BACKGROUND OF THE INVENTION TECHNICAL FIELD OF THE INVENTION This invention relates to semiconductor photo transceiver arrays with silicon circuitry permitting selectable routing of detectors to emitters, and to methods for routing and mapping data channels in opto-electronic semiconductor devices : and in particular to methods for routing. mapping, rerouting and re-mapping internal and optical inter-connections in mufti-array semiconductor devices and systems.

BACKGROUND ART The technology associated with electronics has evolved extremely rapidly over the last 40 years. Computers and related peripheral equipment. satellite and communication systems are becoming ever more sophisticated and powerful. However, data transfer into and out of processors remains a gating capability. The combination of increased parallelism and optics is the focus of optical interconnect technology.

One approach to optical interconnect technology uses so-called flip-chip techniques where the advantages of silicon process technology are combined with the optical properties of IIl-V semiconductor materials. Results to date indicate that this combination will lead to orders of magnitude increases in data transfer rates. These successes suggest that there will be enormous benefits to resolving the remaining issues associated with this technology.

Prior art Fig. I shows an example of so-called flip-chip technology that has been developed to exploit the advantages of CMOS substrates for some aspects of data transfer. and the optical advantages oí III-V semiconductors. In this technology, emitter-detector arrays are fabricated separately from a CMOS substrate. The emitter-detector arrays are then inverted. aligned with the CMOS substrates, and secured in place using solder balls or bumps to form electrical contacts with modest mechanical adhesion properties, and using epoxy to rigidly mount the emitter-detector array to the CMOS chip. In Fig. 1 for clarity only a single pair of fiber optic cables is shown.

The separate arrays of detectors and emitters are interconnected by some suitable light carrying media. such as bundles of liber optic cables.'fhe alignment between the ends the these bundles, or other media. and the arrays causes errors in mapping of the elements of one array with the elements of another array.

With whatever method one uses to connect one array of transmitters to an array of receivers. or one array of transceivers to another array of transceivers, alignment is an issue.

In other words, what is needed is a way to determine that the arrays are aligned to produce a useful device.

Assembly processes are imperfect, so there will be some misaiignment. even if the effect on the performance of the system is negligible, In a production environment the effect of small amounts of misangnment wiH decrease yietd. which costs money in terms of lost revenue.

Prior art Fig. 2 shows a schematic representation of the effect of misaiignment. The darker squares represent liglit from tibers coupled to the transceivers in a remote array. and the tighter squares represent the transceivers in the present array. In Fig. 2A there is some degradation of coupling, but a one-to-one correspondence between transceivers remains.

However, in Fig. 2B. some of the light fibers overlap more than one transceiver, while others fail to overlap any transceivers. In general, the types of misalignment include rotation. linear displacement. scaling, and a combination of these three basic types of misalignment. What is

needed is a way to increase the tolerance to slight misalignment of transceiver arrays when coupling arrays.

Connecting nodes requires the use of fiber bundles or some other point-to-point connection means. However, there will still be some amount of misalignment. which can cause system communication failures. What is needed is a way to increase the tolerance to slight misalignment of transceiver arrays when coupling nodes.

There would be only minimal utility to opto-electronic devices such as the ones shown in Fig. 1. if one array could be connected on only one other array. What is needed is a way to couple more than one array together.

If there is some misalignment between two transceivers, there will be an effect on other transceivers in other nodes. What is needed is a way to determine how the transceiver arrays are aligned in a ring or other network architecture so that channels can be assigned and correct data transfers between nodes in a network can occur.

In some applications, security of data transfer is critical. It is therefore very important to be able to ensure that data transfer is indeed secure. Therefore, what is needed is a way to increase the security with which data is transmitted from node to node.

Though small numbers of transceiver arrays consume modest amounts of power, large arrays consume copious quantities of power, and in so doing generate a lot of heat.

Removing such large amounts of thermal energy is very challenging, but if the thermal load is not controlled the device may degrade prematurely. What is needed is a way to reduce the power consumed by a node.

