Skip navigation
All Places > Keysight Blogs > Oscilloscopes Blog > Blog > Authors BoonCampbell
1 2 Previous Next

Oscilloscopes Blog

20 Posts authored by: BoonCampbell Employee

Have you ever been fighting noise on your measurements and can’t tell where it’s coming from? There are four possible contributors:

  1. Your device under test
  2. Your probe
  3. Your oscilloscope
  4. Or a combination of all three.

 

Learn about the different ways you can minimize noise impacts and improve the quality of your measurements.

 

System Noise Consequences
You should look at your probe and oscilloscope together as one measurement system that can add noise to your measurement. Figure 1 below shows two possible noise sources from your system: one source from your probe and the other from your oscilloscope. Some amount of noise will come from the probe’s amplifier system and ride on your DUT’s signals, which is sent to the attenuator of the oscilloscope. All scopes use an attenuator to vary the vertical scale on your oscilloscope screen. Most oscilloscopes can detect your probe's attenuation ratio and will automatically adjust its vertical scale accordingly. For example, for a 10:1 probe, the oscilloscope will simply amplify both the signal and noise by a factor of ten. Be sure to keep this in mind as you minimize the noise in your signal.

 

 

Signal to noise diagram

Figure 1: Signal-to-noise diagram of the oscilloscope and probe

 

Probe Noise Impacts
Due to added inductance, the probe ground/signal loop formed by the probe and tip contributes to the noise on your signal. This noise can be reduced by:
1) Selecting just enough scope/probe bandwidth to measure your DUTs signals. Excessive bandwidth will contribute to the system’s overall noise.
2) Setting your oscilloscope’s vertical range to the most sensitive voltage range possible while still seeing your DUT’s complete signal on the oscilloscope. This will reduce the amount of gain the oscilloscope needs.

 

Oscilloscope noise floor with no probe connected

Figure 2: Oscilloscope noise floor with no probe connected

 

Oscilloscope noise floor probe connected

Figure 3: Oscilloscope noise floor probe connected

 

In Figure 2, you can see the noise floor of an oscilloscope with no probe attached is 295 µV root mean square (rms).
In Figure 3, with the probe attached to the oscilloscope, the noise increases from 295 to 485 µV rms. So, the probe itself is adding around 200 µV rms (or 67% more noise)! This noise level will reduce when your probe is well grounded, but it is worth noting the increased noise level just by adding a probe. Keep your ground and tip lengths as short as possible to reduce this effect.

 

Probe Attenuation Impacts
The probe attenuation you need is going to depend on the Voltage of the signal you are measuring. The attenuation ratio changes how the signals are fed into your oscilloscope. For example, a 10:1 probe connected to a 1V signal will pass 100 mV to the scope’s input.The oscilloscope will either read (or you can manually enter) the probe’s attenuation ratio. Then the oscilloscope will display the correct signal, factoring in the probe’s attenuation ratio. Having a higher attenuation ratio (100:1, 1000:1) will allow you to view higher voltages, but it will also make the scope’s internal amplifier noise more pronounced. The higher the attenuation ratio, the more scope noise you’ll see. For example, a 10:1 probe will show 10x the noise.

 

One easy way to estimate the amount of your probe noise is to check the attenuation ratio and the probe noise level from the probe’s data sheet or manual. Many probe manufacturers characterize the probe’s noise as equivalent input noise (EIN) and will be listed in volts rms.

 

The probe pictures in Figure 4 (from bottom left going clockwise) are examples of a 10:1 passive; 10:1 single-ended active; 50:1 or 500:1 high voltage; and a 1000:1 high voltage probe. These attenuation ratios are needed to reduce the probed signal down to levels that the oscilloscope attenuators can handle and display on the screen without clipping.

 Examples of probe attenuation and voltage levels
Figure 4: Examples of probe attenuation and voltage levels

 

Common attenuation guidelines and limitations are shown in Figure 5 below. Keep this in mind when determining the attenuation ratio you need.

 

1:110:1100:1
Suitable formeasuring low voltage, low frequency signals (<~25MHz)general purpose measurementhigh voltage measurement
Limitationslimited bandwidth, dynamic rangetypically up to 300 Vhigh probe noise

Figure 5: Probe attenuation guidelines

 

Figure 6 below compares the same signal measured by both a 1:1 and a 10:1 passive probe. The screen shot clearly shows how attenuation from a 10:1 probe can cause the oscilloscope to amplify both the signal and noise. The result is an exaggerated level of noise in the signal (green trace).

 

A 50 mVp-p sinewave measured with bot 1:1 and 10:1 probes - overstated 36%

Figure 6: A 50 mVp-p sine wave measured with both a 1:1 and 10:1 probe.

 

Higher attenuation ratios lead to higher levels of noise shown on the oscilloscope. This might lead you to believe that you should always use a 1:1 probe. But that’s not true! Lower attenuation probes typically have much higher loading on your system and may have lower dynamic range. There are tradeoffs, and you will need to pick the probe that fits your measurement best. You want a probe that can effectively measure the level of voltage on your DUT with the least amount of attenuation and lowest loading effects.

 

Conclusion
Your system noise can be exaggerated by probe and oscilloscope noise levels. Selecting the correct probe for your application with the correct attenuation ratio will lower the probe/oscilloscope added noise. As a result, your measured signal is a cleaner representation of what is on your DUT.

 

Are you falling into oscilloscope probing pitfalls? Avoid making the same mistakes as others with the Oscilloscope Probing Pitfalls eBook

Did you know that when you probe your DUT, the probe becomes part of the circuit? All probes have a loading effect on your circuit to some extent. These effects can manifest into overshoot, ringing, slow rise/fall times, propagation delays and DC offset problems. In addition, the loading impacts vary as you probe from DC voltages through high frequency ranges. Over this large frequency span, your probe impedance can vary greatly.

 

RCRC vs RC Impedance Characteristics

The capacitive and inductive components in a probe are what causes loading on your device. The traditional model off a probe’s looks more like the read trace in figure 1. However, newly developed high-end and high frequency active probes two knee or crossover points (RCRC) and provide different loading responses. Understanding the probe’s input impedance characteristics over frequency enables you to make the best probe selection for the circuits you are testing.

 

Let’s look at some probe input impedance vs frequency curves in Figure 1 below to understand the impact to your measurements. A probe’s input impedance is shown on the vertical axis and frequency is shown on the horizontal axis. Both RC and RCRC probe curves are shown.


The red trace is a typical RC probe response over frequency. Note that from DC to around 10 MHz, the RC probe holds steady at a 50 K? of differential impedance. Higher than 10 MHz, the RC probe’s capacitive reactance comes into play at 210 fF, and the probe impedance continues to decrease as the frequency increases. This is what is called an RC input impedance profile of most conventional probes on the market.

Lower impedance will have accumulative loading impacts on the circuit you are probing.

 

The blue trace is an RCRC probe’s response. Notice from DC to 10 Hz the inductance is at 100 K? and then falls to 1K? from 10 Hz to 10 KHz. The 1 K? inductance will load your circuit more than the RC probe’s 50 K? in this frequency band, but past 10 MHz, the RC loading will be much worse because the RC probes impedance decreases rapidly driven by the capacitance of the probe. The RCRC holds this 1 K? impedance from 10 KHz to around 1 GHz. Past that, the capacitive reactance at 32 fF starts to come into play, reducing the 1 K? impedance further. So, you can see at higher frequencies, above several hundred MHz, the RCRC probe proves to be the better choice because it will decrease loading effects at higher frequencies.

 

input impedance vs frequency of probes
Figure 1: Input Impedance vs Frequency of modern high-performance probes

Red = RC probe example
Blue = RCRC probe example
Pink = RCRC probe example

 

The pink trace is another RCRC probe’s response for additional comparisons. Note that from DC to around 100 MHz the impedance is 100 K?. But from 100 KHz to 10 MHz the probe’s 110 pF capacitive reduces the inductance to 450 ?s. This change in impedance results in a significant amount of additional loading relative to the initial 100 K?s at lower frequencies. And then at 100 MHz and above, the probe’s 65 fF capacitance reduces the impedance further.

 

To summarize the curves in Figure 1, your probe selection for the lowest circuit loading should be:

  • RC probes Higher input impedance for lower loading at mid band (kHz to GHz)
  • RCRC probes – Higher input impedance for lower loading at higher bands (>GHz)

 

Applications tips for each probe type

Use an RCRC probe for:

  • Accurate high frequency content above GHz due to low loading
  • High speed signals with low source impedance, such as a 50 ohm transmission line
  • Reproducing wave shapes with fast edge speeds

 

Use an RC probe for:

  • Mid-band frequencies due to low loading
  • Buses that transition to a “high Z” state such as DDR and MIPI signals
  • Signal sources with high impedance
  • Signals with long time constants

 

Conclusion

A common misconception is that a higher priced, higher bandwidth probe can more effectively measure signals across all bandwidths. However, this is not the case. The best probe for your application will be dependent on what frequencies you are working with. Always factor in the probe loading effects on your measurement.

