A specific example of utilization of the invention is the monitoring of para- meters in every single cell in large battery, for example an emergency current bat- tery or a propulsion battery in a submarine.

However, in principle the present invention represents a quite general sys- tem for, having many stations attached to one and the same single conductor, sending as well as receiving signals between a common central unit and single stations attached to the single conductor, possibly also signals between stations.

The system can be implemented in such a manner that there is galvanic separa- tion between the units. Further, the invention provides possibilities for various met- hods that are quite effective for noise cancellation and error correction of data de- structed by noise.

From US patent no. 5,260,701 is previously known a system for simultane- ous transmission of data between two stations, in which system a central unit transmits an amplitude modulated AC voltage signal, and a slave station receives power as well as control signals from the central unit, and re-transmits a response by modulating its complex impedance. The transmission link is an inductive coupling over a distance of merely a few centimetres, so that the transmitter and receiver antennas may actually be regarded as primary and secondary windings in a transformer.

Further, from US patent no. 5,657,324 is previously known a system for simultaneous transfer of data between plural stations via one single conductor, a central unit transmitting a modulated DC voltage signal, and respective slave sta- tions receiving power as well as control signals from the central unit and re- transmitting responses by modulating their resistance values, via the line.

Norwegian patent application no. 1995 1291 describes a method for simul- taneous transmission of data between stations via a two-wire connection, in which method a central unit transmits an amplitude modulated AC voltage signal by amp- litude modulating only one half period of the signal, and in which the remaining stations receive both power, synchronization and control signals from the central unit, transmitting responses back to the central unit by modulating their impedances on the line in the other half periods. In other words, this Norwegian patent application describes some sort of time multiplexing, in which the signals in respective directions on the communication line are, in principle, not present at the same time, but in different time slots in very close succession.

There exists a need for a more effective manner of communication than exhibited in the prior art. In the present invention, information from the central unit will be modulated onto the data line as an amplitude modulation of a base signal, while the signal from the station is modulated onto the line as combination of cur- rent and phase modulation. Amplitude modulation from the central unit will lead to a change in the line current, because the current draw in the station increases with an amplitude increase on the signal line.

As will be described further down, because the central unit and a station has the same clock, it will be possible to separate a phase modulation due to sig- nalling from the station, from a changing current draw due to changing voltage amplitudes on the line. This means that there may be both an amplitude modula- tion and a current/phase modulation on the data line at the same time, or in the same half periods of the base signal.

Thus, the implementation of the invention will enable two-way communica- tion between a common central unit and many stations attached to a common single conductor data line. Every unit must, in addition to the common single conductor data line, be connected to a common AC earth. For instance in a bat- tery, all coupling terminals situated at different DC voltage potentials, represent such a common AC earthing point.

Hence, in accordance with the present invention there is provided a method such as mentioned in the introduction, and the method in accordance with the in- vention is characterized by the features appearing from the characterizing part of appended claim 1. Special embodiments of the invention appear from the attached dependent claims.

In the following, the invention shall be described in closer detail, and in this connection it is referred to the appended drawings, illuminating special embodi- ments in a non-limiting manner.

Fig. 1 shows schematically a preferred embodiment of a central unit that can be utilized in the method in accordance with the invention; Fig. 2 shows a preferred embodiment of one of several stations connected to a central unit, here in the form of a measuring station, for use in the method in accordance with the invention; Fig. 3 illustrates curve shapes when modulating in a preferred embodiment of the inventive method; and Fig. 4 shows an example of a demodulated data stream from a station.

All stations in the system will in a per se known manner receive their feed voltage and clock signal from the common data line. The station may modulate the draw of current from the line, in a manner that is previously known. However, the circuit solutions enable a phase modulation in addition to a current modulation.

This opens for new and improved qualities.

In a practical construction, the central unit as well as all stations will have a microcontroller for controlling communication and other functions. The communi- cation between station and central unit must necessarily be serial communication.

The feature that the central unit and the station have a common clock signal dur- ing serial communication on a single conductor data line, gives possibilities for a very good data protection and error correction. During serial signal transmission in noise-ridden areas, the traditional transmission systems, for instance RS 232, will be more prone to errors than the system in accordance with the invention. When transmitter and receiver in a serial transmission are controlled from the same clock signal, the receiver will, when communication has been established, always know when the next data bit will arrive. Methods are described for cancellation of noise, as well as methods for creating quite simple, but effective, routines for automatic error correction.

