When should you calibrate your test equipment?

Today I'd like to talk about calibration intervals and how often you actually need to calibrate your equipment. If you are going by the datasheet, most instruments will offer 1 year specifications or maybe 2 years. Is this the recommended cal interval? Why calibrate at all? What the heck is calibration? We'll discuss these questions in today's post. The last three products that I worked on were DMMs, function generators, and counters. My calibration knowledge is skewed towards those types of instruments.

Technically, calibration is defined (in Wikipedia) as a comparison between measurements – one of known magnitude or correctness made or set with one device and another measurement made in as similar a way as possible with a second device. In the test instrument world, this boils down to checking your instrument vs. a traceable standard.

In a more practical sense, I've always considered calibration as having two parts: verification and adjustment. Verification is checking the measurement against a specification i.e. checking your results and comparing them to a datasheet spec. Adjustment is the act of changing the performance of your instrument to be within specification. You don't always need to do an adjustment if your unit is verified to be within specifications, but you might want to depending on your situation.

We're going to discuss two typical use cases when deciding whether or not to calibrate your instrument:
A) You have the instruments in your control i.e. your bench and absolute accuracy isn't your main concern. 
B) The instrument is part of a production system and you rely on results from it for manufacturing

Let's examine the simplest use case first, bench instruments. I'm sure that many of you have equipment on your bench or desk that have not been calibrated for years. They seem to work just fine and you're probably thinking "why would you mess with a good thing?" and you're probably right. Instruments, like a DMM, stabilize the longer that they have been powered on. The mechanical stresses that contribute to reference drift (like thermal expansion and annealing) have settled over time. So how do you determine if you need to calibrate or not? In my experience, the accuracy issue comes out when you are comparing measurements between benches. I measure an output on my bench, but my colleague's measurement is different enough to cause us to question both of our instruments. It's probably time to calibrate (adjust) one of our instruments.

Let's take it a step further and try to figure out which one to calibrate. The easiest answer is to compare the measurements of both instruments against something that is known to be more accurate than our instruments. If we have a 34401A, we can compare the results vs. the output of a Fluke 5720A which is the calibrator of choice for 6.5 digit DMMs. But if you don't have a calibrator available, you might compare the readings to a 3458A. If you don't have one of those, perhaps you can compare the readings against a third DMM that has been calibrated recently. Its all about comparison of measurements in this world. You want to try and compare against something that is known to be more accurate than your instrument. It is sometimes tough to find, but you need a guiding light to show you the way. Comparing against something better will get you there.

There are bench users that do need a regularly calibrated instrument for their work. This typically consists of highly accurate work that will end up in a report or determining a datasheet. In these cases, you need to ensure that your measurements are accurate and calibration is the only way to check your accuracy. In our lab, we do a lot of relative measurements. Calibration affects the absolute accuracy, but not resolution in most cases. Our engineers are able to do quite a bit of work without calibration since all they worry about is the relative accuracy in most measurement. For the final qualification, we use a calibrated system to determine final accuracy and traceability.

Now let's tackle the more complicated production test use case. In one sense it is more simple, you need to calibrate your instrument on a regular cycle. Calibration is the only way that you can ensure that your instrument is providing the accuracy that it is specified to. However, the interesting note here is that calibration only checks that the instrument is in specification during that measurement. It does not necessarily guarantee that your instrument will be in spec for the next calibration cycle. The only way to ensure that your instrument is performing to spec is to compare results between calibrations. It is entirely possible that your instrument could be out of specification shortly after calibration and you wouldn't find out until after the next calibration. If you are out of spec, it would put any measurements between calibrations into question.

Good instrument manufacturers will have characterized the drift of their references and have done their homework on the design to ensure that the instrument design performs to specification or better. Less reputable designers might not have put in the time and engineering to characterize their design long term. In both cases, it is possible that manufacturing mistakes could cause a latent failure or drift once the instrument is in use after a few months. Calibration is really the only way to ensure that your instrument is performing to your expectations.

