Fig. 7 has the characteristic seen at A. After it has been applied to signals that have passed an uncorrected, non-ideal optical link which should emulate a 60 m Cat 5 TP cable and which is assumed to have the characteristic seen at B, the resulting signals has an overall characteristic C that is very similar to the characteristic D of a 60 m Cat 5 TP cable.

The case of a general uncorrected link characteristic being that of a first order low pass filter is here only given as a simplified example to show the principle of converting a particular characteristic to that of a specific length cable. As is seen in Figs. 11a and l ib showing the complete converter units for the direct conversion between electrical and optical signals, there are at least three amplifiers and some other components, each having their own frequency characteristic, for signals passing the optical link. Thus, the total uncorrected link characteristics are more complicated than those of a first order low pass filter. Therefore, the specific correction filter needed will be designed after having determined the virgin characteristics of the optical link to be corrected. Because the optical link consists of at least three amplifiers, the correction filter does not have to be connected at one single location. Instead, hi combination with each transimpedance, transconductance or voltage amplification stage, partial corrections may be performed either directly preceding or directly succeeding or as part of the local feedback loop of each amplifier utilizing their respective relatively high input impedance and/or low output impedance for easy filter design without having to use extra amplifiers. For example if the voltage amplifier in the AGC and cable driver stage has a too low -3 dB point, this may be compensated by modifying the negative feed back of that stage to include a low pass characteristic with a properly chosen -3 dB point. By using proper choices of different frequency characteristics of all the amplification stages only using passive components, a total frequency response very similar to the frequency characteristics of a Cat 5 TP cable of a specific length is obtainable without using a dedicated correction filter connected at a single position. Using this method an optical link having a frequency characteristics close to that of a -16 m long Cat 5 cable with a group delay variation of less than +/-1.5 ns has been obtained. When designing such a filter/filters there is one further issue to consider apart from the frequency characteristics. This is the phase linearity or frequency dependent group delay as a further requirement for the described optical link to be sufficiently similar to a TP cable in all signal aspects. The corrected link including the correction filter must be sufficiently phase linear or have a sufficiently constant, frequency independent group delay within the frequency range used for the signaling system. However, the value of this constant delay is arbitrary as long as full duplex is used. Ideally, group delay variation should be less than a fraction of the period of the highest frequency used for all frequencies. Not fulfilling this will cause signal distortion which would still allow correct detection but would "close the eye" in the eye diagram causing an increased bit error ratio (BER). The amplifiers illustrated in the generic drawing usually have a frequency characteristics similar to that of a first order low pass filter but the phase characteristics is not that of a first order low pass filter. Instead the phase characteristic is almost linear, i.e. there is a substantially constant group delay. hi Fig. 10 the group delay resulting from applying the correction filter according to Fig. 7 is shown. Assuming the rest of the link to be substantially phase linear/having a constant group delay, the total group delay deviation is caused by the correction filter only. Below 2.5 MHz there is no signal for 100BASE-TX and at 2.5 MHz the period time equals 400 ns where thus the deviation, ~8 ns is a very small fraction of the period time. Actual information signals span from ~12 to ~32 MHz plus harmonics. For every frequency, delay deviation is very much smaller than the respective period. Still the remaining fault would cause some distortion of the signal. By relaxing the demands on perfect frequency characteristics match, a better phase linearity could be obtained possibly resulting in improved bit error ratio. Alternatively, a more elaborate filter would have to be used.

In Figs. 11a and lib possible designs of directly converting, cable emulating media converters are shown.

