Electromagnetic interference cancellation for wired receivers with safety features

Abstract

The embodiments of the present disclosure take advantage of the natural nature of RFI noise on wired links. Since the differential RFI noise in the system has some correlation with the common mode noise on the cable, a replica of the RFI noise can be reproduced by an adaptive filter based on information about the common mode noise. The replica RFI is subtracted from the equalizer output prior to the data decision network or slicer. In this way, the system does not require expensive cables, and the equalizer does not incur additional losses due to the RFI notch filter. Since RFI can be detected and mitigated, this information can also be coupled to safety systems to increase functional safety under high EMI conditions.

Description

[Technical field]

[0001] (background) (1. Technical Field) The present disclosure relates generally to mitigating electromagnetic interference in wired receivers. [Background technology]

[0002] 2. Description of Related Art In this increasingly digital world in which we live, the proliferation of data is exponentially exploding, creating a need to increase data rates in both wireless and wired communications. Wired communications typically use baseband transmissions that transmit digital signals over a medium channel, such as an electrical physical link, with the transmissions themselves taking the form of electrical pulses. Because baseband transmissions are wideband in nature and occupy the entire bandwidth, their operation begins to encroach into the ranges of various radio frequencies (RF), microwave, and radar as their signaling speeds reach high megahertz (MHz) and multi-gigahertz (GHz) rates.

[0003] This increasing penetration of digital baseband transmission into the RF and microwave bands presents challenges in meeting electromagnetic compliance (EMC) for both emissions and immunity standpoints. Most digital data transmission has a level of inherent robustness against electromagnetic interference (EMI), also called radio frequency interference (RFI) when in the radio frequency spectrum, due to the use of differential signaling, which is based on interpreting two identically coupled communication lines and identifying data symbols. From an emissions standpoint, differential signaling cancels the emissions of each communication line due to the opposite phases of the signaling. From an immunity standpoint, EMI energy appears as common mode noise on the differential signal, and therefore the receiver does not experience interference.

[0004] However, the quality of the transmitted signal can degrade rapidly at higher data rates due to significant signal attenuation and mode conversion incurred from imbalances in the differential components along the signal path, making it susceptible to EMC problems that are more pronounced in unshielded transmission media.

[0005] EMI in the form of radio waves, also known as RFI, is typically a modulated signal around a carrier frequency. A baseband receiver can suffer long periods of high bit error rates if the carrier signal couples into the medium and does not die out. If the baseband receiver uses adaptive equalization, long-term interference can cause the adaptive equalizer to drift from its optimal setting. Thus, it is possible that after the RFI stops, there may be continuing periods of bit errors due to suboptimal receiver equalization.

[0006] Long-lasting RFI can also present a safety issue in safety-critical applications such as autonomous driving, where a steady stream of uncorrupted video or other data is required to maintain safe operation. A system that can detect and compensate for the onset of RFI or other types of EMI adds a level of functional safety to the overall system.

[0007] One approach to mitigate EMC issues in wired links is to use high-quality cables. However, this increases system costs and is not a preferred solution in cost-sensitive systems. Also, maintaining high-quality cables in many applications, such as automotive, is technically difficult due to the system environment, such as vibration, repeated cable bending, deterioration due to aging, and temperature changes.

[0008] Another approach is to install an in-line common mode choke. Common mode chokes (CMCs) can be effective at reducing common mode signals by 20-30 dB. However, this does not reduce EMI that has already been converted to differential signals by the cable. Also, the CMC itself converts common mode signals (e.g., RFI noise) to differential signals. These are drawbacks that limit the effectiveness of CMCs.

[0009] The third method is the use of a notch filter. This method applies a notch filter that attenuates RFI noise of a specific frequency in the equalizer. This requires frequency detection of the RFI noise and adaptation to it. However, the notch filter also attenuates the signal at the same frequency. Therefore, there is some degradation (lower SNR) when a notch filter is applied.

[0010] Therefore, there is a need for a better approach to reducing the effects of EMI. Summary of the Invention [Means for solving the problem]

[0011] (summary) Certain embodiments of the present disclosure take advantage of the natural properties of RFI noise. Since differential RFI noise in the system has some correlation with the common mode noise on the cable, a replica of the RFI noise can be reproduced by an adaptive filter if information about the common mode noise can be extracted. The replica RFI is subtracted from the equalizer output prior to the data decision network or slicer. With this approach, the system does not require expensive cables, and the equalizer does not incur additional losses due to the RFI notch filter. Since RFI can be detected and mitigated, this information can also be coupled to safety systems to increase functional safety under high EMI conditions.

[0012] In one aspect, a receiver chip for wired communication includes a baseband receiver and EMI cancellation circuitry coupled to the baseband receiver. The baseband receiver receives an input signal transmitted over a wired link. The input signal is encoded with digital data, and the baseband receiver recovers the digital data from the received input signal. However, the input signal may also carry EMI. The EMI cancellation circuitry produces an EMI cancellation signal from the received input signal, e.g., an estimated replica of EMI noise. The baseband receiver receives the EMI cancellation signal and uses the EMI cancellation signal to reduce the impact of EMI on the data recovery process.