Prior art that may provide useful context for the reader, includes the following : US5761350."Method and apparatus for providing a seamless electrical/optical multi- layer micro-opto-electro-mechanical system assembly", illustrates how opto-electronic

interconnections can provide a practical solution to communications bottleneck problems when combining a multitude of information processing units to perform a function. As research activities progress in the field of serial or parallel board-to-board and module-to- module interconnections. some of the research focus has shifted to smaller physical dimensions. such as intra-module interconnections, which combines opto-etectronic interconnections. multi-chip module packaging. and micro-electromechanical systems (MEMS) technologies at the module level. Ihis disclosul-e prcsenis integrated optical input/output (I/O) couplers on multi-chip modules (MCMs) using micro-machined S1IiCcIn mirrors that are used with opto-c : lectronic multi-chip modules (OE-MC'Ms). It uses microstructures that integrate optical wave guide networks, multi-layer electrical transmission line networks, micro-machined silicon mirrors, and C4-bonded photonic devices into a single structure. Using both sides of the silicon wafers, multiple metal layers and optical waveguide layers are fabricated lor all types of metal or optical waveguide materials. The input/output coupling arrangement utilizes a combination of micro-machined silicon mirrors and through- holes across OE-NICI\ integrated together into a single paclSage.

US5625734."Opto-etectronic interconnect device and method of making", describes a waveguide having a core region and a cladding region. A portion of the cladding region forms a first surface and portions of both the core region and the cladding region form an end surface. There is an insulative flexible substrate having an e ! ectricai) y conductive tracing with a first portion and a second portion. wherein the first portion of the insuiative Hexibte substrate is mounted on the end surface of the waveguide.

US5428704,"Opto-electronic interface and method of making", describes an interconnect substrate having a surface with electrical tracings. There is a photonic device with a working portion and a contact electrically coupled to one of the electrical tracings disposed on the interconnect substrate. A molded optical portion encapsulates the photonic device. forming a surface. The surface passes tight between the photonic device and an optical fiber. Alignment of the optical fiber is achieved by an alignment apparatus that is formed in the molded optical portion.

US5420954,"Parallel optical interconnect". describes an optical interconnect that couples multiple optical fibers to an array of opto-electronic devices. The interconnect includes a multiple optical fiber connector and an opto-electronic board. The multiple fiber connector can be mechanicaity attached to or detached from the board.

US5857042,"Optical interconnection arrangements". describes an optical interconnection arrangement consisting of a plurality of parallel optical interconnection channels. In each channel. there are an optical source. an optical receiver, a first lens and a second tens. The first tens conveys tight from the source to the second tens. and the second lens rez'focuses the light at the receiver, Each source and the associated first tens are offset one relative to the other by a predetermined distance in a direction transverse to an optical axis of the first lens. The corresponding receiver and the associated second lens are offset one relative to the other by the same distance but in the opposite direction to the offset between the source and first lens. Each offset is equal and opposite to the corresponding offset in an adjacent channel. With such an arrangement. if a leakage portion of a tight beam from the first lens in one channel impinges on the second lens in an adjacent channel. the leakage portion NA, Ill be refocused at a position which is spaced from the receiver of the adjacent channel.

In one embodiment. in each channel, the first lens and the second lens share a common optical axis and the source and receiver are offset relative to their common optical axis. In an alternative embodiment, the source and the receiver share a common optical axis and the first lens and the second tens are ofl'set relative to their common optical axis. Such optical interconnection arrangements are tolerant of translational or rotational misalignments between the sources and associated first tenses on the one hand and the receivers and associated second lenses on the other hand, which is of particular advantage for free space optical interconnects or couplers.

US5748818."Massive parallel optical interconnect system", describes a massive parallel (MP) connector which includes a fiber optic connector having a polymer ferrule having multiple fibers mounted in V-grooves of the ferrule and beveled edges of the ferrule

providing for alignment of the ferrule when the MP fiber optic connector is mated to a receptacle with an alignment assembly and an alignment member mounted within the alignment assembly to provide for precision alignment. A receptacle assembly has a first receptacle half for receiving a fiber optic connector of a first form factor and a second receptacle half for receiving a fiber optic connector of a second form factor.

US563 1988."Parallel optical interconnect". describes an optical interconnect that couples multiple optical fibers to an array of opto-electronic devices. The interconnect includes a multiple optical fiber connector and an opto-electronic board. The multiple fiber connector can be mechanically attached to or detached from the board. The optical interconnect consists of a multiple optical fiber connector with a holder with a first planar surface. a plurality of optical fibers attached to the holder. each fiber having a first end abutting the first surface so as to expose the first end for receiving or transmitting optical radiation. The first ends of the fibers form a liber array having a list pattern. There is a guiding means disposed in the holder at predetermined positions with respect to the fiber array : and an opto-electronic board consisting of an opto-electronic device array monolithically formed on a semiconductor chip with the same pattern as the first pattern of the fiber array, with aligning means formed on the chip. The aligning means are disposed at substantially the same predetermined positions with respect to the array ot opto-eiectronic devices as the positions of the guiding means rotative to the fiber array. and the aligning means receives the guiding means so as to mechanically align the opto-electronic device arrav with the optical fiber array. whereby each opto-electronic device is aligned to an optical fiber. so that the opto-electronic device emits optical radiation into the fiber array or receives optical radiation from the fiber array.