In the electrical to optical conversion seen in Fig. 11a the signal arriving from the TP cable enters a simple connector and is fed to the primary winding of an isolation transformer. The secondary winding of the transformer is connected, to ground at one end and at the other end to a voltage divider having a total resistance equal to the matching impedance of the cable (if the ratio of the transformer is 1:1, else appropriate change will be made such that the cable sees a matching impedance) thereby accomplishing a differential to single ended conversion. An appropriate fraction of the signal is fed forward via a capacitor, performing DC blockage, and added to a high impedance DC bias network, the DC voltage modulated by the signal from the capacitor fed to a transimpedance amplifier that converts the voltage to a corresponding electrical current flowing through a LED or a laser diode (LD). Even though the current is strictly proportional to the voltage, the light output is not strictly proportional to the current, this resulting in distortion. In order to reduce the distortion the modulation is kept at a low fraction of the average current. The light generated by the LED or LD is coupled to the fiber by an optical connector. In the optical to electrical conversion seen in Fig. lib the light output from the fiber is coupled to a photo diode by an optical connector. The photo diode is back biased. The resulting current which is a replica of the light intensity is converted by a transimpedance amplifier to a proportional voltage. The transimpedance amplifier may or may not have internal automatic gain control (AGC) which is governed by the average light level. This solution implies that varying the modulation of the light in the transmitter does not involve this AGC as varying the modulation does not change the average value. Changing the loss of the fiber, however, will change the average value and thus the AGC will be active to compensate for it. If this AGC function would work completely, there would be no need for any further AGC functionality after the transimpedance amplifier. If, however, it is not complete but only partial or if there is no AGC in the transimpedance amplifier, further AGC action is required to keep the signal level, regardless of varying light loss, within the acceptance levels of the TP receiver in the end equipment following the connected TP cable. This AGC cannot work on the average signal level as the signal now is not DC but AC coupled. Instead, it corrects the amplification in a variable gain amplifier by making the peak value constant determined by a peak detector, hi order to make the frequency characteristics be sufficiently similar to that of a specific length of cable, a correction filter may have to be connected between the transimpedance amplifier and the variable gain amplifier, hi principle this correction filter could be connected anywhere in the chain of components but this position is advantageous not loading the filter due to the high input impedance of the variable gain amplifier. The signal passes the peak detector, and enters a transformer for single ended to differential conversion at the same time as impedance matching to the TP cable can be achieved by a proper ratio selection of the transformer, matching the output impedance of the variable gain amplifier to that of the TP cable connected through a TP connector. It is also in some cases possible to create a sufficient correction without using a specific correction circuit as in Fig. 2h. Then the correction can be made e.g. by carefully selecting the components in the optical link, utilizing the non-ideal characteristics of such specially selected components. The correction can also be made by e.g. adjusting the frequency responses of the amplifiers included in the optical link portion, see Figs. 11a and 1 Ib. As normally the feedback of a transimpedance amplifier is not accessible this adjustment will only involve the transconductance amplifier of the optical transmitter and the AGC amplifier of the receiver.

As the power consumption of amplifiers relates strongly to their -3 dB bandwidth for the same amplification and noise level and type of IC-process, the digital, part-fiber link as described herein needs significantly less power than the amplifiers and drivers in the conventional FX equipment. As the direct conversion as described herein does not use full detection, decoding and re-coding etc., further significantly less power is used as compared to conventional media converters that include TP terminating equipment and protocol conversion.

In conclusion, the part-fiber digital link as described herein implies using the ability of the PHY circuits of all TX receivers to compensate for frequency dependent losses of Cat 5 TP cables using dynamic equalizers. By not allowing the full 100 m Cat 5 TP cable lengths at each side of the optical link but instead only a total of 100 m minus the equivalent length (L_eq) of the optical link, the remaining compensation ability of the TX receivers is used to compensate for the non ideal optical link provided that the loss and phase linearity characteristics of the optical link is sufficiently close to that of a Cat 5 TP cable of that equivalent length and that the optical link does not add more than insignificant amounts of noise and other types of distortion not caused by TP cables to the signal. In such a design the amplifiers and drivers involved will use much less power compared to a conventional FX link. In the vast majority of all optical fiber link installations this restricted Cat 5 TP cable length is not of any concern.

The examples above all relate to Ethernet signaling according to the Ethernet standard 802.3u using MLT-3 multi-level transmission according to the ANSI copper FDDI Physical

Layer Dependent sublayer technology, also called as CDDI (and compared to Ethernet standard 802.3u using the ANSI X3T9.5 FDDI Physical Layer Dependent (fiber PMD) standard for fiber communication). The solution would also work for any other system/communication standard designed for metallic wires including dynamic (or fixed) equalizers compensating for frequency dependent losses of varying (or fixed) lengths of metallic wires. In the fixed length/fixed equalization cases the kind of optical link as described herein would still replace a specific length of metallic wire and the remaining, lengths of metallic wire could still be chosen arbitrarily on either side of the optical link provided that the total length, the sum of the metallic cable parts and the equivalent length (L_eq), is equal to that length for which the fixed equalizers are designed to compensate.

