21 November 2013

Scope Dopes

Written by 
Scope Dopes

When I was in the US Navy in RADAR school I was introduced to the term “scope dope”, a derisive term for the young men who sat in front of the RADAR repeater (scope) and tried to decipher the battle situation from a bunch of greenish dots crawling across the glass CRT. It doesn’t seem a fair label for such a difficult job so I want to move it to another group who richly deserve it.

The test engineer who, like the man who only has a hammer and sees every problem as a nail, the guy who uses an oscilloscope to measure everything he has to test. You know these guys:

• Need to measure RMS signal amplitude? Ask the scope to report Peak to Peak and divide by 2.828. While it’s true that the RMS of a sine wave is Vpp/2.828 they are assuming that the signal that the scope sees is the sine wave they expect, without any noise, harmonics or other unforeseen problems.

• Or the guy that uses a scope to measure noise the same way. No band limiting, no attempt to remove things that can be attributed to other causes, like 60Hz (in America) or 50Hz (Europe and much of Asia). If you want to know the true noise power coming from a device shouldn’t you exclude those sources?

Don’t get me wrong, the oscilloscope is a wonderful tool. I’ve worked for two of the companies that claim to have invented the oscilloscope. However, as useful as it is, an oscilloscope does many things poorly and some things not at all.

For example, most oscilloscopes have very high input bandwidth to make them useful to people looking at high frequency signals. That has a down side too: high bandwidth amplifiers in scopes pass a whole lot of undesired noise into the A/D of the scope. Because they want to be able to capture high speed events, the A/D of the scope must have a fairly low resolution, usually not more than 8 bits.

The timing resolution of a scope is finite and cannot be improved upon using clever sampling techniques available to the savvy engineer who knows a little DSP and sampling theory. Worst of all, while most scopes can do an FFT it will be severely limited in utility because the captured waveform cannot be coherent. I am a huge proponent of the FFT and I use it all the time. I once mentioned the letters FFT one too many times and a young engineer told me “You always mention FFT, but it’s just a button on a scope!” Yeah, scope dope right here in front of me!

The advantages of coherent sampling cannot be ignored and it is my biggest pet peeve against the oscilloscope. If you coherently sample (using an Arbitrary Waveform Generator with a coherently sampled Analog Capture instrument) you can do things no one else has ever dreamed of. You don’t have to window, which every scope with an “FFT button” has to do. Once you eliminate windowing you can see things that no other instrument can see.

• You can use equivalent time sampling (undersampling) to get an infinitely fine timing resolution that no scope can match.

• You can measure noise by sourcing a fixed, known noise signal with your AWG, coherently capture the DUT input and output, spectrally eliminate noise sources that you don’t care about and measure the true noise that your device generates which will improve yields and test repeatability.

• Instead of making a gross measurement of the peak-to-peak amplitude and assuming it’s a sine wave, you can do the FFT and look in the bin of interest and know for certain that what you are measuring IS the sine wave you wanted to see.

The great Matthew Mahoney of LTX years ago wrote the book on “DSP Based Testing” and saw scope dopes in his time, but did not name them as such. But his closing paragraph from Chapter 1 describes them to a tee.

Here is what he said: “In a tutorial I gave overseas several years ago, the real ‘cost’ of using DSP instruments suddenly came clear to one of the engineers midway through the day. He knew only a few words of English, but managed to express his revelation in a way that is all the more expressive for its simplicity. Leaping to his feet with a mixture of excitement and fear, he said ‘But...this means...we must know something!’ Indeed we must.“

The guy who thinks that an FFT is just a button on a scope, or believes that he can measure the RMS amplitude of ANY signal by calculating Max-Min/2.828 is in the class of engineers who are appalled that they might have to “know something”. Scope dopes. But rather than be negative and assume there is no hope for these folks let’s listen to the final sentence from chapter 1 of Mr. Mahoney’s world famous book:

“In short, the real cost of DSP testing may be that it forces us to master the craft of engineering.” That should be the goal of any engineer, to become a better engineer at every opportunity and to move from the ranks of the “scope dopes” to the ranks of those who really understand engineering, not just the faded rote memorizations from their university days.”

Engineer at Applicos