SUMMARY OF THE INVENTION The minima ! device fundamental to the invention consists of a semiconductor photo- array on a silicon substrate with special enhancements. There is fabricated within the silicon circuitry in the substrate the capability to sense the signal strength received at each detector. and to make it available for interpretation in software for mapping of viable optical connections. The circuitry further provides for control through software loi-selectively switching and testing the routing or connections between detectors and emitters within a transceiver array or between the emitters and detectors of two. optical connected arrays. The circuitry may be enhanced, further extending the functionatity and benefits of the invention. by incorporating software control inputs for adjusting the sensitivity or gain of each detector. and for selectively switching power off and on to each of the emitters and detectors.

For the purpose of this disclosure. terms other than"mapping". such as configuring or switching, relate to the act of creating or altering all or part of a signal circuit path or channel.

The term"mapping,"means generally to confirm what is the fully defined present set of connected points or links of each circuit path or channel of a single or multi-channel system.

The term"map"or"mapping"or"re-map"or"re-mappin,"where the context admits. may include a reconliguration or switching of one or more circuit paths within the system of interest, along with a new confirmation or mapping of what is the resulting new set of circuit paths or channels.

The methodology of the invention applicable to a single photo-transceiver array. facilitated by the software and circuitry described above, is the process of routing and rerouting the data transfer channels within a photo transceiver array to overcome device or system defects or failures and provide special operational flexibility. as will be better appreciated when the method is practiced at the next level of complexity.

The methodology of the invention, applied on a larger scale, provides for routing and mapping the data channels of a multi-array opto-electronic semiconductor device, including the intra-nodal connections and the inter-nodal connections that define each data channel or

path through the device. for one or a multitude of purposes, including : adaptation to faulty. redundant, or missing connection paths resulting from lateral or rotational misalignment of arrays ; isolation of detective emitters. detectors, and fiber optic strands : power and heat management : and time-based multiplexing of channel routing for security purposes.

The necessary software and silicon circuitry are initially employed to map a first order set of possible data channels between adjacent nodes and within nodes in the device. selecting out defective emitters. detectors or faulty fiber optic connections, and allocating from among the remaining possibilities a rule-based assignment of connection paths, forming a suitable operational channel set or map for data transfer. The rute may take into account any of several variables of interest in the specific application such as position, signal strength at the detector. plurality of detectors seeing a given emitter, non-useful emitters. architecture of the device or system, and the like.

The invention generally relies on a technique called over sampling, assuming there will be a surplus of detectors available for mapping of optical connections to accommodate misalignment of emitters to detectors and variable topologies, and for mapping and re- mapping of inter and intra-nodal connections for security, power management, and fault correction purposes.

It is an objective of tile invention to provide a process for automatically configuring point-to-point connections between semiconductor photo arrays for increased alignment tolerance. It is a further objective to provide a process for automatically configuring point-to- point connections between the detectors and emitters of each semiconductor photo- transceiver array for better fault tolerance and security of data.

It is yet anther objective to provide a way of operating such devices at less than maximum power consumption during routine operation. while providing for increased bandwidth during peak data-transfer demand times. It is also an objective of the invention to provide a process of automatically configuring data channel connections within the device or system so as to significantly increase fan out of an emitter signal to facilitate the mapping of

star and ring topology devices. and such other complex topologies as may become useful in the development of such devices. for all the same advantages.

An additional objective of the invention is to provide for pixel re-mapping for increased reliability and security of data transmission. for re-routing if a fiber breaks, to improve the reliability and security of transmitted data. and for power reduction and bandwidth management across the global connection map.

Another additional objective is to provide for a useful distribution. such on a proximity-based or generally uniform distribution basis. when using a limited data channel set of emitter and detector allocations within the available array space. for more uniform power and heat distribution across the substrate. A still further objective is to provide means for allowing only the minimum amount of power needed to support the limited channel set. instead of powering connections that are not aligned uscfully. or are not presently needed in the device topotogy. Another objective is to use smaller amounts ol power durillg norlllal operations by turning some useful emitter-detector pairs off. but providing higher throughput during peak demand by turning all useful emitter-detector pairs on for short periods of time.

