Using an oscilliscope

MikeML

Joined Oct 2, 2009
5,444
I know nothing about your HP scope, however, the technique is scope independent.

First, you have to know something about the device you are testing. Presumably, the device's inputs are in the correct state so that a single input switching low-to-high, or high-to-low will cause it's output to switch. You're going to be measuring the time from when that input changes to when the output changes. You need to determine which is the slowest path through the device under test; e.g. on a D Flop, the delay from Clk to Q might be different from Clk to Q-bar, and different from Set to Q, etc

Assuming the scope is fast enough for the measurement, then put one probe on the input of the Device Under Test (DUT). Use that as the input that you measure the prop. delay with respect to. Set the scope to trigger on the rising edge of that signal. Hook a second probe to the output pin of the DUT. Play with the Horiz Time Base until you can see a low-to-high or high-to-low transition on the output pin caused by the input transition. Measure the time from when the input is at the 50% point to when the output switches through the 50% point.

If you can, change the other inputs of the DUT so that you can create all four possible conditions: input rising edge causes output rising edge, input rising edge causes output falling edge, input falling edge causes output rising edge, and input falling edge causes output falling edge. Repeat the time measurement for all four cases, and record the worst case. Then move the first probe to a different input and repeat. Then repeat for all other outputs. Lots of possible combinations on a complex device.
 
Last edited:

someonesdad

Joined Jul 7, 2009
1,583
Mike's description is excellent, but let me add a couple of things. Physically, since such delays relate to things such as the speed of light in the materials and charging capacitors, these effects also apply to the probes and input circuits of the scope.

The delay you're trying to measure will probably be on the order of 10 ns. Let's assume you're using a 1 m scope probe whose cable allows a propagation velocity of 0.75*c or 225 Mm/s. It thus takes about 5 ns for the signal to travel the length of this cable. Clearly, you'll want to use matched scope probes if possible for the measurement; also make sure they're properly compensated. The risetimes of the electronics will also have an effect -- the slower the risetimes, the crummier your measurement's time resolution.

The careful experimenter will repeat the measurements by reversing the probes and the scope channels and looking for and correcting for the differences caused by the measurement equipment. It would be wise to construct a simple mental model of all the delays in the measurement you're making and convincing yourself you have a handle on their magnitudes. Remember Feynman's advice about not fooling oneself -- one must have scientific integrity to make believable and correct measurements.

I have a 100 MHz HP 54901B scope on my bench and it has a maximum sweep speed of 2 ns/div. You may find it challenging to make such a gate delay measurement with one of these digital scopes unless you get the triggering right. Jitter can ruin things and you'll be trying to capture traces near the scope's capabilities. If you can get a stable and repeatable signal without significant jitter, then the scope will probably do what you want. If not, a well-tuned analog scope might be preferable for this measurement.
 
Top