While specific embodiments of the invention have been illustrated and described herein, it is realized that numerous other embodiments may be envisaged and that numerous additional advantages, modifications and changes will readily occur to those skilled in the art without departing from the spirit and scope of the invention. Therefore, the invention in its broader aspects is not limited to the specific details, representative devices and illustrated examples shown and described herein. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents. It is therefore to be understood that the appended claims are intended to cover all such modifications and changes as fall within a true spirit and scope of the invention. Numerous other embodiments may be envisaged without departing from the spirit and scope of the invention.

Claims

1. An optical link including an optical fiber for transmission of signals that from an electrical input terminal to an electrical output terminal has a frequency dependent loss and phase linearity characteristics that are equal or similar, for at least a particular or predetermined frequency

5 region, to a specific length of an electrical cable, in particular a twisted pair cable regardless of the actual length of the optical fiber, within certain limits.

2. An optical link according to claim 1, wherein the frequency dependent loss and phase linearity characteristics are equal or similar to a specific length of an electrical cable .designed for transmission of signals according to a communication standard, in particular the standard

10 100BASE-TX.

3. An optical link according to claim 2, wherein said at least particular or predetermined frequency region is the frequency region used for said communication standard.

4. An optical link according to claim 2, wherein the cable for which the optical link has similar characteristics is of the same type as that cable which is attached to the electrical

15 terminals of to the optical link and which is specified for that communication standard.

5. An optical link according to claim 1, wherein the specific length is chosen so as to be sufficiently smaller than the maximum allowable length for electrical cable in a segment of a communication system where it is used to allow for sufficient lengths of attaching electrical cables for connecting to end equipment.

20 6. An optical link according to claim 5, wherein the specific length is chosen so as to be equal to or smaller than the maximum allowable length for electrical cable in a segment of a communication system where it is used for application where electrical cables are not used, i.e. where the optical link is manifest as optical ports directly interfacing the electronic circuits designed for driving/receiving from electrical cables.

25 7. A signal transmission link for transferring an electric information signal in a medium from an input end to an output end, wherein

- the medium includes at least one electrical link part and an optical fiber link part, the at least one electrical link part introducing in the electric information signal distortions characteristic of the at least one electrical link part, and

30 - the optical fiber link part having at its input and output ends converters converting signals between electrical and optical form, the components of the optical fiber link part being set and/or selected for modifying signals passing through the optical fiber link so that, at the output end, an electric information signal received at the input end has been distorted in substantially the same way as an electric information signal that has passed an electrical link of the same kind as said at least one electrical link part included in the medium.

8. A signal transmission link according to claim 7, wherein the components of the optical fiber link part are set and/or selected to modify an electric information signal to have at the output end a distortion substantially corresponding to the distortion produced when passing an electrical link of the same kind as said at least one electrical link part and having a predetermined length or a length derived from or dependent on the length of said at least one electrical link part.

9. A signal transmission link according to claim 7, wherein said at least one electrical link part included in the medium introduces frequency dependent amplitude and/or phase linearity distortion and the components of the optical fiber link part are set and/or selected to make the optical part have substantially the same kind of distortion.

10. A signal transmission link according to claim 7, wherein the optical fiber link part includes an amplifier function for maintaining the amplitude of the lowest frequency components of an electric information signal that has passed the optical fiber link part at a predetermined level or at a level corresponding to or substantially equal to the amplitude of the lowest frequency components of the electric information signal at the input end of the optical fiber link part or at such a level but reduced by an amount equal to that of an electrical link having an equivalent length.

11. A signal transmission link according to claim 7, wherein said components of the optical fiber link part that are set and/or selected include a signal shaping unit modifying signals passing through the optical fiber link.

12. A signal transmission link according to claim 7, wherein said components of the optical fiber link part that are set and/or selected include electrical amplifiers, the frequency response and/or gain of which are set, such as by selecting values of the passive components, in particular resistors.

13. A signal transmission link according to claim 7 for transferring digital or binary signals, wherein the converters include circuits for a substantially direct conversion between electrical and optical signals, so that the instantaneous signal level in all parts of an electric digitals information signal substantially corresponds, but not necessarily linearly, to a respective instantaneous signal level or light intensity level in the converted optical signal and vice versa.