[0013] In another aspect, the receiver chip includes EMI monitoring circuitry and safety circuitry. The EMI monitoring circuitry produces an EMI detection signal from a received input signal. The EMI detection signal indicates the presence of EMI in the received input signal. The safety circuitry alters operation of the baseband receiver depending on whether EMI is present.

[0014] For receiver chips that implement both EMI cancellation and EMI detection, the EMI cancellation circuitry and the EMI monitoring circuitry may be overlapping or even identical. Signals produced as part of the EMI cancellation may be advantageously used to implement safety features based on the presence or strength of EMI. The EMI detection signal may be derived from the EMI cancellation signal. For example, if the EMI cancellation signal exceeds a threshold, the EMI detection signal may be set to EMI detected.

[0015] Other aspects include components, devices, systems, improvements, methods, processes, applications, computer readable media, and other technology related to any of the above. [Brief description of the drawings]

[0016] Embodiments of the present disclosure have other advantages and features that will become more readily apparent from the following detailed description and the appended claims, when taken in conjunction with the examples in the accompanying drawings.

[0017] [Figure 1A] FIG. 1A is a block diagram illustrating electromagnetic interference (EMI) on a wired communication link.

[0018] [Figure 1B] FIG. 1B is a block diagram of a baseband receiver with EMI mitigation.

[0019] [Figure 2A] 2A-2C are block diagrams of example implementations of a baseband receiver with EMI mitigation. [Figure 2B] 2A-2C are block diagrams of example implementations of a baseband receiver with EMI mitigation. [Figure 2C] 2A-2C are block diagrams of example implementations of a baseband receiver with EMI mitigation.

[0020] [Diagram 3] FIG. 3 is a block diagram of an example implementation of common mode sensing circuitry.

[0021] [Figure 4A] FIG. 4A shows the physical layout of two resistor groups with a common centroid.

[0022] [Figure 4B] FIG. 4B shows the physical layout of the two interdigitated resistor groups.

[0023] [Diagram 5] FIG. 5 shows the RFI pulse response at 1 GHz as filtered through first order low pass filters of different bandwidths.

[0024] [Figure 6]FIG. 6 illustrates the cancellation of an RFI burst.

[0025] [Figure 7] FIG. 7 is a safety state diagram. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0026] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The figures and the following description relate to preferred embodiments by way of example only. It should be noted that from the following discussion, alternative embodiments of the structures and methods disclosed herein will be readily recognized as viable alternatives that may be employed without departing from the principles of the claimed subject matter.

[0027] FIG. 1A is a block diagram illustrating electromagnetic interference (EMI) on a wired communication link 100. The communication link 100 shown in FIG. 1A is bidirectional and includes two transceivers 110A, 110B connected by a wired link 120. Each transceiver 110 includes a transmitter and a receiver. Digital data is transmitted from one transceiver 110A / B to the other transceiver 110B / A via the wired link 120. Electromagnetic interference (EMI) 130 may couple into the physical medium 120. Wired communication uses baseband transmission, which transmits digital signals over the medium channel 120 using pulses. Because baseband transmission is wideband in nature, high data rate transmissions can reach the range of RF and microwave frequencies. EMI 130, including RFI at RF frequencies, can be generated by various sources within the operating bandwidth of the transmitted signal. If coupled into the physical medium 120, these will manifest themselves as noise at the receiver. Each receiver receives an input signal, which is a combination of a desired signal and noise (EMI), and recovers the digital data from the received input signal.

[0028] FIG. 1B is a block diagram of a receiver that mitigates the effects of this EMI. The receive path is the receive portion of the transceiver 110. It includes a baseband receiver 130 and EMI cancellation circuitry 140. The baseband receiver 130 receives an input signal from the wired link 120 and recovers the encoded digital data. The EMI cancellation circuitry 140 produces a signal 145 from the received input signal. The signal 145 is based on the EMI in the input signal. For example, it may be a replica or an estimate of the EMI in the input signal. The baseband receiver 130 uses the signal 145 to mitigate the effects of the EMI on the process for recovering the digital data. For example, if the signal 145 is a replica of the EMI, it may be used to cancel the EMI in the input signal. For convenience, the signal 145 will be referred to as an EMI cancellation signal.

[0029] The EMI cancellation approach described in this disclosure may be applied to many types of communication links 100. For example, the link 100 may be bidirectional or unidirectional, in which case each terminal 110 may include only a transmitter, only a receiver, or both a transmitter and a receiver. The medium 120 is typically an electrical cable. Examples include coaxial cable, twisted pair, parallel pair, and backplane. The cable pair may be shielded or unshielded. Exemplary data rates include 1G (Gbps), 2.5G, 5G, 10G, and higher. The length of the link may vary widely. For example, typical automotive cabling is standardized to be a maximum of 15 meters, but in many practical cases is shorter. Regular Ethernet is specified for a maximum of 100 meters. Note that the longer the cable, the more susceptible this is, as the desired signal is attenuated more over longer lengths, and longer cables present more opportunities for undesired coupling. Many different protocols may be used, for example 1000BASE-T1, 2.5GBASE-T1, 5GBASE-T1, 10GBASE-T1, 1000BASE-T, 2.5GBASE-T, 5GBASE-T, 10GBASE-T, and MIPI APHY. There are also many different end use applications. One example is automotive networking. Other examples include industrial Ethernet, as well as enterprises and data centers. Today's 5G radio towers may have Ethernet and other basebands close to the powerful radio sources. For enterprise and data center vaults, security radio phones and cell phones may cause interference. RFI is especially problematic for automotive and industrial applications. Aviation applications are another exemplary application.