Determining the calibration interval is dependent on your needs. Most of you will not want to calibrate every 90 days due to the cost. As the user of the product, you need to determine which set of specifications will allow you test to the accuracy that you need. It is recommended that you set up a shorter interval test for yourself to ensure that your instrument is not out of specification between your calibration cycle. The worst case scenario for this would be that you need to recall all of your devices that were tested between calibrations because you didn't know when the instrument went out of spec.

Calibration is an essential detail of instrumentation. Understanding the term and figuring out what actions you should take will make your instrumentation life easier in the long run.

In my next post I'll discuss some different types of calibration methods including shorter interval testing.

Video Review of 33500B Series Waveform Generators

Hi there, I'm David. I'm a new contributor to this blog and I hope to help Neil update content one this blog fairly regularly. I've been a test engineer for about 13 years now and I figured that I could help others understand interesting features, issues, and tribulations that are found with test equipment.

For my first post I'm going to direct you to an independent video review of Agilent's 33500B Series Waveform Generators. The 33500B series was introduced in 2012 and people are starting to get their hands dirty with them. This review was published by Watt Circuit Blog which is run by a couple of engineers in England. In this long running video review, the blogger uses the 33522B in their demonstration. He covers many aspects of the 33500B series ranging from unboxing the unit to using it to characterize an IC. A write up with pictures is published on Element 14. He likes it.

Since its a long video, I'm going to link directly to a few points that might be of interest to you.


Unboxing the unit - This section includes first impressions and a front and rear panel walk through. 
Powering up the unit - Using the front panel by setting up a multi-tone frequency modulated signal.
Adding noise to a signal - using summing modulation and visualizing it on a scope
Replacing a pulse generator - setting up fast edge times, phase control between two channels, and pulse bursts.
Setting up a frequency sweep - Using a scope to  to check frequency response of a filter. This uses the frequency sweep function of the generator
Using a function generator to characterize circuit performance - Testing a level shifter IC with a PBRS signal. Characterizes the IC by sweeping frequency and adding noise.
Benchlink SW - discusses and demonstrates the free Agilent Benchlink software to generate your own waveforms.
Importing a .CSV file - demonstrates how to import a comma separated (.csv) file using a USB stick.
LAN connectivity - LAN and web user interface to control the generator
Summary - Gives a nice summary and discusses Trueform technology, robustness of the unit, discusses upgrades, and mentions that the review only scratches the surface of features of these generators.

Agilent's New RTSA Capability on its PXA Signal Analyzers

In this post I wanted to share the new real-time spectrum analysis (RTSA) feature that Agilent added to its PXA line of high performance signal analyzers. The RTSA feature gives the signal analyzer the ability to continuously capture spectra without any dead time between acquisitions. The PXA's maximum capture rate is 3.57us so that means it has a probability of intercept of 100% for signals with a duration of 3.57us or more. This feature is incredibly useful when trying to analyze intermittent or dynamic signals such as those used in radar and electronic warfare applications.

Let me give a quick explanation of what "dead time" means when referring to instruments like scopes, counters, and signal analyzers. When one of the example instruments makes a measurement or acquisition it then displays the results and prepares to make another measurement. The time in between two consecutive measurements (sometimes referred to as re-arm time) is dead time because no measurement is being made. Often, especially in the case of scopes, the dead time is longer then the measurement or acquisition time so any changes that the signal goes through during the dead time is not captured or measured by the instrument. When an instrument such as a signal analyzer has continuous or gap-free measurement capability it has no dead time time between acquisitions it simply continuously acquires data at its set sample or rate.

With the addition of the RTSA capability the PXA also added some advanced display capabilities to show the continuously captured data in a useful easy to interpret way to users. This includes a display persistence setting, mask testing, a histogram display, and a spectra versus time display. Below is an example of a frequency hopping radar signal displayed using the persistence and mask testing capability.

