2 September 2013

Oscilloscope or digitizer

Written by 
Oscilloscope or digitizer


Some friends of mine are working feverishly on a test application and it's not going well. They are trying to measure settling time, with an oscilloscope. Now, don't get me wrong, oscilloscopes are fine instruments, at Tek we just celebrated Howard Vollum's 100th birthday. In case you didn't know Howard, he's the guy who put triggers on oscilloscopes and made them the valuable tools they are today.

However, an 8 bit oscilloscope, no matter how good, is going to have a problem helping you resolve settling time to 1%, because, well, let's do the math. An 8 bit digitizer is good for one part in 256 (about 0.4%). However most engineers understand that you can't really trust those last two bits, so now you are down to one part in 64, and let me think, one part in 100 (1%) requires a little more resolution than a result from one part in 64 instrument.

The problem here is that with a million things to do in his workday an engineer on a critical project will often take the path of least resistance and just pick an off the shelf test instrument like a GPIB enabled o'scope and call it good. Why bother with sampling frequencies, Nyquist, bits, calibration, etc, etc.

But then when time's up and the delivery date is upon you, do you really want to spend endless days, nights and weekends trying to wring the last little bits of accuracy out of a general purpose visual troubleshooting tool?

Or should you spend a little more time up front to do the math and sample with something that has a known resolution, with flexible timing, accurate calibration and a low noise floor? Perhaps a real digitizer is what you need like the PA72G16180, a 16 bit 180MHz digitizer with a noise floor below -70dB.

Do it right, get it done and take some time off, because unlike the guy who grabs a handful of GPIB cables and a bunch of off the shelf instruments, you put some time and thinking into your test plan. You've earned some relaxation!


Engineer at Applicos