[0030] 2A-2C are block diagrams of an example implementation of a baseband receiver with EMI mitigation. FIG. 2A shows only the baseband receiver 230. FIG. 2B adds EMI cancellation circuitry 240, which is mostly below the dashed line, while the baseband receiver 230 is above the dashed line. FIG. 2C adds safety functionality. In this embodiment, the functionality shown in FIGS. 2A-2C is implemented as a single mixed-signal receiver chip with an analog front-end 212 and a digital back-end 214, which includes signal processing and a state machine for functional safety 250, although this is not required. For example, the analog 212 and digital 214 functions may be implemented in separate chips. Additionally, the safety functionality 250 (shown in FIG. 2C) may be implemented in another chip, such as a CPU that runs a software version of the safety state machine, which controls the adaptive controller 258 via register writes and gets interrupts (or polls the RFI detection signal 256 and SNR 259). The receiver has an input pin for the baseband receiver 230, but does not have a separate pin for the EMI cancellation 240, thus saving on pin count. The receiver uses differential signaling, and therefore the incoming signal is a differential signal pair.

[0031] 2A shows a baseband receiver 230. Analog / mixed-signal components include analog signal conditioning circuitry 231. Examples include passive and active analog filters, low noise amplifiers (LNAs), programmable gain amplifiers (PGAs), linear attenuators, DC blocking, baseline drift correction circuits, and signal clamping circuits. This is followed by a high-resolution analog-to-digital converter 233, which converts the incoming analog signal to digital for the digital backend.

[0032] The sample rate of the ADC233 depends on the data transmission rate over the communication link. The sample rate can vary widely, for example, from 125 MHz to 6 GHz or even higher. The sample rate and data transmission rate are also related to the effects of EMI. For example, if the receiver is operating at 125 MHz, EMI from a 1 GHz RFI tone will be filtered out by the basic receiver functionality and will not require special EMI cancellation, whereas EMI from a 50 MHz tone will likely require cancellation, as explained below. Conversely, if the receiver is operating at 6 GHz, 1 GHz EMI will be more of a problem and 50 MHz EMI will be less of a problem.

[0033] The ADC 233 is followed by an equalizer 236, e.g., a feed-forward equalizer, for receiver equalization. The equalized output then goes through a slicer 237, which quantizes the received signal. The filter coefficients for the equalizer 236 are adaptive and are trained to settle on a setting to undo the distortion of the signal in the channel. If the digital data is protected with a forward error correction (FEC) code, the FEC circuitry 239 may correct a certain level of bit errors that exit the slicer 237. The incoming input signal is an analog differential signal. The ADC 233 converts this incoming analog differential signal to a sampled digital form. The signal is differential up to the input of the ADC 233, which is then digitalized.

[0034] When EMI noise interferes with a normally received signal, both the signal and the noise will flow through the equalizer 236 and the EMI noise will distort the equalizer output. A conventional approach to dealing with EMI is to adapt the equalizer 236 to notch out the noise, but this notching will also attenuate good signals, reducing the signal-to-noise ratio (SNR).

[0035] FIG. 2B shows additional EMI cancellation circuitry 240. EMI signals generally manifest themselves on the main data transmission channel to the PHY receiver side in two forms: primarily as common mode signals, and less likely, but more problematic, as differential signals through various mode conversion mechanisms along the data signal propagation. Some examples of mode conversion include imbalances in the communication line pairs, non-uniform windings of the communication lines at the common mode choke, and non-uniform termination impedances. EMI manifested as common mode signals can be attenuated by common mode chokes (CMC) and blocked by the receiver using differential signaling methods. However, EMI manifested as differential signals is not suppressed by CMC and exists as an error signal superimposed on the desired transmitted data signal to the differential receiver, in addition to signal distortion from the channel.

[0036] A dedicated auxiliary EMI sensing receiver may be used to detect common mode movements in the signal. Its function is complementary to that of the main receiver by sensing only common mode perturbations and rejecting differential signals. In a typical conventional solution, the CMC is located off-chip, before the input pin to the analog front-end 212. The EMI sensing receiver would tap the signal before the CMC where the common mode signal energy is larger. However, it would require one or more dedicated ports for the EMI sensing receiver, which would add pins and more circuitry. The approach shown in FIG. 2 eliminates the redundant sensing port by sensing EMI directly at the main signal port and having both the main and EMI signal paths monolithically implemented in the same package or on the same chip. This requires a high level of sensitivity in extracting the EMI signal. If higher sensitivity is provided by EMI cancellation, an external CMC may also become unnecessary. For example, it may be used in an unshielded transmission environment, but may be eliminated for shielded transmission media. The single-chip solution also offers economic advantages and ease of integration.