 

Are you falling into oscilloscope probing pitfalls? Avoid making the same mistakes as others with the Oscilloscope Probing Pitfalls eBook.

The terms “bit error rate” and “bit error ratio” are used interchangeably on many websites and publications. However, the definitions are very different. Understanding the difference will help you effectively analyze your system’s performance.

 

What is a BERT?

To find the bit error rate or bit error ratio of your system, you need a Bit Error Rate Tester or Bit Error Ratio Tester (BERT). BERT refers to a class of test equipment; depending upon the manufacturer or distributor, BERT stands for Bit Error Rate Tester or Bit Error Ratio Tester. A BERT tests the complete transmitter/receiver system for any data loss. It transmits data into a system, and then measures how well a system transmits and receives the data. To do this, a BERT requires a pattern generator and error detector.

 

What is the difference?

There is a very clear difference between the error ratio and rate. Understanding that difference is important to assess your system performance.

 

Bit error ratio (BER) is the number of bit errors divided by the total number of bits transferred during a specific time interval.

Bit error rate (also BER) is the number of bit errors per unit time.

 

Essentially, the bit error rate refers to errors with respect to time, and the bit error ratio refers to errors with respect to the quantity of transferred bits.

 

The bit error ratio is a unitless performance calculation and is expressed as a percentage. It is an estimate of the bit error probability which is the expected value of the bit error ratio. This estimate is more accurate over a longer time interval and when capturing a high number of bit errors.

 

bit error ratio

Why it’s important differentiate?

It’s important to differentiate between bit error rates and bit error ratios. If your BERT pattern generator sends 100 bits to your device under test and your BERT error detector sees 10 errors, the bit error ratio is 10 percent.

 

bit error ratio

The bit error rate is the bit error ratio multiplied by the bit rate. For example, if your BERT pattern generator sends bits to your unit under test at a rate of 100 bits/second and your BERT error detector sees 10 errors every 100 bits, the bit error rate would be the bit error rate equals 10 bits/second.

 

bit error rate

 

The bit error rate is used more often because it tells you how long it will take to encounter an error. For example, using the calculated 0.1 bit error ratio above tells you the ratio between errors received and number of data bits sent.

 

But what does knowing the bit error ratio really tell you about your system performance? Not much - you need to know your data rate. If your system data rate was 1 bit per week, then your system calculated bit error rate would only be one error in 10 weeks. Another example would be if your system data rate was 100 G bits/second, then your bit error rate would be 10,000,000,000 errors every second!

 

Summary

The bit error ratio is the number of bit errors divided by the total number of bits transferred during a specific time interval. Bit error rate is the number of bit errors per unit time. The bit error rate gives you an indication of your system’s performance relative to bits transferred vs bits received. Visit Keysight.com to learn more about Keysight’s bit error ratio test options.

You just bought a high frequency single-ended and differential probe and are feeling good about your measurements. But when using the two probes side-by-side, you notice differences in vertical voltage measurements and rising edge times. This can cause you to question the accuracy of your measurements relative to what your device under test (DUT) signals should look like. Don’t worry, you’re not alone here. Let’s walk through the causes of these differences and the steps needed to correct them.

 

The Problem

The signal path from the DUT to the oscilloscope can create distortions in your signal and result in amplitude and timing differences, which can cause errors in your design. A typical signal path is shown in Figure 1 below. The original signal is what is on your DUT before probing. When you probe your DUT, the probe may add distortions to your signal due to probe capacitance, inductance, or slight impedance mismatches. See my other probing blogs for these effects and how to avoid them. The signal on the probe tip is amplified and sent to the oscilloscope input. The oscilloscope then converts this analog signal to a digitized version using digital signal processing (DSP).

 

Oscilloscope Probing: Simplified signal path.

Figure 1. Simplified signal path.

 

Incorrect skew and amplitude measurements can cause problems with your measurements. Skew is defined as the difference in bit timing relative to the same point on two waveforms. Along with these pesky timing differences, the amplitude differences can cause your DUT’s eye patterns to look more closed than they really are. This may make it seem like there is a lot jitter or noise in your DUT, when in reality it is coming from the signal path. Not seeing a true representation of your DUT’s signals on the oscilloscope can make it difficult (if not impossible) for you to debug or complete your design work.

 

Probe Calibration

Probe calibration is sometimes forgotten, but this is an important step to ensure the most accurate measurement possible. Both single-ended and differential probes come with a calibration certificate. These certificates will give you confidence that the probes will meet their stated specifications. However, the probe and oscilloscope are a measurement system, which means each probe’s output and each oscilloscope’s input characteristics will vary. The coupling needed in the signal path above creates three variabilities: (1) coupling from your DUT to the probe, (2) coupling from the probe output to the oscilloscope input, and (3) coupling from the oscilloscope input through its DSP circuitry. Let’s take some of this variability out of the system by calibrating your probe to improve amplitude and skew.

 

Keysight oscilloscopes can recognize Keysight active probes and give you a message on the screen if your probe is not calibrated to your oscilloscope. Figure 2 below shows a 50 MHz signal from a signal generator before probing, using our E2655C deskew fixture (yellow trace). The signal is also sent through a differential probe and shown on Channel 2 of the oscilloscope (green trace). The generator output on Channel 1 is 1.04 Vp-p (volts peak-to-peak), and the probed signal on Channel 2 is 965 mV (millivolts). In addition, the skew from Channel 1 to Channel 2 is massive (around 3 ms), which means the rise times do not line up at all.

 
Oscilloscope Probing: Generator output and probed signal.

 

Figure 2. Generator output and probed signal.

 

Click on the Channel 2 green button; you will see the probe calibration button in the lower right corner (see Figure 3 below showing a Keysight S-Series screenshot). Click on this button and walk through the amplitude and skew calibrations in less than five seconds each. The oscilloscope will prompt you when each calibration is complete.

 

Oscilloscope Probing: Channel 2 probe calibration selection.

Figure 3. Channel 2 probe calibration selection.

 

Note the calibration results in Figure 4 below. This screenshot is after amplitude and skew calibration. The amplitude is now improved to 972 mVp-p, and the skew has been corrected with both rise times lining up.

 
Oscilloscope Probing: After amplitude and skew calibration.

 

Figure 4. After amplitude and skew calibration.

 

Conclusion

The system links between your DUT, probe, and oscilloscope can cause errors in your measurements relative to what signals are really on your DUT. Always calibrate your probe with the oscilloscope you are using for the best fidelity in your measured signal.

 

Are you falling into oscilloscope probing pitfalls? Avoid making the same mistakes as others with the Oscilloscope Probing Pitfalls eBook.

Many people think that differential probes are made to only probe differential signals. Did you know you can also probe single-ended signals with your differential probe? Today we’ll learn how to maximize the usage of your differential probe and get the best signal fidelity possible. You will need to examine the performance and usability trade-offs between your differential and active probes to make this decision. The simplified models of each probe are shown in Figure 1 below and will be referenced throughout this blog.

 

Differential Probes - Simplified models of single-ended and differential probes.

Figure 1. Simplified models of single-ended and differential probes.

 

Signal Fidelity

Because the physical geometries of the two probes are similar, the values of the capacitors and inductors will also be similar. The ground connection of the single-ended active probe (lg) is sensitive to the length the ground accessory used in the measurement. Since you are not measuring relative to ground with a differential probe (you are measuring relative to another voltage), you are not making an additional LC circuit with a ground lead, so you see much less inductance with a differential probe.

 

Note that the differential probe tips connecting to the DUT (device under test) have a tip resistor on both tips (Zp and Zm). Whereas the single-ended probe has a tip resistor (Zs) on the signal connection and none the ground connection. These tip resistors damp the resonance caused by the probe input inductors and capacitors (ls, lg, and cs in the single-ended probe and lp, lm, cp, and cm in the differential probe).

 

There is a lack of damping on single-ended probe ground connections. Because of this, a voltage can develop between your probe’s ground and your DUT’s ground. This happens at higher frequencies and reduces the input to your single-ended probes amplifier input, thereby reducing your probe’s output. This disadvantage of your single-ended probe can be minimized by keeping your ground lead length as short as possible. 