Other advantages and objectives will be apparent to those skilled in the art. based on the drawings, description of preferred embodiments, claims. and abstract that follow.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a simplified representation of a prior art photo transceiver array on a silicon substrate. with alternate columns of photo emitters and detectors. the array being connected to a further array or device by a bundled liber optic cable transceiver pair.

Fig. 2 is a simplified representation ofmisatignment errors between two sets of photo-arrays common in the prior art : the Fig. 2A set Hiustrating insufficient overlap between emitter and detector. resulting in degraded functionality of the emitter/detector pair. and the Fig. 2B set illustrating misalignment where ad, jacent emitters partially overlap a common detector. and misalignment where some detectors have no overlap an (l no lilnctionalilv.

Fig. 3 is a simplified representation of an inter-noda) connection pair between two arrays or nodes via a fiber optic bund) e : Fig. 3A being the schematic representation, and Fig. 3B being a visual depiction.

Fig. 4 is a simplified sequential representation of the mapping of an emitter array to a detector array. to establish a limited connection set or map. the sequence running from Fig.

4A-Fig. 4D. with the final mapping configuration shown in Fig. 4E.

Fig. 5 is a simplified representation of emitter to mufti-detector mapping from a single emitter signal. enabling a fan-out capability or star-type architecture of a multi-array configuration.

Fig. 6 is a simplified schematic representation of a ring-type array architecture device, to which the methodology of the invention is also applicable.

Fig. 7 is a simplified sequential representation of the mapping and re-mapping of the detector to emitter intra-nodal connections of a 5 node chain of optical transceivers. illustrating a first full connection map and a subsequent reconfiguration of the connection

map where Nodes 2 and 3 have been functionally rewired with CMOS silicon substrate circuitry and software to alter the original detector-emitter mapping.

Fig. 8 is a simplified sequential representation of emitter array to detector array re-mapping of the inter-nodal channel map of Fig. 8A, to replace a faulty fiber optic strand from emitter 1 to detector 2, configuring out the faulty connection and allocating a previously unused connection to redefine the channel map as shown in Fig. 8B.

DESCRIPTION OF THE PREFERRED EMBODIMENT The invention is susceptible of many embodiments, both as to apparatus and method.

The novel and fundamental devices of the invention are fabricated by techniques known to those skilled in the art. based upon introducing the additional capabilities of the invention described and illustrated through out this disclosure, at the design stage. The modifications to the circuitry in the silicon substrate provide for (a) the sensing of photo-detector signal levels and testing of detector to emitter electrical connections. (b) the necessary control inputs. and (c) the software interface necessary to enable a control program to effectuate the methods of the invention. The controllable silicon circuit of the photo-array. prcsumab) y CMOS although the invention is not thereto limited, iilcludes that necessary to adjust each detector's gain. switch power on and off to any of the emitters and detectors, and to reroute or multiplex the routing of detector outputs among two or more emitters.

Providing the necessary circuitry to rewire. connect or route any detector to /n' emitter in the array, while within the scope of the invention. becomes exponentially more challenging as the density and size of the array increases. For this reason. the rerouting capability may be limited, such as on a proximity basis as to a few additional emitters adjacent to the default primary emitter assigned by design to a given detector, or on a row or column basis, dependillg on tlle overall layout of the chip.'The controlling software for effectuating the steps of the invention through the controllable circuitry may be integrated with the circuitry. or connected from an external source. or divided. The division between on-board logic and computing power, and an external software program. is discretionary to the designer and will likely vary with the specifics of the application and tile limits of the related technologies.

A preferred embodiment method is described for automatically configuring point-to- point connections between two arrays, utilizing the technique ofover-sampiing. as disclosed here and in Figs. 3A and 3B.

Figs. 3A and 3B show a simplified diagrammatic presentation of the invention. The structure of Fig. 3B consists of two arrays of transceivers. 1 (1 smd 20. comprising flip-chip mounted transmitters or receivers on a silicon substrate with CMOS peripheral circuitry. along with a communication means connecting the transmitters 12 on array 10 to detectors 22 on the other array, and the transmitters 24 on array 20 to detectors 14 on the first array. Fig.