[0037] In FIG. 2B, common mode sensing circuitry 241 extracts the common mode signal at the main differential signal port, for example, by summing two differential signals in the voltage, current, or charge domain. If CMC is used, the CMC is typically located before the analog front end 212, and common mode sensing circuitry 241 taps the signal after the CMC. FIG. 3 is a block diagram of an example of common mode sensing circuitry 241. The two differential signals are voltages V1 and V2. The output voltage V C is calculated as follows: V C =(V1Z2+V2Z1) / (Z1+Z2) (1) Impedances Z1 and Z2 may be implemented by passive and / or active elements. Passive elements may be, but are not limited to, poly or diffused resistors, MiM (metal insulator metal), or finger MOM (metal oxide metal) capacitors. Active elements may be transistors biased in the triode region or highly matched current sources.

[0038] When impedances Z1 and Z2 are the same, the output Vc=(V1+V2) / 2, which is the correct desired common mode of the signal. However, if impedances Z1 and Z2 are mismatched, the common mode signal Vc will have an error. Because detection is performed at the main signal port, there is high sensitivity to mismatches of the detection elements Z1 and Z2 of the common mode extractor 241. Even in state-of-the-art processes, systematic and random variations exist. One simple example would be non-uniform doping density, or linear (or higher order) gradient effects due to polishing, which can cause resistance variations across physical locations on a silicon wafer. When differences in Z1 and Z2 exist, even if V1 and V2 are perfect differential signals with equal amplitude but opposite phase, this differential signal component can erroneously appear as a common mode signal through mode conversion, thus corrupting the common mode detection.

[0039] One way to meet the required high sensitivity is by using high precision passive or active elements that are proportional and highly matched by placing them in a comb-like and / or common centroid fashion in the physical layout. FIG. 4A shows the physical layout of resistors on a chip, where the array has a common centroid. The common centroid is marked by a "+". The x and y dimensions are vertical along the chip surface. Resistor R1 is connected in a circuit to form an impedance Z1, and resistor R2 is connected in an identical manner to form an impedance Z2. Because the group of resistors R1 has the same common centroid as the group of resistors R2, the variations across the chip tend to cancel between the two groups, and thus Z1 and Z2 are matched despite the variations across the chip. FIG. 4B shows the physical layout, where resistors R1 and R2 are interdigitated (alternated) with each other in the physical layout. This layout will also reduce the mismatch of Z1 and Z2 due to across-chip variations.

[0040] Returning to FIG. 2B, once the EMI signal is extracted from the main input port as a common mode signal using the on-chip common mode detector 241, it is further amplified and processed by analog signal conditioning circuitry 242, which acts on the signal transition edges. Circuitry 242 may provide, for example, signal amplification, filtering, and clamping. The signal is then digitized by a low-resolution oversampling ADC 243. ADC 243 oversamples with respect to ADC 233. That is, ADC 243 produces multiple samples for every sample of ADC 233. For example, ADC 243 may be a 1-bit ADC that samples at N times the sample rate of ADC 233. Alternatively, as shown in FIG. 2B, ADC 243 may be implemented as N 1-bit ADCs, each sampled at the same rate as ADC 233, but offset by 1 / N cycle with respect to each other. For oversampling, N≧2. The ADC 243 is low resolution compared to the ADC 233. Typically, the low resolution ADC 243 will have 1-3 bits of resolution. The digitized output of the ADC 243 contains EMI information in the form of timing or frequency that the downstream digital signal processor (DSP) in the backend 214 uses as a reference for EMI cancellation.

[0041] The approach of FIG. 2B does not attempt to capture and maintain the linear power and amplitude information of the EMI signal, but rather processes and conveys it on the transition or edge information of the EMI signal. This simplifies the design of the EMI front-end receiver 241-243. The analog signal conditioning 242 in the EMI cancellation branch does not rely on maintaining the linearity of the EMI signal over a large dynamic range, but works on the signal transition edges and preserves its fundamental frequency information. Compared with the EMI signal conditioning, which attempts to preserve the signal fidelity of the sensed EMI signal through linear amplification, the analog front-end 241-243 in FIG. 2B does not require highly linear analog circuits such as amplifiers, filters, and high-resolution ADCs. This in turn saves power and area and reduces system complexity.

[0042] One possible implementation of the EMI receiver front-end 241-243 is to immediately convert the sensed common-mode signal to a differential signal. Analog signal processing 242 in differential form necessarily provides its zero crossing as a reference as well as better power supply blocking. The signal can then be amplified by a simpler low-gain amplifier, possibly followed by a high-speed clamping or limiter circuit, and then finally sampled and digitized by one or more single-bit quantizers running at the oversampling rate. With the help of advanced deep submicron silicon process technology, a 1-bit low-noise quantizer clocked at multi-GHz is achievable, capable of regenerating low signals close to differential zero and amplifying them into full rail-to-rail digital bits with high accuracy and power efficiency.

