Keysight Technologies Precision Digital Noise - New Noise Technology and its Application

Transcription

1 Keysight Technologies Precision Digital Noise - New Noise Technology and its Application Application Note Isn t all noise equal? What is the value of different types of noise?

2 Introduction Noise sources are used in many applications. Test and measurement examples include: Noise igure measurements from analog components Jitter tolerance Signal-to-noise ratio tests of digital receivers. It is important to simulate noise based malfunctions to identify additive noise produced by receiving systems. You need to optimize the design, prove that you follow certain standards, or verify the signal is working under real world conditions. Various noise sources have different advantages and disadvantages for different applications. Keysight Technologies, Inc. has developed a new approach for precision digital noise sources, with signiicant advantages over traditional analog and digital sources: Noise distribution is deterministic and controllable. The noise signal is stable and repeatable with a selectable crest factor.

3 03 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Noise Deinition Everybody knows and experiences noise: trafic, the airport, construction sites and even music are all noise sources. Wikipedia deines noise as unwanted sound or noise pollution. Electronic noise is an unwanted signal characteristic of electronic circuits. Encarta deines noise as...random, unpredictable, and undesirable signals that mask the desired information content. Noise is something that no one wants, but it is extremely important to simulate noise situations to understand design and its limitations better. Noise Sources Noise sources are used in many test and measurement applications where we need to verify that components or systems can operate under noise conditions. Jitter and noise cause misalignment of edges and levels, resulting in data errors. Noise is the misalignment on the level and jitter is the misalignment on the edges. Noise on digital data signals causes jitter because it offsets the point in time where the signal crosses the receiver s decision threshold. The peak-to-peak noise amplitude (V noise ) is added to the data signal s DC offset. The jitter increases as the ratio of noise amplitude to signal amplitude gets larger. It also increases with longer transition times. Noise is, by its nature, unpredictable because it can have many different causes, from signal interference caused by sudden voltage changes, to distortions introduced during transmission. It is important to be able to simulate noise-based malfunctions, for example to identify the additive noise produced by receiving systems. It is cheaper to lower the noise igure than to increase the transmitter power.

4 04 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Applications Signal/noise ratio BER tests On the component level, we measure noise igures of analog components, and signal-to-noise (Eb/No) performance. The signal-to-noise ratio (SNR) is deined as the ratio of a signal power to the noise power corrupting the signal. In digital systems it s common to express SNR as energy-per-bit per-noise-power spectral density. A high noise level relative to the signal results in high error ratios. BER is related to the SNR since noise increases occasional bit errors. The ratio of signal strength relative to background noise is usually measured in db. Communication engineers always strive to maximize the SNR to improve the throughput by minimizing noise. Random jitter generation (tolerance tests) Random jitter is unpredictable and is usually caused by physical noise process. Its distribution is assumed (even deined) in many speciications to follow the Gaussian distribution. The extent to which thus is true is a matter of debate. The most relevant RJ mechanisms are thermal noise, shot noise, and licker noise. Thermal noise is also called Johnson-Nyquest noise, and is a form of white noise, which means that its magnitude spectrum is lat on a logarithmic frequency scale. Thermal noise is generated by the luctuation of the electronic current inside an electrical conductor due to the random thermal motion of the charge carriers. An instrument with higher bandwidth is subject to more thermal noise than an instrument with lower bandwidth. Shot noise is another type of electronic noise with a white power density spectrum. It occurs when the inite number of particles that carry energy, such as electrons in an electronic circuit, give rise to detectable statistical luctuations in a measurement. Flicker noise, also called pink noise, has a power spectrum that is proportional to 1/f, where f = frequency. It results from a variety of effects, such as impurities in a conductive channel and generation and recombination noise in a transistor (due to base current). It is always related to direct current. Flicker noise is of concern only at low frequencies usually below 2 khz. Given that many of the RJ sources are Gaussian, the jitter analysis is accepted as Gaussian distribution. There are other types of jitter as well, such as deterministic jitter which is predictable. It contains periodic and data-dependent jitter. Periodic jitter is repetitive with a certain frequency and is caused by electromagnetic pick up for periodic signals such as oscillators or switching power supplies. Data-dependent jitter is correlated to the actual data content. Understanding the impact of the different parts of jitter helps one to improve performance. Jitter tolerance testing is a typical R&D task in the development stage, starting with turning on and debugging a system. This is done on chip tests and perhaps on early ASICs. The second stage is the complete and systematic characterization of deined parameters the design veriication stage. The inal task is to prove complete compliance to a standard or other published speciication (e.g., S-ATA and MIPI). Jitter tolerance DUT BER Figure 2. How much jitter is tolerated by my device under test?