3A is a functionally equivalent block diagram of the structure of Fig. 3B. For simpticity. on ! y a single fiber pair 30 is shown. The communication means could be a fiber optic bundle or other point-to-point communication means. Ajthough the figure shows separate arrays of emitters and detectors. the invention applies to intcr-digitated emitters and detectors as well.

For simplicity. only one emitter in an array on each substrate is shown as being connected to a detector in an array on the other substrate. Of course each emitter in each array is nominally connected to a unique detector in the other array by a fiber optic strand, a single fiber bundle pair is shown here for clarity.

The preferred method of the invention lor automatically configuring and mapping of point-to-point connections between two arrays of transceivers, simply stated. consists of the following steps : 1. Turning on one emitter in the first array. and determining which detector or detectors in the second array senses it. If no detector senses the light, then the emitter in the first array is declared non-useCul.

2. If only one detector senses the light from the emitter in the other array, then that detector is identified as being associated with the first emitter from the first array.

3. If more than one detector senses the light from the emitter in the first array, then all of the detectors that detected the light from the emitter in the first array are designated as possible receivers.

4. Steps (I) thru () are repeated for the remaining emitters of the first array.

5. In the case where more than one detector senses the light from the emitter in the first array, then the detector that detected the strongest signal is designated as the detector corresponding to that emitter.

It will be apparent that step 5 is the application of a preselected rule for resolving a greater number of point to point paths. Other rules could be substituted. Of course. one could stop at step 4. selecting only associated emitter/detector sets for use if sufficient channels are available tor the intended purpose : however the user would be denied, at this stage, most of the benetits of the invention. Assuming the rule-based selection process of the step 5 cases is suitably executed. steps I-5 result in a first fui) round assignment of all unambiguous one to one connections or links between emitters of thc first array and detectors on the second array.

Several approaches for achieving step 5 case resolutions are possible, all within the scope of the invention. For example, the receiver circuitry has gain elements that amplify the current generated by the incoming light. The gain can be adjustable so that the circuitry can be optimized for either low or high light conditions. The gain is typically adjustable to 32 levels. Step &num l of this embodiment would likely have some input pattern such as'0-1-0-1'. and the gain of each receiver would be set at the lowest level and then increased slowly through the software and control circuitry. Lf light hits that detector, then at some gain level, the detector will register the tight : in this example, the ^0-1-0-1 pattern. What is sought is to find the detector that registers the pattern for the lowest gain level. which would correspond to the detector that had the greatest amount of light hitting it.

If multiple detectors see the same amount of light, then one can be chosen by a suitable positional rule of default ; for example, the one most toward the upper right corner of the array is selected. An alternative primary or default rule would sense directly which detector received the most photonically generated photocurrent by accepting photoc : urrcnt from each detector which would charge up, for example, an analog-to-digital converter, to get a numerical value for the light hitting that detector. The detector that registered the highest numerical value for photocurrent would be the winner.

Continuing now with the methodology above : 6. If more than one emitter maps to multiple detectors and some. but not all. of the detectors sense light from more than one emitter, then the unique configurations are selected

out first. Some rule-based approach such as a top-left choice may be used, or one might pick the pixels that spread the'good'pixels out as much as possible to inhibit any tendency for cross talk between adjacent detectors.

7. When there are detectors in the first array and emitters in the second array (whether inter-digitated or not), steps (1) thru (6) are repeated in the reverse direction.

An alternative to step 5, is that all of the detectors that sensed tight from a particular emitter in the first array are lumped together, replicating the received signal. This alternative would make the invention highly tolerant to extremes of vibration and thermal cycling. at least with respect to the receiving array, and the system would be far more tolerant to any system misalignment of emitter/detector connections during operation due to environmental factors.

Configurations having more than the required number of detectors for the necessary number of emitters, enable a technique called over-sampling, which is employed in the embodiment that follows. Fig. 4 shows how the present invention works for the simple case of determining the mapping of three emitters in a first to three detectors in a second array, using a I x 4 array of emitters in Node i and a I x 4 array of detectors in Node 2. The sequence of Figs. 4A to 4D show each emitter of Node) being turned on sequence. and which detectors of Node 2 see the light.

Fig. 4A shows emitter I of Node) being turned on and light from that emitter being detected by the detectors I and 2 of Node 2. Thus. emitter ! coutd be mapped as an optical interconnect line to either detector I or detector 2. In Fig. 4B. only emitter 2 is on, and light from that emitter is sensed by detector 2 and detector 3. Thus. emitter 2 could be mapped to either detector 2 or detector 3. In Fig. 4C, only emitter 3 is on. and light from that emitter is sensed only by detector 3 of node 2. Therefore. emitter 3 can be mapped to detector 3 only.