 

Common Mode Rejection

Unwanted common mode signals can also affect the signal you see on screen. They can be caused by electromagnetic interference, cross talk, and other noise characteristics on or around your DUT. Your probe needs to be able to reject common mode noise in order to give you the most accurate signal. The amount a probe can reject is represented by a ratio: common mode rejection ratio CMRR. But this rejection is only relative to the signals that appear simultaneously and in-phase on both probes’ inputs. You can see in Figure 2 below that the single-ended probe (green line) has less rejection over the frequency band measured than the differential probe (dotted red line). Around 1.5 GHz, the difference is close to 20 dB. This is significant since the common mode (noise) will be amplified by the single-ended probe, causing considerably more noise on its output relative to the differential probe.

 

Differential Probes - CMRR of the single-ended vs. the differential probe.

Figure 2. CMRR of the single-ended vs. the differential probe.

 

Measurement Comparisons

In Figure 3 and Figure 4 below you can see:

  • The input signal to the oscilloscope is show in yellow on all the following figures (un-probed and fed directly into Channel 1 of the oscilloscope). We will call this signal Vin.
  • Figure 1 below shows the single-ended probe measurement of Vin on Channel 2 of the oscilloscope (blue trace).
  • Figure 2 below shows the differential probe measurement of Vin on Channel 2 of the oscilloscope (red trace). Note: one side of the differential probe is connected to Vin and the other is connected to the same ground as the single-ended probe.

 

Single-ended measurement. Differential measurement.

    Figure 3. Single-ended measurement.                             Figure 4. Differential measurement.

 

Notice that the two probe measurements look like they carry the same amount of noise relative to the un-probed signal in yellow. But let’s look at little closer and change the volts per division on the oscilloscope from 200 mV/div (millivolts per division) to 20 mV/div. You can now see in Figure 5 below the single-ended measured signal in blue, and in Figure 6 you can see the differential measured signal in red. Notice that the single-ended measurement in blue has much more noise than the differential measurement in red due to less common mode correction by the single-ended probe.

 

 

 Single-ended measurement. Differential measurement.

     Figure 5. Single-ended measurement.                             Figure 6. Differential measurement.

 

Conclusion

Differential probes can make the same types of measurements that single-ended probes can perform. However, due to the common mode rejection on both inputs of the differential probe, the differential measurements can have significantly less noise. The common mode rejection in the differential probe reduces the amount of noise that goes into its amplifier, resulting in much less noise displayed on the oscilloscope. This allows you to see a better representation of your DUT’s signals and not be misled by random probe noise.

 

Are you falling into oscilloscope probing pitfalls? Avoid making the same mistakes as others with the Oscilloscope Probing Pitfalls eBook.

Did you know that when you probe a circuit, you change the electrical characteristics of that circuit? Oscilloscope probes add resistive, capacitive, and inductive loads to your circuit. These loading affects can change the operation of your circuit under test. Understanding these loading impacts helps you avoid selecting the wrong probe for your specific circuit or system.

 

Figure 1 below shows a circuit under test and the electrical model of a probe connected to it. In a perfect world, Vin (voltage at the input of the probe) should be the same as the Vsource (voltage of your circuit before it is probed). But because of probe loading effects, the impedance of your circuit and probe determines the voltage at the input of the probe. It is a simple resistor divider circuit. Increases in frequency can also become a major source of loading because the probe’s capacitive reactance gets smaller. This loading alters not only the amplitude but also the shape of your original circuit waveform.

 

Oscilloscope Probe Electrical Circuit

Figure 1: The probe’s electrical circuit

 

When the probe is connected to the circuit, the impedance matching of the circuit and probe determines the voltage at the input of the probe.

 


Capacitive Loading

Capacitive loading can be the main culprit of your measurement errors. For general-purpose measurements less than 500 MHz, passive 1:1 and 10:1 high-impedance resistor divider probes are good choices. These passive probes begin to impose heavier capacitive loading as the frequency of the signal being measured increases. As the frequency of the signal goes up, the probe impedance drops and can load your circuit. High-impedance passive probes are a great choice for general-purpose debugging and troubleshooting on most analog or digital circuits below 500 MHz.

 

High-impedance active probes are the best selection below 500 MHz.


Inductive Loading

It is critical to remember that your probe’s impedance is not constant over frequency. Most of the inductance is created from the ground lead you chose for your probe. At DC and low frequency ranges, the probe’s impedance starts out at the rated impedance, but as the frequency goes up, inductance comes into play. The result is higher frequency ringing on your rising edge and across the top of your waveform. Figure 2 below shows the four different types of ground lead solutions’ stepped responses of a Keysight N2796A 2 GHz active probe. The three grounding solutions below decrease in inductance starting with the highest inductance in Case 1 to the lowest inductance solution shown in Case 4. Notice that the Case 4 black line solution has the least amount of overshoot and ringing.

 

Active oscilloscope probe step response

Figure 2. An active probe’s stepped response with different accessories.

 

Resistive Loading

Resistive loading is the least likely to induce nonlinear or low amplitude behavior in your circuit. Your circuit’s output resistance and the probe’s own resistance form a voltage divider circuit. This divider circuit distorts the signal being measured because the probe is seen as a load to the circuit under test.

 

1:1 passive probes can cause resistive loading of you circuit under test above 500 MHz.

 

Passive and Active Probes

The higher the passive probe’s attenuation ratio, the lower the capacitive loading will be. 1:1 passive probes have capacitive loading around 100 pf, while a 10:1 probe is around 10 pf. But there is a tradeoff here. 1:1 probes transfer lower noise levels to the oscilloscope. 10:1 passive probes get both their signal and noise amplified by 10x because the oscilloscope accounts for the fact that the probe output is one tenth of the actual measured signal. 

 

10:1 passive probes increase the noise level on the oscilloscope because both the signal and noise floor are amplified by the oscilloscope.

 

Active probes are another way to reduce probe loading. They have around one tenth the input capacitance of passive probes. Active probes can achieve this lower tip capacitance due to the active circuit at the tip of the probe. See these active vs. passive probe relationships below in Table 1 to aid you in your probe selection.

 

Active probesPassive probes
FeaturesFeatures
Low loadingHigher resistance
High bandwidthHigh dynamic range
High bandwidthRugged
Least intrusiveLow cost
TradeoffsTradeoffs
Higher costBandwidth limited to 500 MHz
Limited input dynamic rangeHeavy capacitive loading

Table 1. Passive vs. active probe selection.

 

For faster frequency of rise time signals, use active probes with lower capacitive loading.

 

 

Higher-End Probes

Higher-end oscilloscopes use digital signal processing to help compensate for probe loading but do not eliminate probe loading altogether. To minimize loading, you need to factor your design parameters with the impedance values of the probe you are using.

 

Conclusion

All probes have some type of impact on your circuit under test. It is up to you to determine what is most important for your tests. Understanding some of the common pitfalls helps you select the right probe. A probe draws a portion of the circuit energy and supplies this energy to the oscilloscope. All probes present a capacitive, resistive, and inductive loading element to your circuit. In order to avoid using a probe that adversely impacts your circuit and changes the signal from its original state, you need to factor in the probe’s resistive, capacitive, and inductive characteristics with the properties of your design.

 

Are you falling into oscilloscope probing pitfalls? Avoid making the same mistakes as others with the Oscilloscope Probing Pitfalls eBook.

 

Probe impedance changes with frequency –
The bigger the probe resistance and smaller the probe capacitance, the less the loading your probe will have.

Bandwidth is one of the key oscilloscope probe banner specifications, and if you pick the wrong bandwidth, you’ll get inaccurate measurements. However, picking the right bandwidth probe isn’t as simple as you might think. Avoid making these common mistakes when selecting the right probe bandwidth, and have confidence that the signal you measure is the best representation of your device under test (DUT).

 

Probe Bandwidth

Probe bandwidth is a point on the frequency response curve where the amplitude has decreased by 3 dB. This 3 dB roll-off is shown below in Figure 1.

 

Diagram showing 3dB roll-off point on probe bandwidth

 

Figure 1. Probe 3 dB roll-off point.

 

For example, measuring a 500 MHz, 1 V peak-to-peak sine wave using a 500 MHz probe results in a measurement error of –3 dB or decrease by –0.3 V peak-to-peak. This 3 dB roll-off diminishes your actual 1 Volt peak-to-peak signal to only 0.7 V peak-to-peak. You can fix this pitfall by calculating the correct bandwidth for your desired frequency. A common rule of thumb is that your probe bandwidth should be three times the sine wave frequency you wish to measure. So, to measure a 500 MHz sine wave, you need to choose a probe with 1.5 GHz or more bandwidth.

 

Rise Time and Bandwidth are Related

Now let’s dive a little deeper into this theory. You need to know more than just your signal frequency. To calculate a more accurate probe bandwidth, you need to know your signal’s rise time. Rise time is the time it takes your signal to get from the 10% level to the 90% level of a rising edge.