5 05 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Total jitter (TJ) Deterministic jitter (DJ) Random jitter (RJ) - Gaussian unbounded - Gaussian bounded Example: thermal noise Data dependent jitter (DDJ) - Duty cycle distortion - Intersymbol interference Example: Bandwidth limit of cables and backplanes Superimposition of all kinds of jitter Periodic jitter (DDJ) - Sinusoidal - Periodic Example: Pickup of other periodic jitter, like power supply Page 25 Thermal noise Flicker noise Shot noise - Johnson-Nyquist Noise - White noise P = kt f P = average power f = Boltzmann s constant T = absolute temperature - Pink noise Power density spectrum is proportion 1/f - White noise Standard deviation is: si = v2ql f q = elementary charge I = average current through the device f = frequency span Figure 3. Types of jitter

6 06 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Mobile industry processor interface - MIPI Test description Mobile industry processor interface (MIPI) is a set of standards for hardware and software interfaces between processors and peripherals (for example phone, camera and display). The increase in functionality requires more bandwidth, high-speed serial architecture, a digital interface and a different power management system. Jitter behavior is especially signiicant does the device still function even if the signal is distorted. The test contains a nominal state and a stress test which contains random jitter, sinusoidal jitter, and customer jitter. Random jitter has a Gaussian probability density function with a theoretically unbounded magnitude. This is why random jitter has so much more leverage on low-probability error performance and is a preferred test set up. Sinusoidal jitter (SJ) is jitter that is periodic and uncorrelated to serial data. It is typically caused by an external deterministic noise source coupling into a system, such as switching power supply noise or crosstalk from a nearby printed circuit board trace carrying a periodic signal. It can be caused by an unstable clock as well. Sinusoidal jitter is easy to create and can be calibrated with high accuracy. The repeatability is high. SJ is used to characterize receiver jitter tolerance along with its jitter transfer function. It reveals how jitter can travel through the entire digital communication system including cascading links. Accurate measurements provide valuable insights into devices under test (DUT). Mobile systems phone, camera, displays, wireless networked devices like handhelds (PDA) TX tests: - Data bus timing - Transition times - DC levels and AC swing - LP/HS mode switching - Jitter RX tests: - Data bus timing - Minimum pulse width - Sensitivity (min/max amplitude) - Differential and common mode, termination switching - Jitter tolerance on clock and data Display RX TX MIPI D-Phy CSI Camera TX MIPI D-Phy DSI DigRF v3 BB-IC RX TX TX RX RX MIPI D-Phy RF-IC 2.5G 3GPP WiMAX All jitter types are available on the 81150A (RJ, SJ, custom J) Stress Stimulus data Generator solution RX Device under test (DUT) Expected data Compare Data analyzer error detector Errors Bit error ratio Bit error ratio tester (BERT) Figure 4. Mobile industry processor interface - MIPI