Besides the new RTSA option capability the PXA signal analyzer line includes a number of industry leading specs and features including:

• –157 dBm at 10 GHz average noise floor which ensures low level signals do not get lost in the "grass"
• 160-MHz real-time bandwidth for looking at wide-band modulated signals
• up to 75 dB spurious-free dynamic range to see small signals in the presence of large ones

If you have any questions from this post feel free to email me and if you have anything to add use the comments section below.

For more information on the new RTSA feature on the PXA click here

Agilent Introduces a New 5 1/2 Digit DMM

Agilent recently introduced the 34450A which is a 5 1/2 digit DMM that starts at under $800! The 34450A includes features such as:

• Fast reading speed at 190 readings/sec
• 11 measurement functions including 4-wire, capacitance and temperature
• High contrast and flexible OLED display
• Up to 50,000 memory points datalogging capability with built-in histogram and basic statistical functions
• Dual measurement display capability
• Multiple choice of connectivity – USB, Serial Interface (RS-232) and optional GPIB

The following video provides an overview of the 34450A's dataloggering capabilities:

For more information on the 34450A click here

Testing for Interleave Distortion in Oscilloscopes

The level of interleave distortion or just sampling distortion in general to expect from an oscilloscope measurement is often not clear from examining an oscilloscope's data sheet. In this post we will look at how to measure and evaluate an oscilloscope's sampling distortion. This post builds on the information from last week's post entitled Interleaved Sampling in Oscilloscopes. Without knowing a scope's level of sampling distortion there is no way to know how accurate a representation of a measured signal the scope is providing. A scope with a high sampling rate may not deliver the best representation of a measured signal compared to a scope with a lower sampling rate. This depends on such factors as the measured signal's bandwidth and if the scope uses an interleaved sampling scheme and how well it was implemented in the design.

The following is a list of tests or methods that we will look at to evaluate interleave distortion:

1. Effective number of bits analysis using sine waves
2. Visual sine wave test
3. Spectrum analysis

Effective number of bits analysis
The closest specification that some scope vendors provide to quantify sampling fidelity is effective number of bits (ENOB). But ENOB is a composite specification consisting of several error components including input amplifier harmonic distortion and random noise. Although an effective number of bits test can provide a good benchmark comparison of overall accuracy between scopes, effective bits is not a very well understood concept, and it requires exporting digitized data to a PC for number crunching. Basically, an effective number of bits test first extracts a theoretical best-fit sinusoidal signal from the digitized sine wave. This sine wave curve-fit algorithm will eliminate any errors induced by oscilloscope amplifier gain and offset inaccuracies. The test then computes the RMS error of the digitized sine wave relative to the ideal/extracted sine wave over one period. This RMS error is then compared to the theoretical RMS error that an ideal ADC of “N” bits would produce. For example, if a scope’s acquisition system has 5.3 effective bits of accuracy, then it generates the same amount of RMS error that a perfect 5.3-bit ADC system would generate. For more detail on ENOB click here.

Visual sine wave test
A more intuitive and easier test to conduct to see if a scope produces ADC interleave distortion is to simply input a sine wave from a high-quality signal generator with a frequency that approaches the bandwidth of the scope. Then just make a visual judgment about the purity of the shape of the digitized and filtered waveform.

Let's look at an example comparison using the visual sine wave test. The below figures show two screen captures from two different scopes of a high quality 200 MHz sine wave. The top screen capture is from 1-GHz bandwidth scope sampling at 4 GSa/s. This scope has a sample-rate-to-bandwidth ratio of 4:1 using non-interleaved ADC technology. The bottom screen capture is from the 1-GHz bandwidth scope sampling at 10 GSa/s. This scope has a maximum sample-rate-to-bandwidth ratio of 10:1 using interleaved technology.