In Fig. 4D. onlv emitter 4 is on. and tight from that emitter is sensed by detector 4 onty.

Since detector 2 is illuminated by emitter I and emitter 2. it cannot I, used to form a unique mapping of the emitters to the detectors. Likewise, since detector 3 is illuminated by emitter 2 and emitter 3. emitter 3 cannot be used to form a unique mapping. I-lowever, detector 4 sensing emitter 4 is unique and must be used. Also, detector I senses light only from emitter 1 and hence this mapping should also be used. Emitter 2 is mapped to detector 3 to avoid cross talk, since detector 2 also senses emitter 1. Thus, the final mapping configuration of optical interconnect lines is as shown in Fig. 4E.

However, note that if the light sensed by detector 2 from emitter) were sufficiently small as compared to the hght detected from emitter 2. then it would not cause enough of a cross talk problem to interfere with using detector 2 for emitter 2. Thus. the explanation above is correct for cases where the hght sensed by two detectors is roughly equal. in other words. detectors 1 and 2 see about the same light with emitter i on. Where this is not the case. for example if detector 2 sees a very small ratio. say one-tenth, of the light seen by detector I when emitter) was on. and detector 3 sees only one-tenth the light of detector 2 when emitter 2 was on, then one could map emitter I to detector 1. emitter 2 to detector 2 and emitter 3 to detector 3 without worrying about cross talk.

One-tenth is not a critical ratio but is used simply to iHustrate that in cases where the detector and related circuitry can clearly distinguish between a good signal and a weak signal, the weak signal can be disregarded. In this case. if the amount of light hitting detector 2 when emitter I was on is much less than the amount of light hitting detector 2 when emitter 2 is on. so that one can distinguish between emitter) and emitter 2, then by setting the gain level on detector 2, one can program detector 2 to ignore the light from emitter I and only switch at the higher light level of emitter 2.

Once the inter-nodal mapping is complete. power to the unused emitters and detectors can be turned off. For example, in Fig. 4. we only wanted to transmit three channels even though our array could in principle, support four pixel channels. One can save power by finding the pixels which optimize the three channel data transfer and, in this case. turn off

emitter 3 of Node I and detector 2 of Node 2. thereby saving considerable power over the case where all of the devices are powered up with less than all devices being used.

The emitters and detectors switch at gigahertz rates. so the entire mapping process takes a very short amount of time. even for targe arrays, in the example cited, there was only one unused emitter and one unused detector. In practice, arrays as large as 256 x 256 emitters and 640 x 480 detectors have been made. To ensure that such large arrays with large numbers of available channels can be captured. a sufficient over-sampling ratio is required. In the example shown in Fig. 4. a 33% over-sampting was used : 4 pixels for 3 channels. This 33% over-sampling rate is sufficient for most arrays envisioned at present. Note that for lareer arrays, smaller over-sampling rates are required because typically as two-dimensional array sizes get larger. there is a greater chance of a pixel receiving some light. Also as the number of available receiver gain levels increases, providing better discrimination capability. the required over-sample ratio goes down.

In addition. in applications where system integrity is particularly important, it may be useful or necessary to ensure that the auto-configuration process produced a mapping with more than the required number of emitter-detector pairs so that if some of the emitters or detectors failed. there are othel emitteli (letector pairs available to take the place of the failed emitters or detectors.

In a further preferred embodiment of the invention, the methodology extends to mapping large fan-out and star type architectures. The mapping of a single emitter to more than one detector can be exploited intentionally as a means of achieving fan-out. Referring now to Fig. 5. the present invention is illustrated as it applies to accomplish fan-out to more than one other transceiver array. The process of automatically configuring point-to-point connections is used to establish the mapping of Node t to Node 2 where a connection to any of the first three detectors of Node 2 results in the first three emitters of Node 2 being turned on. These emitters in turn are connected to other nodes, thereby demonstrating fan-out capability of the present invention. This embodiment also describes how the present invention can be used to render a star architecture more tolerant of misalignment. Node 2 of

Fig. 5 acts like the central node for a star coupler, where nodes t. 3. 4. and 5 represent leaves of the star.

In a variation or extension of the methods of Figs. 4 and 5 the point-to-point configuration of the optical interconnect lines can be applied sequentially from node to node.