 

Diagram showing 10 percent and 90 percent rise time points

Figure 2. 10% and 90% rise time points.

 

Let’s use the universally accepted formula that states: bandwidth times the rise time equals 0.35 when evaluating a rising edge from 10% to 90%.

Or you can configure the same formula another way:

If you are working with a communication standard, often the rise time specification is listed along with other banner specs. For example, if your rise time for the 500 MHz clock signal is 350 ps (pico seconds), use the formula below to calculate the probe bandwidth you need.

The calculated bandwidth now equals 1 GHz. This means the real bandwidth of your signal is 1 GHz, not 500 MHz. 500 MHz is your clock frequency for an entire cycle, but your clock’s rising edge is much faster at 1 GHz. You need a probe with a bandwidth specification of 3 times your calculated 1 GHz bandwidth, or a 3 GHz probe. By doing this, you avoid the pitfall of attenuating any of your signal’s frequency components.

 

The Harmonic Nature of Square Waves

You have been calculating your bandwidth needs based on measuring a sine wave. Let’s transition into the pitfalls associated with measuring square waves. If you remember your schooling on Fourier series, you recall that it is a way to represent a function as the sum of sine waves. In Figure 3 below, an original signal is shown in yellow and is made from its fundamental harmonics. Note that the first harmonic in green has the same period and duty cycle, but its rising edges are slower, and the corners are more rounded. The first and third harmonics combined in purple have faster rising edges, and the corners are crisper. The first, third, and fifth harmonics combined in pink have faster edges, crisper corners, and detail on the top and bottom.

 

Square wave signals with harmonics below it

Figure 3. Square wave signal with its harmonics below it.

 

Now let’s take this harmonic knowledge a little farther and use a 100 MHz square wave clock example. Figure 4 below shows the result of measuring the 100 MHz square wave with a 100 MHz probe. The harmonics of the 100 MHz signal are well below the 3 dB point of the probe, resulting in what looks like a sine wave. Any measurements made with this signal will be inaccurate.

 

Distorted square wave signals due to wrong bandwidth probe selection

Figure 4. Distorted square wave due to the wrong bandwidth probe selection.

 

Now let’s measure the same 100 MHz square wave with a 500 MHz probe. The resulting crisp square wave is shown in Figure 5 below because the critical harmonics of the 100 MHz signal are captured in the measurement. You can see how much difference it makes when you use a probe with the appropriate bandwidth.

 

Square wave with the correct bandwidth probe selection

Figure 5. Square wave with the correct bandwidth probe selection.

 

System Bandwidth, The Weakest Link

It is also worth noting that you should consider your entire system bandwidth. You need to factor in both the bandwidth of your probe and your oscilloscope to determine the bandwidth of your probing system (probe + scope). See the formula for your probing system bandwidth below.

 

Let’s say both your oscilloscope and probe bandwidths are 500 MHz. Using the formula above, the system bandwidth would be 353 MHz. You can see that the system bandwidth degrades greatly from the two individual bandwidth specifications of the probe and oscilloscope. Now, let’s say that the probe bandwidth is 300 MHz and the oscilloscope bandwidth is still 500 MHz. Using the above formula, the system bandwidth reduces further to 257 MHz. You can see that the total system bandwidth is always lower than your weakest link or lowest system component bandwidth.

 

Accessory Choice

There is always a trade-off between measurement flexibility, usability, and resulting bandwidth. Many probe tip accessory solutions are available, but all have an impact on your bandwidth. The added inductance of longer ground leads brings with it stray capacitance, creating RC circuits that resonate within your measured bandwidth. These circuits reduce your bandwidth and the ripple on the DUT’s signals. For example, users can use longer lead solutions for quick checks to ensure the circuits are functioning. However, for critical measurements, lead length (both on the signal and ground side) should be kept to a minimum.

 

Measuring Your Probe’s Bandwidth

People often use a VNA (vector network analyzer) to measure an oscilloscope probe’s bandwidth, but VNA’s are expensive and require some training to operate them effectively. In addition, because the VNA is a 50 Ω system and passive probes should be terminated into 1 MΩ, the VNA is not a good test solution for passives.

 

Another way to test bandwidth is to use a sine wave source, splitter, and power meter to sweep the response. This method requires a remote interface such as GPIB or USB and programing skills in MATLAB.

 

An easier way to measure the bandwidth of probes with bandwidths below 1 GHz is the time domain approach, utilizing only an oscilloscope with a built-in step signal source, and ‘differentiate’ and ‘FFT’ functions. Apply a step function to your system, then apply the differentiate (or derivative) to this step response. You now have an impulse response and can take the FFT of the impulse response to obtain the system’s frequency response.

 

Conclusion

Choosing a probe with adequate bandwidth is crucial to making accurate measurements and good engineering decisions. Understanding the impact of rise times, harmonics, and system constraints can help you determine your bandwidth limitations.

 

Are you falling into oscilloscope probing pitfalls? Avoid making the same mistakes as others with the Oscilloscope Probing Pitfalls eBook.

Knowing key Arbitrary Waveform Generator (AWG) specifications lets you make the best purchase decision, saving you time and money. Learn how to compare these characteristics across arbitrary waveform generator types and vendors: memory, sample rate, dynamic range, and bandwidth. Let’s discuss these specifications in detail.

 

Memory Size

Memory size is the amount of memory available for storing long strings of user-defined waveforms. This specification is listed in giga samples (GSa). Data is fed into the digital-to-analog converter (DAC), which creates a voltage stair-stepped representation of you desired signal. High sample rates and large memories are needed to accurately create the defined signal.

 

Sample Rate

Sample rate is the number of samples the DAC can take in a given time interval. The specification is listed in giga samples per second (GSa/s). The sample rate determines the maximum frequency component of the arbitrary waveform generator output signal. Other industry terms for sample rate are “clock rate” and “sample access rate.” A key formula to understand the relationship between memory size and sample rate is shown below.

 

Memory / sample rate = play time

 

You can see from the formula above that as the sample rate increases, more memory is used and play time decreases. Play time determines the total length of the unique waveform an arbitrary waveform generator can generate. This play time length is also called time before repeating. For example, a memory size of 256 kSa and a sample rate of 64 GSa/s results in a play time of 4 micro seconds. This is not a very long pattern and is why large amounts of memory are needed for more play time.

 

Dynamic Range - Vertical Resolution (ADC Bits)

This is the output of the DAC, and it is expressed in voltage as vertical bits of resolution. The value is listed in decibels relative to carrier amplitude (dBc). For example, an 8-bit DAC can output two to the eighth bits of vertical resolution or 256 different voltage levels creating the desired waveform. When comparing arbitrary waveform generator ADC bits specification across different brands, it is important to know that for every ADC bit added, the vertical resolution is doubled.

 

Bandwidth

Arbitrary waveform generator outputs are limited to a specific upper-end output frequency. The bandwidth of the arbitrary waveform generator is the range of frequency outputs that it can reliably provide. The value is also called “data rate” and is listed in giga bits per second (Gb/s). Note that the bandwidth is determined by the sample rate, but it will not be a one-to-one correlation. Let’s walk through the reasons why this is the case:

  1. The DAC must accurately create the signal in memory and needs to create at least two data points per period. This is called the Nyquist Theory. So, a sample rate of 1 GHz results in a DAC output of 500 MHz or one half the sample rate.
  2. The DAC output signal is not a smooth sinewave but is a voltage stepped representation of the pattern in memory. Because of this, the DAC output needs to be filtered. Within the arbitrary waveform generator, this filtering is done by what is called a reconstruction filter, which creates a smooth sinewave. However, there is a price for this filtering in the form of an additional 10% loss to the sample-rate-to-AWG-output frequency ratio. You can reference the formula below. For example, the 1 GHz sample rate mentioned above results in an arbitrary waveform generator maximum output frequency of 400 MHz.

 

Maximum arbitrary waveform generator output frequency = sample rate x 40%

 

Spurious Free Dynamic Range (SFDR)

This is measured in the frequency domain and is the distance in dB from the selected frequency to the highest visible spur or harmonic within the stated bandwidth. The value is listed in decibels relative to the selected frequency amplitude. The screen shot in Figure 1 below shows an example of an AWG output frequency in the center of the screen with a spur to the left that is 94.54 dB lower than its amplitude.

 

 Arbitrary Waveform Generator (AWG): Center frequency relative to a spur.

 

Figure 1. Center frequency relative to a spur.

 

 

Effective Number of Bits (ENOB)

The effective number of bits is derived from the DAC bits. It is a lesser value than the DAC bits due to impacts of harmonics, spurious signals, and the AWG noise floor. Note that this specification changes over the bandwidth of the arbitrary waveform generator, and you should look at ENOB vs. frequency plots like the one below to see the value associated with your signal out frequency of choice. Note the plot in Figure 2 below references a 14-bit system. But after the impacts of receiver signal to noise and distortion (SINAD), you can see that at 1.5 GHz, the actual bit of resolution is deceased to around 9 bits. 