In the following, preferred exemplary embodiments of the method in accord- ance with the invention shall be discussed, in the form of function descriptions in connection with the drawings.

CENTRAL UNIT FUNCTION Fig. 1 shows schematically a central unit.

O is an oscillator, for instance a sine oscillator having a frequency in the range 100-500kHz T is a transformer R1 is a resistor A1 is an amplifier GND is a fixed DC voltage potential, for instance the 0 volt of the central unit Udc is a fixed AC potential. The DC voltage component may possibly be changed DM is a phase and current demodulation circuit K1 is the microcontroller in the unit.

Microcontroller K1 can amplitude modulate the oscillator signal. R1 and A1 amplify the current in the primary side of transformer T into a readable signal for demodulation circuit DM. DM calculates the phase deviation between the oscillator voltage and the transformer current. DM may e. g. provide a digital on/off signal in- dicating when the phase displacement is larger than a certain threshold (in other words, when a station sends data bits). Fig. 3 shows a time diagram for phase modulation. This diagram shall be commented later. The units on the transformer primary side must be connected to common earth or 0 volt. This is designated in the drawing as GND. One terminal of the transformer secondary side is connected to the data line, while the other terminal must be connected to a fixed AC voltage potential (Udc). When Udc is connected to the same AC potential as the stations along the data line, the stations will be able to generate an AC current in the line.

MEASURING STATION FUNCTION Fig. 2 shows a measuring station.

Vf is a common potential, for instance the negative terminal of a battery cell C1 is a capacitor connecting the station to the data line SL is the data line D2 is a rectifying diode for generating feed voltage in the station C2 is a capacitor for current supply to the circuitry of the station D1 is a diode that charges C1 in the half period after current flow in D2 T1 is a transistor circuit converting the AC line voltage to a digital clock signal

K2 is the station microcontroller or logic control unit R3 is a resistor, the base resistor for microcontroller governing of T2 T2 is a transistor drawing current when modulated by logic"1" R2 is a resistor causing, together with T2 and C3, phase displacement of the current C3 is a capacitor/complex impedance coupling modulation onto the line D3 is a diode charging C3 in the half period after current flow in T2 R4 is a resistor, for instance 1 Mohm VK is a comparator delivering a signal to K2 when DC voltage on SL is within a certain range.

The station receives its feed voltage by rectifying the AC voltage between the data line and the fixed potential Vf of the station. The AC voltage is coupled into the station via C1, the diode D2 single-rectifies the signal, and capacitor C2 then stores a feed voltage to the station.

In a simple transistor circuit T1, the AC voltage signal across D1 is straight- ened, so that microcontroller K2 will see square pulses on its clock input, which square pulses have the same amplitude as the DC voltage across C2.

The microcontroller (K2) may have analog and digital inputs to be read. The results from the readouts are converted to a suitable series code, and modulated onto the common data line via T2, R2, R3, C3 and D3 as complex current changes.

A more detailed explanation of the consequences from this modulation, is given later on.

Further, the measuring station has an amplitude measurement circuit (AM) with the ability to determine if the amplitude of the signal on the line is greater or smaller than a value selected in advance. This circuit makes relative measurements, which means that it determines for instance if instantaneous values of the amplitude is larger/smaller than a percentage of the average of the maximum amplitude during a longer previous period of time.

MANNER OF OPERATION FOR COUPLING IN STATIONS The system presumes that only the central unit and one station communi- cate at the same time. Hence, the stations must be selectively addressable. The stations also fetch their feed current from the data line. This current will, when the

central unit demodulates the phase/current modulation, show up as a"background current". It is desired that the ratio between this background current and the modu- lation signal is as large as possible. Therefore, it is desirable that only one station draws current at a time.

Many possibilities exist, and only a few shall be mentioned: A modern microcontroller draws very little current, if for instance analog/di- gital converters are not used. The station controller may be in some kind of a rest- ing mode, waiting for the central unit to transmit the identity number of the station onto the data line. Then, the station will execute its current-drawing measurement tasks, and send data in return, either for a certain time interval, or until an identity number is entered onto the data line from the central unit. This methodology takes for granted that the stations are pre-programmed with an identity number.