[0043] RFI signals tend to be more coherent in frequency and are typically dominated by the carrier signal of an interfering source in close proximity. Figure 5 depicts the output waveforms of a 1 GHz RFI pulse filtered through a simple RC low-pass filter with various cutoff frequencies from 1 MHz to 10 GHz. The top waveform 510 is the input waveform of a 1 GHz RFI pulse with a fundamental period of 1 ns. Waveform 521 is the output after filtering through a 1 MHz RC low-pass filter. Waveforms 522-525 are the output after filtering through 10 MHz, 100 MHz, 1 GHz, and 10 GHz RC low-pass filters, respectively. The signal power may be degraded at the low bandwidth output, but the fundamental frequency of the RFI signal is maintained. The same is true for RFI signals at various frequencies through a fixed channel.

[0044] This periodicity may be exploited to simplify the RFI detector to simply detect the frequency of the interfering common mode signal. A 1-bit or low resolution quantizer may be used to capture the transition edges, and an appropriate oversampling rate is used to determine the range of RFI frequencies of interest. This information may then be relayed as a reference signal to a DSP in the normal receive path 230, and the exact power of the interferer may be extracted and cancelled, for example, using intelligent signal processing, as described below.

[0045] In FIG. 2B, the outputs of analog receivers 241-243 contain sampled and digitized signals that contain the EMI information. These signals are fed to an adaptive FIR filter 246. The coefficients of the FIR filter are adapted to minimize the mean square error of the error signal. In this embodiment, circuitry 247 generates an error signal 248 based on the operation of slicer 237. Slicer 237 receives an input signal and slices the input into a finite number of output levels based on the thresholds used by the slicer. The difference between the input signal and the allowed slicer output levels is the error signal 248.

[0046] In FIG. 2B, the adaptive FIR filter 246 is implemented as follows: The oversampling ADC 243 produces N samples, offset by 1 / N cycles. These N digital streams are received by N FIR filters 246A-N with the samples offset by 1 / N cycles. Any value of N, including N=1, can be used to trade off complexity and performance. For example, if N=4, then the samples are each offset by 1 / 4 cycle. The output of each FIR is added and then subtracted from the output of the equalizer 236. Oversampling mitigates false adaptation when the frequency and phase of the RFI are in a certain relationship to the ADC 233 clock. When N=2 or higher, oversampling ensures that at least one FIR input has a strong signal if one of the sampled inputs is close to a zero crossing (and therefore is a weaker signal).

[0047] The EMI cancellation circuit 240 replicates the sampled EMI noise as an EMI cancellation signal 245, which is then subtracted from the output of the equalizer 236 output before proceeding into the slicer 237. Because cancellation is not performed inside the equalizer 236, good signals do not suffer from being attenuated.

[0048] Different types of adaptation may be used for the adaptive component equalizer 236 and the FIR 246. Different types of adaptive components may also be used. One algorithm is Minimum Mean Square Error (MMSE), which optimizes the signal-to-noise ratio (SNR) performance at the data slicer input. Another algorithm is the Zero Forcing algorithm (ZF), which removes inter-symbol interference (ISI) independent of noise. A combination of these algorithms can be used for optimization of system performance, which depends on noise and channel loss.

[0049] In the example of FIG. 2B, MMSE is used to adapt the RFI canceller 240 and ZF is used to adapt the baseband receiver 230. If the baseband receiver 230 and the EMI canceller 240 use the same algorithm, they may conflict with each other when optimizing the filter coefficients. Using MMSE for one and ZF for the other significantly reduces or eliminates the conflict. Thus, system performance can be maintained in a harsh RFI environment.

[0050] FIG. 6 illustrates the operation of EMI cancellation. FIG. 6 shows three plots 610, 612, 614 of various signals over time. In this example, the signal is a four-level signal. Plot 610 shows the signal sampled at the input of slicer 237. The signal is centered around four signal levels. When properly equalized, the input signal can be properly assigned to one of the four levels by slicer 237 when comparing the input signal against a threshold 611 shown by the dashed line. A second plot 612 shows the EMI noise coupled into it. Note that this noise is not directly observable by itself, other than being coupled into it with the good signal, and is shown here for reference only. A third plot 614 is the output of EMI canceller 240, which is the EMI cancelled signal.

[0051] There are two EMI bursts 621, 625. The first burst 621 is not cancelled, while the second burst 625 is cancelled. Without cancellation, the input 610 to the slicer is corrupted for the duration of burst 621. With cancellation, the corruption only occurs for a short period at the start 626 and end 627 of the burst, because it takes a finite time for the filter to detect and adapt when the EMI burst begins and ends.

[0052] If the signal is protected with a forward error correction (FEC) code, the corruptions at the beginning 626 and end 627 of the burst can be corrected. Note that the FEC 239 can only correct errors for a limited duration. Therefore, the burst protection capability of the FEC and the adaptation time of the EMI canceller 240 are matched such that the FEC burst protection is longer than the adaptation time of the EMI canceller.