 

Arbitrary Waveform Generator (AWG): An ENOB curve over frequency.

 

Figure 2. An ENOB curve over frequency.

 

ENOB is a great specification to see the actual performance of the arbitrary waveform generator after factoring in these effects on signal quality. ENOB can be measured or calculated. (The formula is below.) Note that SINAD is the ratio of total signal power to unwanted signal noise.

Effective Number of Bits (ENOB) formula

Jitter

A waveform’s jitter can cause misalignment of edges and voltage levels. This can cause the AWG to inject data errors into your system. The jitter value is usually listed in ps peak-peak between the sync clock and the direct data output.

 

Conclusion

Know your key arbitrary waveform generator specifications to properly select the AWG that fits your application:

  • Memory, sample rate, and play time are interrelated.
  • Your bandwidth will not match your sample rate but will be 60% of that number.
  • ENOB is a better representation of resolution than ADC bits.
  • Factor in the jitter specification to account for true signal fidelity.

 

To learn more about arbitrary waveform generators, check out: A High-Performance AWG Primer

There are many cases where certain signals can cause your device to malfunction. This may be a problem your customer ends up finding if you don’t properly test during product development. Designers and test engineers frequently use an Arbitrary Waveform Generator (AWG) to simulate worst-case conditions during design verification. An AWG is the ideal tool for creating degraded or stressed signals to verify product performance limits. System or product noise susceptibility, timing problems, signal-level abnormalities, bandwidth loss, harmonic distortion, or a host of related maladies can be determined.

 

The AWG is a very powerful tool and can create waveforms or waveform bursts needed for your specific application. An AWG combines the capabilities of a function generator with that of a pulse generator, modulation source, noise generator, sweep generator, and trigger generator. It is a good tool for everyday use in the design lab or test environment. You can create custom solutions for a wide range of applications spanning many industries. AWG applications range from high dynamic range to high bandwidth output requirements.

 

 Arbitrary Waveform Generator (AWG) applications

Figure 1. Arbitrary Waveform Generator (AWG) applications.

 

Below is a list of common applications covered in this blog:

  • Radio Frequency (RF) signals
  • Radar signals
  • Environment signals
  • Coherent optical
  • Generic Orthogonal Frequency-Division Multiplexing (OFDM)
  • High-speed serial
  • Simulating real-world aberrations in 100Base-T physical layer
  • Dual Tone Multi-Frequency (DTMF)
  • Pacemaker
  • Automobile suspension testing
  • Power line testing

 

Radio Frequency (RF) Signals

Creating the signals required for RF conformance and margin testing is increasingly difficult. Digital RF technologies require wide-bandwidths and fast-changing signals that other generators cannot produce. These types of signals are seen in RF communications and ultra-wide band radio applications.

 

Radar Signals

Radar signals demand AWG-level performance in terms of sample rate, dynamic range, and memory. AWGs can oversample the signal in instances where phase and amplitude quadrature signal generation is desired. This improves signal quality, creating a spurious free dynamic output. AWG’s also provide Linear Frequency Modulation (LFM), Barker and Polyphase codes, step FM, and nonlinear FM modulation signals. They also generate pulse trains to resolve:

  • Range and doppler shift ambiguity
  • Frequency hopping for electronic counter-counter measures
  • Pulse-to-pulse amplitude variation

 

Environment Signals

Radar signals must coexist with commercial signals and not affect each other. Use your AWG to thoroughly test all the corner case issues at the design or debug stage. An AWG can be programmed to output many industry-standard signals:

  • WiMAX
  • WIFI
  • GSM
  • GSM-EDGE
  • EGPRS-2A
  • EGPRS2B
  • CDMA
  • WCDMA
  • DVB-T
  • Noise
  • CW radar

 

You can also define the carrier frequency, power, start time, and duration of these signals. This allows control of the level of signal interaction or interference.

 

Coherent Optical

Today's web driven world is pushing the demand for high-speed short and long haul coherent optical solutions. Phase modulation, high baud rate, high sample rate, high bandwidth, and high resolution are all critical to optical applications. Multiple synchronized AWGs can be used to generate many desired coherent optical signals.

 

Generic Orthogonal Frequency-Division Multiplexing (OFDM)

OFDM has become the modulation method of choice for transmitting large amounts of digital data over short and medium distances. Wide bandwidths and multiple carriers are needed to test RF receivers in today’s wireless world. AWG OFDM packets can specifying the spacing between the symbols or frames or stressed by adding gated noise.

 

High-speed Serial

Serial signals are made of binary data (simple ones and zeros). These signals have begun to look more like analog waveforms with analog events embedded in the digital data. The textbook zero-rise time and flat top of the theoretical square wave no longer represent reality. Today’s serial communication environments are negatively impacted by noise, jitter, crosstalk, distributed reactances, and power supply variations. Your arbitrary waveform generator can create all these signals!

 

Using direct synthesis techniques, AWGs can simulate the effects of propagation through a transmission line.

 

 

Rise times, pulse shapes, delays, and aberrations can all be controlled by your AWG. You can also create a variety of digital data impairments such as jitter (random, periodic, sinusoidal), noise, pre/de-emphasis, duty cycle distortion, inter-symbol interference, duty cycle distortion, and spread spectrum clocking.

 

Simulating Real-World Aberrations in 100Base-T Physical Layer

To simulate physical layer test signals for 100Base-T transceivers, your AWG will create several analog parameters:

  • Undershoot and overshoot
  • Rise and fall time
  • Ringing
  • Amplitude variations
  • Specific timing variations such as jitter

 

AWGs provide an efficient method for generating signal impairments like these for testing product margins.

 

Dual Tone Multi-Frequency (DTMF)

Touch-tone signals on push button telephones are created by combining a low frequency and a high-frequency signal. Simulating the superimposed frequencies creates a special challenge if the frequencies are not harmonically related. An arbitrary waveform generator can generate these signals along with controlled levels of noise and harmonic content.

 

Pacemaker

A simple square wave or sine-wave pulse was used to test pacemakers in the past. Today’s AWGs can create a simulated heartbeat waveform that pacemakers are designed to detect.

The arbitrary waveform generator can customize pacemaker testing for particular heart rate types.

 

Automobile Suspension Testing

An AWG can simulate automobile sensor outputs just as a car would when it hits a bump. The suspension’s response and reliability can be tested under virtually any simulated road condition because the size of the “bumps” can be precisely controlled.

 

Power Line Testing

Multichannel AWGs can simulate three-phase power. Transients or glitches can be created to simulated problematic waveforms. For example, you could simulate a transient on one phase and signal dropout on another.

 

In addition to all the applications above, there are many more across several different industries, and the arbitrary waveform generator will support them all:

  • Sequencing and deep memory
  • Creating long scenario simulations
  • Leading edge physics, chemistry, and electronics research
  • Validation and compliance testing of high-speed silicon and communications devices
  • Stressing testing receivers with a wide array of signal impairments
  • Generating high Baud rate baseband signals with higher order, complex modulation
  • Radar, satellite, electronic warfare, and multilevel signals
  • Jitter margin testing for analog-to-digital converters

 

Conclusion

We have now covered the importance of an arbitrary waveform generator to ensure your device is working properly for your specific application. As you can see, AWGs excel in creating mixed-signal waveforms that can mimic real world conditions. To learn more about arbitrary waveform generators, check out: A High-Performance AWG Primer.

Quick note: We usually post oscilloscope tips and tricks to this blog, but today we want to share with you about another test & measurement tool often used with or alongside scopes.

 

In my previous post, I outlined the different types of signal generators in the market today, and what you need to consider when selecting the right fit for your application. I also highlighted why the arbitrary waveform generator (AWG) is my recommendation for you to simulate real world stimulus.

 

Arbitrary waveform generators (AWGs) are the most versatile signal generators available. An AWG can generate any mathematically-characterized signal, including sine wave, pulse, modulated, multitone, polarized, and rotated signals. The AWG is commonly seen as the workhorse piece of test equipment and can perform the functions of any other generator type. A typical block diagram of an AWG is shown below. The signal flow through the functional blocks starts with a numeric description of a waveform stored in memory. Then the selected waveform samples are sent to a digital-to-analog converter (DAC), filtered, conditioned, amplified and output as an analog waveform.