One may envisage a second data line. The central unit commands the first station in the row to start. The station executes its task, and gives a start command to the next station in the row, itself returning to a sleep mode. This is a previously known technique, also named"Daisy Chain". In particular when the invention is a system used for monitoring a battery, this is a methodology that is very cost effective. One station is connected to each cell in a battery that consists of many single cells. All of the stations are therefore at different and successively rising potentials. The station will normally be in a sleep mode.

Fig. 2 shows a window comparator VK in the station, measuring the voltage difference between the station connecting point and the DC voltage potential of the data line. When this voltage is within a predetermined range, for instance 0,2 to 1,0 volt, the comparator VK will provide a signal for the microcontroller K2, and the microcontroller will then execute the intended tasks. The input resistance R4 of VK can be made quite high-ohmic, for instance 1 Mohm. This means that there will still be, in practical terms, a"galvanic separation"between station and data line. In fig. 1, transformer T is connected to a fixed potential Udc. Let us assume that Udc may vary, for example within 10 seconds, from the lowest potential of the battery to the highest battery potential (from 0 volt to 48 volt). During such a sweep, all stations will now successively connect themselves in and out.

OPERATING MODE, SIGNAL FROM STATION TO CENTRAL UNIT Let us suppose that a station has been addressed, and is re-transmitting information. The controller (K2) may, as modulation/signalling, control the current draw on the line via R3, T2, R2 and C3.

This modulation signal can in principle be connected to the line via C1, but this will set a limit to the amount of phase shift that can be obtained. If too much current is drawn in the modulation across a common capacitor C1, the feed voltage to the station as well as the clock signal will be destroyed. The use of two separate capacitors distinguishes the present invention substantially above Norwegian application no. 1995 1291.

When the current draw is connected to the line via a capacitor or a complex impedance, the current draw in the group interface will not only have a larger amplitude, but it may also become phase shifted to a very significant degree, relative to the oscillator signal. The modulation is controlled from microcontroller K2 in such a manner that the actual change in current/phase takes place when the AC voltage signal on the data line is zero. Thereby, the emitted noise spectrum will be at a minimum.

The data line may at the same time be amplitude modulated from the central unit. Then, the current in the data line will change in proportion to and in step with this amplitude. However, since the station in itself, when not current/phase modulating, is approximately ohmic, an amplitude modulation on the data line will only cause a small phase modulation. Therefore, the central unit will be able to distinguish the information-carrying phase modulation rather simply.

Fig. 3 illustrates curve shapes during modulation. Curve (a) is the signal from the oscillator in the group interface (i. e. the central unit), (b) is a digitised signal based on a, and is situated in time in the same place relative to signal a, (c) is the signal from A1 when the station does not modulate with increased current draw, and (d) is the signal with an increase in current draw. During current/phase modulation, curve d will exhibit a significant voltage during time slot b. The demodulation circuit (DM) in the group interface detects, on the basis thereof, both current and phase changes.

OPERATING MODE, SIGNAL FROM CENTRAL UNIT TO STATIONS The central unit can amplitude modulate the oscillator"O"in a quite ordinary manner. The demodulation circuit in the station is designed so as to provide a digital output signal at a high amplitude. AM is designed in such a manner that the high/low threshold is not determined by absolute values of the line voltage, but is based for instance on a deviation from the average value of the maximum amplitude on the line during a period. This makes the demodulation circuit quite insensitive to for example changes in the general line amplitude.

NOISE CANCELLATION.

The system makes possible a very special and powerful method for data protection, noise cancellation and automatic error correction.

Fig. 4 outlines an example of a demodulated data stream from a station.

The figure shows one byte of information, with a somewhat special synchroniza- tion bit (SB) at the front edge. In the time T used for transmission of a byte, the oscillator 0 will have run through a significantly higher number of periods than the number of data bits. The time T and the oscillator frequency must be adapted so that one data bit will be at least 3 periods of the oscillator frequency. In practice a larger number will be chosen, for example 16 or 32. The figure indicates 8 data bits between every synchronization bit (SB), and fully drawn pulses show actual "1"logic bits, while pulses shown in broken lines only indicate the positions of bits actually on logic"0", which is on the lower level. The sketch shows that it has been chosen to put time slots between the data bits. This is not necessary. The synch- ronization bit has a special form, two short and one long pulse.