200-MHz sine wave captured on a 

1-GHz bandwidth oscilloscope sampling at 4 GSa/s

200-MHz sine wave captured on a 

 1-GHz bandwidth oscilloscope sampling at 10 GSa/s

Although we would intuitively believe that a higher-sample-rate scope of the same bandwidth should produce more accurate measurement results, we can see in this measurement comparison that the lower sample rate scope actually produces a much more accurate representation of the 200 MHz input sine wave. This is not because lower sample rates are better, but because poorly aligned interleaved real-time ADCs negate the benefit of higher sample rates.

Spectrum analysis comparison tests

The visual sine wave test doesn't really prove where the distortion is coming from. It merely shows the effect of various error/components of distortion. However, a spectrum/FFT analysis will positively identify components of distortion including harmonic distortion, random noise, and interleaved sampling distortion. Using a sine wave generated from a high-quality signal generator, there should be only one frequency component in the input signal. Any frequency components other than the fundamental frequency detected in an FFT analysis on the digitized waveform are oscilloscope induced distortion components. The figure below shows an FFT analysis of a single-shot capture of a 2.5 GHz sine wave using an oscilloscope sampling at 40 GSa/s. The worst-case distortion spur measures approximately 90 dB below the fundamental. This component of distortion is actually second harmonic distortion, most likely produced by the signal generator. And its level is extremely insignificant and is even lower than the scope’s in-band noise floor.

FFT analysis of 2.5-GHz sine wave captured on a sampling at 40 GSa/s scope

The figure below shows an FFT analysis of a single-shot capture of the same 2.5-GHz sine wave using a different scope with the same sample rate (40 GSa/s). The worst-case distortion spur in this FFT analysis measures approximately 32 dB below the fundamental. This is a significant level of distortion that would most likely be visually apparent in the time domain. The frequency of this distortion occurs at 7.5 GHz. This is exactly 10 GHz below the input signal frequency (2.5 GHz), but folded back into the positive domain. The next highest component of distortion occurs at 12.5 GHz. This is exactly 10 GHz above the input signal frequency (2.5 GHz). Both of these components of distortion are directly related to the 40-GSa/s sampling clock and its interleaved clock rates (10 GHz). These components of distortion are not caused by random or harmonic distortion. They are caused by real-time interleaved ADC distortion.

In this post we looked at three methods for measuring or evaluating interleaved sampling distortion and general sampling distortion. The three methods included ENOB, visual test, and spectrum analysis. This type of test is necessary when you need to evaluate a low noise signal, such as a digital clock, and you want to ensure that you are actually seeing the measured signal noise and not the noise of the measuring device. If you have any personal insights to add to this post please use the comments section below and if you have any questions feel free to email me.

To read a post on determining how much scope bandwidth you need click here

For more information on Agilent Oscilloscopes click here

Interleaved Sampling in Oscilloscopes

In this post we will look at interleaved real-time sampling in oscilloscopes and how interleaved sampling errors can result in a distorted representation of a signal. After reading this post you will understand why choosing an oscilloscope based solely on sampling rate does not always give you the best representation of the signal you are measuring.

As digital signals continue to increase in speed, oscilloscope vendors are under constant pressure to increase the sampling rate and bandwidth of their products so that engineers have a means to measure the digital signals in their design. The most intuitive way to increase the sample rate of a scope is to simply use a higher sampling rate ADC in the scope, but this technique is not always cost effective or practical. To solve this design challenge a common technique adopted by all major scope vendors is to interleave multiple real-time ADCs.

The below figure shows a block diagram of a real-time interleaved ADC system consisting of two ADCs with phase delayed sampling. In this example, ADC 2 always samples ½ clock period after ADC 1 samples. After each real-time acquisition cycle is complete, the scope’s CPU or waveform processing ASIC retrieves the data stored in each ADC acquisition memory and then interleaves the samples to produce the real-time digitized waveform with twice the sample density (2X sample rate).