[0053] As shown in plot 614, the output of EMI canceller 240 is an easily observable signal in the system. This signal can be used as a proxy for the presence of EMI and can be used to enhance functional safety by providing an early alert that a given link is being exposed to excessive EMI. This early alert would give the system time to compensate for a possible failure (e.g., switching to an alternate sensor or other component, bringing the system to a safe state, etc.). On the other hand, the system should not trigger a false alert because the link will possibly be exposed to some degree of EMI.

[0054] 2C adds safety functionality to the receiver chip. In this embodiment, the safety circuitry 250 implements a state machine, as described in more detail below, although other approaches may also be used. An EMI detection signal 256 (e.g., in Table 2 below, variable rfi_active) is provided as an input to the safety circuitry 250. The monitoring circuitry 252 converts the EMI cancellation signal 245 to the EMI detection signal 256 by comparing against a threshold. If the EMI cancellation signal 245 is above the threshold, the EMI detection signal 256 is set. The state machine 250 also takes an SNR 259 as an input (e.g., in Table 2 below, variables good_SNR and low_SNR). The SNR is calculated from the square of the error 248, and the corresponding variable may be determined by a threshold. Circuitry 258 is used to control the adaptation of the equalizer 236, for example turning the adaptation on or off based on a control signal 257 from the safety circuitry 250.

[0055] 7 shows the state diagram of state machine 250. The state diagram implements the following safety functions: Temporarily disable the adaptive filter in the equalizer 236 from training when EMI is present, and therefore do not train on a noisy signal. This reduces the degradation of SNR that occurs when adapting on noisy data. Sending an alert ("Alert" output in FIG. 2C) to provide an early warning that EMI may exceed the limits of what the EMI canceller 240 can compensate for. If EMI noise that cannot be canceled to a sufficient degree continues to be present for an excessive amount of time, disable the link (in Figure 2C, the “drop_link” output) to prevent the propagation of erroneous data.

[0056] The state diagram of FIG. 7 uses the states listed in Table 1. [Table 1]

[0057] The state transitions are governed by the variables listed and defined in Table 2. [Table 2]

[0058] Figure 7 is interpreted as follows: The text inside each state indicates the action taken for that state. In NORMAL state, pause is set to 0 (false), retrain is set to 0, etc. In PAUSE state, timers 1, 2, 3 are started, pause is set to 1, retrain is set to 0, etc. State transitions are triggered by combinations of the variables shown. For example, a state transitions from NORMAL to PAUSE depending on ~good_SNR|rfi_active, which is (good_SNR=0 OR rfi_active=1). "|" means OR, "*" means AND, and "~" means NOT.

[0059] In normal operation without EMI, the state machine is in the NORMAL state. Once EMI activity is detected (rfi_active=1) or good SNR is lost (good_SNR=0), the state machine transitions to the PAUSE state. Here, equalizer adaptation is paused to prevent training on noisy data. In FIG. 2C, this is accomplished by sending a signal 257 to the adaptation controller 258, which then turns off the adaptation of the equalizer 236. If after a certain timer2 time, poor SNR and EMI activity are no longer present, the state machine proceeds back to the NORMAL state. Otherwise, if timer1 expires and EMI activity is no longer present, the state machine transitions to the TRAIN state. Otherwise, if timer3 expires and the SNR is low, the state machine transitions to the DROP state.

[0060] In the TRAIN state, EMI noise should not be present and the equalizer is forced to adapt since the SNR is not good. If the SNR is low in this state, an early warning (alert) will be sent. If EMI activity is seen, the state machine transitions back to the PAUSE state. Otherwise, if timer4 has expired and the SNR is at least not low, the state machine transitions to the NORMAL state. Otherwise, if timer3 has expired, the state machine transitions to the DROP state.

[0061] In the DROP state, the link is dropped to prevent erroneous data from being received.

[0062] Note that the safety functions may be implemented using techniques other than finite state machines. The safety functions may also be implemented independently of the EMI cancellation. In that case, the circuitry used to monitor for EMI may include components 241-246, or simplified versions of those components. In FIG. 2C, much of the circuitry shown in 241-246 is used for both EMI cancellation and monitoring.

[0063] Although the detailed description contains many details, these should not be interpreted as limiting the scope of the present invention, but merely as illustrating different examples. It should be understood that the scope of the present disclosure includes other embodiments not discussed in detail above. Various other modifications, changes, and variations that will be apparent to those skilled in the art may be made in the arrangement, operation, and details of the methods and apparatus disclosed herein without departing from the spirit and scope as defined in the appended claims. Therefore, the scope of the present invention should be determined by the appended claims and their legal equivalents.

Claims

1. 1. A receiver chip for wired communications using baseband transmission, the receiver chip comprising: a baseband receiver configured to receive an input signal transmitted over a wired link, the input signal being encoded with digital data, the baseband receiver recovering the digital data from the received input signal; an electromagnetic interference (EMI) cancellation network coupled to the baseband receiver; Equipped with The EMI cancellation circuitry produces an EMI cancellation signal from the received input signal, the EMI cancellation signal being based on EMI in the received input signal; the baseband receiver receives the EMI cancellation signal from the EMI cancellation circuitry and reduces the effect of the EMI on recovering the digital data based on the EMI cancellation signal; the input signal comprises a differential input signal; The EMI cancellation circuitry includes: an analog common mode sensing circuit for extracting a common mode signal from the differential input signal; a digital adaptive filter that produces the EMI cancellation signal from the common mode signal; an analog-to-digital converter (ADC) coupled between the analog common mode sensing circuit and the digital adaptive filter; a feedback loop from the baseband receiver to the digital adaptive filter; Equipped with A receiver chip, wherein the baseband receiver comprises an ADC that samples at a frequency based on a data transmission baud rate of the digital data, and the ADC in the EMI cancellation circuitry oversamples with respect to the frequency of the ADC in the baseband receiver but has a lower resolution than the ADC in the baseband receiver.

