CHASING THE ILLUSIVE MILLIVOLT & NANOSECOND – PART III
In search of more performance limits, I looked at RT (Rise time). RT can be theoretically determined from the frequency. The 2465B has a specified frequency response of 400 MHz (at the -3 DB roll off point). Using the generally accepted formula of .35/Freq = RT, we have .35 divided by 400 MHz. This = a rise time of .875 ns.
Should also point out that -3 DB in signal amplitude means the peak voltage at 400 MHz will read 29.3% less! To maintain reasonable amplitude accuracy, of say -3%, the upper frequency limit would be about .30 x 400MHz, which brings the usable frequency (for amplitude accuracy) down to 120 MHz! A good Tektronix article on this and the approximation of RT can be found at -
http://www2.electron.frba.utn.edu.ar/~jcecconi/Bibliografia/06%20-%20Osciloscopios%20de%20Almacenamiento%20Digital/Understanding_Oscilloscope_BW_RiseT_And_Signal_Fidelity.pdfIdeally, to measure RT accurately, one needs a pulse generator with a RT of roughly 10x the .875 ns RT of the scope, or 87.5 ns. Unfortunately, the best I could do was to borrow a Tek SG503 pulse generator. It’s specifications give a rise time of 1 ns or better, which places it roughly equal to the scope.
When one begins to measure RT, a number of other questions arise. Tektronix specifies it to be measured between the 10% & 90% of amplitude. That’s easy in a perfect world where pulses have square clean corners. But there is the “pre-shoot and overshoot of the leading edge of the displayed pulse. I wrestled for some time with how to deal with that in making the RT measurement. In the article above, they illustrate the answer which I like because it simplifies the measurement.
In order to minimize the leading edge distortions, I found the best results by connecting a 40 inch piece of 50 ohm coax directly between the SG503, using its internal 50 ohm output option, and the internal 50 ohm input to the scope. When using a 50 ohm terminator feed through into the 1 meg ohm port, or a 300 MHz (unfortunately) scope probe into the 1 meg ohm port, the displayed pulse was not as clean.
All this wordiness was needed to qualify the following measurements.
The 1st photo below shows the displayed pulse with the amplitude adjusted to sit on the reticle 0% dotted line and the stabilized top on the 100% dotted line. The under & overshoots pass outside those limits.
In the 2nd photo, the same pulse is shown with the sweep speed at max using the “x10 MAG”. Measuring the time between the 10% & 90% crossings on the reticle, the cursors indicate 1.55 ns. But the 1st photo shows the leading edge of the pulse rounding over before it plateaus across the top. This will lengthen the measured RT. Following some thought and experimentation, I came to the conclusion that, this distortion is coming from the source rather than the scope. If so, then eliminating it would be fair.
The 3rd photo shows the elimination of the rounding and distortion across the top of the pulse. This was done by reducing the pulse width to its minimum. But this also shortens the height of the pulse.
The 4th photo shows where the pulse has been increased in height to again align with the 0% & 100% dotted lines.
In the 5th photo the sweep speed is increased to its max and the cursors are set on the 10% & 90% lines to indicate a rise time of 1.04 ns.
Letting the scope measure the rise time via the red “HELP” button, it indicates a rise time of .89 ns or .93 ns, depending on the repetition rate of the pulse generator. A little discouraging, but again underscores the manually set cursors as being the most accurate.
Using the formula, .35 / 1.04 ns = 337 MHz. But of course, that is the speed of the combined rise times including the scope, generator and cable. But we can estimate the individual RT from the composite. A published formula for this is - the square root of the sum of the individual squared rise times.
In this case – if we assume the cable is perfect for simplicity, it leaves us with 2 unknowns, the RT of the scope and the RT of the generator. Since the scope RT is spec’d at .875 ns, and the generator is spec’d at equal to or better than 1 ns, we can further simplify by assuming they are both the same. Using the formula above, then gives us an individual RT of .73 ns. And since .73 ns is a safe margin below the spec’s of both instruments, I think it’s safe to say that the scope RT is equal to or better than .875 ns. If we were to factor in something for the non-perfect cable, we would have an even larger margin.
This is not as perfect as I would like, but lacking a better pulse generator, it’s the best I can do.