Fig. 6 illustrates a 5-node system in ring architecture, to which the invention is applicable.

The method for configuring the emitter-detector mapping for a ring architecture conststs uf the following steps : (I) Apply the point-to-point configuration procedure to the emitters in Node 1 and the detectors in Node 2.

(2) Apply the point-to-point configuration procedure to the emitters in Node 2 and t1. detectors in Node 1.

(3) Repeat steps (I) and (2) sequentially around the ring to determine unique configurations for each node pair.

There is now described a significant extension or alternate preferred embodiment of the methodology of the invention, providing opportunities for enhanced security and otlrcl benefits. which includes provisions for automatically changing the intra-nodal configuration of detectors to emitters within one or more of the arrays or nodes of a mufti-array device or system, at predetermined intervals. The process involves the following steps : (1) Mapping the connections of each detector to each emitter within each array. so that there is a known one-to-one mapping of the intra-nodal connections as well as the inler-nodal, or array to array connections of the data transmission system ; (2) At a calculated or predetermined interval, changing the data routing scheme by switching the routing of some or all of the detector-emitter data transfer connections at one or more nodes in the data transmission process. and mapping the system.

For example. instead of automatically re-transmitting the data out using the transmitter immediately next to the detector that received the data, the data could be routed to another transmitter. Then the data could be re-routed again.

Fig. 7 shows an example of the use of intra-nodal detector/emitter switching of the routing links at selected nodes to enhance the level of security or confidence that a given set of data being transmitted cannot be readily accessed or decoded by unauthorized third parties.

A fully functional 1 x 4 bit array is shown for simplicity. It can be assumed that different size arrays may be used, with inactive or unused emitters as described above. In this simplified example, data is being transferred from Node I to Node 5. In the top row, according to a first configuration map being utilized at time Tl, data is cross-routed between adjacent emitters 2 and 3 in Node 2 before it is transmitted to Node 3. At some other time T2, shown in the bottom row. the cross routing within Node 2 is reversed so that data flow is"normal". being transferred from detectors to adjacent or default emitters. I lowever also at T2. the data routing in Node 3 is re-formed from tlle Tl mapping to create new routing links connecting detector) to emitter 3. detector 2 to emitter 1, and detector 3 to emitter 2.

Chip topography and circuit design limitations may restrict the number of emitters to which connections from any one detector may be mapped, particularly in large arrays. Even a small subset of available emitters connectable to a given detector, or a convenient. repetitive grouping of detectors and emitters within which such multiplexing is available. provides very significant advantages. The intra-nodal re-routing procedures can be initiated and conducted according to an infinite variety of schemes or rules, but is effectuated in accordance with the invention by using the software in conjunction with the silicon circuitry sensing and control points described. Re-routing can be done after a pre-determined number of clock cycles. including verification that the counter value was valid. or at specified intervals, or when a particular bit in an array became active. and so on. For most of these timing methods, a synchronized clock for all of the nodes is required.

More than one intra-nodal re-routing can be carried out in a given re-mapping exercise. For example, all but the last detector in an array can be re-routed to all but the first

emitter. Then. the last detector could be routed to the first emitter. More than one node can be directed to re-route or re-map the data transfer, and the extent of the re-routing depends only on the number of available bits that can be permuted. False data mapping can also be employed, i. e.. data could be sent to a known non-functional detector (with redundancy provided elsewhere in the system) so that if the data were being monitored. there would be spurious noise present in the signal.

A yet further embodiment of'tile invention provides tolerance for disruptions in the integrity of some of the inter-nodat emitter-detector opticat connections, as in the event of a failed fiber optic connection, and also for the possible failure of an emitter. Figs. 8A and 8B show a sequential inter-nodal re-mapping ot a I x 8 array I of emitters to a I x 8 array 2 of detectors, originally mapped to accommodate a non-tunctional emitter 3 and non-functional detector 1. As is illustrated. the remaining detectors and emitters of the first 4 of each. have been auto-aligned so that a 1 x 3 subset of the emitters has been mapped to a I x 3 subset of the detectors. which is all that is required of this channel map at this time. In addition to the) x 3 subset. another four emitter-detector pairs have been mapped so that if any of the emitters or detectors or the optical connections in the original I x 3 subset fails, a replacement channel is available. The original auto-ahgnment results are shown in Fig. 8A.