2. The receiver chip of claim 1 , wherein the EMI cancellation signal is an estimate of the EMI in the received input signal.

3. The receiver chip of claim 1, wherein common mode noise induces coupled noise in the differential input signal, and wherein the EMI cancellation circuitry extracts the coupled noise from the differential input signal.

4. The receiver chip of claim 3 , wherein the wired link comprises a twisted pair that carries the differential input signal.

5. The receiver chip of claim 3 , wherein the baseband receiver comprises a common mode choke, an output of the common mode choke coupled to an input of the EMI cancellation circuitry.

6. 4. The receiver chip of claim 3, further comprising two external pins that receive the differential input signal and provide the differential input signal to both the baseband receiver and the EMI cancellation circuitry.

7. The receiver chip of claim 1 , wherein the EMI cancellation circuitry converts the extracted common mode signal to a differential form.

8. The receiver chip of claim 1, wherein the ADC in the EMI cancellation circuitry is a 1-bit ADC.

9. 2. The receiver chip of claim 1 , wherein the ADC in the EMI cancellation circuitry comprises N parallel ADCs sampled at the same frequency but offset in phase with respect to one another, where N≧2.

10. The digital adaptive filter comprises: N FIR filters receiving samples from the N parallel ADCs; an adder for adding outputs from the N FIR filters; The receiver chip of claim 9 , comprising:

11. The receiver chip of claim 1 , wherein the EMI cancellation signal produced by the adaptive filter is an estimate of the EMI in the received input signal.

12. The baseband receiver includes: Equalizer and circuitry for subtracting the EMI cancellation signal from the output of the equalizer; The receiver chip of claim 11 comprising:

13. 13. The receiver chip of claim 12, wherein the feedback loop produces an error signal based on a difference between (a) the output of the equalizer minus the EMI cancellation signal and (b) a set of reference levels for the recovered digital data.

14. 2. The receiver chip of claim 1, wherein the baseband receiver comprises forward error correction to correct errors in the recovered digital data, and the digital adaptive filter adapts to bursts of EMI before the forward error correction loses effectiveness in correcting errors.

15. 1. A method for wired communication using baseband transmission, the method comprising: receiving a differential input signal transmitted over a wired link, the input signal being encoded with digital data; Producing an electromagnetic interference (EMI) cancellation signal from the received input signal, the EMI cancellation signal being based on EMI in the received input signal, the producing comprising: extracting an analog common mode signal from the differential input signal; applying a first analog-to-digital (A / D) conversion to convert the extracted common mode signal from analog form to digital form; applying a digital adaptive filter to the common mode signal in digital form to produce the EMI cancellation signal; and recovering the digital data from the received input signal, the recovering comprising: applying a second analog-to-digital (A / D) conversion sampling at a frequency based on a data transmission baud rate of the digital data, the first A / D conversion oversampling with respect to the frequency of the second A / D conversion but having a lower resolution than the second A / D conversion; Using the EMI cancellation signal to reduce the effect of the EMI on recovering the digital data. and providing feedback from the step of recovering the digital data to the digital adaptive filter; A method comprising:

16. The method of claim 15 , wherein the EMI cancellation signal is an estimate of the EMI in the received input signal.

17. The method of claim 15, wherein common mode noise induces coupled noise in the differential input signal, and producing the EMI cancellation signal includes extracting the coupled noise from the differential input signal.

18. The method of claim 15, wherein producing the EMI cancellation signal further comprises converting the extracted common mode signal to a differential form.

19. The method of claim 15, wherein the first A / D conversion is a 1-bit A / D conversion.

20. The method of claim 15, wherein restoring the digital data further includes applying forward error correction to correct errors in the restored digital data, and the digital adaptive filter adapts to bursts of EMI before the forward error correction loses effectiveness in correcting errors.

21. 1. A receiver chip for wired communications using baseband transmission, the receiver chip comprising: a baseband receiver configured to receive an input signal transmitted over a wired link, the input signal being encoded with digital data, the baseband receiver recovering the digital data from the received input signal; EMI cancellation circuitry configured to receive the input signal and produce an EMI cancellation signal from the received input signal, the EMI cancellation signal being an estimate of the EMI in the received input signal and used by the baseband receiver to reduce the effect of the EMI on recovering the digital data; and EMI monitoring circuitry that produces an EMI detection signal from the EMI cancellation signal, the EMI detection signal indicating the presence of EMI in the received input signal; safety circuitry coupled to the EMI monitoring circuitry and to the baseband receiver, the safety circuitry modifying operation of the baseband receiver in response to the EMI detection signal indicating the presence of EMI; A receiver chip comprising:

22. 22. The receiver chip of claim 21, wherein the baseband receiver is adaptive, and the safety circuitry stops adapting the baseband receiver in response to the EMI detection signal indicating the presence of the EMI.