7 07 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Serial advance technology attachment Deterministic jitter (RJ) - Channel 1 sine wave - Channel 2 noise - Channel add - eliminates power divider Random jitter - Choose crest factor 7 The 81150A provides SJ and RJ in one box. RJ source SJ source Framed COMP pattern generator (81134A) TX SMA cable Setup calibration DSO80000 oscilloscope SMA to SATA fixture Same equipment used for 00B tests Turn on test mode 00B & BISTL source Frame error rate counter C-H-S FER TX RX RX S-ATA cable RX 00B detect TX 10/8b 10/8b Scramble Retiming Scramble PUT Does your transceiver/receiver function correctly, even under stress? Figure 5. Serial advance technology attachment interoperability program testing Serial advance technology attachment (S-ATA) is a computer bus designed to transfer data between a computer and mass storage devices. The S-ATA-IO test veriies receiver and transceiver functionality under stressed signal conditions. The test setup should ideally include a source providing sinusoidal jitter and random jitter. A crest factor of 7 is needed for random jitter in the compliance test. On the top right side of Figure 5, a crest factor of 7 is recommended. The instrument steps are: Press the noise button Press the probability density function Identify the crest factor. Generating deterministic jitter is just as simple: Conigure a sine wave on one channel Conigure the noise on the second channel Add channel 2 to channel 1. The set up is very simple and straight forward. The noise is added to a data generator; in this case, the 81134A, via the delay line. Baseband simulations Noise sources can be used as a Gaussian modulating signal source to mimic real world conditions. With this it is possible to predict how communication devices will react under real world conditions but in a repeatable laboratory test. Another example of baseband simulation is CDMA in a spread system. CDMA (code division multiple access) technology emerged as an alternative to GSM cellular architecture. Europe chose GSM for frequency hopping, while the US and part of Asia opted for spread-spectrum technology which uses wide bandwidths, and noise-like signals that are hard to detect and dificult to intercept or demodulate.

8 08 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Jamming and encryption Noise sources can be used to simulate jammers for RF systems, such as missile systems. With the usage of GPS systems for guidance, the need to develop anti-jamming systems has increased. Testing GPS receiver systems necessitates the simulated jamming of signals by incorporate white Gaussian noise in the receiver s frequency band. The noise has a lat frequency response and a Gaussian distribution. Other noise source applications include sample dithering in high-speed analog-to-digital converters and high security encryption systems. Electrical thermal noise sources Electrical thermal noise sources are more random than any other source in nature. Sampling a noise source voltage at any time creates a random result that can be used in encryption. Despite all the pseudo-random encryption efforts being designed into wireless and wired communication and computer networks, the codes can still be cracked. Using a true noise source is perhaps the only way to trump the hackers. Random noise is used in many different applications and is absolutely critical in technology. The ideal noise source All the applications require different types of noise to simulate different issues. The ideal noise source should be calibrated, deterministic, and with well known properties. Otherwise, measurements would not be repeatable, and therefore utterly meaningless. On the other hand, we require that the noise signal be random. To ask for deterministic behavior in a random process is an interesting contradiction. White Gaussian noise Most real noise, including white noise, has a Gaussian distribution. Interestingly, the German 10 Deutche Mark bill sports the Gaussian distribution function (graph and equation) as well as Carl Friedrich Gauss himself. This is the Gaussian frequency distribution equation: 1 ( (x - µ) 2 ) exp σ 2 π 2σ 2 where x is the independent value sigma square is the variance (RMS) value μ is the mean of the distribution The probability density function of a Gaussian process is unbounded, limited only by the value of x. The probability is never zero. The frequency spectrum of an ideal Gaussian noise source is lat from DC to ininity. The lat noise spectrum is sometimes called a white noise spectrum, in contrast to pink, brown, and other noise types where the signal amplitude is a function of the frequency. Figure 6. The Gaussian distribution function on the German 10 DM bank note