If. for example during the course of operation the first emitter fails, then that emitter and the corresponding detector are no longer available ior use. Upon recognition of the failed connection through the circuitry and software or though data channel monitoring means external of the invention but communicating with or recognized by the software control program, re-mapping is executed, the results being illustrated in Fig. 8B.'rhe preterred embodiment methodology of the invention rotating to this aspect of the invention consists of the steps of : (I) Automatically configuring point-to-point connections as previously described to define a full configuration map ; (2) Using the system as designed :

(3) Periodically or continuously testing through the circuitry and software of the invention to see if an emitter to detector connection has failed ; (4) If a failure has occurred as indicated by the testing or from an external indicator. then automatically configure point-to-point connections again and remap the system : (5) If there has been no failure, continue using the original system configuration map.

An alternate method or process for implementing this re-mapping consists of the steps of : (1) Automatically configure point-to-point connections and map a current system configuration. incorporating information about any un-used but functional emitter- detector pairs as ready replacements within the configuration map. for substitution where suitable without requiring system wide mapping ; (2) Use the system as designed : (3) Periodically test to see if an emitter-detector connection has failed ; (4) If a connection has failed, then designate one of the functional un-used emitter- detector pairs as a replacement, so long as unused emitter-detector pairs recognized in the first system mapping are available and suitable for the required substitution, and continue data transfer operations per the same system configuration map : (5) If there has been no failure, continue using the step 1 system configuration map..

Fig. 8 also demonstrates a way of using the present invention to reduce power requirements during routine operation. This method consists of the steps of : (1) Automatically configuring point-to-point connections adequate tor off-peak performance, and saving information about any un-used but functional emitter- detector pairs and switching off power to these emitters and detectors : (2) Using the system as designed ; (3) During peak demand times, switching on power and using the additional emitter- detector pairs to supplement data throughput.

An alternate step (3) to the above method that exercises all available connections, is to cycle repetitively through the available. functional emitter-detector pairs, switching power and transmission duty of each emitter-detector pair on and oft'periodically on a time share basis, so that the heat generated is more uniformly distributed throughout the transceiver array.

It will be readily apparent that the exercise of the methodology ol the invention requires a correspondingly capable device or system. Accordingly, there is an example within the scope of the invention, a photo-transceiver semiconductor array of photo emitters and detectors and supporting circuitry fabricated on a silicon substrate, with a capability in the supporting circuitry for forming and un-forming an electrical circuit between a detector and at least one emitter, It may have circuitry for sensing signal current in the electrical circuit, as well as for sensing signal light level at the detector. It may have circuitry for adjusting the gain of each detector of the array, and for switching power between on and off for the detector or the emitter or both.

As a further example of a device or system of the invention. there may be included or associated with the device or system a software controt program and implementing circuitry for receiving input from the supporting circuitry in the silicon substrate of the array that is sensing the signal current and the signal light level in the electrical circuits between detectors and emitters. The software through its implementing circuitry may aiso controi the supporting circuitry in the silicon substrate that adjusts the detector gain and switches power to the de L, and emitters, and also control the supporting circuitry that forms and un-forms the electrical circuits between the detectors and emitters.

In summary, according to the invention, silicon circuitry in the substrate of a photo transceiver semiconductor array is provided for sensing the received light signals and signal levels of all photo-detectors, switching power on and off to each optical unit. meaning all photo-emitters and photo-detectors whether one or both are present on the chip. adjusting the sensitivity of each detector, and selectively and repetitively forming or routing and testing or sensing new detector to emitter electrical connections within each array. These hardware sensing, testing, and control capabilities are interfaced with a suitable software control

program or programs, whether local or remote, to exercise the control functions described herein on the photo emitters and detectors of a device or system using one or more of such photo-transceiver arrays. When combined with the technique ofover-sampting. where arrays are provided that have surplus emitters and/or detectors relative to the total bandwidth requirement of the application. the invention thus enables system mapping, monitoring and selection of available emitter to detector inter-array optical connections, and routing or multiplexing of detector to emitter internal array connections whenever needed or desired, for configuring and controlling the data transfer channels through a single node or multi-noode optical transmission semiconductor device.

A host ot advantages and benefits derive from this capability, including : the ability to adapt and correct for misalignment of mating transceiver arrays, the ability to bypass or isolate faulty emitters. detectors, internal connections and optical connections ; the ability to exercise power management within the device and gain more uniform heat dissipation across each chip ; and to engage in enhanced security modes for routing and re-routing the data streams.

Other and various embodiments within the scope of the invention illustrated. disclosed and claimed herein will be readily apparent to those skilled in the ul.