 

Diagram of a common Arbitrary Waveform GeneratorA common AWG block diagram

 

 

A Closer Look at Each Arbitrary Waveform Generator (AWG) Functional Block

 

1. Memory

A digital representation of a waveform is loaded into AWG memory through a variety of software applications, such as MATLAB, LabView, Visual Studio Plus, IVI, and SCPI. The memory is clocked at the highest sampling rate supported by the AWG. The size of the memory will dictate the amount of signal playback time available. A rule of thumb to determine the playback time is: memory depth divided by sample rate equals playback time. The faster your sample rate, the quicker you will use up the available memory.

 

2. Sequencer

The sequencer circuitry can solve memory depth limitations by arranging (sequencing) the waveform into segments to create your desired waveform. Memory sequencing (or memory ping-pong) does this by only enabling memory during critical waveform portions and then shutting off. You can think about it like this: when recording a round of golf, imagine how much recording time you would save if you only recorded the players striking the ball and not all the walking and setup time. The sequencer does the same thing by only recording waveform transitions and not idle time. Synchronization is maintained by the trigger generator, which enables the waveform. Trigger events can be internal, external, or linked to another AWG.


3. Markers and Triggers

Marker outputs are useful for triggering external equipment. Trigger inputs are used to alter sequencer operation, resulting in the desired waveform entering the DAC. Hardware or software triggers can be used for applications requiring exact timing, like wideband chirp signals. They can also be used where multiple AWGs are synchronized together and need to be triggered simultaneously.

 

4. Clock Generator

The timing of the waveform is controlled by an internal or external clock source. The memory controller keeps track of waveform events in memory and then outputs them in the correct order to the DAC. The memory controller saves space by looping on repetitive elements so that the elements are listed only once in the waveform memory. Clocking circuitry controls both the DAC and the sequencer.

 

5. Digital-to-Analog Converter (DAC)

Waveform memory contents are sent to the DAC. Here the digital voltage values are converted into analog voltages. The number of bits within the DAC will impact the AWG’s vertical resolution. The higher the number of bits, the higher the vertical resolution and the more detailed the output waveform will be. DACs can use interpolation to reach an even higher update rate than what was supplied by the waveform memory.

 

6. Low Pass Filter

Because the DAC output is a series of voltage stair steps, it is harmonic-rich and requires filtering for a smooth sinusoidal analog waveform.

 

7. Output Amplifier

After the signal passes through the filter, it will enter an amplifier. The amplifier controls both gain and offset. This gives you the flexibility to adjust output gain and offset depending on your application. For example, you may need high dynamic ranges for radar and satellite solutions or high bandwidth for high-speed and coherent optical solutions.

 

Use this blog’s functional building blocks to help you understand just what is happening within your AWG and fully utilize the AWG’s capabilities. For a deeper understanding of arbitrary waveform generator fundamentals, I recommend that you download the comprehensive "Fundamentals of Arbitrary Waveform Generation" guide. 

 

Takeaway and the Demands of Our Connected World

Image of a connected worldInternet technologies have driven advanced AWG solutions

 

Our connected world demands increased speed and data complexity. To support this demand, today’s AWGs must:

  • Reach higher frequencies while providing wider bandwidth
  • Handle complex modulation techniques that cram more data into available bandwidths
  • Work with ideal and real-world signals
  • Generate signals that stress devices to their limits
  • Provide reliable and repeatable results

 

More on this topic in my posts to come. 

Quick note: We usually post oscilloscope tips and tricks to this blog, but today we want to share with you about another test & measurement tool often used with or alongside scopes.

 

To develop or test your electronic design, you need to stress it beyond its real-world application. This will insure your device will operate flawlessly for your customers. In some cases, you may find real world stimuli, but most of the time you will need to use instrumentation.

 

Figure shows the stimulus test model with instrumentation to simulate real-world application

Figure 1. The stimulus test model

 

The instrumentation needed to develop and test today’s technologies has grown into many signal generator types over the years. The most popular signal generators provided by test and measurement manufacturers are:

  • Function Generators
  • RF Generators
  • Pulse and Pattern Generators
  • Arbitrary Waveform Generators (AWG)

There are many generators types out there, so choose yours wisely.

 

Function generators

These are the most well-known and cost-effective signal generators. But, they can only provide a limited set of waveforms such as sine, square, and triangle. They are designed to be very easy to use for simple waveforms with limited memory. You can also adjust the generator’s frequency, offset and other output variables as well as the types of modulation.

Function generators are a good general-purpose source. Use function generators when you need a stable and repeatable stimulus signal. You can use them in applications requiring only periodic waveforms such as stimulus response testing, filter characterization and clock source simulation. Some of the more modern function generators are even capable of generating simple AWG waveforms.

 

Radio Frequency Signal Generators (RF Signal Generators)

RF signal generators produce continuous wave tones with variable output power levels. RF generator outputs typically range from a few kHz to 6 GHz while microwave signal generators cover from 1 MHz to 20 GHz. Use them to service radio receivers and other RF applications. Keep in mind that there are many types and sub categories of these generators, and Keysight provides solutions from 9 kHz all the way up to 25 GHz.

 

Pulse Generators and Pattern Generators

Pulse generators and pattern generators have an advantage over function generators because the output repetition rate and pulse widths can be varied. Their internal circuits may be digital, analog, or a combination of both in order to create the desired outputs. The benefits of direct digital synthesis yield precise frequencies. Use them when working on digital circuits, whereas you should use a function generator primarily for analog circuits.

 

Arbitrary Waveform Generators (AWGs)

An arbitrary waveform generator (AWG) can create all the repetitive waveforms that the three generators above can provide. Plus, an AWG can provide single shot pulses and interpolate between defined points. This is helpful when you need to create triangular waveforms. By harnessing the power and versatility of an AWG’s digital signal processing techniques, you can create whatever signals you need to fully exercise your device under test. With this flexibility, you can either confirm proper operation or pinpoint faults within your system. Basically, with your AWG you can create custom solutions for a wide range of applications.

 

Image of Keysight M8195A Arbitrary Waveform Generator

Figure 2. M8195A Keysight AWG

 

Below is another view of the diverse generator market :-

Signal sourceCharacteristicsWave shape
RF Signal GeneratorsCW (continuous wave) sinusoidal signals over a broad range of frequencies. Modulation types include amplitude, frequency, phase and pulse modulation. May include the ability to sweep the output frequency over a user-set range for frequency response testing.Sine
Modulated sine
Swept sine
Vector Signal GeneratorsDigitally-modulated RF signals that may use any of a large number of digital modulation formats such as QAM, QPSK, FSK, BPSK, and OFDM.Sine
Modulated sine
Pulse GeneratorsPulse waveforms or square waves. Used for testing digital and pulsed systems.Rectangular pulse
Data or Data Pattern GeneratorsMultiple logic signals (i.e. logic 1s and 0s) used as a stimulus source for functional validation and testing of digital circuits and systems.Rectangular pulse
Function GeneratorsSimple repetitive waveforms like sine wave, saw tooth, step (pulse), square, and triangular. May include a modulation function such as amplitude modulation (AM), frequency modulation (FM), or phase modulation (PM).Sine
Rectangular pulse
Square wave
Triangle
Ramp/saw tooth
Modulated waveforms
Noise
Arbitrary Waveform Generator (AWG)Digitally-based signal source generating any waveform, within published limits of bandwidth, frequency range, accuracy, and output level.All the above

 

Use your AWG to simulate real world stimulus

In many scenarios, you need real-world, non-repetitive stimulus to fully test your product. To do this, you need an AWG. In addition, you want to simulate the real-world signals early in the development cycle and fully test the robustness of your designs. An AWG allows you to do this and catch any intermittent or inherent design issues quickly and efficiently. This results in less revisions and gets products to market quicker.

 

Once the desired waveform is loaded into the AWG’s memory, it can be used to generate an output waveform. You can then adjust your frequency, amplitude, and DC offset and use tools such as triggering, gating, bursts, and modulation to further customize your stimuli. The advanced stimulus created by your AWG allows you to verify product performance limits under worst case conditions. Because of this, the applications for AWGs span virtually all industries.

 

What kind of signal generator do you need? Function, RF, Pulse or AWG? For real-world, non-repetitive stimuli creation, look to a function or arbitrary waveform generator. To fully utilize your AWG, you should have a basic understanding of the instrument's controls, features, and operating modes. To learn more about AWGs, download the “Fundamentals of Arbitrary Waveform Generation” guide.

Comparing different manufacturers’ oscilloscopes and their various specifications and features can be time-consuming and confusing. Streamline this process and select the oscilloscope that best fits your application with these top 10 considerations.

 

Streamline this process and select the oscilloscope that best fits your application with these top 10 considerations.

 

Bandwidth

The primary oscilloscope specification is bandwidth, and it is critical that you select an oscilloscope that has sufficient bandwidth to accurately capture the highest frequency content of your signals. Most oscilloscopes with bandwidth below 8 GHz have a Gaussian, or single-pole low-pass filter frequency response. The oscilloscope’s bandwidth is the frequency at which the input signal is attenuated by 3 dB. Because of this, you cannot expect to make accurate measurements near your oscilloscope’s specified bandwidth frequency.