23. 23. The receiver chip of claim 22, wherein the safety circuitry initiates adaptation of the baseband receiver in response to the EMI detection signal no longer indicating the presence of the EMI.

24. 23. The receiver chip of claim 22, wherein the safety circuitry enables adaptation of the baseband receiver in response to the EMI detection signal no longer indicating the presence of the EMI.

25. 22. The receiver chip of claim 21, wherein the safety circuitry is further configured to generate an alert in response to the EMI detection signal indicating the presence of the EMI.

26. The receiver chip of claim 21 , wherein the security circuitry implements a state machine.

27. 27. The receiver chip of claim 26, wherein the state machine includes a NORMAL state for normal transmission over the wired link and a PAUSE state for pausing adaptation of the baseband receiver.

28. 30. The receiver chip of claim 27, wherein the state machine further includes a TRAIN state for forcing adaptation of the baseband receiver.

29. 22. The receiver chip of claim 21, wherein the safety circuitry alters operation of the baseband receiver in response to an indication that the EMI cancellation circuitry is operating near a cancellation limit.

30. 22. The receiver chip of claim 21 , wherein the input signal comprises a differential input signal, common mode noise induces coupled noise in the differential input signal, and the EMI cancellation circuitry extracts the coupled noise from the differential input signal.

31. The input signal comprises a differential input signal, and the EMI cancellation circuitry comprises: a common mode sensing circuit for extracting a common mode signal from the differential input signal; an adaptive filter that produces the EMI cancellation signal from the common mode signal; a feedback loop from the baseband receiver to the adaptive filter; 22. The receiver chip of claim 21 comprising:

32. A receiver chip for wired communications using baseband transmission, the receiver chip comprising: a baseband receiver configured to receive an input signal transmitted over a wired link, the input signal being encoded with digital data, the baseband receiver recovering the digital data from the received input signal; EMI monitoring circuitry for producing an EMI detection signal from the received input signal, the EMI detection signal indicating the presence of EMI in the received input signal; safety circuitry coupled to the EMI monitoring circuitry and to the baseband receiver, the safety circuitry modifying operation of the baseband receiver in response to the EMI detection signal indicating the presence of EMI, and the safety circuitry initiating a shutdown of the wired link in response to the EMI detection signal indicating the presence of EMI and the baseband receiver indicating an unacceptable quality of digital data recovery; A receiver chip comprising:

33. A receiver chip for wired communications using baseband transmission, the receiver chip comprising: a baseband receiver configured to receive an input signal transmitted over a wired link, the input signal being encoded with digital data, the baseband receiver recovering the digital data from the received input signal; EMI monitoring circuitry for producing an EMI detection signal from the received input signal, the EMI detection signal indicating the presence of EMI in the received input signal; safety circuitry coupled to the EMI monitoring circuitry and the baseband receiver, the safety circuitry modifying operation of the baseband receiver in response to the EMI detection signal indicating the presence of EMI, the safety circuitry implementing a state machine, state transitions of the state machine depending on a set of variables indicating whether transmission quality is acceptable, whether EMI is present, and whether EMI cancellation is sufficient; A receiver chip comprising:

34. 1. A method for wired communication using baseband transmission, the method comprising: receiving an input signal transmitted over a wired link, the input signal being encoded with digital data; producing an EMI cancellation signal that is an estimate of the EMI in the received input signal; producing an EMI detection signal from the EMI cancellation signal, the EMI detection signal indicating the presence of EMI in the received input signal; recovering the digital data from the received input signal, the recovering including using the EMI cancellation signal to reduce the effect of the EMI on recovering the digital data; modifying the recovery of the digital data in response to the EMI detection signal indicating the presence of EMI; A method comprising:

35. 35. The method of claim 34, wherein restoring the digital data is adaptive, and modifying the restoration of the digital data includes ceasing adaptation in response to the EMI detection signal indicating the presence of the EMI.

36. 35. The method of claim 34, further comprising generating an alert in response to the EMI detection signal indicating the presence of the EMI.

37. The method of claim 34, further comprising initiating a shutdown of the wired link in response to the EMI detection signal indicating the presence of the EMI and unacceptable quality of digital data recovery.

38. The method of claim 34, wherein modifying the digital data reconstruction is based on a state of a state machine, the state transitions of the state machine depending on a set of variables indicating whether transmission quality is acceptable, whether EMI is present, and whether EMI cancellation is sufficient.

39. The method of claim 34, wherein the input signal comprises a differential input signal, common mode noise induces coupled noise in the differential input signal, and producing the EMI cancellation signal includes extracting the coupled noise from the differential input signal.

40. The method of claim 39, wherein restoring the digital data is adaptive, and modifying the restoration of the digital data includes ceasing adaptation in response to the EMI detection signal indicating the presence of the EMI.

Images