Real-time sampling system consisting of two interleaved ADCs

Scopes with real-time interleaved sampling must adhere to two requirements. For accurate distortion-free interleaving, each ADC’s vertical gain, offset and frequency response must be closely matched. Secondly, the phase-delayed clocks must be aligned with high precision to ensure equally spaced samples. That means ADC 2 must be delayed precisely 180 degrees after the clock that samples ADC 1. Remember equally spaced sampling is part of the Nyquist sampling theorem for accurately recreating a signal. in order to satisfy Nyquist’s rule 2 that dictates equally spaced samples. For the rest of the post we will focus on sampling spacing errors due to poor phase delayed clocking.

The timing diagram shown in the below figure illustrates incorrect timing of interleaved samples if the phase-delayed clock system of two interleaved ADCs is not exactly ½ sample period delayed relative to each other. This diagram shows where real-time digitized points (red dots) are actually converted relative to the input signal. But due to the poor alignment of phase-delayed clocking (purple waveforms), these digitized points are not evenly spaced, thus a violation of Nyquist’s sampling theorem.

Timing diagram showing non-evenly spaced samples

When the scope’s waveform processing engine retrieves the stored data from each ADC’s acquisition memory, it assumes that samples from each memory device are equally spaced. In an attempt to reconstruct the shape on the original input signal, the scope’s Sin(x)/x reconstruction filter produces a severely distorted representation of the signal, as shown in the figure below.

Timing diagram showing distorted reconstruction of waveform

Since the phase relationship between the input signal and the scope’s sample clock is random, real-time sampling distortion, which is sometimes referred to as “sampling noise,” may be interpreted mistakenly as random noise when you are viewing repetitive acquisitions. But it is not random at all. It is deterministic and directly related to harmonics of the scope’s sample clock.

Unfortunately, oscilloscope vendors do not provide their customers with a specification in their data sheets
that directly quantifies the quality of their scope’s digitizing process. This means if you just grab the scope in your lab with the highest sample rate assuming that it will deliver the best representation of your signal you may be wrong. For instance lets say you want to capture a 1 GHz digital clock signal. If you have a 6 GS/s scope and a 10 GS/s scope to choose from it makes sense to assume the latter will deliver the best representation of signal. But if the 6 GS/s scope has a single ADC and the 10 GS/s scope uses an interleaved ADC scheme, then the former may deliver a better representation of the signal, but there may not be a clear way to know this by comparing the scope's data sheets. In next week's post we will look at methods to test and evaluate a scope's sampling distortion. If you have any personal insights to add to this post please use the comments section below and if you have any questions feel free to email me.

To read a post on determining how much scope bandwidth you need click here

For more information on Agilent Oscilloscopes click here

Dynamic Current Measurements: Current Probe vs Seamless Ranging

In the following video a colleague of mine compares two methods for measuring varying or dynamic currents: using a current probe versus using seamless ranging. Seamless ranging is an Agilent only technology found in the N6781A and N6782A SMU modules. Seamless ranging technology allows the power source to measure the current flowing through it and transition from one current measurement range to another as the current varies. All with no glitches or discontinuities in the output power. It provides the user with a tremendous amount of dynamic current measurement range which can be useful when doing power and battery optimization measurements on devices that go through various standby (low current) and transmit (high current) cycles. The figure below shows a graphical representation of using seamless ranging to measure dynamic current.  

From the video we can see that although current probes are great tools for measuring high current they are a poor solution for measuring low level current and small current transitions. Seamless ranging on the other hand provides a very high resolution look at a varying current profile over a wide dynamic range. 

To see more of seamless ranging in action check out the YouTube video Agilent Battery Drain Demo: Seamless Measurement Ranging on the N6781A. If you have anything to add to this post please use the comments section below and if you have any questions feel free to email me. 

For more information on the N6781A click here

For more information on the N6782A click here

For more information on the N6705B click here