There are many prior art methods for noise cancellation and data correction. In the following, we will describe a methodology that is very efficient for the system in accordance with the invention, due to the common clock signal in central unit and station, as well as the feature that the transmission is made in real time, and is not time multiplexed.

Fig. 4 shall in the following text be described as communication from a station to the central unit, but in principle the communication could also take place the opposite way.

The synchronization bit has been chosen with two very short pulses and one long pulse. The central unit reads SB for example in 6-8 different positions, in

order to provide a very reliable verification of presence as well as accurate position in time for SB.

It is assumed that an SB has been read without error. Thereby, 100% synchronization has been established between station and central unit. The central unit now knows exactly where to receive data bits in time. This is deter- mined through the code in the station, which is of course known to the central unit.

The controller K1 in the central unit can now count along in the message, reading a first data bit location in for example 3 places. Because this is a synch- ronous transmission, these 3 points will be were the data bit appears, or where it would have appeared, if the data bit has been destroyed by noise.

If K1 does not read 3 equal values, then the value appearing the most times, will be selected as the correct value. In this manner, the counting as well as possible error correction will continue through the data byte.

The station may be programmed for instance for transmitting measurement values from channel 1-4, and thereafter repeating the same message over and over, until the station is turned off via the addressing methodology. In such a cycle, the central unit microcontroller may, when it has to correct errors in one or several data bits, attempt to measure the same data bits once again without error when the message for channels 1-4 is received in the next sweep.

The synchronization bit SB has been chosen to be difficult to read, due to its 4 short pulse widths ("1"and"0"). SB is not an information carrier, which means that as long as SB can be read, the data bits will be readable with a better signal to noise ratio.

It is desirable that if there is a lot of noise on the line, K1 will find it difficult to recognize SB before having trouble with reading the proper data bits.

However, if SB is difficult or impossible to read without errors, the counter in K1 knows where SB should be, and will still be able to maintain the synchroni- zation in the transmission. Criteria can be entered, for instance that up to 3-4 SB may be missing before K1 starts rejecting a byte message. As soon as an SB can be read again without errors, full synchronization has been re-established. This methodology is entered to ensure that possible erroneous counts in the station controller K2, as a consequence of noise on the data line, are eliminated. Noise on the data line may result in missing clock pulses, or possibly additional clock pulses to the station controller.

Especially for the invention utilized with a battery, a very efficient methodo- logy for noise cancellation is outlined. For a battery, the expected noise pattern is known. For example, noise pulses may occur from a 3-500 Hz inverter recharging or discharging the battery.

When a station is connected to the line, it will make measurements and thereafter send back data. The data are re-transmitted over and over again, until the station is switched off.

By selecting oscillating frequency as well as the number of pulses per data bit, in a suitable manner, the following can be achieved: There will never be more than one noise pulse between two SB. Further, it can be ensured that the distance between SB and noise pulses is never the same, or mutual multiples. Then, noise pulses will not be able to destroy the same SB or the same data bit in two consecutive bytes with the same information. This is an error correction option that distinguishes the present invention above the methodo- logy described in Norwegian patent application 1995 1291.

4-WAY COMMUNICATION In fig. 4 appears a primary data stream, a primary communication, with controlled intervals between the information-carrying data bits. In these intervals it is of course possible to enter a secondary data stream that is synchronized from the common oscillator"O". Thereby it will be possible to obtain two data streams in each direction, in a mixture of amplitude modulation, phase modulation and time multiplexing. The data streams, up to 4 thereof, do not need a"handshake", they can be transmitted fully asynchronously and independent from each other. This is also a quality that clearly distinguishes the invention above the methodology described in Norwegian patent application 1995 1291.

COMMUNICATION BETWEEN STATIONS Because the system allows two data streams at the same time, one in each direction, information can be transmitted from station to station in real time. A station may transmit a certain code, requesting communication to an other station.

The central unit will then start transmitting all received phase modulation infor- mation on to the line as the same information, however by means of amplitude modulation. In this information there must be included addressing messages as well as messages regarding how long the central unit should operate in this"mirror mode".