9 09 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note The crest factor The crest factor is an indicator of the signal quality. A sine wave has a crest factor of 1.4. Higher crest factors indicate more noise on the signal. The crest factor is a calculation of voltage peak-to-peak (Vpp) over RMS (root mean square) or voltage peak / RMS. Both calculations are common in the industry. Incidence Figure 7. Text Crest factor = Vpp/RMS = Vpeak/RMS V p V RM Mean δ Voltage Some standards, e.g. S-ATA, require a crest factor of 7 / 14 to achieve a BER of 10 to 12. Only with this functionality can the compliance test pass. If you test your receiver with a smaller crest factor, the device might pass even if it is a bad device. Incidence Different crest factors The graphical representation of bit error rates is the bathtub curve (see Figure 9). It measures BER values versus the sample delay offset using a linear scale for the BER values. Because of the characteristic shape of this plot, it is often called a bathtub curve. A smaller crest factor has a steeper decrease. CF CF Mean δ Figure 8. Text BER and the different crest factors Voltage The eye opening is different as well, the higher the crest factor is the closer the eye is. If the curve is broader, the eye closes more. Therefore the higher crest factor achieves more stress. BER CF = 5 CF = 7 Figure 9. Bath tub curve The eye diagram with different CFs Figure 10. Comparison of different noise sources analog noise

10 10 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Comparison of different noise sources: analog noise Analog noise sources use an effect that is unwanted in virtually all other applications - namely resistive materials are used to produce a small amount of noise. Caused by random thermal movements of electrons in the material, the thermal noise power is proportional to the temperature of the material. Thermal noise comes very close to the ideal noise signal: its distribution function is Gaussian over a wide range, and the frequency spectrum is lat over the entire frequency range. The original analog noise signal needs to be ampliied with a very high gain in order to get usable signal amplitudes. The required analog RF components make analog noise sources expensive, and hard to calibrate because of non-linearity and other effects in ampliiers. If the source is constructed carefully the generated noise signal has a very high bandwidth and excellent crest factors. Arbitrary waveform generator Traditional digital noise sources are based on a large sample memory that is illed with pre-calculated random values, and a digital-to-analog converter plus optional digital signal processing. While inexpensive, digital sources have limited crest factors and a non-ideal frequency spectrum. Gaussian noise is the standard waveform of commercial AWGs and due to the nature of its generation the crest factor could be set by entering the appropriate waveform into the memory. You just have to know the waveform. But even if you have such a waveform which is not very likely there are other limitations. Due to limited memory within AWGs only a low crest factor is practical, which makes them relatively useless as a random voltage source for real test scenarios. For 10 Mwords of memory, a crest factor of 5 can be achieved. To reach a crest factor of 7, which is a meaningful measurement, unreasonable deep memory of more than 1 Tbit would be required which is not practical. Linear address generation Sample memory Digital analog converter (DAC) Figure 11. Arbitrary waveform generator

11 11 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Precision digital noise source The block diagram is very similar to traditional memory-based noise generation, but there is one difference: the sample memory holds a sample distribution rather than pre-calculated samples. The sample memory is read using a random address generator. The advantage of this approach is that the voltage domain is separated from the time domain. The memory content determines the distribution of the output signal, while the address generation determines the repetition rate and randomness of the output signal. Due to the huge virtual address range, the generated noise signal repeats after 26 days. In triggered mode, the random address generation will be reset to its start conditions whenever the trigger event occurs. This allows you to generate the same random noise sequence again without the need to wait for the repetition time to elapse. The gated mode will block the output signal when the gate signal is inactive. The noise source will continue generating random addresses while the output signal is blocked. Depending on your gate you will get different parts of the noise signal. Using a mathematic calculation allows one to use either a Gaussian or arbitrary distribution. Random address generation Sample memory Digital filter (FIR) Digital analog converter (DAC) Figure 12. Precision digital noise source