 

Rule of Thumb:

  • Analog applications: Choose a bandwidth at least 3x higher than the highest sine wave frequencies you will measure
  • Digital applications: Choose a bandwidth at least 5x the highest clock rate in your system.

 

This will allow you to capture the fifth harmonic with minimum signal attenuation. This fifth harmonic of the signal is critical in determining the overall shape of your digital signals. This 5-to-1 rule-of-thumb does not take into account lower clock-rate signals that may have relatively fast edge speeds. These clock signals may contain significant frequency components beyond the fifth harmonic and require even higher bandwidths.

 

Keysight’s S-Series oscilloscope has bandwidth options of 1, 2, 2.5, 4, 6, and 8 GHz  

 

 

Figure 1 – Keysight’s S-Series Oscilloscope from 500 MHz to 8 GHz

 

Sample Rate

A digital oscilloscope can spend a lot of time calculating between the trigger event, the signal displayed, and the next trigger. This can result in only a few captures of your signal each second. Your oscilloscope will fail to capture intermittent errors or faults in your signal without a high enough update date.  This happens because the oscilloscope is busy calculating the last acquisition captured instead of acquiring. The higher the update rate, the higher your chances are of capturing that rare event.

 

Memory Depth

Closely related to an oscilloscope’s maximum sample rate is its maximum available acquisition memory depth. You should select an oscilloscope that has sufficient acquisition memory to capture your most complex signals with high resolution. Even though an oscilloscope’s banner specifications may list a high maximum sample rate, this does not mean that the oscilloscope always samples at this high rate. Oscilloscopes sample at their fastest rates when the time base is set on one of the faster time ranges. But when the time base is set to slower ranges to capture longer time spans across the oscilloscope’s display, the scope automatically reduces the sample rate based on the available acquisition memory. Maintaining the oscilloscope’s fastest sample rate at the slower time base ranges requires that the scope have additional acquisition memory. Determining the amount of acquisition memory you require is based on the time span of your signal and your desired maximum sample rate:

 

Acquisition Memory = Time Span x Required Sample Rate

 

Even though an oscilloscope’s banner specifications may list a high maximum sample rate, this does not mean that the oscilloscope always samples at this high rate. 

 

To better understand the relationship between bandwidth, sampling rate, and memory depth, let’s look at a real-world example. Consider trying to capture one frame data that lasts 1 ms and has serial data transmitted at 12 Mbps. So let’s assume that we have to capture a 12 MHz square wave for 1 ms.

  • Bandwidth — to measure the 12 MHz signal, we need an absolute minimum of 12 MHz, however, this will give a very distorted signal. So a scope with at least 50 MHz bandwidth should be selected.
  • Sampling rate — to reconstruct the 12 MHz signal, we need around 5 points per waveform, so a minimum sampling rate of 60 MS/s is required.
  • Memory depth — to capture data at 60 MS/s for 1 ms requires a minimum memory depth of 60,000 samples.

 

Triggering

Triggering allows you to synchronize the oscilloscope’s acquisition and display particular parts of your signal under test. Most digital oscilloscopes trigger on simple edge crossings, but you should select your oscilloscope based on the types of advanced triggering needed to help you isolate your most complex signals. Some oscilloscopes have the ability to trigger on pulses that meet a particular timing qualification. For example, trigger only when a pulse is less than 20 ns wide. This type of triggering (qualified pulse-width) can be very useful for triggering on unsuspected glitches. Pattern triggering is also very common and allows you to set up the oscilloscope to trigger on a logical/Boolean combination of highs (or 1s) and lows (or 0s) across two or more input channels. More advanced oscilloscopes even provide triggering that can synchronize on signals that have parametric violations. In other words, trigger only if the input signal violates a particular parametric condition such as reduced pulse height (runt trigger), edge speed violation (rise/fall time), or perhaps a clock to data timing violation (setup and hold time trigger).

 

Most digital oscilloscopes trigger on simple edge crossings, but you should select your oscilloscope based on the types of advanced triggering needed to help you isolate your most complex signals.

 

Display Quality

The quality of your oscilloscope’s display can make a big difference in your ability to effectively troubleshoot your designs. You should select an oscilloscope that provides multiple levels of trace intensity gradation in order to display subtle waveform details like noise distribution, jitter, and other signal anomalies. For the highest oscilloscope display quality in the industry, go to Keysight.com. An example is shown in Figure 2 below.

 

 

Figure 2 – High levels of display quality are required when viewing complex modulated signals such as video

 

Serial Bus Applications

To help you debug your designs faster, select an oscilloscope that can trigger on and decode serial buses. Serial buses such as I2C, SPI, RS232/UART, CAN, USB, etc., are pervasive in many of today’s digital and mixed-signal designs. Verifying proper bus communication along with analog signal quality measurements requires an oscilloscope. Many of today’s oscilloscopes have optional built-in serial bus protocol decode and triggering capabilities. If your designs include serial bus technology, then selecting an oscilloscope that can decode and trigger on these buses can be a significant time-saver to help you debug your systems faster.

 

If your designs include serial bus technology, then selecting an oscilloscope that can decode and trigger on these buses can be a significant time-saver to help you debug your systems faster. 

 

Connectivity and Documentation

Selecting an oscilloscope that meets your hardware connectivity, test automation, and electronic documentation requirements are key. Automated testing requires that the oscilloscope’s ports be programmable. So make sure any measurement that can be performed using the oscilloscope’s front panel and menu controls can also be programmed remotely via LAN or USB connectivity.

 

Keysight’s InfiniiVision X-Series and Infiniium Series oscilloscopes are all fully programmable via SCPI commands as well as National Instruments IVI drivers. Saved images (screen-shots) and data (waveforms) can also be easily imported into various word processors, spreadsheets, and applications such as MATLAB. Keysight’s N8900A InfiniiView offline analysis software lets you easily capture waveforms on your oscilloscope, save them to a file, and recall the waveforms into the application. With Keysight’s > 50 standard automated measurements with statistics and 16 independent math functions, you’ll be able to analyze a wide variety of tests.

 

Probing

Your oscilloscope measurements can only be as good as the data your probe delivers to the oscilloscope’s BNC inputs. Always keep in mind that when probing your circuit, your oscilloscope and probe become part of your device-under-test. This means it can change the behavior of your device-under-test signals due to capacitive or inductive loading. Select an appropriate probe for your measurement to minimize these loading effects. This will prevent disturbance of the input signal and deliver a true representation of your signal as it existed in your circuit before the probe was attached.

 

Select an appropriate probe for your measurement to minimize these loading effects.

 

Ease of Use

Your oscilloscope should be user-friendly and intuitive. This can be just as important as specified performance characteristics.  Oscilloscopes have evolved over the years with many additional features and capabilities but ease of use should not be compromised. Although most oscilloscope vendors will claim that their oscilloscopes are the easiest to use, usability is not a specified parameter that you can compare against in a product’s data sheet. Ease-of-use is subjective, and you must evaluate it for yourself. However, there are a few things you can look for when evaluating ease of use:

  • Built-in help menus, which reduce the need to reference manuals
  • Large color displays, which allow views of waveforms and measurement data at the same time
  • Voice control, which allows for hands-free control

Always request a demo when picking out your oscilloscope with a reputable company providing field engineers to help demo the scope or go to trade shows that provide hands-on demos.

 

Learn more about these and other oscilloscope selection topics:

If you’re an Infiniium user, you’ll definitely want to keep reading. The 6.1 software update just launched and it’s packed with new features that will improve your testing efficiency. Use this as a guide to make use of the new tools and enhancements that you’ll find in this software update.

 

These updates include:

  • New Jitter Decision Feedback Equalization
  • Improved Clock Recovery and Mask Testing for PAM4 and NRZ
  • New Software Update Analysis Tool
  • New Impedance Warning Function
  • New Power Integrity Analysis Application

 

Jitter Decision Feedback Equalization

Technological advances towards achieving greater Ethernet speeds presents two design possibilities, NRZ and PAM4, and each comes with a unique set of challenges. NRZ (Non-Return-to-Zero) has evolved over 50 years to 100G (25/28G, 4 lanes) and 400G (56G, 8 lanes). From a time domain perspective, NRZ consists of 1’s and 0’s and can be referred to as PAM2 (pulse amplitude modulation, 2-level) with two amplitude levels that contain 1 bit of information in every symbol. The NRZ eye diagram provides timing and voltage used to measure link performance and contains a single eye. But this single eye technology requires advanced technology in order to achieve the higher 400 Gb/s data rate.

