3
\$\begingroup\$

Judging from this question, it seems that different people/sources have different definitions for measurements like amplitude.

I personally encountered people disagreeing on the definition of the rise/fall time, some stating that it should be computed using 10% and 90% of the amplitude, other stating that it should use the peak-to-peak values.

Since this can lead to confusion at best (thinking about contract issues) is there some sort of recognized standard? Or should I explain how each measurement is computed?

\$\endgroup\$
  • 3
    \$\begingroup\$ Using pp for rise/fall seems useless, but there are indeed various definitions on how to measure it, some 10/90, some 5/95 some 20/80 and for every serious measurement, it should be stated how it was done. \$\endgroup\$ – PlasmaHH Jun 28 '17 at 8:57
  • \$\begingroup\$ that's my point, currently i simply define everything, but it would be simpler to simply write "all measurement computed using ieee standard n°XXX" or somthing similar. \$\endgroup\$ – Sclrx Jun 28 '17 at 9:18
  • \$\begingroup\$ measurements is a complicated thing, and the procedures and results are within a context, and there it can be more useful to use one or another way to measure things. It can also depend on the tools you have available, not everyone can do every kind of measurement the same way. Keep in mind that even "Measured 3.14159V" doesn't mean that there were 3.14159V, but that your instrument configured in a specific way and attached in a specific way displayed it. \$\endgroup\$ – PlasmaHH Jun 28 '17 at 9:24
  • 2
    \$\begingroup\$ The fact you have to ask this question would indicate that if such a standard does exist saying that you used it wouldn't help. Stating that you are using the levels defined in a standard that no one knows doesn't aid clarity unless they dig it out and could make things worse unless you were explicit on the correct subsection of the standard used depending on the complexity and number of revisions to the standard. At that point it's far easier to simply state the levels that you used. \$\endgroup\$ – Andrew Jun 28 '17 at 9:28
  • \$\begingroup\$ @Andrew: plus that an awful lot of standards are only available for a (hefty) fee \$\endgroup\$ – PlasmaHH Jun 28 '17 at 9:35
6
\$\begingroup\$

There is no standard for exactly what rise and fall times mean. Even if there were, good documentation defines its terms.

In the case of rise and fall time of a digital signal, the appropriate thresholds depend on how the signal is intended to be received. The first thing to look at are the levels are of the minimum guaranteed logic high and maximum guaranteed logic low. For any specific receiver, these are the levels that matter. Of course you need these levels to be guaranteed at the receiver end of whatever transmission line is between the sender and receiver, so there needs to be some margin beyond the min/max levels at the sender end.

If the spec is for something more general and various receivers could be used in the future to detect the signal, then you usually get more conservative. Even receivers with Schmitt trigger (hysteresis) inputs usually have their min/max levels no wider than 20% and 80% of the supply. In that case, you might specify rise and fall times between the 10% and 90% thresholds.

For analog signals, rise and fall times take on a whole different meaning. These are often specified to settle within some minimum error of the final steady state. If the signal needs to be accurate to within 1%, then it hasn't "settled" until it gets to within 1% of its final value. You'd usually add some margin there too, since there are other sources of errors in the system. You might specify "settled" as within ½% of the final steady state value, for example.

Again, don't leave things like this open to interpretation. Good specs include definitions for anything that matters that could be interpreted differently.

\$\endgroup\$
  • 1
    \$\begingroup\$ nice answer! I think it's really important to make the point you're stressing: Document hell out of what you're measuring. Even when measuring according to a specific standard, it doesn't hurt, and in fact, it's not that rare that standards get revisions, or that standards aren't 100% precise, or that one simply does make a mistake at understanding the standard (which would not be apparent if the actually followed measurement routine wasn't documented, and also, if someone found out, it would completely invalidate the measurement instead of allowing for interpretation). \$\endgroup\$ – Marcus Müller Jun 28 '17 at 11:13
1
\$\begingroup\$

You'll not find a single standard for signal measurements in general.

Think about it: Signals don't exist for their own good, they always exist to make to things cooperate (typically: communicate).

And thus, you'll find a lot of documents describing requirements for the signal shape for a specific standard – be it USB (in all its variations), Gigabit Ethernet, SDRAM, FireWire, PCIe, JESD204B, SATA, or many many others – whenever devices are meant to be generally compatible with each other. In that case, both ends, the transmitter and the receiver need to adhere to that standard, and if they do, communication should work.

If you're not dealing with a standardized link, well, then the requirements of the receiver become your standard.

If you don't know your receiver yet, then the best you can do is probably do a measurement following the methods defined in one of the link standards that is close as close to your application as sensible.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for?Browse other questions tagged or ask your own question.