12 12 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Keysight s offering The Keysight 81150A pulse function arbitrary noise generator sets the standard for the next generation of labs for fast, accurate insight into your design and device under test. It s a pulse generator with precise signals for performance veriication and characterization. It is also a function arbitrary generator for versatile signal generation to optimize testing and for modulation to shape the signal to the DUT s needs. Furthermore, it s a noise generator to distort signals and build worst-case scenarios. The 81150A provides deterministic Gaussian white noise, with signal repetition of 26 days. You can decide on any arbitrary distribution, and trigger the noise to start when you need it. You can select one of four crest factor values an indicator of signal quality depending on the standard being veriied. Noise sensitivity tests can be done with the touch of a inger, by just pressing a different probability density function. Three instruments in one: What you need: Pulse generator 1 µhz 120 MHz pulse with variable rise/fall time; trigger and clock up to 120 MHz; coupled/uncoupled channels Test the DUT and not the source Superior precision pulses with high timing stability guarantee reproducible tests. Function arbitrary generator 1 µhz 240 MHz sine; 14 bit, 2 GSa/s arbitrary waveforms; FM, AM, PM, FSK, PWM up to 10 MHz modulation frequency, internally or externally Stress your device to its limits Versatile waveforms and modulation capabilities adapt the signal to any real world signal. Noise generator Unique! Crest factor (peak/rms) selectable 3.1, 4.8, 6, 7; noise type: repeatable, random and triggerable; signal repetition after 26 days Repeatable and random noise Combines two required extremes: Repeatable noise with long repetition rates for simple problem identification. Figure 13. Noise generator features and values

13 13 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note 81150A set up screen for the different probility density functions Noise with a crest factor of A noise set up screen Noise with a crest factor of A set up screen for non-gaussian distribution Exponential rise distribution Figure 14. Different noise setups for probability density function

14 14 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note Related Keysight Literature Publication title Pub number Pulse Pattern and Data Generators Brochure E Keysight 81150A Pulse Function Noise Generator Demo Guide Keysight 81150A Pulse Function Noise Generator Demo Guide Keysight 81150A Pulse Function Arbitrary Noise Generator Flyer Keysight 81150A Pulse Function Arbitrary Noise Generator Application Booklet EN EN EN

15 15 Keysight Precision Digital Noise - New Noise Technology and its Application - Application Note mykeysight A personalized view into the information most relevant to you. AdvancedTCA Extensions for Instrumentation and Test (AXIe) is an open standard that extends the AdvancedTCA for general purpose and semiconductor test. Keysight is a founding member of the AXIe consortium. ATCA, AdvancedTCA, and the ATCA logo are registered US trademarks of the PCI Industrial Computer Manufacturers Group. LAN extensions for Instruments puts the power of Ethernet and the Web inside your test systems. Keysight is a founding member of the LXI consortium. PCI extensions for Instrumentation (PXI) modular instrumentation delivers a rugged, PC-based high-performance measurement and automation system. Three-Year Warranty Keysight s commitment to superior product quality and lower total cost of ownership. The only test and measurement company with three-year warranty standard on all instruments, worldwide. Keysight Assurance Plans Up to five years of protection and no budgetary surprises to ensure your instruments are operating to specification so you can rely on accurate measurements. Keysight Technologies, Inc. DEKRA Certified ISO 9001:2008 Quality Management System Keysight Channel Partners Get the best of both worlds: Keysight s measurement expertise and product breadth, combined with channel partner convenience. For more information on Keysight Technologies products, applications or services, please contact your local Keysight office. The complete list is available at: Americas Canada (877) Brazil Mexico United States (800) Asia Paciic Australia China Hong Kong India Japan 0120 (421) 345 Korea Malaysia Singapore Taiwan Other AP Countries (65) Europe & Middle East Austria Belgium Finland France Germany Ireland Israel Italy Luxembourg Netherlands Russia Spain Sweden Switzerland Opt. 1 (DE) Opt. 2 (FR) Opt. 3 (IT) United Kingdom For other unlisted countries: (BP ) This information is subject to change without notice. Keysight Technologies, Published in USA, July 31, EN