 

The current 400 Gb/s challenges include totally closed eyes, shorter unit intervals (UI), tighter jitter requirements, and the mandatory use of forward error correction (FEC). These closed eye issues require enhanced receiver equalization such as continuous-time-linear equalization (CTLE) and decision feedback equalization (DFE) to correct. Moreover, new communication standards are requiring increased receiver sensitivity (down to 50 mV) and jitter budgets are even tighter for 400G at 17ps.

 

This software update adds a Jitter Decision Feedback Equalization function (DFE) to meet these increasing demands.

 

Improved Clock Recovery and Mask Testing for PAM4 and NRZ

If you’re working in 400G, you’ll be glad to hear of some major enhancements to the existing PAM4 solution. In the new software update, an improved PAM4 clock recovery algorithm was added along with jitter measurements on Decision Feedback Equalization (DFE) for NRZ signals.

 

A “Draw Mask” feature was also added to allow users to draw their own polygon mask specific to their eye pattern, enabling exact mask tests to customer specified limits.

 

New Software Update Analysis Tool

If you’ve ever had difficulty determining if an application or protocol is compatible with your scope’s software, you’ll be interested in the new Software Update Analysis tool rolling out in this software update. This tool will help you quickly determine if all applications and protocols on an oscilloscope are compatible with the latest software. This feature also allows software updates to oscilloscopes in secured laboratories that cannot have access to internet-supplied software.

 

New Impedance Warning Function

Safety is always a top concern. We are always working to improve product quality and reliability, and with this comes a new Impedance Warning function. With the new software installed, you will be notified when you have selected a lower voltage 50 Ohm input, and you'll see the max input level allowed on screen. This will help to reduce the chances of user-damage to the oscilloscope input.

 

New Power Integrity Analysis Application- N8846A

The new Power Integrity Analysis Application-N8846A was added to Keysight’s already industry-leading set of N8833A/B cross talk applications. This application was specifically designed to target power supply-induced cross talk.

 

Download Now > Infiniium 6.10 System Software for Windows 7 | Keysight   

IEEE Standards for 400G optical Ethernet links have been in development for many months, and the specifications are finally stabilizing as of mid-2017. The result will be the first optical standards to employ PAM4 modulation. This requires a new set of measurements, with the TDECQ (Transmitter and Dispersion Eye Closure Quaternary) measurement getting the most attention. Increasingly more cutting edge companies are ready to evaluate their 400G products and need the TDECQ capability now.

 

TDECQ is a new and significantly easier method of calculating the penalty for transmitters that have unequal sub-eyes. This software calculation requires only an oscilloscope and are achieved by a direct measurement of the transmitter eye diagram.

 

It is simpler, faster and less expensive than older TDP (transmitter dispersion penalty) measurements, which would need a reference transmitter or an optical enabled BERT.

 

In its simplest definition, the TDECQ measurement creates two vertical histograms measured on an eye diagram like Figure 1 below. The histograms are centered at 0.45 and 0.55 unit intervals and each spans all modulation levels of the PAM4 eye diagram.

 


Figure 1: Illustration of the TDECQ Measurement

 

The amount of noise captured in the histogram is compared to an ideal receiver, and the dB difference in noise levels represents the power penalty for the transmitter under test.

 

One common mistake is to confuse the TDECQ result in dB with BER measurements.

 

We must keep in mind that TDECQ is a measurement of transmitter eye opening quality relative to an ideal transmitter and not a bit error rate.

 

Keysight engaged early in the development of a TDECQ solution with the IEEE Standards Association, sharing many of our hardware evaluations with the 400G committee. As a result, our TDECQ solution is easy to set up and creates fast measurements for use in both R&D and manufacturing environments. Our competition has also developed TDECQ capability, but their solution is separated from the scope and runs much slower.

  

The Keysight solution was designed for easy integration into a manufacturing system and can be quickly updated at the customer site if any changes are made to the IEEE 400G Standard.

 

Due to its low noise and fast sampling, the primary Keysight TDECQ solution is the low-cost N1092 DCA-M sampling scope module. It has predefined TDECQ reference receivers for both 26 GBaud and 53 GBaud transmitters. In addition, the Keysight 86105D-281 and 86116C-025 plug-in modules can also be used with the 86100 mainframes with the TDECQ option for the same result. 

 

View more information & order your solution today >>

N1092A DCA-M Sampling Oscilloscope

86105D 34 GHz Optical, 50 GHz Electrical Module

86116C 40 to 65 GHz Optical and 80 GHz Electrical Plug-in Modules

86100D Infiniium DCA-X Wide-Bandwidth Oscilloscope Mainframe

 

Get a free 400G poster!

Understanding the effects of crosstalk in high speed communication circuitry and pin pointing the root causes have been extremely hard for designers, until now. Keysight’s industry leading N8833A and N8833B software not only identifies the crosstalk but also allows the user to determine the sources. In addition, it can remove the effects of the crosstalk and determine the recovered margin.

First let’s discuss the many types of crosstalk and their origins.  Crosstalk has become an important issue as data rates have increased, and more and more lanes are being packed into smaller and smaller spaces. The amplitude interference of crosstalk impacts the signal fidelity of a communication eye diagram, essentially causing the eye to become more closed. The majority of crosstalk is caused by capacitive or inductive coupling between multiple transmission lines and/or power delivery networks. Prominent sources are near end crosstalk, far end crosstalk, power supply induced jitter, voltage dependent amplitude noise and simultaneous switching noise. Transmission line crosstalk is the result of electromagnetic interference between electrical components and is mainly caused by capacitive or inductive coupling. Forward traveling or far end crosstalk travels the same direction as the aggressor signal and its energy grows and it travels down the transmission line resulting in amplitude bulge in one area of the eye pattern. Reverse traveling or near end crosstalk is constantly moving away from the aggressor edge and is spread somewhat evenly over the transmission line resulting in a smearing of the entire eye pattern. Power supply aggressor crosstalk is created by noise on the power rail supply and caused phase noise changes or jitter. Voltage dependent amplitude noise crosstalk adds noise to the voltage and ground bus and causes non-linear effects on each logic level. A power supply can also be a victim of crosstalk due to simultaneous switch noise on serial lines and this is caused parasitic inductances lying between board and system ground and is also known as ground bounce. The above crosstalk origins, effects and sources may seem overwhelming at first but Keysight’s N8833 application greatly simplifies both the user knowledge and effort needed to get to root cause.

 

Legacy methods of determining crosstalk digital communications systems has relied on the process of selectively disabling some channels while enabling others. This process usually took significant time and effort. Power supply noise adds yet another analysis hurdle creating a non-linear transfer on the serial data timing called the Time Interval Error and has been difficult to solve and correlate. Past troubleshooting methods also required special design and test modes to analyze the crosstalk. Another challenge is that many times crosstalk aggressor signals are created within a package or in a system that is not accessible to probing.

 

The new Keysight crosstalk analysis application meets all of the above challenges by:

1) identifying the sources of crosstalk affecting the victim,

2) quantifying how much each aggressor is disrupting the victim,

3) removing the effect of crosstalk from the victim signal for analysis and

4) checking how much design margin is recovered when crosstalk is removed from victim.

 

Other features of the Keysight software are:

1) analyzes up to four signals (aggressors or victims) at once,

2) requires no crosstalk simulation or model,

3) identifies and reports the amount of crosstalk present on victims from each aggressors,

4) plots waveforms without crosstalk,

5) compares them with the original waveforms using scope tools such as eye diagram and jitter separation to see how much margins can be recovered.

 

The key types of crosstalk that can be analyzed are:

1) transmission line aggressors: Near-End Crosstalk (NEXT) and Far-End Crosstalk (FEXT),

2) power supply aggressors: Power Supply Induced Jitter (PSIJ) and voltage-dependent amplitude noise,

3) power supply victim: Simultaneous Switching Network (SSN).

 

Finally the software can report the results in different ways:

1) Inter-Symbol Interference (ISI) magnitude of the victim on itself,

2) crosstalk magnitude of the transmission line aggressors on the serial data victim,

3) crosstalk magnitude and jitter of the power supply aggressors on the serial data victim,

4) crosstalk magnitude of the transmission line aggressors on the power supply victim.

 

The Keysight crosstalk application N8833A/B is the most comprehensive solution in the market enabling a closed loop design cycle, saving designers both time and money. This application effectively solves the challenges mentioned above by enabling designers and engineers to: 1) identify which signals are coupling into your victim signal, 2) quantifying how much error each aggressor signal adds to your victim signal, 3) see what the victim signal would look like without the crosstalk and how much eye margin can be recovered without the crosstalk, 4) determine if the existing crosstalk justifies a design change and where to improve the circuit or system design.