How selecting the right oscilloscope can help you get to market faster
01 September 2019
As data rates continue to increase, jitter budgets continue to shrink, real time eyes are closed and need to be opened again, and the utilisation of processing and analysis has become increasingly important for your design.
This article was originally featured in the September 2019 issue of EPDT magazine [read the digital issue]. Sign up to receive your own copy each month.
To make matters even more difficult, CAPEX budgets are shrinking, so mistakes cannot afford to be made with large capital purchases. This is especially true in the oscilloscope market, where there is a multitude of choice and the cost of the instrument is high. Here, Brig Asay, Director of Strategic Planning for the Internet Infrastructure Group at T&M expert, Keysight Technologies, explains how selecting the wrong piece of test & measurement equipment can exacerbate the problem.
For example, for the latest PCIExpress 5.0 signal running at 32 gpbs, the unit interval (UI) has now shrunk to less than 35 ps. This miniscule UI does not even include the added jitter and other impairments that can shrink your margins further. In the case of PCIExpress 5.0, this means a random jitter budget of less than 450 fs for transmit test. Making it even more difficult is the fact that board materials continue to be used that are high loss based, such as FR4, to save costs, further eroding precious margins. For PCIExpress, the eye height is specified at less than 15mV peak to peak after de-embedding and equalisation techniques are applied. So, choosing the wrong piece of test and measurement equipment, such as real time oscilloscopes, will only make your problem worse.
Many don’t realise that when equalisation and de-embedding techniques are used, the jitter and noise added via test and measurement equipment is boosted to remove the loss of the channel. Real time oscilloscopes have important known characteristics, including noise floor, jitter measurement floor and bandwidth. All of these limitations erode crucial margins in your design. No longer can oscilloscopes, and their signal integrity specifications, be discounted as insignificant contributors to design margins.
What to look for when selecting your oscilloscope
As you can imagine, there are several characteristics to consider when you are purchasing an oscilloscope. For high-end oscilloscopes, almost every hardware specification can and will impact margins. In addition to the hardware specifications, knowing the trade-offs of the different oscilloscope vendors’ software tools, such as jitter, equalisation and waveform transformation tools, will affect margins. Choosing the best combination of software and hardware will improve time-to-market, ultimately making your company more successful.
Oscilloscope noise and jitter matter
Every oscilloscope vendor will discuss different specifications, and which specs are the most important, including noise floor, jitter measurement floor, bandwidth, effective number of bits, sine wave repeatability, and more. Most recently, we are starting to see RF measurements make their way into oscilloscope literature, such as EVM, SINAD and TOI. When you are looking at depicting your signal as closely as possible to the real-life signal, all these measurements are valid ways of showing how oscilloscope hardware performance degrades your signal.
Let’s look at noise: oscilloscope noise erodes eye height and impacts eye width, as well as contributing to your random jitter bucket. Oscilloscopes measure noise floor using the following method: the vendor sets the oscilloscope to a known sensitivity setting and disconnects all cables and probes from the oscilloscope. As a result, the only signal that is displayed on screen is the oscilloscope noise. A histogram of the noise is then enabled and the oscilloscope noise can be measured. The readings you get are then compared to different vendors using the same sensitivity and bandwidth. For instance, Keysight’s UXR Series oscilloscope set to a bandwidth of 33 GHz and a sensitivity setting of 50mV/div will measure a noise floor of 1.3 mVrms, which is about 80% less than Keysight’s own Z-Series oscilloscope, which measures at 2.3 mVrms (see figure 1).
Another oscilloscope specification that matters is its jitter measurement floor. Oscilloscope vendors spend significant time designing a time base with low phase noise to minimise its jitter measurement floor. As jitter buckets shrink, a lower jitter measurement floor means your jitter budget can truly be your device, and not the residual effects of your measurement instrument.
Consider an oscilloscope such as Keysight’s UXR, with a measurement floor of 40 fs (compared to over 100-200 fs for most other real time oscilloscopes). The 100-200 fs of jitter and higher noise floor directly translate to random jitter and total jitter. If your design is marginal, the extra Random Jitter (RJ) and Total Jitter (TJ) can mean failing compliance – and ultimately taking longer to get to market. Saving a few ps of random jitter and 10s of ps of total jitter can make or break the success of a product. The measurement difference could mean that one oscilloscope passes, while the other one fails the device-under-test.
Measuring the jitter measurement floor of an oscilloscope is a simple process. One must use a high accuracy sine wave source, such as Keysight’s PSG. Then measure the sine wave through the oscilloscope. For the greatest accuracy, compare the results of the real time oscilloscope with a sampling oscilloscope (which has less than 50 fs of jitter measurement floor). Once the sine wave is displayed, turn on the time interval error measurement of the oscilloscope (see figure 2). Specify the constant clock recovery as a sine wave, so it is easy for the clock recovery to lock on to the given data rate.
Maximising margins after purchase
After the purchase of an oscilloscope, you can achieve even more accurate measurements by following some simple techniques. For example, only use the bandwidth you need. If you are looking at PCIExpress 5.0 signalling, you need 50 GHz of bandwidth. If you have purchased a 70 GHz oscilloscope, that 20 extra GHz of bandwidth is just additional noise on your signal; the extra bandwidth works as an inhibitor to maximising the margins. A standard feature on today’s oscilloscopes is the ability to specify how much bandwidth is needed. When you choose this option, a low pass filter is applied to the oscilloscope to limit the bandwidth. By limiting the bandwidth of the latest technology, measurement accuracy can be increased. Bandwidth limiting not only lowers the oscilloscope noise, but also decreases the oscilloscope’s measurement jitter noise floor.
Today’s market demands require the most accurate measurements possible to minimise error caused by instrumentation. For the high-speed digital market, this means choosing an oscilloscope with a low noise and jitter measurement floor. It also means using advanced features of the oscilloscope to enable better measurement accuracy. When you purchase the right oscilloscope with the features you need, you can reduce time-to-market and differentiate against your competitors.
Contact Details and Archive...